On the impact of network slicing on

5G radio access networks

Icaro Leonardo Da Silva, Gunnar Mildh Alexandros Kaloxylos, Panagiotis Spapis
Wireless Access Networks European Research Center
Ericsson Research Huawei Technologies, Duesseldorf GmbH,
Stockholm, Sweden Munich, Germany

Enrico Buracchini, Alessandro Trogolo Gerd Zimmermann, Nico Bayer

Wireless Access Telekom Innovation Laboratories
Telecom Italia Lab Deutsche Telekom
Turin, Italy Darmstadt, Germany

Abstract— This paper presents an analysis of the impact that and should also address the needs of different 5G use cases
the network slicing concept may have on the 5G radio access with highly diverging requirements. A network slice is
networks (RAN) in order to fulfill NGMN’s vision and support envisioned to support the communication service of a
the wide diversity of services from different industries. Network particular connection type possibly with a specific way of
slicing addresses the deployment of multiple logical networks as handling control plane and user plane for this service. To this
independent business operations on a common physical end, a “5G slice” could be composed of a collection of 5G
infrastructure. The concept has initially been proposed for the network functions (NF) and specific air interface / radio access
5G core network (CN) and has evolved to end-to-end (E2E) technology (RAT) settings that are combined together for a
network slicing. However, it has not been investigated yet what
specific use case and/or business model (see Fig. 1).
network slicing would represent to the design of the RAN. The
paper explains how network slicing may impact several aspects of
the 5G RAN design such as the protocol architecture, the design
of network functions (NF) and the management framework that
needs to support both the management of the infrastructure to be
shared among the slices and the slice operation.

Keywords—network slicing; radio access network; RAN design.

I. INTRODUCTION
Worldwide initiatives in the mobile industry have started
in the past few years in order to define the requirements and
principles of what is to be standardized in the 5th generation of
mobile networks (5G). These initiatives (research projects,
industry fora, etc.) have acknowledged the demand of mobile
networks that are capable to cope with a much higher diversity Fig. 1. Network slicing concept as envisioned by NGMN [2].
of use cases in order to unleash new business opportunities to
the telecom industry. Among these initiatives are the Mobile According to [2] [3], a slice is seen from an end customer
and wireless communications Enablers for the Twenty-twenty perspective or slice customer as an independent network.
(2020) Information Society II (METIS-II) [1], a research However, in contrast to deploying an independent network
project focused on standard related problems associated to the infrastructure, each slice will be realized together with other
5G RAN design, and the Next Generation Mobile Networks slices on a common infrastructure (also referred to as “virtual
(NGMN) Alliance [2]. network”), also including assets such as licensed spectrum. In
this way, the infrastructure and assets utilization will be much
Network slicing is an important part of the NGMN’s vision more cost and energy efficient compared to present
for the overall 5G architecture [2] that addresses the realizations.
deployment of multiple logical networks as independent
business operations on a common physical infrastructure. The The concept of network slicing has initially been proposed
ultimate goal would be to provide “network slices on an as-a- for the 5G core network (CN) [4]. By exploring software-
service basis” and meet the wide range of use cases that the defined networking (SDN) and NF virtualization (NFV)
2020 timeframe will demand e.g., for the different industries principles a fully virtualized CN instance optimized per
[3]. The creation of network slices is mainly business driven business purpose could be defined. The concept has been
evolved by NGMN to end-to-end (E2E) network slicing. This  RAN should support protection mechanisms for slice
is somewhat underlined through statements in 3rd Generation isolation so that events (such as congestion) within
Partnership Project (3GPP) study items such as “the network one slice do not have a negative impact on another
slicing primarily targets a partition of the CN, but it is not slice;
excluded that RAN may need specific functionality to support
multiple slices or even partitioning of resources for different  RAN should support efficient management
network slices” [5]. mechanisms e.g. to efficiently setup new slices and to
efficiently operate new business/services.
The work in 3GPP about the overall 5G system has
recently been kicked-off with an approved study item in the
3GPP Working Group SA2 [6]. The support for E2E network A. Utilization of RAN resources should be maximized
slicing appears as one of the key requirements. Even though Network slicing should make possible to support several
that the first 5G Radio Access Network (RAN) study items in different virtual networks on the same physical network
3GPP start in 2016 and despite the introduction of the notion infrastructure in order to reduce costs and energy consumption
of E2E slicing, it has not been investigated yet what network compared to deploying separate physical networks for the
slicing would represent e.g., to the RAN design, which different use case or business scenarios. In order to fully
comprise both network side and User Equipment (UE). exploit this benefit, it is required that the slicing concept
allows for efficient usage of common resources such as radio
This paper aims to provide a relevant contribution to the resources and infrastructure, and transport links between the
5G standardization process by a detailed analysis of a set of slices such as fronthaul and backhaul. In any case, this
requirements to the 5G RAN that have been identified in order requirement does not preclude the possibility to assign
to support network slicing. These requirements, discussed in dedicated (static) resources.
Section II, are intended to guide the design of a future proof
RAN enabling the smooth introduction of new features and
services. In this perspective the paper identifies the aspects of
the 5G RAN that are envisioned to be affected by the RAN
requirements related to network slicing. The first is the
“protocol architecture”, i.e., how is the split between functions
among the layers of the protocol stack and how different
functions may be relevant or not to different slices and/or
services. The second aspect is the “design of NFs”, i.e.,
defining which NFs should be common to multiple slices (e.g.
in the case of some kind of resource sharing) and which NFs Fig. 2. Radio resources shared in a dedicated and dynamic manner.
could be independent (possibly with some level of
coordination). The third aspect relates to the “management of For example, static splitting of the available resources and
the infrastructure to be shared among multiple slices” and “the users into two equal slices, as illustrated in Fig. 2, could,
management of individual slices to enable a proper business depending on the service bit-rate requirements, easily lower
operation”. These three aspects are discussed in Section III. the overall (busy hour) capacity by 20%, while splitting into 5
Finally, the findings of our analysis are summarized in Section equal slices lead to a loss of more than 50%. In the case
IV. multiple slices are sharing the physical resources it would be
preferred that the sharing is dynamic, making it possible to
II. RAN REQUIREMENTS RELATED TO NETWORK SLICING adapt to rapid changing traffic patterns as typical for many
Internet services. It is expected that sharing mechanisms being
A first step to understand the impact of network slicing to able to operate on Time to Transmit Interval (TTI) level will
the 5G RAN design has been given by identifying RAN- be better than mechanism that operate on 100 ms level.
specific requirements needed to fulfill the network slicing Considering the above example with dynamic sharing of
vision. The derived set of requirements is the following: resources between different slices it is assumed that the slices
 Utilization of RAN resources, such as radio resources could either use common or coordinated radio and transport
schedulers. In this way one slice can utilize the full bandwidth
(e.g. time, frequency, power) and hardware (HW) /
if there is no traffic in the other slice achieving maximum
software (SW) platforms, should be maximized
efficiency.
among multiple slices;
In summary, slicing should be supported by static (i.e.,
 RAN should be slice-aware via some explicit or dedicated) and dynamic resource sharing in the RAN and
implicit identification (e.g., based on an abstraction transport network. Dynamically shared resources raise for
model); example the requirements on common or coordinated
scheduling mechanisms between slices, access channels and/or
 RAN should support mechanisms for traffic
other common channels in general. On the other hand,
differentiation in order to be able to treat different
dedicated resources per slices should be carefully considered
slices differently and/or different services within the since it could lead to an inefficient usage of radio resources.
multi-service slices;
B. RAN should be slice-aware E. RAN should support the management of the infrastructure
Current 3GPP networks have support for extensive quality The slicing concept should efficiently enable fast set-up of
of service (QoS) mechanisms such as bearers, QoS class new services and applications or modification of existing ones.
indicators, allocation and resolution priority values) making it In order to achieve it, it should be needed to efficiently
possible to prioritize different services and signalling in a manage assets (such as licensed spectrum) and efficiently use
different way. The question arise if these mechanisms can also network topology including the HW and SW resources
be used to support E2E network slicing as is and thus avoiding available at the different sites (processing, networking, and
the need to make the RAN slice-aware. storage capabilities). It shall be also possible to properly and
efficiently take into consideration the corresponding
Most likely, existing mechanisms are not enough to performance and business requirements for the different slices
support all possible slicing scenarios. For instance let us such as the number of the devices, their mobility and traffic
assume a slice A with both high and low priority traffic but patterns, the delay, and the jitter requirements, the device
where the traffic at congestion is not allowed to consume say capabilities, etc.
more than X% of the total system resources (the rest is given
to other slices). In this case there might be a will to treat the F. RAN should support the management of a slice
high priority traffic of slice A with absolute high priority up to It is also envisioned that the vertical industries, mobile
X%, but after that the traffic from other slices should have virtual network operators (MVNOs), over-the-top (OTT)
higher priority. These types of mechanisms are currently not service providers or mobile network operators (MNOs)
directly supported using existing QoS concept since there is no themselves should be allowed to manage aspects of a network
mechanism to apply specific policies in the RAN to a group of slice as if it was a separate dedicated network. The main
users. For this reason it is required that the RAN should be purpose should be to efficiently operate their business which
made slice-aware. This can be done e.g. by providing could be realized via available application programming
information such as some form of Slice-ID, another interfaces (API). The management framework for network
abstraction framework or by extended individual data packet slicing therefore should provide slice specific performance
tagging on the interface between RAN and CN (see e.g. [7]). management (PM) incl. counters, key performance indicators
(KPI), as well as configuration management (CM) and fault
management (FM). For the slice specific CM and FM
C. RAN should support mechanisms for traffic differentiation
functions it is expected that there will be close interactions
One of the drivers for network slicing is the ability to with the CM and FM of the whole network which governs
support 5G use cases with diverging requirements in different (based on policies) what the slice specific CM and FM
slices. Therefore, in addition to make the flows and the notion functions are allowed to see and to do.
of slicing visible to the 5G RAN, it should support
mechanisms to prioritize different traffic over the radio and III. IMPACT OF NETWORK SLICING TO THE 5G RAN DESIGN
transport interfaces. Current 3GPP systems already support
such mechanisms using radio schedulers and Diffserv etc. It is The first 5G RAN study items in 3GPP start in 2016,
where the definition of the RAN architecture (or design) is one
assumed that similar mechanisms are also required for the 5G
of the main topics to be addressed. Three of the aspects that
RAN. Further investigations are ongoing in order to verify
will be impacted by the requirements discussed in Section II
whether the slicing in the 5G RAN requires any additional are the following:
functions that are not currently in place.
 Protocol architecture, which defines the NFs per
D. RAN should support protection mechanisms protocol necessary to fulfill the 5G requirements. A
special attention is given to the air interface
In addition to prioritization of traffic over the radio and
protocols. However, network slicing might also
transport interface, mechanisms are required to protect impact the CN / RAN functional split, i.e., which NFs
common channels or resources used for UEs accessing system are logically defined as CN or as RAN NFs.
so that congestion in one slice does not have a negative impact
on another slice. Currently in 3GPP system there is some  Design of NFs, which defines which of the NFs
support for protecting common control channels for extensive should be common to multiple slices and which NFs
load from different services. These mechanisms include could be independent (possibly with some level of
Access Class Barring, Enhanced Access Barring, Service coordination between them).
Specific Access Barring, as well as implementation specific  Network management, which defines how the
admission control etc. current RAN management frameworks (e.g., PM,
It needs to be investigated whether existing barring and CM, FM) could potentially be extended to enable
admission control mechanisms used in LTE are enough to both an efficient management of the infrastructure
support E2E network slicing, or if additional slice specific and the network slice for proper business operation.
barring and admission control mechanisms are required to
provide enough isolation and protection between slices.
A. Protocol Architecture integrity of the data to be transmitted under specific delay
The requirement that multiple slices should be served by constraints is essential. In such slice the NFs related to
the same network infrastructure leads to a solution based on a handover, location updates, fragmentation and reassembly
common protocol architecture addressing multiple slices. The may be possibly tailored (e.g., by introducing grouping
protocol architecture defines the set of NFs per protocol schemes to assist their operation) or even proven unnecessary
necessary to fulfil the end-user and system requirements. since the service requires the transmission of very small
Network slicing will likely affect the design of this common packets from static nodes.
protocol architecture since there will be slices associated to
different 5G use cases or services, e.g., massive Machine- Fig. 3 provides an example where some of the LTE NFs
Type Communication (mMTC), Ultra-Reliable MTC (uMTC) are omitted from the smart grid slice and the benefit that could
or Extreme Mobile Broadband (xMBB). arise from such omission using as reference the values
provided by 3GPP [9]. Obviously this example should be
This has been concluded by the fact that not all the 5G considered valid under specific assumptions (e.g., message
services may benefit in the same way by the same set of NFs size, ciphering taking place in the application layer,
defined for a given protocol. It is likely that slices associated retransmissions being controlled and executed on the transport
to mMTC services (e.g., supporting very low cost sensors) layer etc.).
might require different set of NFs, protocols and/or protocol
configuration(s) compared to slices associated to xMMB 2) Slice-tailored NFs optimizations
services (e.g., requiring higher spectral efficiency) or a slice In addition to the selection and/or omission of NFs per
associated to uMTC services (e.g., requiring very high slice, different slices and/or services should possibly benefit
reliability with bounded latency). from slice-tailored NFs optimizations. Regarding the random
1) Selection / omission of NFs per slice access of a massive number of devices, the collision rate could
The fact that the set of required NFs could differ from one be drastically reduced by the introduction of grouping
slice to another is mainly driven by the fact that different schemes with various prioritizations in the devices trying to
services may have quite different sets of performance access the wireless medium, especially in the case of static
requirements and overall characteristics (e.g., traffic and groups where these is no major cost in setting up these groups.
mobility patterns). That could also be driven by different In [10] the authors propose a group based access of the
Service Level Agreements (SLA). wireless medium where the cluster heads’ communication is
coordinated (Fig. 4), whereas in [11] focus is on the omitting
LTE Optimization for Smart grids of redundant message transmissions.
Required time Required time
Step Description Explanation
for the operation for the operation

0 UE wakeup time Implementation dependent

1 Average delay due to RACH scheduling period 5ms 5ms

2 RACH Preamble 1ms 1ms
Preamble detection and transmission of RA
response (Time between the end RACH
3 5ms 5ms
transmission and UE’s reception of scheduling
grant and timing adjustment)
UE Processing Delay (decoding of scheduling
4 grant, timing alignment and C-RNTI assignment + 2.5ms 2.5ms
L1 encoding of RRC Connection Request)

5 TTI for transmission of RRC Connection Request 1ms 1ms

6 HARQ Retransmission (@ 30%) 0.3 *5ms 0ms No HARQ
7 Processing delay in eNB (Uu –> S1-C) 4ms 1ms No ciphering, RohC
Ts1c (2ms –
8 S1-C Transfer delay Ts1c (2ms – 15ms)
15ms)
MME Processing Delay (including UE context
9 15ms 15ms
retrieval of 10ms)
Ts1c (2ms – Ts1c (2ms –
10 S1-C Transfer delay
15ms) 15ms) Fig. 4. Example of how service-tailored random access optimization can be
11 Processing delay in eNB (S1-C –> Uu)
TTI for transmission of RRC Connection Setup
4ms 1ms No ciphering, RoHC
beneficial in mMTC using cluster based access [10].
12 1.5ms 1.5ms
(+Average alignment)
13 HARQ Retransmission (@ 30%) 0.3 *5ms 0ms No HARQ
14 Processing delay in UE
TTI for transmission of L3 RRC Connection
3ms 0.6ms No ciphering, RoHC 3) QoS framework for slice awareness
15
16
Complete
HARQ Retransmission (@ 30%)
1ms
0.3 *5ms
1ms
0ms No HARQ
Another aspect of the protocol architecture that may need
Total LTE IDLE  ACTIVE delay (C-plane
establishment)
47.5ms + 2 *
Ts1c
34.6ms + 2 * Ts1c enhancements is the QoS framework. The envisioned
scenarios for network slicing may lead to cases where a slice
Fig. 3. Example of performance benefits by the omission of NFs.
may be composed by multiple services which may have
different treatment in the RAN. Therefore, in that case, the
For instance, let us assume a “slice provided for uMTC QoS framework needs to take into account some slicing level
smart grid network” compared to a “slice provided for information in addition to service level information. That
xMBB”. Smart grids are static nodes that have to transmit only could be explicit information or some abstraction (such as an
small messages with E2E delay of less than 8 ms, whereas for enhanced identity, like the QCI). In the example of Fig. 5 a
the xMBB use cases such as the Dense Urban Information Slice-ID is reported from the CN to the RAN.
Society the acceptable E2E delay is of up to 50 ms for the
human oriented services [11]. In smart grid networks the
 In addition to that, the omission or selection of NFs therefore the task becomes more challenging. For the 5G RAN
described previously could lead to enhancements to the QoS design it could be envisioned to further reduce the number of
framework where different flows per slice / bearer types could common channels allowing for different NFs, protocol
be mapped to subset of NFs. configurations, or even protocols to be used in parallel. One
could also try to harmonize as much as possible the protocol
stack to maximize the reuse of different NFs in order to speed
up standardization and product (UE and RAN) development
and testing.

NF3 NF4 NF5

Fig. 5. One way to make RAN slice-aware is via a Slice-ID.
NF 1 NF2

4) CN / RAN functional split NF3’ NF5’

One of the assumptions in 3GPP concerning the overall 5G
The level of the separation of two slices could vary depending on the considered case
system is the CN /RAN functional split [6]. This would allow
an independent evolution of the RAN and CN functions in
Fig. 6. Common NFs providing services to independent/dedicated NFs.
order to speed up introduction of new technology and services
and facilitate multi-vendor CN/RAN interoperability. One of
the options in [6] is to keep the same functional split in the C. Impact to Network Management
Evolved Packet System (EPS). This seems the a reasonable 1) Performance monitoring and optimization per slice
choice, at least for xMBB slices, however, there could be new As described earlier, the PM solution at the 5G RAN may
5G services and use cases that could benefit from a different consist of counters, traces and/or KPIs possibly aggregated at
CN/RAN split, e.g., static devices benefiting from a RAN- least in a per slice granularity (in addition to existing
based paging, which is currently a core NF. granularity e.g. per service, flow, etc.). Such monitoring could
be essential to verify the fulfillment of SLAs and/or to
In order to support the tight integration of LTE to the 5G
properly operate the different business associated to different
RAN, a common CN/RAN interface denoted S1* that evolves slices. In the case of CM, some features could be tuned, turned
from the existing S1 is considered as one alternative [12]. on/off and/or possibly configured differently from different
Network slicing may impact the design of protocols for that slices. That may also affect the way Self-Organizing Networks
interface that may need to support potentially different (SON) functions are designed.
functional splits per slice.
2) Sharing of HW/SW platforms for multiple slices
B. Impact to the Radio NFs Novel management schemes for radio resources
reconfiguration may be adopted for their more efficient use
In order to fulfill the requirement regarding the sharing of
RAN resources among different slices, it is required to e.g. in order to adapt the network to the dynamic behavior of
efficiently use radio and transport network resources between the traffic and to globally maximize the capacity. These
different slices (including both common channels and mechanisms are further enabled and facilitated if the resources
dedicated channels). From that perspective some radio NFs are shared among different slices, possibly distributed
(such as schedulers) are either defined as “common NFs” or geographically and having distinct traffic patterns over time.
“independent/dedicated NFs” with (at least some) In fact, given a cell set in a certain area, the traffic of
coordination mechanisms between NFs from different slices. different slices / services e.g. on a specified air interface may
This means that the radio NFs of one slice do not operate in
change from one sub-area to the other according to the day
complete isolation from another slice. Possible
implementation of such scenario could include a set of period (e.g., in a stadium before, during and after an event
common radio NFs which provides slicing as a service to such as football match or a concert). It often happens that
higher layers (see Fig. 6). It is also possible to have more some areas may be congested (with high blocking
dedicated radio NFs running that communicate with some percentages) in some particular area (the so called hot-spots as
common resource management framework. One possible the stadium) in which the traffic is more consistent, while
example that could be a set of common radio NFs related to surrounding cells are less loaded or characterized by low
radio resource management (e.g., scheduling, random access, blocking percentages. Moreover, in case of deployment of two
etc.) and a set of dedicated NFs e.g. core NFs. or more air interfaces in the same area, the offered traffic of
different services on each air interface may also be differently
The choice among common or dedicated NFs may be distributed in time and space.
affected by the way radio resources are used. In LTE it is
theoretically possible to introduce different protocols and/or A management framework, enabling the use cases just
NFs on dedicated channels as soon as contention resolutions described, includes both radio reconfiguration management
have been performed. On the other hand, when it comes to and reconfigurable HW. The radio reconfiguration
protocols related to the access of common channels such as management functionality spans different air interfaces
the random access channel (RACH) the same or compatible managing and controlling the nodes inside the network, with
protocols need to be used to allow efficient multiplexing and the goal of self-adapting towards an optimal mix of supported
air interfaces and frequency bands. This function could act on IV. CONCLUSION
the basis of some input parameters, such as the available The paper presented a set of requirements that should be
resources (spectrum and HW), the traffic demand, the
followed during the 5G RAN design in order to support the
capabilities of the UEs within the network (supported air
interfaces, frequency bands, etc.), the requested services concept of E2E network slicing, envisioned by NGMN. These
(bandwidth, QoS, etc.), etc. In addition, this functionality requirements should drive the design of a common protocol
could exploit a collaborative radio reconfiguration architecture supporting multiple slices and services but
management scheme, where the decision making functions are allowing slice-tailored optimizations and NFs usage (where
shared among different network nodes. The main features can different subsets of NFs can be possibly used per slice). The
be summarized as follows (see Fig 7): most efficient way to realize and/or standardize this common
architecture is still under investigation. It has also been
 The dynamic self-adaptation of the network concluded that the identified requirements may also impact the
configuration towards an optimal mix of supported CN / RAN interface (e.g., to possibly support different
air interfaces (AI) and frequency bands can be functional splits) and the QoS framework (e.g., to possibly
achieved by the exploitation of the reconfigurable
support slice awareness and traffic differentiation within a
hardware and the application of radio reconfiguration
slice). Finally, the paper also provided the envisioned impact
management functionalities;
to the management framework of the 5G RAN. Two aspects
 The dynamic self-adaptation (e.g. network have been investigated, the need to support the management of
configuration) can be based on the traffic patterns the infrastructure to be shared and the management of the
variations in time and space for the different different slices that may be associated to different business
deployed AIs or according to specific needs (i.e. IoT purposes.
service provision, integration of high frequency air
interfaces, etc.); ACKNOWLEDGMENT

 Ability to provide sufficient information to the This work has been performed in the framework of the
terminals for initiating a communication session H2020 project METIS-II co-funded by the European Union
appropriately in a dynamic context (e.g. wireless (EU) [1]. The views expressed are those of the authors and do
control channels). not necessarily represent the project. The consortium is not
liable for any use that may be made of any of the information
contained therein.
REFERENCES
[1] METIS-II project (“Mobile and wireless communications Enablers for
Twenty-twenty (2020) Information Society II”), see https://5g-
ppp.eu/metis-ii/.
[2] NGMN Alliance, “5G White Paper”, February 2015.
[3] Ericsson, “5G Systems White Paper”, January 2015, see
http://www.ericsson.com/res/docs/whitepapers/what-is-a-5g-system.pdf
[4] Ericsson, “The Real-Time Cloud White Paper”, February 2014, see
http://www.ericsson.com/res/docs/whitepapers/wp-sdn-and-cloud.pdf
[5] 3GPP TR 22.891, “Study on New Services and Markets Technology
Enablers”, September 2015.
Fig. 7. Resource re-configuration for shared infrastructure. [6] 3GPP S2-153651, “Study on Architecture for Next Generation System”,
October 2015.
[7] Deutsche Telekom, T-Mobile USA, “Deutsche Telekom’s view on 5G“,
According to the evolution of the technologies and the 3GPP RAN 5G Workshop, Sep 17th-18th, 2015, Doc. no. RWS-150033.
deployment strategies, from a concrete implementation [8] ICT-671680 METIS-II, Deliverable 1.1 Version 1 “Refined scenarios
perspective, baseband pooling based on centralized-RAN and requirements, consolidated test cases, and qualitative techno-
solutions (which could be based on cloud platforms) can be economic feasibility assessment”, February 2016 (to be published).
exploited to accomplish the proposed objectives, allowing [9] 3GPP TS.25.912, “Feasibility study for evolved Universal Terrestrial
high degree of flexibility in managing and reconfiguring radio Radio Access (UTRA) and Universal Terrestrial Radio Access Network
(UTRAN)”, chapter 13.2. “C-plane latency”, 2014.
resources and related parameters, jointly with open evolved [10] Konstantinos Chatzikokolakis, Alexandros Kaloxylos, Panagiotis
SON features & open API/interfaces availability in order to Spapis, et al., “On the Way to Massive Access in 5G: Challenges and
increase operational efficiency and effectiveness of RAN Solutions for Massive Machine Communications”, CrownCom 2015.
management and optimization. [11] Yunyan Chang, Chan Zhou, Oemer Bulakci, “Coordinated Random
Access Management for Network Overload Avoidance in Cellular
Machine-to-Machine Communications”, European Wireless 2014.
[12] I. Da Silva, G. Mildh at al., “Tight integration of new 5G air interface
and LTE to fulfill 5G requirements,” IEEE 81st Vehicular Technology
Conference (VTC Spring), Glasgow, May 2015.