Relaying in The Internet of Things IoT A

This article has been accepted for publication in a future issue of this journal, but has not been
fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2021.3112940, IEEE Access
Date of publication xxxx 00, 0000, date of current version xxxx 00, 0000.
Digital Object Identifier 10.1109/ACCESS.2017.DOI
Relaying in the Internet of Things (IoT):

A Survey
UYOATA UYOATA1, , (Member, IEEE), JOYCE MWANGAMA2 , (Member, IEEE), RAMONI
ADEOGUN3 , (Senior Member, IEEE)
1
Department of Electrical and Electronics Engineering, Modibbo Adama University of Technology, Yola, Nigeria e-mail: uyoataue@mautech.edu.ng.
2
Department of Electrical Engineering, University of Cape Town, Cape Town, South Africa email: jb.mwangama@uct.ac.za
3
Wireless Communication and Network Section, Department of Electronic Systems, Aalborg University, Denmark email: ra@es.aau.dk
Corresponding author: Uyoata Uyoata (e-mail: uyoataue@mautech.edu.ng.)
"This work was supported in part by the University of Cape Town Research Committee. Ramoni Adeogun is supported by the Danish
Council for Independent Research, grant no. DFF 9041- 00146B. ”
ABSTRACT The deployment of relays between Internet of Things (IoT) end devices and gateways can
improve link quality. In cellular-based IoT, relays have the potential to reduce base station overload. The
energy expended in single-hop long-range communication can be reduced if relays listen to transmissions of
end devices and forward these observations to gateways. However, incorporating relays into IoT networks
faces some challenges. IoT end devices are designed primarily for uplink communication of small-sized
observations toward the network; hence, opportunistically using end devices as relays needs a redesign
of both the medium access control (MAC) layer protocol of such end devices and possible addition of new
communication interfaces. Additionally, the wake-up time of IoT end devices needs to be synchronized with
that of the relays. For cellular-based IoT, the possibility of using infrastructure relays exists, and noncellular
IoT networks can leverage the presence of mobile devices for relaying, for example, in remote healthcare.
However, the latter presents problems of incentivizing relay participation and managing the mobility of
relays. Furthermore, although relays can increase the lifetime of IoT networks, deploying relays implies
the need for additional batteries to power them. This can erode the energy efficiency gain that relays offer.
Therefore, designing relay-assisted IoT networks that provide acceptable trade-offs is key, and this goes
beyond adding an extra transmit RF chain to a relay-enabled IoT end device. There has been increasing
research interest in IoT relaying, as demonstrated in the available literature. Works that consider these issues
are surveyed in this paper to provide insight into the state of the art, provide design insights for network
designers and motivate future research directions.
INDEX TERMS Cooperative communication, Energy harvesting, Internet of Things, Federated learning,
IoT relaying, Relay networks, Relay selection, Secure relaying, SWIPT, UAV, Machine learning, Artificial
intelligence.
I. INTRODUCTION that by 2023, there will be 14.7 billion M2M connections,

He Internet of Things (IoT) paradigm allows for the which signifies a 2.4-fold increase from the number of M2M
T connection of many devices to one another and to
the Internet. These devices are usually low-powered sen-
connections in 2018.
The IoT is not a standalone concept but builds on existing
sors deployed to sense the environment and report sensed technologies. The nature of the IoT makes it a candidate for
information through a gateway for information processing the use of already existing technologies such as machine-type
and decision making at a central network server using an IP communication, device-to-device (D2D) communication [3]
link. Indeed, IoT is being envisioned to be a universal utility and cognitive radio networks [4]. Moreover, the IoT extends
[1]. IoT networks are applied in smart agriculture for soil earlier technologies such as radio frequency identification
quality monitoring, smart home applications for basic home (RFID) and wireless sensor networks (WSNs), solutions that
automation and intelligent road transport. It is envisaged that are application-specific with limited interoperability between
the IoT will connect billions of devices and will change the them [5], [6].
way humans and devices communicate. Cisco [2] projected Communication between IoT end devices and a gateway
VOLUME 4, 2016 1
This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
U. Uyoata et al.: Relaying in the Internet of Things: A Survey
can occur directly or through an intermediary node (a relay). could dent the low-cost target of IoT networks. Some of these
Direct transmission when the distance between an IoT end challenges and more have been considered in the literature.
node and the gateway is large can cost the network more This survey reviews the research conducted in the area of
power. For practical deployment of the IoT, low-power wire- IoT relaying, highlighting the challenges and the approaches
less access network (LPWAN) technologies such as SigFox, adopted to manage them.
LoRa and NB-IoT are designed for at least 1 km of single-hop
transmission [7]. Such long-range single-hop communication A. RELATED SURVEYS
relies on the quality of a single link and consequently does Despite the increasing interest in relay-enabled IoT networks,
not enjoy the diversity gain that relayed communication can there is a paucity of surveys in this research area. There are
offer. In [8], it is reported that actual deployments of single- several surveys in the literature that provide overviews and
hop IoT networks (using LoRa technology) in rural areas categorization of various aspects of IoT networks; see, e.g.,
experience connectivity deterioration because a clear line of [12]–[24] and the references therein. However, to the best
sight (LOS) between the gateway and end devices is not of our knowledge, this survey is the first to focus on the
easily achieved. To ensure reliability, relays can assist in challenges of - and approaches to relaying in IoT networks.
forwarding intended signals from an IoT end device to its We now present a review of selected related survey papers.
gateway. This can be achieved through two or more hops [9]. The selection has been made to have complete coverage of all
For short-range IoT connectivity solutions, such as BLE and aspects of the IoT with particular emphasis on their relation
Wi-Fi-Hola, that use ISM unlicensed bands (which are sub- to relay-enabled networks.
ject to interference), relays can help extend the transmission In [25], the current challenges facing in-band full-duplex
range of sensors. relaying are discussed, and open research questions are
There are differences between relaying in the IoT and also highlighted. The authors categorized full-duplex relay-
relaying in cellular networks. First, most IoT networks are ing based on relaying strategy, antenna design and spatial
designed for single-hop communication with end devices streams. It also enumerates relaying schemes designed to
configured for mostly uplink transmissions of their obser- overcome the limitations of half-duplex relaying to include
vations to the gateway. Therefore, using IoT end devices successive relaying and two-way relaying, among others.
that are enabled to act as relays requires physical interface Traditional relaying protocols and clustering techniques are
modifications. Second, IoT end devices are designed to trans- discussed in light of machine-to-machine communication in
mit minimal payloads for short durations, and each gateway [26]. Selfish and malicious behavior of nodes are surveyed in
serves several end devices. Therefore, if end devices are used [27], where the impact of such behavior on wireless relay
for relaying, modifications to their duty cycle and messaging networks is presented. Various types of attacks that mali-
windows need to be made while maintaining a lean end cious nodes can launch and the approaches to detect them
device. Such technical constraints make IoT relaying unique. are also discussed. Incentive mechanisms to encourage data
Third, as energy-constrained networks, IoT relays need to forwarding in such human-centric networks are discussed as
balance their energy needs with the energy efficiency con- well. Different from the work in [25], our survey is not only
straint of the network. Last, unlike relays in cellular networks focused on in-band full-duplex relaying and so has a broader
that could be part of the network rollout, IoT relays may be scope. In IoT networks, user-provided machines/devices can
remedial measures where the need arises. Consequently, their serve as relays forming the so-called wireless relay network
introduction to the network should be transparent to both [27], but relaying in the IoT is not limited to such a scenario;
gateway and end devices. hence, our survey provides a wider scope. Moreover, our
IoT relaying poses some questions. For example, should survey does not have selfish behavior as its singular focus but
IoT relays be part of network planning or should they be also considers such behavior as an aspect of user-provided
rolled out as an after-thought if the single-hop deployment relaying for which incentive mechanisms can be proposed,
encounters connectivity deterioration? Moreover, should re- as shown in Section IV. The work in [28] limits its focus
lays in IoT networks be centrally controlled by either a to relay-assisted wireless body area networks (WBANs).
gateway/network server or should they be transparent relays Therein, the authors discuss network architectures, relay
whose entry and exit from the network go unnoticed by other node selection and point out the unique quality of service
network components? As research in human-in-the-loop in- requirements of relay-based WBANs. In Table 1, key related
teractions [10] gains traction, implementing user-owned ma- surveys and their focus areas are given to highlight how these
chines (such as drones, cars and smartphones) as relay nodes surveys differ from our work.
is also becoming an option. With humans involved, providing In Table 2, a list of abbreviations used in this article and
incentives to encourage sharing communication resources for their meanings are given, and some key features of cellular
IoT data forwarding becomes an issue in addressing and man- and IoT relaying are provided in Table 3. Our survey is not
aging the mobility of user-owned machines. Furthermore, it limited to a particular IoT connectivity technology because
remains to be seen whether relays for IoT networks should be the use of relays has been considered for various connectivity
modified end devices or if they should simply be additional solutions, such as LoRaWAN [30], NB-IoT [31], and Blue-
gateways [11]. The latter approach incurs some cost, which tooth low energy [32].
2 VOLUME 4, 2016
TABLE 1: Selected Surveys and Their Focus Area.

Article Type Area Technologies Application Year
Published
[25] Survey In-band full Relaying strategies, antenna patterns General relaying 2015
duplex relaying and spatial streams (non-IoT focus)
[13] Survey Buffer-aided Improved relay General relaying 2015
relaying selection mechanisms (non-IoT focus)
[21] Survey Key enabling 5G, semantic Web, IoT focus 2017
technologies for IoT cognitive radio, cloud computing focus
[26] Review Cooperative Clustering methods and M2M 2018
Communication relay selection communication
[17] Concepts and Key enabling 5G IoT enablers (D2D, mmWave IoT focus 2018
overview technologies for IoT , relaying, wireless SDNs, NFVs) 2018
[27] Survey Noncooperative Strategies in overcoming selfish D2D 2018
relaying and malicious behaviors communication
in human-centric networks
[16] Concept and Cooperative NOMA-based D2D D2D communication 2018
survey communication communication with focus on
industrial IoT
[22] Survey IoT architectures End-to-end architecture IoT focus 2019
descriptions of cyber-physical systems
[14] Survey Cooperative Simultaneous wireless information and MIMO, WSN, cognitive 2019
communication and power transfer with cooperative relay radio, vehicular ad hoc network,
and future challenges NOMA beamforming
techniques and IoT
[23] Concepts and QoS-aware buffer Relaying strategies, link Wireless Body Area 2019
overview -aided relaying selection mechanisms Networks, Healthcare IoT
[20] Survey IoT applications End-to-end taxonomy for IoT applications 2020
IoT applications Smart Cities, Industrial IoT
[29] Survey Key enabling technologies for IoT in 5G enablers IoT in 5G 2020
IoT in 5G systems (5GNR, HetNets, LPWAN) systems
[28] Survey Relay node selection Cooperative diversity, WBANs 2021
and Quality of Service demands cooperation communication WBANs
This Survey Relaying in IoT Selection algorithms, secure IoT focus 2021
paper relaying approaches,
energy harvesting-based
relaying and applications
of relaying in IoT
B. KEY CONTRIBUTIONS the literature that has proposed solutions to these chal-
As stated above, the focus of this paper is to provide a holistic lenges;
survey of works that are related to relaying in IoT networks. • We provide a concise review of machine learning (ML)
This represents a significant deviation from the few existing and artificial intelligence (AI) techniques and their ap-
survey articles, most of which consider specific aspects of plications in relay-assisted IoT networks. Some insights
IoT networks and architectures. We believe that researchers, into other potential issues that can be solved using ML
IoT designers and industries will benefit from having a single are also provided.
reference survey covering applications, challenges, advances, • Considering the interest that UAV relaying in IoT net-
solutions and open problems in IoT relay networks. Consid- works is garnering, we thoroughly review aspects of
ering the foregoing, the main contributions of this survey are UAV relaying related to IoT networks.
summarized as follows: • We highlight key application areas and use cases of
relay-assisted IoT networks and provide a survey of
• We provide a taxonomy of the challenges of relaying in highlighted application areas, and
the IoT and the research approaches to these challenges • We discuss open research questions that can provide a
as proposed in various studies. To accomplish this, road map for future research in the area of relay-assisted
we review current literature in the subject area while IoT networks.
also building on more established research contribu-
tions in relay networks. Particularly, the approaches are The aim of this article is hence to provide a resource where
classified as relay selection, secure relay-assisted IoT the merits and demerits of various IoT relaying approaches
networks, incentive-based relay-assisted IoT networks, in the literature are surveyed. We specifically highlight the
energy harvesting in relay-assisted IoT networks, relay challenges of relaying in the IoT and the solutions researchers
physical interface design and UAV relaying in IoT net- have proposed. We also group the various research works into
works; classes to enable a structured presentation. This will help IoT
• We provide a comprehensive survey of the identified network designers in their decisions regarding the choice of
challenges to relaying in the IoT and thoroughly review a design approach. Moreover, researchers can identify future
VOLUME 4, 2016 3
research directions to provide a focus for their work. a rich research area; therefore, the choice of routing protocols
The organization of the remainder of this paper is as fol- and the taxonomy of routing protocols are outside the scope
lows. In Section II, a description of relaying in IoT networks of this survey.
is given with emphasis on classifications of relays and relay-
ing protocols. In addition, a brief background on relaying and B. RELAYING PROTOCOLS
a description of relaying strategies (amplify-and-forward, Relays perform operations on the signal they are meant to
decode-and-forward and compress-and-forward protocols) retransmit. Based on the operations that relays carry out on
are also given in this section. In Section II, challenges of their received signals, relaying protocols can be categorized
relaying in the IoT are also highlighted as a forerunner to the as follows:
approaches in the literature to solve these challenges. Section
III thoroughly reviews the approaches to the challenges of Amplify and Forward (AF)
relaying in the IoT adopted in the literature. In this section, In the amplify-and-forward protocol, the relay retransmits an
there is a subsection that focuses on specific research solu- amplified version of the signal it receives from the source.
tions, such as edge caching and NOMA. Section IV is ded- The simplicity of implementation of the AF strategy makes it
icated to the literature that has considered machine learning attractive, although it has the downside of noise and interfer-
and AI approaches for IoT relaying, and in Section V, we ence amplification.
describe key application areas of relay-enabled IoT networks.
We conclude the paper in Section VII after detailing the open
Decode and Forward (DF)
research questions in Section VI.
Here, the relay extracts the received signal and re-encodes it
II. A SHORT DESCRIPTION OF RELAYING IN THE IOT
before re-transmission. In the DF protocol, a condition for
In this section, we give a general background on relaying relay selection can be the successful decoding of the received
and then go on to explain in brief detail the popular relaying signal by a relay.
strategies. Against this backdrop, IoT relaying is described. A
key feature in this section is the categorization of relays in the Compress and Forward
IoT into classes to enhance the discussion in the subsequent In the DF strategy, the relay can decode the received signal,
sections. whereas the compress-and-forward protocol allows the relay
to send a compressed/scaled-down version of the received
A. BRIEF BACKGROUND ON RELAYING signal to the destination. In the reviewed literature for this
The idea of using an intermediary device or cluster of de- paper, AF and DF relaying strategies are the most widely
vices to assist the transmission of information for another used protocols.
device or set of devices (referred to as source(s)) has been The performance of AF and DF relaying protocols has
shown to offer several gains and has been studied for a long been studied in cellular networks as a coverage extension
time in academia and applied in industry. Extensions of the strategy [39], cognitive radio networks as a cooperative sens-
basic three-node network (having a source node, destination ing strategy [40] and wireless sensor networks (WSNs) as a
node and a relay) have also been discussed in the literature. coverage extension and energy-saving strategy [41]. Similar
Early work on the relay channel was reported in [33], where to WSNs, IoT devices are power constrained, and deploy-
the classic three-terminal transmission setup was introduced ing relays for forwarding observations from end devices to
and its capacity bounds were presented. The capacity of a gateways can help IoT networks reduce their overall network
Gaussian degraded relay channel was studied in [34]. This energy consumption. Both AF and DF strategies can be
classic three-terminal communication channel has also been applied in IoT networks, although the reduced complexity of
extended to include multiple relays [35], multiantenna relays AF relaying makes it a preferred choice.
[36], and buffered relays [37]. In [38], the diversity gain
obtained when a source and a relay cooperate to transmit C. RELAYING TOPOLOGIES
information is demonstrated. Cooperative communication is In wireless networks, relays can assume various topologies
possible when the source-destination and the source-relay- depending on the nature of the applications. Fixed relays are
destination links are available. In such a case, there must common in mobile communication networks and in wireless
be a suitable combining strategy at the receiver. Cooperative sensor networks. Network designs with fixed relays require
communication also occurs when multiple sources jointly prior planning before deployment and tend to be rigid. Re-
transmit each other’s information such that all the sources act laying topology-based fixed nodes are easier to model but
as relays for a single source iteratively. In scenarios where challenging to modify. On the one hand, relaying topologies
the source-destination link is unavailable, relaying can still based on mobile nodes are more dynamic and adaptive to
be achieved without the relay cooperating with the source changes in network structure. With increasing interest in
device. When multiple relays are deployed in a network and UAVs, the performance of relay topologies based on such
multihopping from source to destination is enabled, the prob- high mobility nodes has been analyzed. UAV-based relays
lem of selecting suitable routing protocols arises. Routing is in IoT nodes are flexible and can be quickly deployed in an
4 VOLUME 4, 2016
TABLE 2: Table of Abbreviations

Abbreviation Meaning Abbreviation Meaning
BSI Battery State Information CSI Channel State Information
SU Secondary User AF Amplify and Forward
DF Decode and Forward PHY Physical layer
MAC Medium Access Control HD Half duplex
IoT Internet of Things D2D Device-to-Device
UAV Unmanned Aerial Vehicle NOMA Nonorthogonal Multiple Access
OFDM Orthogonal Frequency Division Multiplexing BER Bit Error Rate
SNR Signal-to-Noise Ratio SWIPT Simultaneous Wireless Information and Power Transfer
eNB enhanced Node B RSSI Received Signal Strength Indicator
R-D Relay-Destination CDI Channel Distribution Information
VG-AF Variable Gain Amplify and Forward OMA Orthogonal Multiple Access
RS Relay Selection S-R-D Source-Relay-Destination
LTE-A Long Term Evolution Advanced LPWAN Low-Power Wide Area Network
Wi-Fi Wireless Fidelity ML Machine Learning
WBAN Wireless Body Area Network SIC Successive Interference Cancellation
SOP Secrecy Outage Probability FD Full Duplex
CRT Chinese Remainder Theorem PLC Power Line Communication
NB-IoT Narrow Band Internet of Things PB Power Beacon
PS Power Splitting TS Time Switching
R-E Relay-Eavesdropper MISO Multi-Input-Single-Output
MIMO Multi-Input-Multi-|Output 5G Fifth Generation
FFN Feed Forward Network DNN Deep Neural Network
RNN Recurrent Neural Network DT Decision Tree
SVM Support Vector Machine RL Reinforcement Learning
RS Relay Selection AS Antenna Selection
PHS Physical Layer Security CAR Cache-Aided Relaying
AC Access Control PA Power Allocation
BW Bandwidth POS Position
PTH Path FW Fixed Wing
PWR Power AoI Age of Information
RG Range DL Delay
EE Energy Efficiency S/L R Sum/Link Rate
OP/BER Outage Probability /Bit Error Rate CP Coverage Probability
Nu Number of Users RW Rotary Wing
TABLE 3: IoT Relaying and Cellular Relaying cally serve as relays. Relays in the IoT can also be dedicated,
IoT Relaying Cellular Relaying where the nodes are originally designed to forward data or
Energy constrained Plugged into constant they could be opportunistic where their presence in a network
power source
Currently Standardized
is fluid. Furthermore, relays can be grouped as mobile relays
Nonstandardized (LTE L1, L2 or fixed relay nodes. The boundaries of these classifications
and L3 relays) overlap because fixed network-provided relays can also be
Dynamic topology Mostly fixed topology
mobile, as in the case of an access point mounted on a
Multihop-enabled Mostly dual hop
to the nearest BS vehicle. The fixed and mobile relay categories are broader
Hierarchical Hierarchical and more encompassing. A diagram showing various IoT
Planned or ad hoc Mostly planned classes is shown in Fig. 1, and our classification of relays is
deployment deployment
For mobile relays Mobility of UEs managed given in Fig. 2. In network-provided relays, a base station or a
there is no mainstream, through network controlled fixed relay node handles the allocation of radio resources and
adopted mobility handovers coordinates interference, as in the case of LTE-A relays [42],
management strategy
and pico-BSs act as relays to enable communication between
IoT devices [43].
The addition of relays in IoT networks can improve the
emergency. In this paper, Section III-D is devoted to UAV reliability of networks, increase network lifetime, save en-
relaying in IoT networks. ergy through reduced transmit power of the end device, and
decrease the cost of multiple gateway deployments, among
D. RELAYING IN THE IOT other gains. It is still debatable whether relays in IoT net-
Relays in the IoT can be categorized based on various param- works should be IoT end devices modified to serve as relays
eters. Relays can be categorized as being network-provided or whether they should be gateways that forward IoT end
or user-provided entities. In network-provided relays, the device observations to the network in a multihop fashion. In
relay nodes are part of the network rollout, whereas user- the literature reviewed for this paper, the form of relays for
provided relays are user-owned devices that can opportunisti- the IoT depends on the design goals of the research paper.
VOLUME 4, 2016 5
have a limited duty cycle of approximately 1%. Deploying

relays in IoT networks should not significantly increase the
energy consumption of networks. In uplink communication,
relays consume energy when (1) listening for signals from
IoT end devices and (2) forwarding signals to the gateways.
Although with the deployment of relays, the transmit power
of IoT end devices can be reduced, there are trade-offs
between the increased reliability that relays can offer vis-
a-vis the power consumption of relays. Whereas gateways
are mostly connected to a constant power source and end
devices are battery-powered, relays, on the other hand, need
power sources that allow them sufficient capacity to forward
received signals from IoT end devices without being tethered
FIGURE 1: Cellular-provided and user-provided IoT to a power source such as a gateway.
Approaches to addressing the energy constraint of IoT
relays have focused on enabling relays to harvest energy from
the transmit signal or from other sources (such as solar) in the
environment. Research has proposed simultaneous wireless
and information power transfer (SWIPT) methods for IoT
relays, whereas little research emphasis has been placed on
the use of other forms of energy to boost the residual energy
of relays. Despite the research efforts on SWIPT in IoT re-
laying, implementation challenges can hinder actual testbed
evaluations. For IoT networks that use already available
cellular infrastructure, energy availability may not be a major
constraint given that there may be a constant power source.
Subsection III-A presents a survey of SWIPT techniques for
relay-aided IoT networks in the literature. Energy constraints
at relays can also be addressed by designing IoT architectures
with energy-efficient relay nodes. Such architectures limit the
FIGURE 2: Classification of Relays amount of transmit power of the relay or allow the relay to
only participate in message forwarding instead of combining
sensing and message forwarding.
A possible architecture for a relay-assisted IoT network is
shown in Fig. 3. 2) Relay Selection
Relay selection is a problem that comes up when there
is more than one potential relay available to forward data
between an IoT end device and a gateway/IoT end device,
as demonstrated in Fig. 5. A straightforward method would
be to select a relay for which the data rate of the source-
relay-destination (S-R-D) link is maximum, i.e., for N relays,
FIGURE 3: Possible Relay-Aided IoT Network Architecture where the data rate of the S-R-D link of the ith relay is given
by Ri , and the selection criteria are given by:
Despite the gains and benefits of having relays in IoT
networks, which include energy efficiency, diversity gain
and increased lifetime of the network, various hurdles can max (Ri ) (1a)
1<i<N
impair the implementation of relaying in IoT networks. In
this section, such challenges are studied. A taxonomy of the A selection criterion is key in choosing a relay. The problem
challenges of relaying in the IoT and the approaches given in of relay selection becomes more complicated when more
the literature to solving them is given in Fig. 4. than a single criterion is used to select a relay. Using the link
quality alone as in equation 1 as the selection criterion can ig-
1) Energy Constraint on Relay Nodes nore the battery life of the relay and other upper layer perfor-
One of the design goals of the IoT is energy efficiency. To mance metrics. Similarly, for buffer-aided relays, if only the
achieve this goal, there are limitations on the amount of link quality is used for relay selection, the relay buffer state
energy that IoT end devices can consume. In fact, some con- can be ignored, leading to packet losses. Hence, for relay-
nectivity technologies for the IoT (such as LoRa and SigFox) enabled IoT networks, relay selection is a challenge. IoT
6 VOLUME 4, 2016
FIGURE 4: Taxonomy of Challenges and Approaches in IoT Relaying Research. *Commercial Off-The-Shelf
case of vehicle relaying for smart transport), there then

needs to be a way of motivating relay participation. In such
cases of human participation, end devices can use short-
range connectivity solutions, such as BLE or Wi-Fi, to send
data to participating human-held relays that help forward
received data to gateways or access points in a time-slotted
manner. Selfish behavior among relay-capable devices owned
by humans can deteriorate the performance of IoT networks
because of the humans in the loop. This is because such
users may not be willing to use their devices to forward
observations of IoT end devices and thus lead to dropped
packets. Designing incentives that are commensurate with
FIGURE 5: Relay selection setup the relay services offered is a challenge for relaying in the
IoT given the nature of data transmitted by IoT devices.
IoT sender nodes that are observing the environment may
not be transmitting regularly, and the data sizes may not be
networks are often designed with a centralized architecture
large enough to warrant high-value incentives to spur relay
where IoT end devices connect to a gateway/base station in a
participation. Incentive mechanisms also pose a challenge
star topology. When relays are introduced into IoT networks,
for standalone IoT networks because where there are many
a problem to address is whether to use a centralized relay
noncompatible proprietary IoT networks, porting or moving
selection algorithm where the network server or gateway
the incentive of one network to another becomes cumber-
selects a relay or whether end devices should select relays
some. Popular incentive mechanisms used in other wireless
in a distributed fashion, an approach that does not require
networks fall into the categories of game theory-based and
complete knowledge of channel state information [44]. In
nongame theory-based approaches.
this survey, the relay selection algorithms proposed by recent
research contributions for relay-assisted IoT networks are
reviewed in subsection III-B. 4) Security and Trust
The broadcast nature of wireless networks makes them sus-
3) Incentive Mechanism ceptible to security infractions. Relays have been shown to
The ’things’ that make up the IoT may not have self- increase the secrecy rate of wireless networks [45] without
interested motives, so in selecting them as relays, incentives the use of higher-level cryptography functions such as the
may not be a priority. When these things are carried around exchange of secret keys. However, the presence of user-
by humans, for example, mobile devices and cars (in the owned relays (machines, objects, smartphones, vehicles, etc.)
VOLUME 4, 2016 7
in IoT networks poses a security risk. Such risk arises when to allow two-way communication. Subsection III-E surveys
the relay is untrusted. Untrusted relays, although helpful the approaches to practically design and deploy relays in IoT
in information forwarding, can be malicious. This can be networks.
problematic when the data to be forwarded are confidential.
For private information such as patient data in medical IoT III. PROPOSED APPROACHES TO THE CHALLENGES
networks, unauthorized access to such data constitutes a OF RELAYING IN IOT
security challenge. Physical layer security has been proposed In the previous section, the various challenges that relay-
to ensure that malicious eavesdroppers and untrusted relay enabled IoT networks face have been highlighted and ex-
nodes do not compromise the network. Cryptographic ap- plained. In this section, a review of the approaches proposed
proaches have also been proposed where keys are shared in the literature to address these challenges is presented.
between the relay nodes and the IoT nodes. Some low-power We first discuss energy harvesting solutions for relay-
technologies, such as LoRaWAN, already have end-to-end enabled IoT networks before reviewing relay selection al-
data encryption schemes built into their IoT network [46]. gorithms proposed for IoT networks. Incentive mechanisms
However, for the work in this survey, current research works proposed for IoT relays and secure relay-enabled IoT net-
on physical layer security for relay-assisted IoT networks are works are discussed next. These are followed by a survey
reviewed. Cryptographic approaches are beyond the scope of of papers focused on other broader concepts such as radio
this paper. resource allocation, edge caching, NOMA and full-duplex
relaying in IoT networks. There is also a subsection on relay
5) Mobility physical interface design.
Where relays are not stationary infrastructure relays, their
positions in a network vary, and this variation can lead to A. ENERGY HARVESTING (EH) IN RELAY-ENABLED IOT
loss of connection for the stationary IoT end devices that the NETWORKS
relays are meant to support. To keep tabs on mobile relays in To cater to the energy needs of relays in IoT networks,
a network, the relay devices can be registered with the net- some contributions have proposed the use of the energy har-
work to ensure authentication and provide incentive action. vested from wireless signals through wireless power transfer
Where such registration is done, mobility apart from creating (WPT). In WPT, the relay either replenishes its embed-
connectivity challenges will also cause redundant registration ded energy source using the transmission from a dedicated
with the server. Since the nodes are mobile, some may not power beacon (PB) transmitter or extracts energy from the
complete the transmission of data and thus can leave the source IoT device for which it is relaying while receiving
network coverage area, leading to the storage of redundant data signals as well (Fig. 6). The latter approach, which is
data. This can be overcome by setting a time-to-live threshold called simultaneous wireless information and power transfer
for registered relays to assist in IoT communication. Where (SWIPT), is mainly achieved through a power splitting (PS)
relay nodes are human-held devices, incentives can help method, a time switching (TS) method or a hybrid of both
to motivate relaying and stem random mobility. Managing modes at the relay. Antenna switching and separate receiver
relay mobility for IoT devices, especially when the relays architectures are other ways of achieving SWIPT, although
are third-party owned devices, is a challenge. For cellular TS and PS are mostly used in research works on IoT re-
network-based IoT, mobility management through handovers laying. Practically, it is not feasible to harvest energy and
is supported by the network. The research contribution in [47] decode information concurrently, hence the use of splitting
studied time-varying IoT networks assisted by relays with and switching techniques to achieve SWIPT. Apart from elec-
time-varying locations. Recently, there has been a surge in tromagnetic radiation, solar energy and vibration are other
research interest in UAV relays. UAV relays are not only sources from which relay nodes can harvest energy. In this
mobile but are elevated above the IoT nodes they serve. subsection, contributions to energy harvesting for IoT relays
Hence, they present a good example to study relay mobility in are discussed and categorized. A summary of approaches
IoT networks. In Section III-D, a thorough review of research to energy harvesting-based relay-enabled IoT networks is
works in UAV relays in IoT networks is presented. presented in Table 4, and the key features of TS and PS are
given in Table 5.
6) Physical Interface Design
One key challenge of relaying in the IoT is the physical 1) Power Splitting in SWIPT Relay-enabled IoT
interface design of the relay. In cellular networks, infrastruc- When SWIPT is implemented using PS, the relay divides the
ture relays are scaled-down base stations to which mobile power received from the source device into two portions; a
devices can associate for data forwarding. The challenge portion is used to replenish the energy reservoir of the relay,
of designing relays in the manner of cellular networks will whereas the remainder is used for information processing and
imply having additional low-power gateways. This adds to forwarding. Power switching has been considered for IoT
the cost of IoT network rollout. Conversely, using end de- networks with SWIPT relays [48], [49] and [50].
vices as relays would require some modification to allow an D. Asiedu et al. [48] proposed a PS ratio for SWIPT relays
increased receive window and an additional radio interface without external energy sources in a downlink multihop IoT
8 VOLUME 4, 2016
TABLE 4: Summary of Energy Harvesting-Based Relay- information forwarding comes from harvested RF energy, the
enabled IoT Networks destination node is treated as an information receiver. Not
Harvesting Article Approach so in [49], where a relay powered by harvested energy from
technique the source device forwards data to one destination and RF
SWIPT [48] PS ratio energy to a second destination. The research also used the
(PS) optimization
[49] PS ratio Lagrangian multiplier method to solve an energy efficiency
optimization maximization problem subject to harvested energy at the
[50] PS ratio second destination node. This is done to obtain the optimal
optimization
SWIPT(TS) [51] Power and transmission strategy of the relay, although the relaying strat-
subcarrier allocation egy is not clear and the work assumes perfect knowledge
[52] TS ratio and of the associated channels. In formulating the optimization
relay power
optimization problem, the relay transmit power is not bounded by an upper
[53] Transmission time limit.
optimization
[4] Outage probability In [50], Y. Zou et al. considered an IoT network enabled
minimization by a SWIPT relay using the PS protocol. The work presented
[54] Sum-rate an outage probability analysis of a proposed optimal PS
maximization
[55] Outage probability and with relay selection. Hence, a relay that has a PS ratio
Ergodic probability that maximizes the capacity of communication is selected.
analysis Through simulation and theoretical formulations, the authors
SWIPT [56] Error
(Dual PS and TS) probability show the gains of employing their approach over the equal PS
analysis ratio method. The proposed approach may be biased toward
[57] Outage allocating more power to signal decoding than to energy
probability
analysis harvesting and thus ignores the residual power needs of the
[58] Outage energy harvesting relay. The proposed PS ratio technique was
probability tested in DF and AF relay networks. The outage performance
analysis
Green energy [59] Distortion of the EH with AF relays consistently outperformed that of
harvesting reduction DF relays for various numbers of relays, SNRs and energy
maximization conversion efficiencies.
[43] Bandwidth and
power optimization
Key Insight
TABLE 5: Features of Power Splitting and Time Switching
Downlink system models are considered in [48] and [49],
Power Splitting Time Switching
where the SWIPT relay harvests energy from the transmis-
EH and Information reception Switches between EH
happens simultaneously and Information reception sion of the base station/gateway in one time slot and uses the
Requires additional hardware No additional hardware next time slot to transmit information to the IoT end device.
to divide the received signal into for TS apart from a Such a time-slotted approach where the relay is a half-duplex
harvested energy and information switch at the receiver
decoding relay is also used by Y. Zou et al. [50]. The objective of the
EH and Information decoding TS ratio can be optimized work in [48] and [50] includes optimizing the PS ratio to
can be jointly optimized improve the system-level data rate and outage probability,
Consumes less Harvest more
time resources power respectively. The assumption of the availability of perfect
Lower SNR for Shorter frame length for channel state information (CSI) in [50] is extended in [48]
information decoding [56] information transmission [56] to consider the existence of CSI errors. The surveyed works
in PS SWIPT use simulations, whereas in [48] and [50],
closed-form expressions are derived for the optimal power
network using DF protocols at the relays. Source transmit splitting ratio. The considered works similarly assumed that
power minimization and system throughput maximization the relay has a battery that stores harvested energy, although
problems were formulated subject to energy and power ratio [50] actually characterized the stored energy. The optimal
constraints. The authors demonstrated the power splitting PS ratio improves the system rate and reduces the energy
ratio that can result in reduced source transmit power by consumption of a relay-enabled IoT network over a fixed PS
performing simulations and offer an improved system data ratio. Moreover, for a fixed data rate, the energy conversion
rate. It was also shown that an optimal number of relay nodes efficiency of a SWIPT relay limits the outage probability of
exists for a multihop half-duplex IoT system. An optimal a relay IoT network.
PS ratio was also proposed for a case where the nodes had
Implementing SWIPT using TS has also been studied in
imperfect CSI. Analogous to most research efforts in SWIPT,
relay-enabled IoT networks.
the work in [48] dwelt on optimizing the PS ratio.
In relay networks where the energy that a relay uses for
VOLUME 4, 2016 9
FIGURE 6: (a). Relay energy harvesting from downlink transmission (b). SWIPT using power splitting (c). SWIPT using time
switching
2) Time Switching in SWIPT Relay-Enabled IoT posed scheme offers system throughput gains in comparison
Relays that use TS-based SWIPT alternate between energy to a fixed time allocation approach. Although both [54] and
harvesting and signal forwarding, as shown in Fig. 6c There [53] enable backscattering between an IoT device and the
are several research works on the IoT that study TS in SWIPT relay, in [54], the direct path between the IoT source device
relays [52], [53], [4], [54] and [55]. and the receiver is considered absent.
Unlike the work in [53] and [52], where one-way commu-
The papers on relaying in IoT networks in which the relay
nication between the relay and the destination is studied, [4]
uses the time-switching technique for energy harvesting can
derived closed-form expressions for the outage probability
be loosely grouped into two categories. These are dedicated
of a 2-way relaying setup with Nakagami-m faded channels.
PB transmitter-based [52], [53], [54] and nondedicated PB
In the setup, two secondary IoT devices act as DF relays
transmitter-based [4] and [55] approaches.
that depend on harvested energy from a pair of primary
Optimal solutions are derived for a sum-throughput max- users and forward information to them. The accuracy of
imization problem in [52], where a relay’s energy need is the derived analytical expressions is verified by simulations
met by harvested energy from a dedicated access point (AP). to demonstrate the effect of channel fading parameters on
In the modeled network, the AP is the receiver of the trans- the system outage probability. Moreover, the frame struc-
missions of IoT end devices through the relay. The research ture of the proposed setup allows switching between en-
work showed that the fairness performance and throughput ergy harvesting and information processing. Although some
of the proposed system depend on the scheduling of IoT practical IoT connectivity technologies, such as Weightless
end device transmission. [53] derives closed-form solutions [60], enable cognitive radio technology, the limited transmit
for transmission time minimization problems. The presented window of some IoT applications can be a drawback for
system model involves a dedicated power beacon (PB) trans- two-way communication. Similarly, T. Nguyen et al. [55]
mitter and an IoT source device that backscatters PB signals presented the performance of a two-way relay-assisted IoT
to a relay and a destination IoT device (gateway). In so network. The work does not assume the availability of a
doing, cooperation between the source device and the relay dedicated PB transmitter since the relay harvests energy from
is enabled. The proposed relay cooperation scheme offers the transmission of the source IoT device. It extends the
improved throughput in comparison with a case where the two-way relay setup to consider a Rician fading channel be-
relay has an embedded energy source. Both works [52] and tween the source and destination IoT devices, deriving outage
[53] consider system models with a dedicated PB transmitter probability expressions for both delay-tolerant and delay-
serving a single relay network. This is advantageous because limited cases. This is useful considering the varied delay
more energy can be harvested from a PB transmitter than requirements of IoT applications. Performance metrics such
from a low-duty-cycle, energy-constrained IoT end device. as outage probability, ergodic capacity and throughput are
The proposed relay harvesting approach in [52] does not en- studied for various system parameters, and their performance
joy the cooperative gain proposed in [53]. Another difference also proves the accuracy of the theoretical derivations.
between the two works is that the relay in [53] switches
between information forwarding, backscattering and energy Key Insights
harvesting, whereas in [52], the relay switches between en- In the reviewed contributions to IoT relaying where the relay
ergy harvesting and information forwarding. uses TS-based SWIPT, dedicated PB transmitters are as-
A novel relaying scheme is presented in [54], where gate- sumed in works that consider one-way relaying, and the sys-
ways are powered by the energy harvested from AP relay tem performance is analyzed for parameters such as through-
messages between the AP and batteryless IoT devices. The put [53], [52], [55] and ergodic capacity [55]. Theoretical
work derived closed-form solutions to the problem of max- analysis mostly accomplished through closed-form solutions
imizing a formulated sum-rate maximization problem. This to formulated optimization problems is the approach that the
was accomplished by jointly optimizing the time scheduling, reviewed papers use. Moreover, a DF relaying protocol is also
energy beamforming and power allocation at the AP. The pro- adopted by the surveyed papers. The relaying protocol is key
10 VOLUME 4, 2016
to attaining target throughput in SWIPT relay IoT networks. Key Insights

This is because DF relays tend to offer improved performance Using a hybrid of TS and PS protocols can help exploit
relative to AF relaying for parameters such as throughput, the advantage each offers by switching to the TS protocol
outage probability and energy consumption. Combining PS when more energy is needed and switching to PS when
and TS protocols can ensure that the gains of both protocols the transmission is delay-sensitive [56]. The downside of
are derived by the relay-enabled IoT network. a hybrid approach may be hardware complexity, and this
has not been studied. Combining NOMA and EH can result
3) Dual Power Splitting and Time Switching in SWIPT in improved network throughput [58]. The NOMA-enabled
Relay-enabled IoT relaying IoT network is considered separately in Section IV-
PS and TS SWIPT protocols have gains that can be exploited F
by using both protocols together. Certain research works have Studies have been performed on relay nodes harvesting
concurrently considered PS and TS protocols for SWIPT energy from sources other than wireless signals. As future
relays in IoT [56], [57] and [58]. Y. Hu et al. [56] study the networks are envisaged to be more energy-efficient, employ-
reliability performance of a hybrid PS-TS SWIPT-enabled ing renewable energy sources for energy harvesting relays is
IoT network in the finite blocklength regime. Combining TS also attractive.
and PS was shown to offer better error probability perfor-
mance than using each protocol individually. Using TS-based 4) Green Energy Harvesting-based Relay-enabled IoT
SWIPT in relays offers improved throughput and outage SWIPT techniques are not the only approach to energy har-
probability over PT-based SWIPT [57]. Moreover, jointly vesting studied in relay-enabled IoT. Harvesting energy from
optimizing the blocklength allocation and SWIPT parameters sources other than RF, such as solar energy [59], [43], has
results in a slight improvement in error probability perfor- also been studied.
mance over nonhybrid SWIPT protocols [56]. Furthermore, In [59], the authors present a search method for obtaining
average throughput analysis is performed in [61] for an the optimal relay selection and power allocation for relays
IoT network assisted by a relay node selected based on the assisting sky cameras in an energy harvesting wireless sen-
source-relay data rate. Unlike most works, the paper consid- sor network. The proposed search algorithm is designed to
ered finite blocklength codeword transmission and proposed solve a distortion reduction maximization problem subject
an optimal and suboptimal design of the transmission data to energy constraints. Simulations have demonstrated the
rate based on an approximated closed-form expression of the improved outage performance of the proposed algorithm over
throughput of the IoT network. a so-called nearest relay selection algorithm. However, the
Relays in IoT networks are often modeled as nodes that relaying protocol is not clear, and the complexity of the algo-
are nondata generating and thus only wait for data from a rithm could increase with an increased number of relay hops.
source that requires forwarding services. Where relay nodes Similarly, [43] considered harvesting solar energy to power
are modeled as data-generating, their data transmission is fixed pico BS relays. Different from the research works on
usually assumed not to occur during the forwarding phase SWIPT relaying that assume batteryless relays, a dual bat-
of relay communication. A case where a relay has data to tery architecture is proposed in [43], and the architecture is
transmit during its forwarding phase might occur when the shown through simulations to result in higher residual energy
source and relay target the same destination device with compared to single battery relays. Providing additional solar-
their distinct signals or where both the relay and the source powered batteries for IoT relays incurs costs, especially for
devices target separate destinations. For the latter case, there standalone private application-specific IoT deployments.
has to be a way to tell each destination to treat as noise or
interference the signals not intended for it. Such an indicator Key Insights
can be added to the overhead when signaling. Additional Energy-efficient relaying is important in relay-enabled IoT
overhead to inform the destinations can be avoided by using networks since a relay uses its energy resources to cater to
nonorthogonal multiple access (NOMA), which allows the the data forwarding needs of neighboring IoT devices. Hence,
superposition of separate signals and allocating power to harvesting energy is a key strategy to maintain the battery
the separate signals based on the link quality [58]. In [58], energy of relay nodes. In a cellular-provided IoT network,
the effect of interference is studied for a relay-aided IoT where a fixed relay is used, a hybrid energy source that uses
network in which a NOMA-based relay harvests energy from both on-grid power and harvested energy can be used to
the transmission of the source IoT device. TS and PS are harness the green energy potential of energy harvesting [43].
considered for the half-duplex single relay that not only
forwards the source signal to a destination but also has its B. RELAY SELECTION IN RELAY-ENABLED IOT
own signals to send to a separate destination. The work em- NETWORKS
ploys a golden search method to solve outage probability and The use of relays provides diversity gains and energy ef-
throughput optimization problems. This approach assumes ficiency improvement because a source IoT device may
that the relaying node has a signal to send concurrently with not need to transmit at full power to reach a gateway or
the source devices data. destination IoT device. The lifetime of an IoT network is
VOLUME 4, 2016 11
also enhanced when relays participate in data forwarding. To the IoT sender device. Hence, determining a mobility range
exploit these gains, schemes to select a relay or a cluster of for mobile relays [63] can help exclude some relay-enabled
relays to assist in data forwarding are necessary. In the works devices from the potential relay set. In [63], an iterative
reviewed for this survey, various relay selection techniques steepest descent algorithm that solves an outage probability
were proposed, and their results are herein discussed. Re- minimization problem (to obtain the optimal relay position
lay selection algorithms proposed in the surveyed research and relay transmit power) is proposed. A relay selection
contributions for relaying in IoT networks can be loosely algorithm is also proposed that selects a relay link for which
classified into (a) physical layer selection algorithms and (b) the solution to the optimization algorithm converges fastest.
cross-layer selection algorithms. In the former approach, the In formulating the optimization problem, there is a constraint
research works focus on selecting a relay based on the phys- on the mobility of the relay, which is not counterintuitive
ical layer parameters of the participating links, whereas in since if a relay is too far from the source and destination IoT
the latter approach, the authors employ additional parameters devices, the outage probability of the network will increase.
from upper communication layers to determine the selection However, it is not clear how the mobility range is obtained.
metric. In this section, a review and classification of proposed For IoT deployed in a cellular network, the base station
approaches in the literature are given. A summary of works can be tasked with allocating fixed relays or mobile relays to
on relay selection techniques for IoT networks is given in assist in data forwarding for IoT devices. In such a case, re-
Table 6. layed communication can use already available technologies
such as D2D and M2M communication. Relay-enabled IoT
1) Physical Layer-based Relay Selection deployed within a cellular network was studied in [64], [65],
Selecting relays based on a measured physical layer parame- [31] and [67]. The IoT deployed within a 5G heterogeneous
ter has been studied for IoT networks in [62], [50], [63], [64], network is considered by N. Dao et al. [64]. In the considered
[65], [31], [66], [67], [68] and [69] setup, an IoT source that experiences intercell interference
The lifetime of relay-enabled networks depends on the uses D2D communication to discover nearby idle IoT termi-
residual energy of the relays, and if a relay is selected solely nals. These discovered IoT devices then report the received
using the link quality as the metric, the network may experi- power from the interested IoT source to a central eNB, which
ence outage if the relay’s energy source is severely drained. assigns a suitable IoT device and a pico BS with a sufficient
H. Kawabata et al. [62] proposed a relay selection algorithm resource block as relays for the IoT source device. The
that is based primarily on three metrics: 1) the residual energy proposed selection method offers increased network through-
of relays, 2) the channel distribution information (CDI) of put and number of served IoT devices. However, since the
the channel gain between the relays and the destination IoT approach uses D2D communication, D2D discovery and link
device and 3) the distribution of the distance of each relay establishment increase the overhead of relay transmission.
from the destination IoT device. Using stochastic geometry Channel state information, battery energy level and distance
for modeling the IoT network, the paper derives the closed- from the BS are the metrics employed by J. Lianghai et al.
form outage probability for the network, and through numeri- [65] to select relays for cellular IoT communication. The pro-
cal analysis and simulations, the proposed approach is shown posed system model initially clusters nearby devices using
to offer improved outage probability performance over a se- K-means clustering and determines their transmission mode
lection scheme that uses the channel gain mean. However, the (either cellular or D2D) based on the measured performance
algorithm proposed in [62] falls below an instantaneous CSI- metric. The relay selection approach is centralized and may
based selection approach in outage probability performance. result in increased overhead.
The relaying protocol employed is the variable gain amplify- Specifications for NB-IoT, a low-power IoT technology
and-forward protocol. Y. Zou et al. [50] proposed a relay developed by 3GPP, have been created for cellular network-
selection scheme that selects a relay that has a power splitting provided IoT [70]. Using NB-IoT, the work in [31] proposed
ratio that best maximizes the overall channel capacity of an a relay selection algorithm for a formulated total network en-
IoT network. The scheme showed improved outage probabil- ergy minimization problem. The algorithm selects relays in a
ity performance over an equal power slitting ratio selection manner in which the relay with the least energy consumption
algorithm. Both [62] and [50] consider energy harvesting in comparison to the energy consumption of the direct link is
IoT networks where the relay splits the received signal from selected. Moreover, for the selected relay, the source-relay (S
the source IoT device into energy for signal decoding and â R) distance must be less than the relay-destination (R â D)
energy harvesting. Both also use outage probability as their distance. The proposed relay selection algorithm was shown
performance evaluation metric. to consume less energy than direct communication. Consider-
The availability of mobile relays within the forwarding ing a cellular-based IoT network, S. Hsu et al. [67] proposed a
distance of source and destination IoT devices is advanta- relay transmission order, data partitioning method and a relay
geous, as their presence can be exploited to improve the selection algorithm for a formulated optimization problem
performance of an IoT network. However, since the move- targeted at minimizing the maximum total data transmission
ment of these mobile relays is random, the probability of time of the IoT network. The source IoT device selects a
outages can increase when they move out of the coverage of relay set for which the S-R data rate exceeds the relay â
12 VOLUME 4, 2016
base station data rate. The proposed selection algorithm is considered statistical CSI instead of exact CSI. The avail-
shown to offer reduced transmission time and is suitable ability of instantaneous CSI can help improve diversity gain
for mission-critical IoT communication. It requires perfect [62], although in practical IoT networks, acquiring CSI can
channel knowledge and would find application in cellular- be overhead intensive. Furthermore, relay selection schemes
based IoT with a computationally capable central entity (that formulated as joint optimization allow for better performance
is, a BS). trade-offs when compared to single optimization schemes.
When perfect CSI is not available, channel uncertainty can In designing relay selection algorithms, physical layer pa-
affect the channel estimates and consequently the achievable rameters such as channel gain, data rate, and secrecy capacity
data rate of an IoT setup. For such a system model where can be used to choose relays from a set of idle nodes. Physical
channel uncertainty is considered, an adaptive transmit power layer parameters relate directly to physical layer performance
strategy that is channel uncertainty-aware can be employed to metrics, although the performance of relays can also be
overcome the effect of channel uncertainty [71]. In modeling affected by upper layer conditions such as the queue state
channel estimation errors, an approach is to present it as a information of the buffer of a relay or the delay in a relay link.
Gaussian distributed random variable that is added to the Therefore, cross-layer approaches are necessary to capture
channel estimate [68], [72]. In [68], a channel estimation layer-specific performance metrics.
error-aware relay setup is presented. The work proposed a
way to select devices from a relay set in a manner that 2) Cross-layer-based Relay Selection
maximizes the capacity of the S-R-D link in the presence of In [47], [73], [74], [75], [76] and [77], cross-layer relay
an eavesdropper with better link quality than the S-R-D link. selection algorithms were proposed for relay-enabled IoT
It also proposed that a jamming device can be selected to networks.
transmit artificial noise to limit interception of target signals S. Redhu et al. [47] considered a mobile IoT network
by the eavesdropper. However, transmitting a jamming signal and proposed a method of selecting a relay based on the
for closely positioned devices may require knowledge of link reliability and the latency performance of the relay link.
the artificial noise sequence at the destination IoT device To capture the mobility of nodes in the network, a way-
and thus contribute to overhead. A machine learning (ML) point mobility model was used. The relay selection problem
selection technique was proposed for an IoT network assisted was formulated as a joint minimization of the packet loss
by multiple relays in [66]. Channel coefficients were used risk and the link latency parameters and solved after some
as data for model training, and the so-called iterative sparse reformulation. Using relay data sets, it was demonstrated
relay selection algorithm was used for relay selection. The that there was a linear dependence of the network latency
selection metric is basically the channel coefficients, and on the node mobility variance. Furthermore, the proposed
the paper considers jointly optimizing the received signal- algorithm was shown to offer less overhead than two other
to-noise ratio (SNR) at the receiver and the beamforming routing protocols. The work assumed precise knowledge
coefficients at the relays. Using ML is attractive considering of the positions of the nodes. Different from the works in
that the IoT will interconnect many devices, resulting in con- [47] and [78], where only data forwarding relay devices
siderable data generation, but the resources required for ML are selected, some works have proposed relay and source
implementation may only be suitable for centrally controlled device selection algorithms. A case where source selection
IoT networks. becomes necessary is where measurement of a parameter
Two relay selection schemes were proposed by M. Farooq (such as soil moisture level, among others) is being reported
et al. in [69], namely, a selection scheme that selects a relay by more than one IoT source device. In such applications,
nearest to the IoT source device and a selection scheme that there must be a way to schedule which IoT device should
selects a relay that enhances the progress of forwarded data report its measurement to ensure fairness. C. Zhang et al.
toward the destination. The work studied a massive IoT net- [73] proposed a method to select source devices and relays
work assisted by multihop relays and used the Poisson point in a manner that optimizes a defined fairness index and the
process to model the distribution of the considered devices. end-to-end data rate for an AF relay-assisted IoT network.
The numerical results showed that the strategy based on near- It is shown that the proposed schemes offer improved fair-
ness to the transmitter resulted in higher transmission success ness performance when compared to an outage probability-
probability, a result that highlights the gains of short hops. only-based selection algorithm. The approach requires the
Moreover, the research contribution in [69] also showed that acquisition of the CSI between nodes in the IoT network.
the spatial frequency reuse increased as the carrier sensing Methods to select the source device are also proposed for an
threshold was increased, an intuitive conclusion. IoT network having an untrusted AF relay in D. Chen et al.
[79] without relay selection. It was shown that the so-called
Key Insights optimal scheduling scheme offers improved secrecy through-
Relay selection algorithms in the surveyed literature that are put and secrecy outage probability over random scheduling
based on physical layer parameters have mostly used CSI- or and threshold scheduling techniques.
CSI-related metrics for relay selection except in [50], where Redundancy from repeated transmissions from IoT end
the power splitting ratio was used. In using CSI, only [62] devices can improve reception quality at a relay, although
VOLUME 4, 2016 13
there must be trade-offs between improved reception quality latency as upper layer metrics in addition to a physical
and the energy consumed by the relay node. This problem layer metric. When compared to noncross-layer approaches,
is addressed in [80], where a Euclidean distance-based sim- the performance improvement of cross-layer relay selection
ilarity test is used to measure the redundancy between the techniques is most visible when the parameters of assessment
received signals at the relay and to determine the wake-up are upper layer performance metrics, for example, data la-
time of relays in LoRa-based IoT networks. tency [47] and fairness [73]. Buffered relays that can opt to
The increasing miniaturization of storage chips indicates receive from IoT source devices or forward received signals
that storage capabilities can be integrated into sensors. This to gateways are promising because the limited transmit and
allows the integration of buffers into relays. Having buffers receive windows of IoT devices (which can be reconfigured
in relays allows packets to be temporarily stored for future as relays) do not need to be severely altered to be employed
transmission. When relays are equipped with buffers, there as relays.
are more degrees of freedom in choosing a relay because
not only is the link quality used for relay selection but also C. INCENTIVE-BASED RELAY-ENABLED IOT
the buffer state can be used as a metric to select appropriate NETWORKS
relays. Cross-layer relay selection algorithms with a buffer User-owned relay-capable devices (such as cars in vehicle-
state-dependent metric are studied in [74], [75] and [76]. based IoT networks and drones) can, in theory, serve as
The performance of a nonorthogonal multiple access relays in IoT networks. However, practically, owners of these
(NOMA)-enabled IoT network with relays equipped with devices may be uncooperative because of their self-focused
buffers is presented in [74]. In this work, six relay transmis- nature. Where the reliability of an IoT network depends on
sion modes are considered, including a NOMA mode. The an uncooperative relay, there is a high probability of network
proposed selection scheme selects a DF relay to transmit if outages. In the works discussed in subsections III-A and -
its buffer state meets a system capacity threshold condition. B, a general assumption is that relays are cooperative. This
The scheme is shown to offer improved throughput over assumption may not hold when the devices are third-party
orthogonal multiple access (OMA) and max-min selection owned, so appropriate incentives are necessary to motivate
approaches. However, the selection scheme in [74] requires relay participation. In this subsection, incentive mechanisms
global knowledge of the CSI. Having buffers in relays can proposed to motivate relaying in IoT networks are reviewed.
also allow the use of one relay or a set of relays for the recep- The works in [83], [81], [84] and [85] proposed ways
tion of information from the source and the use of another to motivate relay participation. Table 7 summarizes the ap-
relay or set of relays for forwarding of the signals to the proaches used by these works.
destination [75]. In [74], relays are assumed to have the same The research in [83] and [81] considered relay-assisted
buffer size, although in practical applications, this may not IoT networks where a direct link between the source IoT
be the case [75]. Such an asymmetric buffer size approach is device and destination IoT device exists to enable cooperative
considered in [75], where a relay selection method based on communication. X. Zhang et al. [83] considered a setup in
buffer occupancy and the data rate of the S-R and R-D links which a relay-enabled IoT network is at risk of an eavesdrop-
are used as metrics for selecting half-duplex relays. Consid- ping attack. To motivate relay participation, a Stackelberg
ering relays with more than one antenna, the work linked game is designed that allows relays and source IoT devices
each buffer space to a specific antenna and showed through to define their utilities. Since relays exhaust their energy to
simulation that the proposed approach achieved gains in forward messages for IoT source devices, the game seeks
reduced outage probability as the number of buffer spaces to improve the price of transmit power that the relay uses
and relays increased. Both research contributions, that is, the for data forwarding. Similarly, source IoT devices seek to
contributions in [74] and [75], use half-duplex DF relays and improve the amount of power purchased from the relays.
assume that the relay has a sufficient energy supply to power The simulation results showed that the proposed incentive
its transmissions. J. Xia et al. [76] instead considered a single mechanism motivated relays to compete for improved utility.
relay equipped with a buffer and proposed a transmission Furthermore, when the number of relays in the network is
mode selection technique that chooses either to receive from increased, the unit power price is reduced, giving IoT source
the source IoT device or to transmit to the destination device. devices a selection opportunity. To solve the problem of
The selection metric used is the state of the relay buffers motivating energy harvesting access points (EAPs) to help
and the CSI of the relay link. Outage probability analysis charge sensors that report observations to a data access point
in [76] showed improved performance over conventional (DAP), the work in [87] also proposed a Stackelberg-based
opportunistic selection. The works [74], [75] and [76] use incentive mechanism. A Stackelberg game was used to model
outage probability as one of their performance metrics and the interaction between DAPs and EAPs and to capture the
employed a DF relaying protocol. information asymmetry due to the DAP not knowing the
channel conditions of the EAPs. The problem of incentive
Key Insights design was reformulated as a contract between the two enti-
Cross-layer approaches to relay selection for IoT networks ties. Through simulations, it was demonstrated in [87] that
have mostly used relay buffer states, fairness index and link the effect of not having complete CSI at the DAP can be
14 VOLUME 4, 2016
TABLE 6: Relay Selection Techniques in IoT Networks

Article Objective RS Metric Protocol Layer Relay Strategy Approach Mode
[50] Minimize outage probability Power splitting ratio PHY AF and DF Simulation and HD
of the S-R-D link √ analysis
[62] Minimize outage R-D CDI and VG-AF Analysis and HD
probability Residual energy Simulation
[63] Outage probability Convergence time of PHY AF Theoretical HD
minimization relay link analysis and
simulations
[64] Maximization of network Received power PHY Not clear Simulation HD
resource availability
[47] Minimize data latency and latency and MAC and PHY Not clear Simulation and HD
packet loss risk link reliability analysis
[65] Service availability CSI, relay battery energy PHY No relay Simulation HD
maximization and battery relay distance to BS processing
life improvement
[66] Destination’s received S-R-D CSI PHY AF Simulation HD
SNR maximization
[67] Maximum total
transmission time S-R data rate PHY Not clear Simulation HD
minimization
[68] Security-reliability Secrecy capacity of PHY DF Theoretical HD
trade-off relay link analysis and
optimization simulation
[73] Maximize data rate and Data rate and PHY AF Simulation and HD
CDF of the channel CDF of the channel gain analysis
[74] Maximize throughput Link capacity and PHY and DF Simulation and HD
buffer state MAC analysis
[81] Minimize secrecy and link BSI and CSI PHY AF Simulation HD
outage probabilities
[76] Outage probability Relay buffer state PHY and DF Theoretical HD
minimization and CSI of relay link MAC analysis and
simulation
[77] Latency minimization Transmission Probability PHY Not Clear Simulation HD
[82] SOP minimization R-D Data rate PHY AF Simulation HD
[75] OP and delay S-R-D Data rate PHY and DF Analysis and HD
minimization and Buffer occupancy MAC simulation
[31] Energy minimization S-R Distance and PHY Not clear Simulation HD
energy consumption
[83] Secrecy capacity R-D/ R-E channel PHY AF Simulation HD
maximization gain ratio
TABLE 7: Summary of Incentive Approaches in Relay En- a Vickrey auction incentive mechanism to motivate SUs to
abled IoT Networks relay the data of an SU pair in which the source SU is the auc-
Article Incentive Approach tioneer and the relays are the bidders. As in [83], the work in
[83] Optimized relay Stackelberg- [81] shows that having more relays reduces the bid price and
defined utility game
[81] Bandwidth Vickrey
consequently decreases the utility of the relays. Similarly,
allocation auction when auctioneers require higher utility, the payment the relay
mechanism enjoys increases. The Vickrey auction mechanism enforces
[84] Transferable Blockchain
tokens smart contracts
positive utility. The incentive mechanism proposed in [81]
[85] low-power Analysis is distributed, whereas for the work in [83], knowledge of
channel allocation associated channels (except the wiretap channel) is necessary
[86] Transferable Nash Bargaining and may require a central controller.
credits
To reward multihop relay nodes in an IoT network, [85]
proposed low-power channel allocation to the relays to com-
pensate for their energy expenditure in forwarding signals for
mitigated using contract theory. In effect, [87] extended the an IoT S-D pair in a linear network. The allocation is done
Stackelberg game for incentive design [83] in IoT networks in a manner that ensures equal energy expenditure in all the
to capture the case of information asymmetry. relay nodes. The approach offered reduced energy consump-
Cooperative cognitive relaying is the approach put forward tion compared to a random channel assignment method. It
in [81] for an energy harvesting IoT network. In this work, is not clear if the approach increases relay participation. A
secondary users (SUs) act as AF relays to forward data for a routing protocol for a multirelay IoT network is proposed in
primary user (PU) pair and are rewarded by using the PU [84]. It exploits the public ledger property of blockchain to
spectrum for SU communication. The work also proposed design smart contracts for relay requests and relay request
VOLUME 4, 2016 15
acceptance. The proposed protocol was shown to demon- relays, the security risk increases. Secure relay-assisted IoT
strate less overhead relative to the ad hoc on-demand dis- communication can be viewed from two broad perspectives,
tance vector (AODV) protocol. The proposed protocol uses namely, the case of a trusted relay in the IoT network whose
multihop relaying. To motivate relaying among neighboring communication can be compromised by an eavesdropper and
nodes, tokens are transferred to relays as earnings for pro- (2) the case of an untrusted relay whose forwarding services
viding forwarding service. Incentive design to motivate relay are required in IoT networks, as shown in Fig 7.
participation in a mobile crowdsensing system is studied in
[86], where a data collector selects an intermediary device to 1) Secure Relay-enabled IoT Networks with Trusted Relays
forward sensed data to a requestor in a manner that improves
The use of artificial noise is proposed in [89] for an IoT setup
the reward of the collector. The problem is modeled as a two-
that is at risk of an eavesdropping attack. In this work, a
person cooperative game and solved using Nash equilibrium.
multiantenna EH relay uses beamforming and artificial noise
to improve the secrecy performance of the IoT network.
Key Insights
Two cases of eavesdropper configurations are considered: a
Incentive mechanisms proposed to motivate the participation
passive case where the IoT transmitter does not have the CSI
of relays in IoT networks can be broadly categorized into
of the eavesdropper’s link and an active case in which the
1) game theory-based incentive mechanisms such as the
eavesdropper’s CSI is available at the transmitter. Proposed
Stackelberg game [83], [87] and Nash bargaining [86], 2)
solutions to the secrecy sum-rate optimization problems
auction-based mechanisms [81] and 3) token-based mecha-
achieved a higher rate, which improved with an increased
nisms [84], [85]. These mechanisms are also either deployed
number of antennas at the relay.
centrally with global knowledge of CSI assumed, or they are
The availability of location information and the CSI of
distributed, in which case a central entity is assumed to be
eavesdroppers is an assumption made in some research work
nonexistent.
in secure relay-enabled IoT . Acquiring such information is
D. SECURE RELAY-ENABLED IOT NETWORKS
difficult and can contribute to overhead. Hence, [90] inves-
tigated the secrecy outage probability (SOP) performance of
The presence of relays in the IoT network presents both
an AF relay-enabled IoT network exposed to eavesdroppers
an opportunity and a risk. Appropriately exploiting the re-
with an uncertain location. The work used the Poisson point
sources of relays can offer improvement in the lifetime of the
process (PPP) to model the random locations of the eaves-
network and increase the reliability of the network. However,
droppers and showed through simulations optimal codeword
user-owned or third-party relays can be a source of a security
rates and power allocation to improve the SOP of the stud-
breach in a case where malicious attacks are launched by
ied setup. Cases of single and multiantennas at relays and
such relays. This can compromise the information to be
eavesdroppers were studied. The relays used a randomize-
forwarded or, in a worst-case scenario, compromise the entire
and-forward protocol. The SOP of a trusted relay-assisted
IoT network.
IoT network is also considered in [91]. The use of relays with
Registering and authenticating prospective relays can help
hybrid AF and DF capabilities is studied in [91] to provide
streamline the number of acceptable relays by providing
physical layer security in an IoT network at the risk of a
an identification method for these relays. Security measures
single eavesdroppers attack. The relay with a link that offers
become necessary when the presence of relays in IoT net-
maximum secrecy capacity is selected to forward signals to
works is opportunistic and their entry into and exit from
a destination. Closed-form expressions for SOP are derived
the network is random. Table 8 gives a summary of the
and verified through simulations.
approaches used in the literature that focus on secure relay-
enabled IoT networks. In this subsection, a review of recent
research contributions to secure relay-enabled IoT networks 2) Secure Relay-enabled IoT Networks with Untrusted
is conducted with an emphasis on physical layer security Relays
approaches proposed and studied to date. At the physical layer, the use of jammers can be employed as
Physical layer (PHY) security has gained attention due in [92], which presents an analysis of an AF relay-aided IoT
to its unique features, such as eliminating the need to use network having an untrusted relay. To prevent the relay from
encryption and the exchange of keys between large-scale decoding the signal it is forwarding, the relay’s reception is
IoT devices. With the increased computational capacity of jammed by a dedicated jammer device. At the destination
devices, the use of cryptography is not completely foolproof, IoT device, the relayed signal and the direct signal (from
as eavesdroppers can acquire high capacity devices to break the IoT source device) are combined to improve reception
encryption. Hence, physical layer security continues to gain quality. The performance of the proposed scheme showed im-
traction [88]. Through cooperative communication, relays proved SOP for various detection schemes at the destination
can provide PHY security in the presence of eavesdrop- node. Specifically, maximum likelihood and minimum mean
pers by increasing the secrecy capacity of the S-R-D link. squared error estimators at the receiver showed similar BER,
However, in scenarios where the relay is untrusted, as in SOP and ergodic secrecy throughput performance over the
the case of third-party relay infrastructure or user-owned considered range of SNR.
16 VOLUME 4, 2016
FIGURE 7: IoT Relaying in the Presence of an Eavesdropper
Where potential relay nodes in the vicinity of an IoT TABLE 8: Summary of Secure Relay-enabled IoT Networks
source are not network registered and thus pose a risk of Article Objective Source of Approach
eavesdropping, (i.e., the relays are untrusted), employing the security risk
[92] Minimize SOP, Untrusted Signal
link quality difference between the S-D link and the S-R link Maximize ergodic relay jamming
can reduce the SOP of the IoT network [82]. secrecy throughput
In [82], the work considered a setup having a cluster of [97] Maximize secrecy Eavesdropper Cooperative
capacity jamming
IoT source devices and a group of relay devices that harvest [89] Sum-rate Eavesdroppers Artificial noise,
energy from surrounding radio frequency signals. The relays maximization MIMO
that are AF devices are considered to be eavesdroppers, so [82] Minimize SOP Untrusted Source selection and
relays Optimal transmission
the target of the research is to select the source devices and time
the relays in a manner that minimizes the SOP of the setup. [93] Minimize SOP Untrusted Transmit antenna
Hence, the authors propose a strategy that selects the source relay selection and
relay selection
node and the relay node for which the source-destination [90] SOP Eavesdropper Beamforming and
(controller) link and the R-D link have the highest measured optimization Codeword rate
quality. Through simulations, they demonstrate that an en- optimization
ergy harvesting duration exists within which an optimal SOP
can be achieved. The proposed approach assumes that all
the sensors within the cluster have the same information to generally assume that the relay will use the AF protocol and
send or that the priority of the sensors’ data is link quality consequently not be able to decode the relayed information-
independent. bearing signal. The features of blockchain technology make it
The works in [92], [82] and [93] studied networks with an attractive tool for relay-assisted IoT communication. The
untrusted relays. Whereas jamming with artificial noise is diversity of operators of IoT applications poses data security
used in [92], the work in [82] used source and relay selection and privacy issues for which the immutability and trans-
to improve the secrecy of relayed information. Similarly, in parency features of blockchain can be an asset [94]. Running
both works, relays were modeled to be AF relays instead blockchain algorithms on UAVs (acting as relays for ground
of DF relays. Doing so prevents decoding of relayed infor- IoT devices) may be computationally demanding considering
mation, although the relaying mode may be out of control the power limitations of UAVs. Blockchain algorithms can
of the IoT network operator for cases of third-party owned be run on mobile edge computing (MEC) servers that receive
relays. However, the research contribution [93] proposed measured acquired data from IoT devices via a UAV relay
joint transmit antenna selection and link quality-based relay [95]. The work in [96] proposed storing the interactions
selection algorithms to improve the downlink SOP of an IoT between D2D devices and potential relays in blocks within
network. The network was modeled as having a multiantenna the devices and showed through simulations that the pro-
base station and EH AF relays. The SOP and throughput per- posed block-chain-based relay selection technique can result
formance showed improvement using the proposed selection in higher utility. However, the computational time of their
algorithm. solution was not compared with other benchmark works.
Key Insights E. IOT RELAYING BASED ON MOBILE NODES: UAV

In secure relay-based IoT network research, there has been RELAY-ASSISTED IOT NETWORKS
more research contribution toward ensuring that an untrusted UAVs are flying devices that have been shown to be capable
relay does not successfully eavesdrop on the communication of coverage extension and relaying the data of fixed ground
of the IoT device pair [92], [82] and [93]. These works sensors that are cut off from the network due to emergent
VOLUME 4, 2016 17
situations such as a natural disaster. When obstruction occurs solving the problem of selecting UAV relays that ferry obser-
between IoT transmitters and their destination, UAV relays vations between sensors and a BS. They formulated UAV en-
can fly to the transmitters to collect the data and fly to the des- ergy consumption minimization and maximum transmission
tination to deliver the acquired data. UAVs are a typical ex- time minimization problems. For these problems, energy-
ample of mobile relays, and unlike mobile devices or vehicle- aware and delay-aware algorithms were proposed. The re-
mounted relays, their elevation can be dynamically increased sults showed that having more UAVs to select from reduced
to achieve line-of-sight communication with ground devices. the consumed energy of the network and the operational time
UAVs can help energy-constrained IoT ground nodes reduce when the target objective functions are energy efficiency and
transmit energy consumption as well. In [98], a framework operational time, respectively. Similar to [101], the work by
for analyzing the performance of a D2D communication Y. Chen et al. [102] also considered multiple UAV relays,
setup assisted by a UAV relay is presented. Specifically, although without UAV relay selection. The work instead
the work derived expressions for the system sum rate and compared the performance of UAVs forming a single mul-
coverage probability and showed through simulations the op- tihop link versus UAVs forming various dual-hop multilinks.
timal UAV height for optimizing these performance metrics. It first derived the optimal positions of the UAVs and showed
Furthermore, it showed the dependency of the number of that the DF relaying protocol results in improved BER and
UAV stop points on the density of device pairs. probability of outage for both multihop and dual-hop relay-
Y. Ma et al. [99] demonstrated through a developed pro- ing. The results also demonstrated that the multihop single
totype that a relay mounted on a drone can improve the UAV relaying case ensures better BER performance when the
read range of RFID tags by approximately 10 times the case distance between source and destination IoT pairs is larger
where a relay is nonpresent. Their results also showed that than for short distances.
the relay was also able to offer a reduced localization error L. Kong et al. [103] proposed a method for UAV relays to
of approximately 19 cm. To achieve these results, the authors determine the optimal relay position that maximizes the link
designed an RFID relay on a single printed circuit board that quality to solve the disconnectivity problem in a mmWave
served as a transparent intermediary device between an RFID network. In the proposed approach, the UAV samples the
tag and an RFID reader. The drone flies on a predetermined channel states at few locations and uses 3D matrix com-
path; hence, path planning was not part of the research. This pletion to estimate the rest of the channel state along its
onboard full-duplex relay uses baseband filters to exploit flight path. This compressive sensing approach was shown
the guard band between uplink and downlink transmissions through simulation to achieve higher accuracy but incur more
to overcome interlink self-interference at the filter. Further- time costs than the K-nearest neighbor and tensor recovery
more, it uses an out-of-band full-duplex communication. For algorithms.
hardware, the relay is a custom-made PCB, and the drone is In [104], a rate maximization problem is formulated to
a Parrot Bebop2 drone. Although the work proposes methods optimize the UAV position, transmit power and allocated
for maintaining the phase of the uplink and downlink signals, bandwidth. Considering a single UAV, a polyblock algorithm
it assumes that the drone-mounted relay is stationary in the was proposed for reformulated subproblems of the original
air while receiving backscatter signals from the RFIDs, so problem. The rate throughput performance of the proposed
the effect of drone speed on the range performance is not algorithm outperforms an equal power allocation-based UAV
considered. The path of the UAVs is also assumed to be placement method. The problem of UAV relay placement
already optimally planned. is also studied in [105], where a method is proposed to fly
Different from the work presented in [99], in [100], op- a UAV relay to an optimal position without having global
timal deployment of UAVs for mobile ground IoT nodes knowledge of the possible ground nodes for which it is relay-
is performed with the aim of collecting ground data in an ing. In the proposed method, the signal strength and the angle
energy-efficient manner. To cater to the time-varying nature of arrival of the signal between the ground units and the UAV
of the network resulting from the mobility of ground IoT relay are used to determine the optimal position of the UAV.
devices, clusters of devices are created and dynamically Both mobile and stationary ground units were considered.
updated. To ensure reliability in the mobile IoT network, Apart from theoretical analysis, the work used a quadrotor,
discrete transport theory was used to model the interaction and for ground units, laptops were used to demonstrate the
between mobile ground IoT nodes and UAVs. The resulting effectiveness of their proposed method, which allows the
optimization problem was solved using the revised simplex quadrotor to move from a given initial position to the centroid
method. The results showed that the ground IoT devices of the ground units.
spent 56% less transmit power for uplink communication Time-constrained or delay-limited data need to reach dis-
than the case where UAVs are fixed. Although both [100] and tantly located controllers before these collected data be-
[99] considered UAVs collecting data from IoT nodes, [99] come stale or outdated. There are IoT applications such
focused on range improvement and localization accuracy, as public safety focused, intelligent transportation focused
whereas [100] focused on energy-efficient uplink transmis- and mission-critical applications that require timely report
sion and optimizing the UAV flight path. of sensor measurements. For such applications, the currency
N. Motlagh et al. [101] presented two algorithms for or timeliness of updates from sensors is key to achieving a
18 VOLUME 4, 2016
correct response and facilitating decision making. In [106], problem, the authors showed that improvements to the energy
the age of information (AoI) for an IoT pair is characterized efficiency can be achieved. B. Ji et al. [113] derived the
as a metric for performance analysis of a UAV relay-assisted outage probability and BER expressions for a UAV relay-
IoT network. In this paper, an AoI minimization problem assisted IoT network with energy harvesting at the UAV
is formulated that jointly optimizes the UAV trajectory and relay for Nakagami-m-affected channels. In studying the
the allocation of energy (for the sender IoT and UAV relay). performance of the setup, both TS and PS energy harvesting
To solve the AoI minimization problem, the paper proposed protocols were considered, and simulations were used to
iterative UAV trajectory optimization, energy allocation and verify the derived expressions.
packet service time optimization. The proposed solution re-
sults in a lower average peak AoI compared to a direct trajec-
tory between the IoT pairs. The authors in [107] maximized
the total IoT devices that a UAV relay serves by optimizing F. RELAY PHYSICAL INTERFACE DESIGN
the UAV trajectory, bandwidth and transmit power allocation.
To allow for the limited storage capacity of IoT ground nodes, As there is a growing body of work focused on theoretical
the UAV is a full-duplex relay and is cache-enabled. For the analysis and algorithm designs for relay-assisted IoT net-
formulated problem, an iterative solution was proposed and works, there are also research contributions that have studied
showed an increased number of served IoT devices as the the practical deployment of IoT relays. O. Flauzac et al. [114]
cache size of the UAV relay was increased. Similarly, M. considered the practical use of a modified LoRa node as a
Samir et al. [108] maximized the number of served IoT de- relay to extend the coverage of a LoRa gateway to an isolated
vices by jointly optimizing the UAV trajectory and the radio IoT node. They proposed the use of LoRA between the IoT
resource allocation for a UAV-assisted IoT communication end device and the relay and the use of the LoRAWAN
setup. To solve the mixed-integer nonconvex problem, the au- protocol to send the data from the relay to a gateway. The
thors employed successive convex approximations to achieve relays first joined network gateways through join requests
suboptimal solutions. For the special case where the take- before being attached to IoT end nodes through synchro-
off and landing locations of the UAV are known, a distance nization signals and data request messages. The network
minimization algorithm was proposed. To address the case considered in [114] is a linear network that is application-
where sensed data do not have uniform deadlines, O. Ghdiri specific. Similar to the work in [114], a two-hop relay is
et al. [109] investigated using multiple UAVs to serve clusters also considered an addition to a LoRa-based IoT network
of IoT devices. Specifically, the work in [109] focused on after network deployment in [8]. Although similar to end
optimizing the number of required UAVs, UAV trajectories devices in hardware, the relay does not perform sensing
and cluster formation to minimize the total energy consumed and aggregation (unlike the relay in [114]). Furthermore,
for data collection. To cluster the IoT sensors, the authors asymmetric synchronization between the relay and IoT end
proposed an improved K-means algorithm, and for cluster devices is proposed without a real-time clock to guarantee
head positioning, the algorithm minimized the distance be- time synchronization. Unlike the works in [8] and [115],
tween the UAV dockstation and the cluster head. The authors where modified IoT end devices were used as relays in
modeled the multi-UAV trajectory optimization problem as practical relay enabled IoT networks, in [32], a prototype
a directed graph and showed that tabu search obtained the of a mobile phone relay was presented. In so doing, no
preferred multi-UAV trajectory design. The problem of UAV modification is needed on the relay’s physical features. The
relaying for data collection in IoT networks becomes more work in [32] used a smartphone to forward medical data from
tasking when the ground nodes are mobile [110]. a sensor to a central server without the use of a gateway. BLE
Three optimization problems are the focus of the work in is used as the connectivity technology between the sensor
[111], where practical considerations such as sparse distri- and the smartphone that accesses the central server using
bution of IoT sensors and the limited transmission range of an IP network. In this work, time synchronization between
a full-duplex-rotary-wing UAV are made. Specifically, the the sensor and smartphone relay is not considered since the
work sought to maximize the system sum throughput and smartphone does not require wake-up and sleep times but
minimize both the total energy consumed and the total time rather needs to have a health app installed. The hardware of
required for end-to-end communication. Secrecy communi- the end device (an EFM32GG-STK3700 Giant Gecko [116])
cation is considered in [112], wherein a UAV not only serves and a gateway (iC880A concentrator board and a Raspberry
as a relay to the ground node but also acts as a jammer. Pi 3) are combined to create a forwarder node in [117]
Specifically, the full-duplex UAV uses jamming signals to for a LoRaWAN-based IoT network. To minimize packet
limit a potential eavesdropper from intercepting confidential collision, the wait time of the end device was increased to
information from the IoT ground node. The authors aim give the relay some wait time to obtain a reply from the
to maximize the energy efficiency of UAV-assisted secrecy gateway before sending it on the downlink to the end device.
communication by optimizing the UAV trajectory and the The testbed was shown to reduce packet loss in comparison
transmit powers of the ground units and of the UAV. Using to a bad direct link. Table 10 summarizes the contributions
an iterative approach to solve the formulated maximization regarding the aspect of relay physical interface design.
VOLUME 4, 2016 19
TABLE 9: Summary of Key Contributions in UAV-based IoT Relaying

Optimization Parameter Performance Metric Wing Type
Contribution
BW POS PTH RS PWR AoI RG DL EE S/L
√R OP/BER CP
√ Nu RW
√ FW
[98] √ √
[99] √ √ √
[100] √ √ √ √
[101] √ √ √
[102] √ √ √
[103] √ √ √ √
[104] √ √ √
[105] √ √ √ √
[106] √ √ √ √ √
[107] √ √ √ √
[108] √ √ √ √
[109] √ √ √
[111] √ √ √ √
[112] √ √
[113]
TABLE 10: Practical Relay Hardware Design

[114] [8] [115] [32] [117]
Hardware Not clear Arduino ProMIni Waspmote Pro Samsung Galaxy EFM32GG-STK3700 Giant Gecko
Note 8 and iC880A concentrator board
+ a Raspberry Pi 3
µC Not clear ATMega328P ATMega1281 Exynos 8895/ EFM32GG990F1024
Snapdragon 835
Synchronization Symmetric Asymmetric Symmetric Symmetric Symmetric
Processing None None None None None
Connectivity LoRa LoRa LoRa BLE LoRa
Technology
Key Insights creasing attention in relation to relay-assisted IoT networks.

From the surveyed literature, practical relays for IoT net- In this subsection, other perspectives on relaying in the IoT
works can be modified IoT end devices [8], [115]. In actual are discussed.
relay deployment, relays can be strictly receive-and-forward
devices. That is, no further processing is carried out on the 1) Radio Resource Allocation in Relay-enabled IoT
relayed signal [114], [8], [115]. Although there is extensive A closed-form expression for the energy efficiency of a relay
research work on EH relaying, in actual relay deployments, enabled massive IoT network is derived in [118], where a
the relays have residual energy sources [8], [115]. The hard- DF relay equipped with multiple antennas helps forward the
ware combination of the end device and gateway can be data of a large number of IoT device pairs. The aim of the
used as a relay [117]. Software modifications can be made work in [118] is to allocate relay transmit power, the number
to IoT end devices and gateway hardware to make them act of relay antennas and the number of IoT device pairs in a
as relays [117]. Furthermore, from the reviewed literature, a manner that optimizes the derived energy efficiency of the
testbed for relaying in an IoT network that uses NB-IoT as a network. For the formulated problem, a resource allocation
connectivity technology has not been studied. strategy to minimize the lower bound of the energy efficiency
was shown through simulation to be less computationally
G. BROADER PERSPECTIVES ON RELAYING IN IOT demanding than the exhaustive search algorithm, although
Apart from the approaches to relaying challenges in the with a reduced energy efficiency penalty.
IoT that have been discussed in subsection III-A-E, allo- Mobile relay-assisted IoT networks or relay-assisted mo-
cating radio frequency resources to relay transmission has bile IoT devices have a dynamic topology due to the mobility
also been considered in a few studies [118], [78]. The low of the relay nodes or IoT nodes, respectively. To perform
computational capacity of IoT devices would greatly favor a relay selection or channel assignment for such a dynamic
centralized resource allocation undertaken by a gateway or setup may require CSI that is being updated often. Acquiring
a BS. A distributed approach to resource allocation would updated CSI within short intervals can lead to signaling
require lean algorithms and some knowledge of the link overhead due to the number of training symbols or the chan-
information of the associated channels. Full-duplex relaying nel feedback resulting from the constantly varying topology.
has also been studied for the IoT, an attraction being the To combat this challenge, resource block assignment based
reduced number of time slots it requires in comparison to on the Chinese Remainder Theorem (CRT) is proposed for
half-duplex communication [119]. There are other research cellular network-provided relay-assisted IoT networks [78].
areas (such as NOMA and edge caching) that are gaining in- The assignment does not require the availability of CSI and
20 VOLUME 4, 2016
central coordination from the BS because the relay terminals placed at the edge of the network to ensure proximity to
generate the sequence for resource block assignment in a the end users. Such proximity provides the opportunity to
distributed manner. The approach was shown to reduce the reduce the load on the core network by prefetching popularly
probability of neighboring relays assigning the same RBs, requested user content to relays from the core network during
which can deteriorate the performance of the IoT network. off-peak periods. This relevance of edge caching becomes
The approach requires a level of coordination between relays more glaring when ultrareliable and ultralow latency appli-
and the IoT devices, and the rigor of sequence generation cations are considered. In IoT networks, the performance
may not suit the energy bounds of mobile devices. of cache-assisted relaying has been studied for static relays
Although both [118] and [78] consider resource allocation, [124] [125]and mobile relays [126], [127] [128] and [129],
the work in [118] does not focus on resource block allocation and common performance metrics have included outage
but rather on allocating a number of relay antennas and relay probability and error rate.
transmit power. Both works considered cellular networks The effect of increasing the computational task of a net-
to provide IoT networks with fixed relays in [118] and work can be reduced when a relay is cache-enabled. This is
mobile relays in [78]. U. Tefek et al. [120] proposed two shown in [130], wherein a cache-enabled relay assists source
relaying techniques to enable massive access of machine- devices in offloading computational tasks to destinations.
type communication (MTC) devices. In the first technique, With cache-enabled relays, source devices are less likely to
which is signal-to-interference (SIR) relaying, DF relays only be in outages, and latency is reduced in comparison to cache-
forward the signals of devices with strong SIRs, whereas in free cases. The work in [124] considered cache-enabled
the location-based technique, the signals of MTC devices infrastructure FD relays that provide popular multimedia
closest to the relays are forwarded to the base station. The content to users within its coverage. In the network consid-
density of the MTC devices was shown to determine the ered in [124], caching is also performed at the user levels
comparative outage probability and the transmission capac- where devices can fetch cached content from nearby users
ity performance of the relaying schemes. The work also through D2D communication. Such a setup has potential for
demonstrated that there exists an optimal frequency resource smart city IoT networks, where devices can keep contents
partitioning between the devices to the relay link and the downloaded from a cache relay. The user devices can offer
relay to the BS link. their downloaded content on request through. To reap the
benefits of caching, the density of cache-enabled relays needs
2) Full-duplex Relay-enabled IoT to be high in response to high user density. However, where
Half-duplex relaying is popular in relay network research. the density of the relays cannot meet the download request
However, with recorded success in self-interference cancel- of massively deployed users, the BS can compensate for
lation (SIC) [121], interest in full-duplex relays is increasing. the shortfall through the relays [124]. For aerial relays in
In [119], full-duplex multihop DF relaying is analyzed with IoT networks, cache-enabled UAVs can wait a predetermined
each successive relay node experiencing self-interference, time before flying toward the destination. In so doing, the
relay interference and interference from neighboring active energy efficiency of the network can be improved when com-
nodes. Different from other full-duplex relaying works, the bined with an appropriate trajectory optimization strategy
Markov chain model is used to model the end-to-end error [128]. Cache-enabled aerial relays are studied by B. Jiang
probability of the IoT setup. Through analysis and simula- et al. [131], who optimized the 3D location of UAVs and
tions, it was demonstrated that the choice of FD over HD the optimized file caching location to obtain improved data
relaying depends on the level of self-interference that the throughput.
network can accommodate and where there is cluttering Equipping relays with caches implies that where a user
from randomly placed interferers. Short-distance multihop requests a file that is among the files cached at the relay, dual-
communication is preferred, although the availability of CSI hop communication is reduced to a single hop, assuming
is assumed, and the relays despite being randomly positioned no D2D communication. It also implies that the interference
are static. The work in [122] studied the outage probabil- experienced by the two-hop communication is reduced. In
ity and network capacity performance of a cellular-based particular, the destination alone becomes the interference
massive machine-type communication (MTC) assisted by in- target [125]. Such a scenario clearly plays out in [125],
band full-duplex (IBFD) relays. The network capacity of where cache-enabled relay-assisted networks demonstrate
the IBFD relay-based network did not outperform an HD- improved outage performance over relay-assisted networks
based system when the considered nodes were modeled as without caching at the relay.
homogeneously and independently distributed Poisson point
processes (PPPs). 4) NOMA-assisted Relaying in the IoT
Orthogonal multiple access technologies have been used to
3) Edge Caching-assisted Relays in the IoT provide access to users in cellular networks. The emergence
The advantage of equipping relays with memory capability of IoT networks that will be characterized by ultradense
to enable the offer of additional services beyond signal for- deployment stretches these orthogonal technologies to their
warding has been highlighted [123]. Fixed relays are mostly resource limits. Recently, nonorthogonal multiple access
VOLUME 4, 2016 21
(NOMA) has received much research interest as an access propagation [145]–[148], wireless signal identification [149],
technique that can permit densely deployed devices to use access control and routing protocols [150] and radio resource
the same orthogonal time-frequency resource by exploiting management [151], [152]. ML- and AI-enabled relaying in
the power and code domains [132], [133]. NOMA has been IoT networks has also received significant research attention.
combined with various other technologies, specifically, cog- The focus has been on the application of various ML algo-
nitive radio [134], SWIPT [135], MIMO [136], D2D commu- rithms to solve different challenges associated with relaying
nication [137] and cooperative communications [137], [138], in IoT networks.
[139]. When combined with cooperative communication, the We now present a review of selected existing work on
relay is mostly modeled as providing NOMA access to IoT ML for relaying in IoT networks. For clarity, we group
devices [140]. In the literature, NOMA improves spectral these works according to the ML techniques used. For each
efficiency, provides access to more users and improves the technique, we provide a concise introduction to the method
EE of considered networks. followed by a review of related papers. A summary of the ex-
In [141], uplink and downlink secure IoT communication isting works applying ML to relaying in the IoT is presented
based on NOMA is demonstrated. The achievable secrecy in Table 11.
rate performance showed that the NOMA system, although
it outperformed an OMA equivalent, suffered degradation in A. SUPERVISED LEARNING
performance in a short packet scenario. In [135], derivations In supervised learning, an ML model is trained to approx-
of the analytical expressions for outage probability and er- imate an arbitrary function using examples that are either
godic capacity of a bidirectional relay-assisted IoT setup are collected from real measurements or generated synthetically.
presented. The relay in the proposed model aids the commu- Depending on the type of output, supervised learning models
nication between two groups of NOMA users and harvests can be classified into regression or classification models.
energy from their communication. Considering hardware While the latter is used in predicting the probability of a given
impairment of communication components, specifically the input belonging to a particular class or classes, the former is
resulting in-phase and quadrature imbalance, X. Li et al. used to approximate continuous functions. In designing su-
[142] showed that such impairment can limit the outage prob- pervised learning-based algorithms, there are multiple stages
ability and ergodic capacity performance of NOMA-based that can be grouped into training and execution phases. In
relaying in IoT networks. NOMA-based relaying can also be the training phase, examples (sets of input and output values)
susceptible to eavesdropping attacks when the eavesdropper are used to optimize the weights of the model using gradient
has a better channel than the target receiver. To address such a descent algorithms. In the execution phase, the trained model
challenge, transmission of artificial noise by a two-way full- is applied to perform predictions or to label new data.
duplex relay was proposed by B. Zheng et al. [143]. Different supervised learning algorithms, including sup-
port vector machines (SVMs), feed forward neural networks
IV. MACHINE LEARNING AND AI FOR IOT RELAYING (FNNs), deep neural networks (DNNs) and decision trees
Future generation communication networks are envisioned (DTs), have been used in existing studies considering relay-
to support applications that have stringent quality of service ing in IoT networks; see, e.g., [153]–[167]. A key bottleneck
(QoS) requirements, such as ultrareliable low latency com- in conventional relaying is the acquisition of accurate CSI.
munication. Next-generation networks are also expected to Conventional relaying based on mathematical optimization
be dynamic owing to the diverse applications that will be requires global CSI. However, with large-scale device de-
supported. To meet these requirements, artificial intelligence ployment in future networks and real-time IoT applications,
and machine learning are key technologies that have been CSI acquisition can add to communication overhead and
proposed. Massive Internet of Things devices generating delays. Supervised learning combined with deep neural net-
large volumes of data will require the power of machine works can offer real-time relay selection algorithms in IoT
learning and intelligence to coordinate their activities. This networks [153]. Considering the nonlinearity in EH relays,
intelligence will not be operational at the core network only [153] developed a DNN-based model for relay selection by
but at the edge as well. using throughput-dependent parameters such as SNR, num-
As a consequence of extreme requirements, next- ber of users and relay position for offline training. One-time
generation wireless networks will feature unprecedented offline model training, as in [153], results in lower complex-
complexity, thereby limiting the applicability of classical ity selection algorithms and can offer a real-time selection
mathematical model-based design methodologies for net- advantage; however, offline training still needs the resources
work design, deployment and network resource optimization of central entities such as access points or base stations. Neu-
[144]. Machine learning (ML) and artificial intelligence (AI) ral network relay selection was shown to offer throughput
are therefore expected to play a pivotal role in the design performance improvement over an SVM approach in [168].
of all aspects of wireless networks. This has led to a surge
in the number of published works exploring ML/AI-based B. REINFORCEMENT LEARNING
data-driven solutions for solving challenges associated with As stated above, supervised learning methods require the
different aspects of wireless network design, including radio availability and/or generation of labeled data sets to train
22 VOLUME 4, 2016
a neural network. In many applications, it may be nearly

impossible to obtain such data sets. This is particularly true
for most problems associated with relaying in IoT networks
due to the nonavailability of labeled data. Reinforcement
learning (RL) ameliorates this problem by instead allowing
neural networks (often referred to as agents) to learn through
a trial-and-error procedure involving interactions with the
environment. During this interaction, a carefully designed
reward signal is used to guide the agent(s) toward learning
to successfully perform a given task. For this reason, RL
methods, including multiarm bandit, Q-learning and deep Q-
RL, are becoming increasingly popular for many wireless
communication applications. As seen in Table 11, research FIGURE 8: Relay-aided health monitoring IoT setup
on RL for relay-enabled IoT is still limited; however, we
foresee a surge in its application in the near future.
In [155], a low-complexity mechanism for relay schedul- parameters are required may not be in the region of an access
ing in cooperative IoT networks using a stateless RL method point and might need to utilize mobile devices as relays
- the multiarmed bandit (MAB) - is investigated. The authors to forward data. Hence, mobile relaying is a key area for
utilized the MAB framework to learn relay scheduling using research in healthcare-oriented IoT.
only the acknowledgments (and negative acknowledgments) Opportunistic relays can ensure connectivity for patients
of packet transmissions. Despite the limited information used who require forwarding services. A relay-enabled IoT net-
for scheduling decisions, this method still shows comparable work using Bluetooth low energy (BLE) technology is imple-
performance to optimal scheduling based on full-CSI but mented to forward medical data to an IoT server using third-
with lower complexity. party mobile relays in [32]. In the work [32], the mobile relay,
which is selected based on the best-received signal strength
V. APPLICATIONS OF RELAY-ENABLED IOT (RSSI), establishes a secure connection to an IoT server and
Relay-enabled IoT finds applications mostly when the trans- is rewarded after successful completion of the forwarding
mitting node or sender in an IoT setup requires assistance in process. The testbed performance showed that the probability
delivering its data to the gateway through which the network of meeting a mobile relay increased with the increase in the
server can be accessed. In such cases, adjacent or neighboring arrival rate of mobile relays into the network.
nodes could be selected to deliver the data. Such adjacent Another issue apart from mobility management in medical
nodes could be static or mobile. Additionally, the use of IoTs is the issue of maintaining the confidentiality of the
relays could be opportunistic in which an available node is transmitted information in the case of untrusted relay nodes.
selected based on some criteria, as discussed in subsection [32] proposed having the relay nodes register with a server
III-B, or it could be preplanned in which case the relays through an application installed on the mobile device.
are included in the network rollout or deployment. In the The abovementioned works under relaying in medical or
reviewed research contributions for this paper, some research healthcare IoT assume that the medical information to be
work focused on relay-aided IoT networks tied to specific relayed through a network for response is of equal priority.
applications. In this section, such works are reviewed to This could be so when a patient has only one sensor that
showcase the use cases for relaying in IoT networks. sends unique priority information. Where there is multipri-
ority information to be relayed, the scheduling needs to be
A. MEDICAL MONITORING OR REMOTE HEALTHCARE a QoS-aware implementation. In [180], the authors proposed
Sensors strapped to the body can report health indicators such a QoS-aware relaying technique for a wireless body access
as the blood pressure and sugar level to a remote server that is network (WBAN). On-body sensors can also serve as relays
accessible to health care providers. Such reports can be criti- in medical IoT in WBAN applications. These sensors can
cal in offering timely diagnoses and providing interventions, help forward the signals from body implants to a local
especially for remotely located patients. In [179], a relay- collection point, such as the mobile device of the patient
assisted IoT setup was proposed for healthcare parameter with a specialized medical monitoring app. This information
monitoring for patients in rural areas. Simulations were used can then be sent through the cellular network to medical
to analyze the end-to-end delay, throughput and energy con- personnel for a prompt response. Fig. 8 shows a set up for
sumption of the proposed setup. The work considered a case medical IoT.
where a large array of sensors opportunistically transmits
the information of a source to a relay that sends the same B. SMART TRANSPORT
information to the network. This approach largely assumes a The IoT is envisioned to change the way people move around.
static relay and does not factor in the mobility of the source This will be accomplished by adding a level of intelligence
nodes or the relay. Sometimes the patients from whom health to the transport system in cities. The IoT is looked upon
VOLUME 4, 2016 23
TABLE 11: Summary of existing contributions on machine learning for relay-based IoT networks.
Machine Learning Tools Main Applications

Existing Contributions
FFN DNN RNN DT/SVM RL RS AS PHS CAR AC PA
[153], [157], [164], [169] X X
[154], [170] X X X
[171]–[173] X X
[155] X X
[156] X X X
[174] X X
[160] X X X
[162], [175] X X X
[163] X X X X
[167], [176] X X X
[177] X X
[178] X X
to help cities plan roads based on the data obtained from bility and BER of the setup were derived with accompany-
road-installed sensors and the data reported from sensor ing performance analysis. Using simulations and numerical
nodes installed in vehicles. Armed with such data, city analysis, the gains of the proposed setup were shown. The
transport agencies can plan better, and vehicle owners can relay strategy is DF, and although relay selection is not
also reserve parking spots in advance. Car owners can also performed, relay interface selection between wireless and
avoid crowded routes for alternative less congested routes. PLC is performed based on the acquired SNR. Unlike the
The application of relay-aided IoT in transportation finds work in [183], which considered the channels in the IoT
expressions where road-installed sensors can take advantage setup to be experiencing Rayleigh fading, two distributions
of the presence of relay nodes installed in vehicles to forward are used in [184], where the S-R and R-D channels follow a
messages to a server. This application can take advantage Nakagami-m and a lognormal distribution, respectively. The
of the available energy of the car that powers the sensor to single relay considered is a PLC and wireless hybrid AF
forward data. Such an approach was studied in [181], where relay assisting a source IoT device to forward its information.
the use of vehicle-mounted relays to assist in forwarding Analytical expressions are derived for the IoT setup for
delay-sensitive data between IoT devices and servers was metrics including outage probability and BER, which are
considered. The authors demonstrate through simulation that verified through simulations.
having many vehicles participate in NB-IoT networks can Similar to the work in [184], Z. Chen et al. [185] analyzed
reduce the message loss probability and increase the energy a hybrid fading scenario having both a wireless and PLC
efficiency of IoT networks. As vehicles are mostly driven interface for the S-R and R-D links, respectively. The wire-
by humans (who are self-focused), an incentive mechanism less channel was modeled as a Nakagami-m fading channel,
to motivate relaying needs to be designed in the proposed whereas the PLC channel was represented with a lognormal
framework. In [182], UAV-assisted vehicle-to-vehicle com- distribution for a single AF relay. The work analyzed the
munication is modeled as a Markov decision process, and performance of the relay setup for a scenario where the
a total throughput maximization problem is formulated for channel gains of both S-R and R-D links are approximated to
the UAV-to-vehicle downlink. This paper proposed a deep have gamma distributions and a scenario where both channels
reinforcement learning-based algorithm to allow the UAV to are approximated to have lognormal distributions. Similar to
determine the optimal policy to optimize the total through- the work in [183] and [184], the work in [185] considered the
put. Specifically, the UAV learns its optimal 3-dimensional presence of a single fixed relay.
position and transmission control (i.e., bandwidth and power
allocation). D. ULTRARELIABLE LOW LATENCY COMMUNICATION
(URLLC) APPLICATION
C. POWER LINE COMMUNICATION Relays are also useful in URLLC applications, which are
An intelligent electricity grid is a thriving research area delay-sensitive applications that involve the transmission
wherein methods have been developed to make the network of short packets. Although introducing relays into URLLC
respond better to consumer-side demand. IoT-based power applications may slightly increase the delay in packet re-
line communication (PLC) can enable smart grid commu- ception, relays can guarantee meeting reliability targets of
nication. The research in [183] considered a hybrid power such critical applications. It has been shown that for a fi-
line and wireless sensor network assisted by relay nodes. nite blocklength scenario, which is characteristic of URLLC
In this work, closed-form expressions of the outage proba- applications, two-hop relaying results in better reliability
24 VOLUME 4, 2016
performance than direct communication [186], [187] and A. FEDERATED LEARNING IN RELAY-ENABLED IOT
[188]. NETWORKS
Examples of such URLLC applications are factory au- The current approach to machine learning in relay-enabled
tomation [187], military operations, augmented reality, IoT networks is centralized except for the work in [195],
cyber-physical systems, and intelligent transport systems where relaying is offered to edge mobile devices as a ser-
[189]. To meet the stringent performance requirements of vice. Centralized machine learning raises the concern of user
URLLC, approaches have included optimizing the packet data privacy. Federated learning overcomes this hurdle by
length for the application [190] and reducing the latency enabling the central server to send models to users who train
experienced by the considered system [189], [191]. these models based on their local data. Updates are then
The research in [190] proposed iterative algorithms to sent by the users to the global model. Federated learning
solve a UAV decoding error minimization problem for a assures user data privacy, as users only send model updates.
frontline operation. The algorithm iteratively found the op- In federated learning over wireless networks, selected edge
timal packet blocklength for fixed UAV location and vice devices that carry out local training may be in an outage state.
versa. It was shown that it matched the performance of an ex- In such a scenario, relays can assist in forwarding the model
haustive search algorithm in error decoding probability, albeit updates and ensure model accuracy and link reliability. Such
with lower complexity. Traditionally, research in wireless re- a scenario results in joint optimization problems that include
laying has assumed infinite packet blocklength transmission. optimizing relay selection and model accuracy, among other
However, for URLLC, there are stringent limits for packet parameters. These problems, often NP-hard and nonconvex,
sizes. Considering fixed packet sizes, the work in [189] are likely to require problem decomposition and the use of
proposed VLC communication between traffic infrastructure heuristics. Relay-assisted federated learning for IoT networks
and vehicles via a DF vehicle relay. The analysis showed is an exciting area that requires further investigation.
improved latency performance (sub-milliseconds) and packet
error rate performance over an RF-based vehicle-to-vehicle B. RELAY-ENABLED IOT NETWORK TESTBEDS
(V2V) relaying system.
Although there is teeming literature on relay-enabled IoT
Pilots that are inserted into packets for channel estimation networks, there has been much focus on the analytical
purposes can contribute to increased packet sizes, especially and theoretical framework, whereas actual demonstration
in delay-sensitive cases such as URLLC. However, pilot through hardware implementation has not kept pace with the
length optimization for URLLC involves reaching a com- theoretical analysis. A few works have demonstrated through
promise between reliability through good channel estimation testbeds [32], [8] and [115] the various use cases of relay-
and low delay using reduced pilot symbols if the delay intro- enabled IoT. Testbeds can help show a proof of concept,
duced by the relay is ignored [191]. Achieving the delay and especially in the area of EH relay-enabled IoT networks
reliability requirements of relay-based URLLC also requires considering the energy constraint of IoT nodes. Hence, more
a suitable relay selection technique. Dense deployment is research effort needs to be employed in developing testbeds
common in URLLC applications, and consequently, selecting to determine whether the actual deployment of IoT relays
a set of good relays for short packet communication is impor- matches the theoretical analysis.
tant [192]. Whereas selection techniques that assume static
channels within the coherence time can offer improvement
C. MODELING MOBILITY OF MOBILE RELAYS
in error rate, relay selection algorithms that are channel-
dynamics-aware can offer an approximate 10% improvement When relay nodes are not static devices but rather mobile
in error rate [193]. Furthermore, when the duplex method devices, the connectivity they provide to source-destination
is considered, URLLC applications fare better (in terms of nodes may be erratic due to the mobility of the relay nodes.
block error rate performance) with full-duplex relays than This can be a problem if the source node being assisted is a
with half-duplex relays [194]. patient’s on-body sensor that reports urgent data through the
relays. To capture the variations that such mobility brings to
the network, [47] used time-varying network graphs, which
VI. OPEN ISSUES would require frequent updating. Moreover, variations in
From the research contributions that have been surveyed, device position can also be presented as uncertainty in CSI
various investigations have been made into the feasibility of acquisition and added as an error term to the channel gain.
having relays in an IoT network. Some of the challenges of For this, an uncertainty-aware relay power allocation algo-
IoT relaying, such as relay selection, have received greater rithm can be used, except that for serious uncertainty cases,
focus than others, such as relay mobility. These challenges the energy store of the relay will be drained. Hence, mobility-
and the approaches put forward in the literature stimulate aware relay-enabled IoT is an area of future research. An
open issues that can inspire future research directions. In this approach could be the use of ML algorithms that are trained
section, some of the open issues are discussed. by the mobility history of user-held devices.
VOLUME 4, 2016 25
D. OPTIMAL RELAY PLACEMENT laid out. Open issues in relaying in the IoT have also been
Some works have shown that with one or multiple relays, the discussed.
reliability of an IoT network can be improved; for example,
see [69]. However, optimal positioning of these relays has REFERENCES
not been investigated. For a single relay, a straightforward [1] J. A. Stankovic, “Research directions for the internet of things,” IEEE
approach would be to position the relay midway between the Internet of Things Journal, vol. 1, no. 1, pp. 3–9, February 2014.
IoT end device and the gateway. However, when there are [2] Cisco, “Cisco’s annual internet report (2018 - 2023)
white paper,” Cisco, Tech. Rep., 03 2020. [Online].
many IoT end devices, such as smart agriculture occupying a Available: https://www.cisco.com/c/en/us/solutions/collateral,executive-
large geographical area, it remains to be shown whether mid- perspectives/annual-internet-report/white-paper-c11-741490.html
way positioning would be optimal. Such optimal placement [3] J. Lianghai, B. Han, M. Liu, and H. D. Schotten, “Applying device-to-
device communication to enhance IoT services,” IEEE Communications
can be a metric to select a mobile relay when it is in a defined Standards Magazine, vol. 1, no. 2, pp. 85–91, 2017.
optimal region. [4] D. S. Gurjar, H. H. Nguyen, and H. D. Tuan, “Wireless information and
power transfer for IoT applications in overlay cognitive radio networks,”
E. LEAN CSI ACQUISITION TECHNIQUES IEEE Internet of Things Journal, vol. 6, no. 2, pp. 3257–3270, 2019.
[5] M. R. Palattella, M. Dohler, A. Grieco, G. Rizzo, J. Torsner, T. Engel,
Traditional or conventional relaying relies on the availabil- and L. Ladid, “Internet of things in the 5G era: Enablers, architecture, and
ity of CSI to make relay selection decisions. The channels business models,” IEEE Journal on Selected Areas in Communications,
for which CSI acquisition is critical include the relay-to- vol. 34, no. 3, pp. 510–527, March 2016.
[6] V. Petrov, K. Mikhaylov, D. Moltchanov, S. Andreev, G. Fodor,
destination channels. With massive deployment of devices, J. Torsner, H. Yanikomeroglu, M. Juntti, and Y. Koucheryavy, “When
the number of devices from which a relay can be selected IoT keeps people in the loop: A path towards a new global utility,” IEEE
increases. For such many devices in a relay set, acquiring Communications Surveys Tutorials, 2018.
[7] K. Mekkia, E. Bajica, F. Chaxela, and F. Meyerb, “A comparative study
perfect CSI can potentially weigh heavily on the overhead. of LPWAN technologies for large-scale IoT deployment,” ICT Express,
Moreover, for URLLC applications, where short packet com- vol. 5, no. 1, pp. 1–7, April 2019.
munication is a key feature, perfect CSI acquisition can cause [8] M. Diop and C. Pham, “Increased flexibilty in long-range IoT deploy-
ments with transparent and light-weight 2-hop LoRa approach,” in 2019
severe delays. Despite the gains of NOMA, as demonstrated
Wireless Days (WD), 2019, pp. 1–6.
in spectral efficiency and secrecy probability gains, NOMA [9] F. Adelantado, X. Vilajosana, P. Tuset-Peiro, B. Martinez, J. Melià-
requires knowledge of the existing channels. Therefore, an Seguí, and T. Watteyne, “Understanding the limits of LoRaWAN,” IEEE
exciting area for research is efficient CSI acquisition in relay- Communications Magazine, vol. 55, no. 9, pp. 34 – 40, September 2017.
[10] L. Duan, L. Huang, C. Langbort, A. Pozdnukhov, J. Walrand, and
enabled IoT networks. ML/AI-based relaying in the IoT L. Zhang, “Human-in-the-loop mobile networks: A survey of recent
has been shown to offer reduced complexity; specifically, advancements,” IEEE Journal on Selected Areas In Communications,
complexity reduction can be achieved using deep neural vol. 33, no. 4, pp. 813–831, April 2017.
[11] B. Ooi, Z. Kong, W. Lee, and S. Shirmohammadi, “A collaborative IoT-
networks. However, these approaches are largely centralized gateway architecture for reliable and cost effective measurements,” IEEE
approaches that may be challenging due to the distributed Instrumentation Measurement Magazine, vol. 22, no. 6, pp. 11–17, 2019.
nature of certain IoT applications. [12] C. Stergiou, K. E. Psannis, B. B. Gupta, and Y. Ishibashi, “Security,
privacy & efficiency of sustainable cloud computing for big data & iot,”
Sustainable Computing: Informatics and Systems, vol. 19, pp. 174–184,
VII. CONCLUSION 2018.
Due to obstructions and fading, the quality of the direct link [13] N. Nomikos, T. Charalambous, I. Krikidis, D. N. Skoutas, D. Vouyioukas,
between source and destination devices might require that a M. Johansson, and C. Skianis, “A survey on buffer-aided relay selection,”
IEEE Communications Surveys & Tutorials, vol. 18, no. 2, pp. 1073–
relay be coopted to help forward signals to the destination. 1097, 2015.
The availability of relay nodes provides both opportunities [14] M. A. Hossain, R. M. Noor, K.-L. A. Yau, I. Ahmedy, and S. S. Anjum,
and challenges. The gains of relaying indicate that relays can “A survey on simultaneous wireless information and power transfer with
cooperative relay and future challenges,” IEEE access, vol. 7, pp. 19 166–
be built into the IoT architecture or that the presence of relays 19 198, 2019.
can be exploited opportunistically for data forwarding. [15] J. Tournier, F. Lesueur, F. Le Mouël, L. Guyon, and H. Ben-Hassine, “A
Resource constraints in relay nodes, specifically energy, survey of iot protocols and their security issues through the lens of a
can be relieved using energy harvested from radio frequency generic iot stack,” Internet of Things, p. 100264, 2020.
[16] R. Elouafadi and M. Benjillali, “Cooperative noma-based d2d communi-
or from green energy sources, including solar energy. Har- cations: A survey in the 5g/iot context,” in 2018 19th IEEE Mediterranean
vesting energy and transmitting information concurrently, Electrotechnical Conference (MELECON). IEEE, 2018, pp. 132–137.
however attractive, comes with its hardware complexity con- [17] M. M. Alsulami and N. Akkari, “The role of 5g wireless networks in
the internet-of-things (iot),” in 2018 1st International Conference on
straint. Relay selection, a classic problem in relay networks, Computer Applications & Information Security (ICCAIS). IEEE, 2018,
has a spectrum of algorithms put forward to offer near- pp. 1–8.
optimal and optimal solutions. In this survey, the above as- [18] A. Vyas, S. Pal, and B. K. Saha, “Relay-based communications in wbans:
A comprehensive survey,” ACM Computing Surveys (CSUR), vol. 54,
pects of relaying in the IoT were comprehensively reviewed. no. 1, pp. 1–34, 2021.
Topical classifications of current approaches employed in [19] J. Ding, M. Nemati, C. Ranaweera, and J. Choi, “Iot connectivity tech-
the literature have been discussed. Possible application areas nologies and applications: A survey,” arXiv preprint arXiv:2002.12646,
of relay-enabled IoT have also been surveyed. Incentive- 2020.
[20] R. Hassan, F. Qamar, M. K. Hasan, A. H. M. Aman, and A. S. Ahmed,
based approaches to relaying have also been examined, and “Internet of things and its applications: A comprehensive survey,” Sym-
the application areas of relay-aided IoT networks have been metry, vol. 12, no. 10, p. 1674, 2020.
26 VOLUME 4, 2016
[21] S. K. Goudos, P. I. Dallas, S. Chatziefthymiou, and S. Kyriazakos, “A [44] B. Wang, Z. Han, and K. Liu, “Distributed relay selecction and power
survey of iot key enabling and future technologies: 5g, mobile iot, sematic control for multiuser cooperative communication using buyer/seller
web and applications,” Wireless Personal Communications, vol. 97, no. 2, game,” in IEEE INFOCOM 2007 - 26th IEEE International Conference
pp. 1645–1675, 2017. on Computer Communications, 2007, pp. 544–552.
[22] P. P. Ray, “A survey on internet of things architectures,” Journal of King [45] J. Huang and A. L. Swindlehurst, “Cooperative jamming for secure
Saud University-Computer and Information Sciences, vol. 30, no. 3, pp. communications in MIMO relay networks,” IEEE Transactions on Signal
291–319, 2018. Processing, vol. 59, no. 10, pp. 4871–4884, July 2011.
[23] G. Cai, Y. Fang, J. Wen, G. Han, and X. Yang, “Qos-aware buffer-aided [46] T. Rosati and E. Wood, “End-to-end secure LoRaWAN: Secure devices
relaying implant wban for healthcare iot: Opportunities and challenges,” with key management for provisioning to operations,” Tech. Rep., March
IEEE Network, vol. 33, no. 4, pp. 96–103, 2019. 2020.
[24] A. Hezam, D. Konstantas, and M. Mahyoub, “A comprehensive IoT [47] S. Redhu, M. Anupam, and R. M. Hegde, “Optimal relay node selec-
attacks survey based on a building-blocked reference mode,” Int. J. Adv. tion for robust data forwarding over time-varying IoT networks,” IEEE
Comput. Sci. Appl, vol. 9, 2018. Transactions on Vehicular Technology, vol. 68, no. 9, pp. 9178–9190,
[25] G. Liu, F. R. Yu, H. Ji, V. Leung, and X. Li, “In-band full-duplex relay- Sep. 2019.
ing: A survey, research issues and challenges,” IEEE Communications [48] D. K. P. Asiedu, H. Lee, and K. Lee, “Simultaneous wireless information
Surveys Tutorials, vol. 17, no. 2, pp. 500–524, 2015. and power transfer for decode-and-forward multihop relay systems in
[26] J. W. Raymond, T. O. Olwal, and A. M. Kurien, “Cooperative communi- energy-constrained IoT networks,” IEEE Internet of Things Journal,
cations in machine to machine(M2M): Solutions, challenges and future vol. 6, no. 6, pp. 9413–9426, Dec 2019.
work,” IEEE Access, vol. 6, pp. 9750–9766, 2018. [49] B. Ji, K. Song, C. Li, W. ping Zhu, and L. Yang, “Energy harvest
[27] B. Jedari, F. Xia, and Z. Ning, “A survey on human-centric communi- and information transmission design in internet-of-things wireless
cations in non-cooperative wireless relay networks,” IEEE Communica- communication systems,” AEU - International Journal of Electronics
tions Surveys Tutorials, vol. 20, no. 2, pp. 914–944, 2018. and Communications, vol. 87, pp. 124 – 127, 2018. [Online]. Available:
[28] A. Vyas, S.Pal, and B. K. Saha, “Relay based communications in wban:a http://www.sciencedirect.com/science/article/pii/S1434841117329357
comprehensive survey,” ACM Comput. Surv., vol. 54, no. 1, Jan 2021. [50] Y. Zou, J. Zhu, and X. Jiang, “Joint power splitting and relay selection in
[29] L. Chettri and R. Bera, “A comprehensive survey on internet of things energy-harvesting communications for IoT networks,” IEEE Internet of
(iot) toward 5g wireless systems,” IEEE Internet of Things Journal, vol. 7, Things Journal, vol. 7, no. 1, pp. 584–597, Jan 2020.
no. 1, pp. 16–32, 2020. [51] W. Lu, G. Liu, P. Si, G. Zhang, B. Li, and H. Peng, “Joint resource
optimization in simultaneous wireless information and power transfer
[30] S. Borkotoky, U. Schilcher, and C. Bettstetter, “Cooperative relaying in
(SWIPT) enabled multi-relay internet of things (iot) system,” Sensors,
LoRa sensor networks,” in 2019 IEEE Global Communications Cofer-
vol. 19, no. 11, pp. 25–36, 2019.
ence (GLOBECOM), 2019, pp. 1–5.
[52] P. Ramezani, Y. Zeng, and A. Jamalipour, “Optimal resource allocation
[31] D. Di Lecce, A. Grassi, G. Piro, and G. Boggia, “Boosting energy
for multiuser Internet of Things network with single wireless-powered
efficiency of NB-IoT cellular networks through cooperative relaying,” in
relay,” IEEE Internet of Things Journal, vol. 6, no. 2, pp. 3132–3142,
2018 IEEE 29th Annual International Symposium on Personal, Indoor
April 2019.
and Mobile Radio Communications (PIMRC), Sep. 2018, pp. 1–5.
[53] B. Lyu, Z. Yang, H. Guo, F. Tian, and G. Gui, “Relay cooperation
[32] P. Porambage, A. Manzoor, M. Liyanage, A. Gurtov, and M. Ylianttila,
enhanced backscatter communication for Internet-of-Things,” IEEE In-
“Managing mobile relays for secure e2e connectivity of low-power
ternet of Things Journal, vol. 6, no. 2, pp. 2860–2871, 2019.
IoT devices,” in 2019 16th IEEE Annual Consumer Communications
[54] B.Lyu, D.T.Hoang, and Z.Yang, “Backscatter then forward: A relaying
Networking Conference (CCNC), Jan 2019, pp. 1–7.
scheme for batteryless IoT networks,” IEEE WIreless Communications
[33] E. Van Der Meulen, “Three-terminal communication channels,” Ad-
Letters, vol. 9, no. 4, pp. 562–566, 2020.
vances in Applied Probability, vol. 3, no. 1, pp. 120–154, 1971.
[55] T. N. Nguyen, T. H. Q. Minh, P. T. Tran, and M. Voznak, “Adaptive
[34] T. Cover and A. Gamal, “Capacity theorems for the relay channel,” IEEE energy harvesting relaying protocol for two-way half-duplex system
Transactions on Information Theory, vol. 25, no. 5, pp. 572–584, 1979. network over rician fading channels,” Wireless Communications and
[35] M. Gastpar, G. Gerhard, and P. Gupta, “The multiple-relay channel: Cod- Mobile Computing, vol. 2018, 2018.
ing and antenna-clustering capacity,” in Proceedings IEEE International [56] Y. Hu, Y. Zhu, M. C. Gursoy, and A. Schmeink, “SWIPT-enabled
Symposium on Information Theory, 2002, p. 136. relaying in iot networks operating with finite blocklength codes,” IEEE
[36] A. Adinoyi and Y. Halim, “Cooperative relaying in multi-antenna fixed Journal on Selected Areas in Communications, vol. 37, no. 1, pp. 74–88,
relay networks,” IEEE Transactions on Wireless Communications, vol. 6, Jan 2019.
no. 2, pp. 533–544, 2007. [57] G. Shabbir, J. Ahmad, W. Raza, Y. Amin, A. Akram, J. Loo, and
[37] B. Xia, Y. Fan, J. Thompson, and H.V.Poor, “Buffering in a three-node H. Tenhunen, “Buffer-aided successive relay selection scheme for energy
relay network,” IEEE Transactions on Wireless Communications, vol. 7, harvesting IoT networks,” IEEE Access, vol. 7, pp. 36 246–36 258, 2019.
no. 11, pp. 4492–4496, 2008. [58] A. Rauniyar, P. E. Engelstad, and O. N. Østerbø, “Performance analysis
[38] J. Laneman, D. Tse, and G. Cornell, “Cooperative diversity in wireless of RF energy harvesting and information transmission based on noma
network: Efficient protocols and outage behavior,” IEEE Transactions on with interfering signal for iot relay systems,” IEEE Sensors Journal,
Information Theory, vol. 50, no. 12, pp. 3062–3080, 2004. vol. 19, no. 17, pp. 7668–7682, 2019.
[39] Huining Hu, H. Yanikomeroglu, D. Falconer, and S. Periyalwar, “Range [59] R. Yao, W. Wang, M. Farrokh-Baroughi, H. Wang, and Y. Qiang,
extension without capacity penalty in cellular networks with digital “Quality-driven energy-neutralized power and relay selection for smart
fixed relays,” in IEEE Global Telecommunications Conference, 2004. grid wireless multimedia sensor based IoTs,” IEEE Sensors Journal,
GLOBECOM ’04., vol. 5, 2004, pp. 3053–3057 Vol,5. vol. 13, no. 10, pp. 3637–3644, 2013.
[40] W. Zhang and K. B. Letaief, “Cooperative spectrum sensing with transmit [60] Weightless, “What is weightless,” Weightless, Tech. Rep., 2015.
and relay diversity in cognitive radio networks - [transaction letters],” [Online]. Available: https://www.weightless.org
IEEE Transactions on Wireless Communication, vol. 7, no. 12, pp. 4761– [61] L.Zhang and Y.Liang, “Average throughput analysis and optimization
4766, 2008. in cooperative IoT networks with short packet communication,” IEEE
[41] S.Fedor and M. Collier, “On the problem of energy efficiency of multi- Transaction on Vehicular Technology, vol. 67, no. 12, pp. 11 549–11 562,
hop vs one-hop routing in wireless sensors networks,” in 21st Interna- 2018.
tional Conference on Advanced Information Networking and Applica- [62] H. Kawabata, K. Ishibashi, S. Vuppala, and G. T. F. de Abreu, “Robust
tions Workshop(AINAW’07), vol. 2, 2007, pp. 380–385. relay selection for large-scale energy-harvesting IoT networks,” IEEE
[42] J. Chen, J. Liang, and Z. Chen, “Energy-efficient uplink radio resource Internet of Things Journal, vol. 4, no. 2, pp. 384–392, April 2017.
management in LTE-advanced relay networks for Internet of Things,” [63] B. Omoniwa, R. Hussain, M. Adil, A. Shakeel, A. K. Tahir, Q. U. Hasan,
in 2014 International Wireless Communications and Mobile Computing and S. A. Malik, “An optimal relay scheme for outage minimization in
Conference (IWCMC), Aug 2014, pp. 745–750. fog-based Internet-of-Things (IoT) networks,” IEEE Internet of Things
[43] X. Liu and N. Ansari, “Green relay assisted D2D communications with Journal, vol. 6, no. 2, pp. 3044–3054, April 2019.
dual batteries in heterogeneous cellular networks for IoT,” IEEE Internet [64] N. Dao, M. Park, J. Kim, J. Paek, and S. Cho, “Resource-aware
of Things Journal, vol. 4, no. 5, pp. 1707–1715, Oct 2017. relay selection for inter-cell interference avoidance in 5G heterogeneous
VOLUME 4, 2016 27
network for Internet of Things systems,” Future Generation Computer [85] A. Das, N. Das, A. Das Barman, and S. Dhar, “Energy incentive for
Systems, vol. 93, pp. 877 – 887, 2019. [Online]. Available: packet relay using cognitive radio in IoT networks,” IEEE Communica-
http://www.sciencedirect.com/science/article/pii/S0167739X17304028 tions Letters, vol. PP, pp. 1–1, 06 2019.
[65] J. Lianghai, B. Han, M. Liu, and H. D. Schotten, “Applying device-to- [86] Y. Zhan, Y. Xia, J. Zhang, and Y. Wang, “Incentive mechanism design in
device communication to enhance IoT services,” IEEE Communications mobile opportunistic data collection with time sensitivity,” IEEE Internet
Standards Magazine, vol. 1, no. 2, pp. 85–91, 2017. of Things Journal, vol. 5, no. 1, pp. 246–256, 2018.
[66] A. Gouissem, L. Samara, R. Hamila, N. Al-Dhahir, L. Ben-Brahim, and [87] Z. Hou, H. Chen, Y. Li, and B. Vucetic, “Incentive mechanism design for
A. Gastli, “Machine-learning based relay selection in AF cooperative wireless energy harvesting-based Internet of Things,” IEEE Internet of
networks,” in 2019 IEEE Wireless Communications and Networking Things Journal, vol. 5, no. 4, pp. 2620–2630, 2017.
Conference (WCNC), April 2019, pp. 1–7. [88] L. Rodríguez, N. Tran, T. Duong, T. Le-Ngoc, M. Elkashlan, and
[67] S.Hsu, C.Lin, C.Wang, and W.Chen, “Breaking bndwidth limitation for S. Shetty, “Physical layer security in wireless cooperative relay networks:
mission-critical IoT using sequential multiple relays,” in IEEE Internet of State of the art and beyond,” IEEE Coomunications Magazine, pp. 32 –
Things Journal, vol. 5, no. 5, 2018, pp. 3316–3329. 39, December 2015.
[68] X. Ding, Y. Zou, F. Ding, D. Zhang, and G. Zhang, “Opportunistic relay- [89] P. Huang, Y. Hao, T. Lv, J. Xing, J. Yang, and P. T. Mathiopoulos, “Secure
ing against eavesdropping for Internet-of-Things: A security-reliability beamforming design in relay-assisted Internet of Things,” IEEE Internet
tradeoff perspective,” IEEE Internet of Things Journal, vol. 6, no. 5, pp. of Things Journal, vol. 6, no. 4, pp. 6453–6464, Aug 2019.
8727–8738, 2019. [90] Q. Xu, P. Ren, H. Song, and Q. Du, “Security enhancement for IoT
[69] M. J. Farooq, H. ElSawy, Q. Zhu, and M. Alouini, “Optimizing mission communications exposed to eavesdroppers with uncertain locations,”
critical data dissemination in massive IoT networks,” in 2017 15th IEEE Access, vol. 4, pp. 2840–2853, 2016.
International Symposium on Modelling and Optimization in Mobile, Ad [91] Caihong Han, L.Sun, R.Ren, and Y.Wang, “Toward secure Internet
Hoc and Wireless Networks (WiOpt), 2017, pp. 1–6. of Things via hybrid forwarding and opportunistic relaying,” in 2016
[70] 3GPP, “3GPP TR 36.802, narrowband Internet of Things (NB-IoT), IEEE/CIC International Conference on Communications in ChinaICCC
technical report TR 36.802 V1.0.0 , technical specification group radio Workshops, 2016, pp. 1–6.
access networks„” 3GPP, Tech. Rep., June 2016. [92] Q. Du, N. Lu, L. Sun, X. Zhang, and B. Sun, “Robust relay in narrow-
[71] U. Uyoata, M. Dlodlo, and J. Mwangama, “Robust multicast device band communications for ubiquitous IoT access,” Journal of Sensors, vol.
- to - device communication,” in 2018 IEEE International Conference 2017, 2017.
on Advanced Networks and Telecommunications Systems (ANTS), Dec [93] V. N. Vo, C. So-In, H. Tran, D. Tran, S. Heng, P. Aimtongkham, and
2018, pp. 1–6. A. Nguyen, “On security and throughput for energy harvesting untrusted
[72] U. Uyoata and M. Dlodlo, “Joint power allocation and relay selection for relays in IoT systems using NOMA,” IEEE Access, vol. 7, pp. 149 341–
relay assisted D2D communication with channel uncertainties,” in 2017 149 354, 2019.
IEEE EUROCON International Conference on Smart Technologies, July
[94] M. Li, F. R. Yu, P. Si, R. Yang, Z. Wang, and Y. Zhang, “Uav-assisted data
2017, pp. 486 – 490.
transmission in blockchain-enabled m2m communications with mobile
[73] C. Zhang, J. Ge, M. Pan, F. Gong, and J. Men, “One stone two birds: A edge computing,” IEEE Network, vol. 34, no. 6, pp. 242–249, 2020.
joint thing and relay selection for diverse IoT networks,” IEEE Transac-
[95] A. Islam and S. Y. Shin, “Buav: A blockchain based secure uav-assisted
tions on Vehicular Technology, vol. 67, no. 6, pp. 5424–5434, June 2018.
data acquisition scheme ininternet of things,” Journal Of Communica-
[74] M. Alkhawatrah, Y. Gong, G. Chen, S. Lambotharan, and J. A. Chambers,
tions and Networks, vol. 21, no. 5, pp. 491–502, 2019.
“Buffer-aided relay selection for cooperative NOMA in the Internet of
[96] Y. Gao, M. Wu, Y. Xiao, P. Yang, B. Fu, and D. Wang, “Blockchain
Things,” IEEE Internet of Things Journal, vol. 6, no. 3, pp. 5722–5731,
enabled distributed cooperative d2d communications,” 2019, pp. 1–6.
June 2019.
[75] H. Nasir, N. Javaid, W. Raza, M. Guizani, N. Alrajeh, and M. S. [97] A. Nair, S. Asmi, and G. Aloor, “Analysis of physical layer security via
Alabed, “Virtual-link relay selection scheme for buffer-aided IoT based co-operative communication in Internet of Things,” Procedia Technology,
cooperative relay networks,” IEEE Access, vol. 6, pp. 74 648–74 659, vol. 24, pp. 896–903, 12 2016.
2018. [98] M. Mozaffari, W.Saad, M. Bennis, and M. Debbah, “Unmanned aerial
[76] J. Xia, D. Deng, Y. Rao, D. Li, F. Zhu, and L. Fan, “When distributed vehicle with underlaid device-to-device communications: Performance
switch-and-stay combining meets buffer in IoT relaying networks,” and tradeoffs,” IEEE Transactions on Wireless Communications, vol. 15,
Physical Communication, vol. 38, p. 100920, 2020. [Online]. Available: no. 6, pp. 3949 – 3963, June 2016.
http://www.sciencedirect.com/science/article/pii/S1874490719304252 [99] Y. Ma, N.Selby, and F. Adib, “Drone relays for battery-free networks,” in
[77] B. Shao and M. S. Leeson, “PaFiR: Particle filter routing – a predictive re- SIGCOMM ’17: Proceedings of the Conference of ACM Special Interest
laying scheme for UAV-assisted IoT communications in future innovated Group on Data Communication, 2017, p. 335–347.
networks,” Internet of Things, p. 100077, 2019. [Online]. Available: [100] M. Mozaffari, W.Saad, M. Bennis, and M. Debbah, “Mobile internet of
http://www.sciencedirect.com/science/article/pii/S2542660519301362 things: Can uavs provide an energy-efficient mobile architecture?” in
[78] Y. M. M. Fouad, R. H. Gohary, and H. Yanikomeroglu, “Chinese re- 2016 IEEE Global Communications Conference (GLOBECOM), 2016,
mainder theorem-based sequence design for resource block assignment pp. 1–6.
in relay-assisted Internet-of-Things communications,” IEEE Transactions [101] N. H. Motlagh, M.Bagaa, and T. Taleb, “Uav selection for a uav-
on Wireless Communications, vol. 17, no. 5, pp. 3401–3416, May 2018. based integrative iot platform,” in 2016 IEEE Global Communications
[79] D. Chen, W. Yang, J. Hu, Y. Cai, and X. Tang, “Energy-efficient secure Conference (GLOBECOM), 2016, pp. 1–6.
transmission design for the Internet of Things with an untrusted relay,” [102] Y. Chen, X.Liu, N. Zhao, and Z. Ding, “Using multiple uavs as relays for
IEEE Access, vol. 6, pp. 11 862–11 870, 2018. reliable communications,” in 2018 IEEE Vehicular Technology Confer-
[80] C. Pham, A. Makhoul, and M. Diop, “Similarity detection for smart ence (VTC Spring), 2018, pp. 1–5.
and transparent long-range IoT relaying,” in 2019 IEEE Symposium on [103] L. Kong, L.Ye, F. Wu, M. Tao, G. Chen, and A. V. Vasilakos, “Au-
Computers and Communications (ISCC), 2019, pp. 1–7. tonomous relay for millimeter-wave wireless communications,” IEEE
[81] Y. Huo, M. Xu, X. Fan, and T. Jing, “A novel secure relay selection Journal on Selected Areas in Communications, vol. 35, no. 9, pp. 3949 –
strategy for energy-harvesting-enabled Internet of Things,” EURASIP 3963, September 2017.
Journal on Wireless Communications and Networking, vol. 2018, 12 [104] R. Fan, J.Cui, S. Jin, K. Yang, and J. An, “Optimal node placement
2018. and resource allocation for uav relaying network,” IEEE Communication
[82] V. Nhan Vo, D. Tran, C. So-In, and H. Tran, “Secrecy performance Letters, vol. 22, no. 4, pp. 808 – 811, April 2018.
analysis for fixed-gain energy harvesting in an Internet of Things with [105] A. Chamseddine, O. Akhrif, G. Charland-Arcand, F. Gagnon, and
untrusted relays,” IEEE Access, vol. 6, pp. 48 247–48 258, 2018. D. Couillard, “Communication relay for multiground units with un-
[83] X. Zhang, P. Huang, L. Guo, and M. Sha, “Incentivizing relay participa- manned aerial vehicle using only signal strength and angle of arrival,”
tion for securing IoT communication,” in IEEE INFOCOM 2019 - IEEE IEEE Transactions on Control System Technology, vol. 25, no. 1, pp. 286
Conference on Computer Communications, April 2019, pp. 1504–1512. – 239, 2017.
[84] G. Ramezan and C. Leung, “A blockchain-based contractual routing [106] M. A. Abd-Elmagid and H. S. Dhilon, “Average peak age-of-information
protocol for the Internet of Things using smart contracts,” Wireless minimization in uav-assisted iot networks,” IEEE Transactions on Vehic-
Communications and Mobile Computing, vol. 2018, 2018. ular Technology, vol. 68, no. 2, pp. 2003 – 2008, January 2019.
28 VOLUME 4, 2016
[107] D.-H. Tran, V. Nguyen, S. Gautam, S. Chatzinotas, T. X. Vu, and B. Ot- SPECIAL SECTION ON GREEN INTERNET OF THINGS, vol. 8, no.
tersten, “Uav relay-assisted emergency communications in iot networks: 2020, pp. 117 503–117 512, 2020.
Resource allocation and trajectory optimization,” arXiv e-prints, pp. [129] Y. Guo, Z. Zhao, R. Zhao, S. Lai, Z. Dan, J. Xia, and L.Fan, “Intelligent
arXiv–2008. offloading strategy design for relaying mobile edge computing networks,”
[108] M. Samir, S. Sharafeddine, C. M. Assi, T. M. Nguyeb, and A. Ghrayeb, IEEE Access: Special Section on Emerging Trends, Issues and Chal-
“Uav trajectory planning for data collection from time-constrained iot lenges for Array Signal Processing and Its Applications in Smart City,
devices,” IEEE Transactions on Wireless Communications, vol. 19, no. 1, vol. 8, no. 2020, pp. 35 127–35 135, 2020.
pp. 34 – 46, 2020. [130] J. Xia, C. Li, X. Lai, S. Lai, F. Zhu, D. Deng, and L. Fan, “Cache-aided
[109] O. Ghdiri, W.Jafar, S. Alfattani, J. B. Abderrazak, and H. Yanikomeroglu, mobile edge computing for b5g wireless communication networks,”
“Energy efficient multi-uav data collection for iot networks with time EURASIP Journal on Wireless Communications and Networking, vol.
deadlines,” in GLOBECOM 2020 - 2020 IEEE Global Communications 2020, no. 1, pp. 1–10, 2020.
Conference, 2020, pp. 1–6. [131] B. Jiang, J. Yang, H. Xu, H. Song, and G. Zheng, “Multimedia data
[110] X. Ma, T. Liu, S. Liu, R. Kacimi, and R. Dhaou, “Priority-based throughput maximization in Internet-of-Things system based on opti-
data collection for UAV-aided Mobile Sensor Network,” Sensors, pp. mization of cache-enabled UAV,” IEEE Internet of Things Journal, vol. 6,
1–24, May 2020. [Online]. Available: https://hal.archives-ouvertes.fr/hal- no. 2, pp. 3525–3532, April 2019.
02617533 [132] M. Elbayoumi, M. Kamel, W. Hamouda, and A. Youssef, “Noma-assisted
[111] H.-T. Ye, X. Kang, J. Joung, and Y.-C. Liang, “Optimization for full- machine-type communications in udn: State-of-the-art and challenges,”
duplex rotary-wing uav-enabled wireless-powered iot networks,” IEEE IEEE Communications Surveys Tutorials, vol. 22, no. 2, pp. 1276–1304,
Transactions on Wireless Communications, vol. 19, no. 7, pp. 5057– 2020.
5072, 2020. [133] Z. Zhang, G. Yang, Z. Ma, M. Xiao, Z. Ding, and P. Fan, “Heteroge-
[112] B. Duo, Q. Wu, X. Yuan, and R. Zhang, “Energy efficiency maximiza- neous ultradense networks with noma: System architecture, coordination
tion for full-duplex uav secrecy communication,” IEEE Transactions on framework, and performance evaluation,” IEEE Vehicular Technology
Vehicular Technology, vol. 69, no. 4, pp. 4590–4595, 2020. Magazine, vol. 13, no. 2, pp. 110–120, 2018.
[113] B. Ji, Y. Li, B. Zhou, C. Li, K. Song, and H. Wen, “Performance analysis [134] S. Arzykulov, G. Nauryzbayev, M. S. Hashmi, A. M. Eltawil, K. M.
of uav relay assisted iot communication network enhanced with energy Rabie, and S. Seilov, “Hardware- and interference-limited cognitive
harvesting,” IEEE Access, vol. 7, pp. 38 738–38 747, 2019. iot relaying noma networks with imperfect sic over generalized non-
[114] O. Flauzac, J. Herard, F. Nolot, and P. Cola, “A low power LoRa- homogeneous fading channels,” IEEE Access, vol. 8, pp. 72 942–72 956,
LoRaWAN relay function with a single input, single output device,” in 2020.
International Conference on Embedded Wireless Systems and Networks [135] A. Rauniyar, P. E. Engelstad, and O. N. Østerbø, “On the performance
(EWSN) 2020, 2020, pp. 283–288. of bidirectional noma-swipt enabled iot relay networks,” IEEE Sensors
[115] A. Sanfratello, “Enabling relay-based communication in LoRa networks Journal, vol. 21, no. 2, pp. 2299–2315, 2021.
for the Internet of Things: design, implementation and experimental [136] Z. Mobini, M. Mohammadi, H. A. Suraweera, and Z. Ding, “Full-
evaluation,” Master’s Thesis, Universit‘a di Pisa, 2016. duplex multi-antenna relay assisted cooperative non-orthogonal multiple
access,” in GLOBECOM 2017 - 2017 IEEE Global Communications
[116] Silicon-Lab, “User manual starter kit efm32gg-stk3700,” Silicon Lab,
Conference, 2017, pp. 1–7.
Tech. Rep., 2013.
[137] R. Elouafadi and M. Benjillali, “Cooperative noma-based d2d communi-
[117] B. Velde, “Multi-hop LoRaWAN: including a forwarding node,” 2017.
cations: A survey in the 5g/iot context,” in 2018 19th IEEE Mediterranean
[118] T. Lv, Z. Lin, P. Huang, and J. Zeng, “Optimization of the energy-efficient
Electrotechnical Conference (MELECON), 2018, pp. 132–137.
relay-based massive IoT network,” IEEE Internet of Things Journal,
[138] Z. Hadzi-Velkov, S. Pejoski, N. Zlatanov, and R. Schober, “Uav-assisted
vol. 5, no. 4, pp. 3043–3058, Aug 2018.
wireless powered relay networks with cyclical noma-tdma,” IEEE Wire-
[119] G. Chen, J. P. Coon, A. Mondal, B. Allen, and J. A. Chambers, “Perfor- less Communications Letters, vol. 9, no. 12, pp. 2088–2092, 2020.
mance analysis for multihop full-duplex IoT networks subject to poisson [139] D.-T. Do, T.-T. T. Nguyen, C.-B. Le, M. Voznak, Z. Kaleem, and K. M.
distributed interferers,” IEEE Internet of Things Journal, vol. 6, no. 2, pp. Rabie, “Uav relaying enabled noma network with hybrid duplexing and
3467–3479, April 2019. multiple antennas,” IEEE Access, vol. 8, pp. 186 993–187 007, 2020.
[120] U. Tefek and T. J. Lim, “Relaying and radio resource partitioning for [140] B. Chen, Y. Chen, Y. Chen, Y. Cao, N. Zhao, and Z. Ding, “A novel spec-
machine-type communications in cellular networks,” IEEE Transactions trum sharing scheme assisted by secondary noma relay,” IEEE Wireless
on Wireless ommunication, vol. 16, no. 2, pp. 1344–1356, 2016. Communications Letters, vol. 7, no. 5, pp. 732–735, 2018.
[121] M. Jain, J. I. Choi, T. Kim, D. Bharadia, S. Seth, K. Srinivasan, P. Levis, [141] Z. Xiang, W. Yang, Y. Cai, Z. Ding, Y. Song, and Y. Zou, “Noma-assisted
S. Katti, and P. Sinha, “Practical, real-time, full duplex wireless,” in secure short-packet communications in iot,” IEEE Wireless Communica-
Proceedings of the 17th Annual International Conference on Mobile tions, vol. 27, no. 4, pp. 8–15, 2020.
Computing and Networking, ser. MobiCom ’11. New York, NY, USA: [142] X. Li, M. Zhao, X.-C. Gao, L. Li, D.-T. Do, K. M. Rabie, and R. Kharel,
Association for Computing Machinery, 2011, p. 301–312. [Online]. “Physical layer security of cooperative noma for iot networks under i/q
Available: https://doi.org/10.1145/2030613.2030647 imbalance,” IEEE Access, vol. 8, pp. 51 189–51 199, 2020.
[122] U. Tefek and T. J. Lim, “Full-duplex SIMO relaying for machine-type [143] B. Zheng, M. Wen, C.-X. Wang, X. Wang, F. Chen, J. Tang, and
communications in cellular networks,” in 2017 IEEE 86th Vehicular F. Ji, “Secure noma based two-way relay networks using artificial noise
Technology Conference (VTC-Fall), 2017, pp. 1–6. and full duplex,” IEEE Journal on Selected Areas in Communications,
[123] T. X. Vu, S. Chatzinotas, and B. Ottersten, “Edge-caching wireless vol. 36, no. 7, pp. 1426–1440, 2018.
networks: Performance analysis and optimization,” IEEE Transactions on [144] M. G. Kibria, K. Nguyen, G. P. Villardi, O. Zhao, K. Ishizu, and F. Ko-
Wireless Communications, vol. 17, no. 4, pp. 2827–2839, 2018. jima, “Big data analytics, machine learning, and artificial intelligence
[124] L. T. Tan, R. Q. Hu, and Y. Qian, “D2d communications in heterogeneous in next-generation wireless networks,” IEEE Access, vol. 6, pp. 32 328–
networks with full-duplex relays and edge caching,” IEEE Transactions 32 338, 2018.
on Industrial Informatics, vol. 14, no. 10, pp. 4557–4567, 2018. [145] R. Adeogun, “Calibration of stochastic radio propagation models using
[125] B. Lu, F. Zhu, J. Xia, X. Li, W. Zhou, and L. Fan, “Interference machine learning,” IEEE Antennas and Wireless Propagation Letters,
suppression by exploiting wireless cache in relaying networks for b5g vol. 18, no. 12, pp. 2538–2542, 2019.
communications,” Physical Communication, vol. 42, no. 2020, 2020. [146] T. Imai, K. Kitao, and M. Inomata, “Radio propagation prediction model
[126] X. Li, K. Chen, H. Hou, L. Deng, and Q. Zhou, “Cache-aided multi-hop using convolutional neural networks by deep learning,” in 2019 13th
uav-relaying networks,” Physical Communication, vol. 33, no. 2019, pp. European Conference on Antennas and Propagation (EuCAP), 2019, pp.
165–171, 2019. 1–5.
[127] B. Jiang, J. Yiang, H. Xu, H. Song, and G. Zheng, “Multimedia data [147] J. Zhang, L. Liu, Y. Fan, L. Zhuang, T. Zhou, and Z. Piao, “Wireless
throughput maximization in internet-of-things system based on optimiza- channel propagation scenarios identification: A perspective of machine
tion of cache-enabled uav,” IEEE Internet of Things Journal, vol. 6, no. 2, learning,” IEEE Access, vol. 8, pp. 47 797–47 806, 2020.
pp. 3525–3532, 2019. [148] A. Bharti, R. Adeogun, and T. Pedersen, “Learning parameters of
[128] K. Song, J. Zhang, Z. Ji, J. Jiang, and C. Li, “Energy-efficiency for stochastic radio channel models from summaries,” IEEE Open Journal
iot system with cache-enabled fixed-wing uav relay,” IEEE ACCESS: of Antennas and Propagation, vol. 1, pp. 175–188, 2020.
VOLUME 4, 2016 29
[149] M. Kulin, T. Kazaz, I. Moerman, and E. De Poorter, “End-to-end learning [168] Z. Zhang, Y. Lu, Y. Huang, and P. Zhang, “Neural network-based relay
from spectrum data: A deep learning approach for wireless signal iden- selection in two-way swipt-enabled cognitive radio networks,” IEEE
tification in spectrum monitoring applications,” IEEE Access, vol. 6, pp. Transactions on Vehicular Technology, vol. 69, no. 6, pp. 6264–6274,
18 484–18 501, 2018. 2020.
[150] F. Tang, B. Mao, Y. Kawamoto, and N. Kato, “Survey on machine [169] ——, “Neural network-based relay selection in two-way swipt-enabled
learning for intelligent end-to-end communication towards 6g: From cognitive radio networks,” IEEE Transactions on Vehicular Technology,
network access, routing to traffic control and streaming adaption,” IEEE vol. 69, no. 6, pp. 6264–6274, 2020.
Communications Surveys Tutorials, pp. 1–1, 2021. [170] T.-T. Nguyen, J.-H. Lee, M.-T. Nguyen, and Y.-H. Kim, “Machine
[151] F. Hussain, S. A. Hassan, R. Hussain, and E. Hossain, “Machine learning learning-based relay selection for secure transmission in multi-hop df
for resource management in cellular and iot networks: Potentials, current relay networks,” Electronics, vol. 8, no. 9, p. 949, 2019.
solutions, and open challenges,” IEEE Communications Surveys Tutori- [171] N. Nomikos, S. Talebi, R. Wichman, and T. Charalambous, “Bandit-
als, vol. 22, no. 2, pp. 1251–1275, 2020. based relay selection in cooperative networks over unknown stationary
[152] H. A. Shah and L. Zhao, “Multiagent deep-reinforcement-learning-based channels,” in 2020 IEEE 30th International Workshop on Machine Learn-
virtual resource allocation through network function virtualization in ing for Signal Processing (MLSP), 2020, pp. 1–6.
internet of things,” IEEE Internet of Things Journal, vol. 8, no. 5, pp. [172] X. Du, H. Van Nguyen, C. Jiang, Y. Li, F. R. Yu, and Z. Han, “Virtual
3410–3421, 2021. relay selection in lte-v: A deep reinforcement learning approach to
[153] T.-V. Nguyen, T.-N. Tran, K. Shim, T. Huynh-The, and B. An, “A heterogeneous data,” IEEE Access, vol. 8, pp. 102 477–102 492, 2020.
deep-neural-network-based relay selection scheme in wireless-powered [173] M. E. Morocho-Cayamcela, H. Lee, and W. Lim, “Machine learning to
cognitive iot networks,” IEEE Internet of Things Journal, vol. 8, no. 9, improve multi-hop searching and extended wireless reachability in v2x,”
pp. 7423–7436, 2021. IEEE Communications Letters, vol. 24, no. 7, pp. 1477–1481, 2020.
[154] M. A. Salem, A. Bin Abd Aziz, H. F. Al-Selwi, and M. Yusoff Bin Alias, [174] Y. Alghorani, A. S. Chekkouri, D. A. Chekired, and S. Pierre, “Improved
“Cooperative relay and jammer node selection strategies based on feed- s-af and s-df relaying schemes using machine learning based power
forward neural network,” in 2020 International Conference on Computing allocation over cascaded rayleigh fading channels,” IEEE Transactions
and Information Technology (ICCIT-1441), 2020, pp. 1–6. on Intelligent Transportation Systems, pp. 1–13, 2020.
[155] H. Zhang, D. Zhan, C. J. Zhang, K. Wu, Y. Liu, and S. Luo, “Deep [175] Z. Deng, Q. Sang, Y. Gao, and C. Cai, “Optimal relay selection for wire-
reinforcement learning-based access control for buffer-aided relaying less relay channel with external eavesdropper: a nn-based approach,” in
systems with energy harvesting,” IEEE Access, vol. 8, pp. 145 006– 2018 IEEE/CIC International Conference on Communications in China
145 017, 2020. (ICCC), 2018, pp. 515–519.
[156] H. Sakran, “Joint relay and jammer selection based on deep learning [176] R. Yao, Y. Zhang, S. Wang, N. Qi, N. I. Miridakis, and T. A. Tsiftsis,
for improving the physical layer secrecy in cooperative networks,” in “Deep neural network assisted approach for antenna selection in un-
2020 International Wireless Communications and Mobile Computing trusted relay networks,” IEEE Wireless Communications Letters, vol. 8,
(IWCMC), 2020, pp. 1124–1129. no. 6, pp. 1644–1647, 2019.
[177] X. Wang, “Decision-tree-based relay selection in dualhop wireless com-
[157] P. Fabian and A. Rachedi, “Dynamic selection of relays based on clas-
munications,” IEEE Transactions on Vehicular Technology, vol. 68, no. 6,
sification of mobility profile in a highly mobile context,” in ICC 2020 -
pp. 6212–6216, 2019.
2020 IEEE International Conference on Communications (ICC), 2020,
[178] C. Koushik and P. Vetrivelan, “Heuristic relay-node selection in op-
pp. 1–6.
portunistic network using rnn-lstm based mobility prediction,” Wireless
[158] J. Zhang, J. Tang, and F. Wang, “Cooperative relay selection for load
Personal Communications, vol. 114, pp. 2363–2388, 2020.
balancing with mobility in hierarchical wsns: A multi-armed bandit
[179] V. M. Rohokale, N. R. Prasad, and R. Prasad, “A cooperative Internet
approach,” IEEE Access, vol. 8, pp. 18 110–18 122, 2020.
of Things (IoT) for rural healthcare monitoring and control,” in 2011
[159] F.-L. Luo, Neural Network–Based Wireless Channel Prediction, 2020,
2nd International Conference on Wireless Communication, Vehicular
pp. 303–325.
Technology, Information Theory and Aerospace Electronic Systems
[160] X. Wang and F. Liu, “Data-driven relay selection for physical-layer Technology (Wireless VITAE), Feb 2011, pp. 1–6.
security: A decision tree approach,” IEEE Access, vol. 8, pp. 12 105– [180] G. Cai, Y. Fang, J. Wen, G. Han, and X. Yang, “QoS-aware buffer-aided
12 116, 2020. relaying implant wban for healthcare IoT: Opportunities and challenges,”
[161] A. Sturaro, S. Silvestri, M. Conti, and S. K. Das, “A realistic model IEEE Network, vol. 33, no. 4, pp. 96–103, July 2019.
for failure propagation in interdependent cyber-physical systems,” IEEE [181] V. Petrov, A. Samuylov, V. Begishev, D. Moltchanov, S. Andreev,
Transactions on Network Science and Engineering, vol. 7, no. 2, pp. 817– K. Samouylov, and Y. Koucheryavy, “Vehicle-based relay assistance
831, 2020. for opportunistic crowdsensing over narrowband IoT (NB-IoT),” IEEE
[162] F. Liu, Y. Wang, and H. Li, “Physical layer security of cooperative Internet of Things Journal, vol. 5, no. 5, pp. 3710–3723, Oct 2018.
network with machine learning-based relay selection,” in 2019 IEEE 6th [182] M. Zhu, X.-Y. Liu, and X. Wang, “Deep reinforcement learning for
International Symposium on Electromagnetic Compatibility (ISEMC), unmanned aerial vehicle-assisted vehicular networks,” 2020.
2019, pp. 1–5. [183] Y. Qian, J. Yan, H. Guan, J. Li, X. Zhou, S. Guo, and D. N. K. Jayakody,
[163] Y. Yengi, A. Kavak, H. Arslan, K. Küçük, and H. Yiğit, “Malicious “Design of hybrid wireless and power line sensor networks with dual-
relay node detection with unsupervised learning in amplify-forward interface relay in IoT,” IEEE Internet of Things Journal, vol. 6, no. 1, pp.
cooperative networks,” in 2019 International Conference on Innovation 239–249, Feb 2019.
and Intelligence for Informatics, Computing, and Technologies (3ICT), [184] Z. Chen, L. Wang, D. Han, and H. Zeng, “A unified performance analysis
2019, pp. 1–5. of relaying communication system for IoT application with hybrid fad-
[164] A. Gouissem, L. Samara, R. Hamila, N. Al-Dhahir, L. Ben-Brahim, ing,” IEEE Internet of Things Journal, vol. 7, no. 1, pp. 570–583, 2020.
and A. Gastli, “Machine-learning based relay selection in af cooperative [185] Z. Chen, C. Ye, J. Yuan, and D. Han, “MGF-based mutual approximation
networks,” in 2019 IEEE Wireless Communications and Networking of hybrid fading: Performance of wireless/power line relaying communi-
Conference (WCNC), 2019, pp. 1–7. cation for iot,” Sensors, vol. 19, p. 2460, 05 2019.
[165] D. Patil, O. Naidu, P. Yalla, and S. Hida, “An ensemble machine learning [186] P. Nouri, H. Alves, and M. Latva-aho, “On the performance of ultra-
based fault classification method for faults during power swing,” in 2019 reliable decode and forward relaying under the finite blocklength,” 2017,
IEEE Innovative Smart Grid Technologies - Asia (ISGT Asia), 2019, pp. pp. 1–5.
4225–4230. [187] B. Singh, O. Tirkkonen, Z. Li, M. Uusitalo, and R. Wichman, “Selective
[166] A. A. A. El-Banna, B. M. ElHalawany, A. B. Zaky, J. Z. Huang, and multi-hop relaying for ultra-reliable communication in a factory environ-
K. Wu, “Machine learning-based multi-layer multi-hop transmission ment,” 2016, pp. 1–6.
scheme for dense networks,” IEEE Communications Letters, vol. 23, [188] P. Nouri, H. Alves, and L. Matti, “Performance analysis of ultra-reliable
no. 12, pp. 2238–2242, 2019. short message decode and forward relaying protocols,” EURASIP Jour-
[167] R. Yao, Y. Zhang, N. Qi, T. A. Tsiftsis, and Y. Liu, “Machine learning- nal on Wireless Communications and Networking, vol. 202, no. 2018, pp.
based antenna selection in untrusted relay networks,” in 2019 2nd Inter- 498–501, 2018.
national Conference on Artificial Intelligence and Big Data (ICAIBD), [189] T. Nawaz, M. Seminara, S. Caputo, L. Mucchi, F. S. Cataliotti, and
2019, pp. 323–328. J. Catani, “Ieee 802.15.7-compliant ultra-low latency relaying vlc system
30 VOLUME 4, 2016
for safety-critical its,” IEEE Transactions on Vehicular Technology,, RAMONI ADEOGUN received a B. Eng. in
vol. 68, no. 2, pp. 12 040–12 051, 2019. electrical and computer engineering from Federal
[190] C. Pan, H. Ren, Y. Deng, M. Elkashian, and A. Nallanathan, “Joint block- University of Technology, Minna, Nigeria in 2007
length and location optimization for urllc-enabled uav relay systems,” and a Ph.D. in electronic and computer systems
IEEE Communication Letters, vol. 23, no. 3, pp. 498–501, 2019. engineering from Victoria University of Welling-
[191] P. Nouri, H. Alves, R. D. Souza, and M. Latva-aho, “In-band pilot ton, New Zealand in 2015. He is currently an as-
overhead in ultra-reliable low latency decode and forward relaying,” sistant professor at Aalborg University, Denmark
2019, pp. 1–5.
and an external research engineer with Nokia Bell
[192] Y. Hu, M. C. Gursoy, and A. Schmeink, “Relaying-enabled ultra-reliable
Labs, Aalborg, Denmark. Prior to joining Aalborg
low-latency communications in 5g,” IEEE Network, vol. 32, no. 2, pp.
62–68, 2018. University, he worked in various positions at the
[193] V. N. Swamy, P. Rigge, G. Ranade, B. Nikoli, and A. Sahai, “Predicting University of Cape Town, South Africa, Izon Science Limited, New Zealand,
wireless channels for ultra-reliable low-latency communications,” 2018, Odua Telecoms Ltd. and National Space Research and Development
pp. 1–5. Agency, Nigeria. His research interests include channel characterization,
[194] Y. Gu, Y. Li, and B. Vucetic, “Ultra-reliable short-packet communica- statistical signal processing, machine learning and AI for communications,
tions: Half-duplex or full-duplex relaying?” IEEE Wireless Communica- IoT, intelligent spectrum access and interference management. He is a senior
tion Letters, vol. 7, no. 3, pp. 348–351, 2018. member of the IEEE.
[195] S. Feng, D. Niyato, P. Wang, D. Kim, and Y.-C. Liang, “Joint service pric-
ing and cooperative relay communication for federated learning,” 2019
International Conference on Internet of Things (iThings) and IEEE Green
Computing and Communications (GreenCom) and IEEE Cyber, Physical
and Social Computing (CPSCom) and IEEE Smart Data (SmartData), pp.
815–820, 2019.
UYOATA UYOATA received his B.Eng. degree in

electrical and electronics engineering from Fed-
eral University of Technology, Yola in 2009, and
his M.Sc. degree in personal, mobile and satellite
communication from the University of Bradford
in 2014. He completed his Ph. D. at the Centre
of Excellence in Broadband Networks, University
of Cape Town in 2019. His research interests in-
clude device-to-device communication, relay net-
works, Internet of Things, network optimization
and emerging wireless technologies.
JOYCE MWANGAMA received her B.Sc. degree

in electrical and computer engineering and her
M.Sc. degree in electrical engineering from the
University of Cape Town (South Africa) in 2008
and 2011, respectively. In 2012 she began her
study toward a Ph.D. in the Centre for Broadband
Networks and joined the Department of Electrical
Engineering at the University of Cape Town as a
lecturer in 2015. Joyce has published her research
work in a number of peer-reviewed publications.
Her research work has also contributed to the Universities for Future Internet
and the Testbeds for Reliable Smart City Machine-to-Machine Commu-
nications international research collaboration projects. Joyce has received
multiple research awards, including the Google Anita Borg Scholarship and
the LOréal-UNESCO Sub-Saharan Women in Science Fellowship.
VOLUME 4, 2016 31

Relaying in The Internet of Things IoT A

Uploaded by

Copyright:

Available Formats

Relaying in The Internet of Things IoT A

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Relaying in The Internet of Things IoT A

Uploaded by

Copyright:

Available Formats

This article has been accepted for publication in a future issue of this journal, but has not been

Relaying in the Internet of Things (IoT):

I. INTRODUCTION that by 2023, there will be 14.7 billion M2M connections,

U. Uyoata et al.: Relaying in the Internet of Things: A Survey

U. Uyoata et al.: Relaying in the Internet of Things: A Survey

TABLE 1: Selected Surveys and Their Focus Area.

U. Uyoata et al.: Relaying in the Internet of Things: A Survey

U. Uyoata et al.: Relaying in the Internet of Things: A Survey

TABLE 2: Table of Abbreviations

U. Uyoata et al.: Relaying in the Internet of Things: A Survey

have a limited duty cycle of approximately 1%. Deploying

U. Uyoata et al.: Relaying in the Internet of Things: A Survey

case of vehicle relaying for smart transport), there then

U. Uyoata et al.: Relaying in the Internet of Things: A Survey

U. Uyoata et al.: Relaying in the Internet of Things: A Survey

U. Uyoata et al.: Relaying in the Internet of Things: A Survey

U. Uyoata et al.: Relaying in the Internet of Things: A Survey

to attaining target throughput in SWIPT relay IoT networks. Key Insights

U. Uyoata et al.: Relaying in the Internet of Things: A Survey

U. Uyoata et al.: Relaying in the Internet of Things: A Survey

U. Uyoata et al.: Relaying in the Internet of Things: A Survey

U. Uyoata et al.: Relaying in the Internet of Things: A Survey

TABLE 6: Relay Selection Techniques in IoT Networks

U. Uyoata et al.: Relaying in the Internet of Things: A Survey

U. Uyoata et al.: Relaying in the Internet of Things: A Survey

FIGURE 7: IoT Relaying in the Presence of an Eavesdropper

Key Insights E. IOT RELAYING BASED ON MOBILE NODES: UAV

U. Uyoata et al.: Relaying in the Internet of Things: A Survey

U. Uyoata et al.: Relaying in the Internet of Things: A Survey

U. Uyoata et al.: Relaying in the Internet of Things: A Survey

TABLE 9: Summary of Key Contributions in UAV-based IoT Relaying

TABLE 10: Practical Relay Hardware Design

Key Insights creasing attention in relation to relay-assisted IoT networks.

U. Uyoata et al.: Relaying in the Internet of Things: A Survey

U. Uyoata et al.: Relaying in the Internet of Things: A Survey

U. Uyoata et al.: Relaying in the Internet of Things: A Survey

a neural network. In many applications, it may be nearly

U. Uyoata et al.: Relaying in the Internet of Things: A Survey

Machine Learning Tools Main Applications

U. Uyoata et al.: Relaying in the Internet of Things: A Survey

U. Uyoata et al.: Relaying in the Internet of Things: A Survey

U. Uyoata et al.: Relaying in the Internet of Things: A Survey

U. Uyoata et al.: Relaying in the Internet of Things: A Survey

U. Uyoata et al.: Relaying in the Internet of Things: A Survey

U. Uyoata et al.: Relaying in the Internet of Things: A Survey

U. Uyoata et al.: Relaying in the Internet of Things: A Survey

UYOATA UYOATA received his B.Eng. degree in

JOYCE MWANGAMA received her B.Sc. degree

You might also like