2014 Fourth International Conference on Communication Systems and Network Technologies

ARIMA - Asymmetric Receiver Initiated Multiple

Access Power Control Protocol for Wireless
Ad-Hoc Networks
Bighnaraj Panigrahi1 , Swades De2 , Hemant Kumar Rath1 , and Anantha Simha1
1
TCS Networks Lab, Bangalore, India, Email:{bighnaraj.panigrahi, hemant.rath, anantha.simha}@tcs.com
2
Dept. of Electrical Engineering, IIT Delhi, India, Email:swadesd@ee.iitd.ac.in

AbstractIn a distributed control wireless ad-hoc network, whereas the receivers interference range is determined by its
network performance is signicantly effected by multi-user in- signal detection sensitivity . Hidden terminals are the nodes
terference. In such networks, transmit power control can enhance that are located outside the CS range of the transmitter, but in-
the energy usage as well as the QoS performance. Although, the
effects of hidden/exposed terminals on the system throughput side the interference range of the corresponding active receiver.
in wireless ad-hoc networks have been studied extensively, the The problem due to hidden terminals is that, these nodes can
IEEE 802.11 with variable transmit power has to be re-looked cause interference to an ongoing reception should they choose
in order to optimize the energy and other QoS measures (e.g, to initiate transmission during an ongoing reception at the
bandwidth, throughput). In this paper, we capture the drawbacks intended receiver. It has been already shown in several prior
of the existing xed and variable transmission-power control
Medium Access Control (MAC) protocols and propose an ef- works (e.g., [2], [3]) that, due to distributed communication
cient energy aware receiver initiated variable transmission-power control, even with VCS hidden/exposed terminals problem
control protocol called Asymmetric Receiver Initiated Multiple cannot be completely eliminated.
Access (ARIMA). It minimizes interference due to the exposed The related works are surveyed in Section II. Problem
terminal carrier sense nodes at the receiver and hence offers denition and assumptions are described in Section III. The
higher network performance. We compare the performance of
our proposed protocol with the existing protocols analytically and proposed receiver initiated transmit power control protocol
through exhaustive simulation studies and observe that ARIMA model is presented in Section IV. Power control protocol
outperforms the existing protocols in terms of throughput and performance is analyzed in Section V. The paper is concluded
packet transmission probability. This is a simple protocol and in Section VI.
can be used in practice.
II. R ELATED W ORKS
I. I NTRODUCTION
Several recent works have investigated the hidden/exposed
Wireless ad-hoc network is typically characterized by dis- terminals issues. The impact of physical carrier sensing range
tributed control, i.e., any node can set up a communication on the aggregated network throughput of one-hop ows was
session with any other node without needing any central studied by Deng et al. [4], where the authors proposed to
coordination. Network solutions based on a xed and full have a tunable carrier sensing range and showed the trade-off
transmission power control approach [1] improve the physi- between spatial reuse and packet collisions.
cal connectivity of wireless networks. However, this goal is
Another set of studies (e.g., [3], [5], [6]) on IEEE 802.11
achieved at the expense of sacricing network capacity and
ad-hoc networks aimed at reduced exposed and hidden nodes
wasting precious transmission power of the nodes. Effective
by exploiting the differential capture capability of modern
transmission power control plays an important role in the
transceivers. However these solutions work well only at lower
design and performance of such emerging application oriented
data rates. Also, they do not address the issues of interference
distributed networks. In addition, it also optimizes the system
associated with power control. In [6], Zhu and Zheng showed
design metrics such as physical/network connectivity, energy that the performance of collaborative relays in large wireless
conservation, and Quality of Service (QoS) (e.g., throughput,
networks degrades signicantly due to interference. The ef-
packet loss and delay).
fects of power control and redundant transmissions were not
Carrier Sense Multiple Access (CSMA), which has the in- considered on the interference and nodal energy consumption.
built Physical Carrier Sensing (PCS) procedure at the trans- Although the interference problem in CSMA-based access
mitter before the actual communication, is popularly used for systems has been identied and several alternative proposals
setting up a communication link between two ad-hoc wireless have been presented in recent research literature [4], [7]; most
nodes. A two-way handshake (Virtual Carrier Sensing VCS) of the prior works are protocol level studies. In our recent
is also commonly used along with PCS for transmission of
longer data packets. In PCS, Carrier Sensing (CS) range of In general carrier sense range of a node could be different from its
the transmitter determines the number of exposed terminals, interference range.

978-1-4799-3070-8/14 $31.00 2014 IEEE 293

DOI 10.1109/CSNT.2014.64
work [8], we have developed a probabilistic model which transmission range and carrier sense ranges respectively. If
characterizes the effects of interference in a wireless scenario the distance between the sender and receiver is less than the
with homogeneous nodal coverage range. transmission range R, then instead of using full transmission
The authors in [9] proposed a Power Control MAC protocol power, the frames can be transmitted at lowest possible power
(PCM), referred as Asymmetric Source Initiated Multiple that allows the receiver just be able to correctly receive the
Access (ASMIA), where RTS/CTS (Request To Send/ Clear data packet. Depending on how much transmit power to be
To Send) signals are transmitted at full power, and the data used in transmitting a packet, two types of transmission is
frames are transmitted at lowest possible power that allows possible such as:
the receiver just to be able to correctly receive the data. The Full transmit power control (FP) : Irrespective of the
transmit signal power is periodically increased to the full level, distance between the transmitter and receiver, the packet
and the periodicity of these high power pulses are chosen will be transmitted using full transmit power P tmax .
optimally to virtually create the same interference environment Variable transmit power control (PC): In this kind
as in IEEE 802.11 Distributed Coordination Function (DCF) of transmission the transmit power for a packet is set
without power control. This protocol achieves power saving to exactly amount that required to achieve the desired
without deteriorating the network performance compared to receiver threshold Rxth at the receiver. Thus, the vari-
the basic IEEE 802.11 standard without power control. How- able transmission power, P tpc (d) is a function of the
ever due to the asymmetric handling of the interfering nodes transmitter-receiver distance, d.
only on the source side, the hidden/exposed problem at the When a packet transmission is being initiated between the
receiver side remains unresolved. transmitter and receiver node, then the transmission and carrier
In this paper, we study and propose an energy efcient
sense zone for full power and variable power cases respectively
variable transmit-power control protocol called Asymmetric
are fp-tx-zone, fp-cs-zone, pc-tx-zone, and pc-cs-zones. The
Received Initiated Multiple Access (ARIMA) that attempts to
respective hidden zones for transmitter and receiver are fp-
minimize the packet dropping probability due to interference
src-intf-zone and fp-dst-intf-zone.
at the receiver. As maximum of interference occur at the
Fig. 1 explains the interference scenario at the receiver node.
receiver side, ARIMA attempts to prevent the system from
S and D are the source and receiver nodes respectively. Due
such interferers.
to transmit power control the CS range of the nodes will
III. P ROBLEM D EFINITION change. Hence, due to the reduced power of the DATA packet
Designing a variable transmit power control system for transmission, nodes in the CS range of the receiver but outside
wireless ad-hoc networks requires signicant understanding of the CS range of the transmitter (red nodes in fp-dst-intf-zone)
the existing protocols and their operation, which we discuss can interfere at the destination.
below. Post that, the new system needs to be analyzed to
CS range reduced
understand its impact from the prospective of energy savings, (Power Control)
physical connectivity, implementation complexity and QoS,
etc., which we discuss in the next and subsequent sections.
A. Assumptions
Nodal distribution: The nodes are distributed uniformly 2R

randomly in a two-dimensional location space, which f p dst intf zone

f p src intf zone
S R D
is approximated as a two-dimensional spatial Poisson
point process with node density . Average number of
neighbors of a node in an area A is N = A.
Channel access: The entire channel bandwidth is avail-
able to all users. Communication is any-to-any, and
trafc is uniform across all nodes. The access protocol is
slotted CSMA/CA with RTS/CTS with exponential back-
off mechanism.
Fig. 1. Interference Zone with Transmit Power Control
Memoryless channel errors: A current error does not have
any bearing on the future errors. Thus, beyond MAC
contention and Gaussian noise, fading dependent errors C. Existing MAC layer Protocols
are not accounted.
We have assumed that carrier sense range of a transmis- We now discuss the existing MAC layer protocols for ad-
sion is double of the transmission range. hoc networks and capture the drawbacks of those protocols.
We then propose a solution to overcome these drawbacks.
B. Full transmit power Vs. Variable transmit power 1) IEEE-802.11 Without Power Control (WPC): In this case
Let R and Rc be the full communication and carrier sense [1], DATA, RTS, CTS and ACK will be transmitted using full
ranges respectively. d and Rcpc are the variable or reduced transmit power. fp-tx-zone and fp-cs-zone nodes of source node

294
stay silent during the transmission of RTS as well as DATA P tmax

packet from the source node. Similarly, fp-tx-zone and fp-cs- S 465s
P tpc
zone nodes of the destination keep themselves silent during
0
CTS and ACK packet transmissions. However, either nodes
from fp-dst-intf-zone (c.f. Fig. 1) during the reception of DATA
10s
packet at the destination node or fp-src-intf-zone (c.f. Fig. 1)
P tmax
nodes during the reception of ACK packet at the source node
D
may create interference. P tpc
2) IEEE-802.11 with Power Control (BASIC-PC): In the 0
basic power control mechanism RTS and CTS are transmitted
using full power whereas DATA and ACK are transmitted
using reduced power. However, during transmission of DATA RTS CTS DATA
packets with reduced power, nodes in the fp-cs-zone of the
destination node which are outside the fp-tx-zone of both, Fig. 2. Packet Transmission Process in ARIMA
source and destination and also outside the fp-cs-zone of the
destination node, can create interference at the receiver node.
3) Asymmetric Source Initiated Multiple Access with power signicantly improves the energy saving and QoS performance
control (ASIMA) : ASIMA [9] is a power control protocol that of the network.
tries to improve the performance of BASIC-PC protocol. In We now analyze the performance of ARIMA analytically
ASIMA, during the transmission of DATA packet, the source and through exhaustive simulations and compares it with
node repeatedly, after every Extended Inter-Frame Space existing protocols.
(EIFS) time instant, varies the transmit power from P tpc (d) V. P ERFORMANCE A NALYSIS
to P tmax and then back to P tpc (d) again for a small duration
of time called beacon period. We call this process as repeated A. Analytical - Packet Success Probability
power transition. By doing so, all the fp-cs-zone nodes of The successful packet transmission probability between S
the source can be prevented from interference. During the and D depends on the successful transmission of RTS, CTS,
DATA transmission however, nodes at the destination side that DATA, and ACK packets. The packet success probability of
are outside fp-cs-zone of source node can cause interference any of these packets depend on the interfering area around the
at the receiver. However, there wont be any collision at the transmitter and receiver nodes which can cause collision of
destination during the reception of ACK packet. the ongoing transmission. We call such an area as vulnerable
zone. The vulnerable zone depends on many factors related to
IV. P ROPOSED M ODEL the transmission such as, transmitter-receiver distance, kind of
Since, interference happens on the receiver side only, we packet (RTS/CTS/DATA), node density, what MAC protocol
argue that the repeated power transition method should be being used. We denote the vulnerable zone as Avu (d) where,
carried out by the receiver node instead of the transmitter (as u and v represent the protocol type and packet type respec-
in case of ASIMA). Thus, at every EIFS period of time the tively. Vulnerable zone for different protocols and at different
receiver changes its mode from receiving to transmitting and types of packets can be easily computed using geometrical
transmits a beacon with maximum power. By doing this, the mathematics, which we could not include in this paper due
interferers in the fp-dst-intf-zone can be kept quite. We now to limited space. We now compute the success probability of
explain our proposed model ARIMA which attempts to solve each of the events.
this interference problem. 1) RTS Success Probability Prts succ : It is the probability
associated with the event when a node in idle state transmits
A. Asymmetric Receiver Initiated Multiple Access with power the RTS and receiver receives it successfully.
control (ARIMA) Let, Ptr be the probability that an idle node attempts
to begin transmission. Collision occurs when even a single
Due the fact that maximum packet drop occurs at the node from the vulnerable zone transmits while RTS is being
destination nodes, ARIMA proposes that the repeated power received. Thus we can write
transition process to be performed by the destination node
instead by the source node. In ARIMA, during the DATA
(Avu (d)) Avu (d)
k
packet reception, the destination node repeatedly, after every (1 Ptr )k
Prts succ (d) = Ptr (1Ptr ) e ,
EIFS time instant, switches from receive mode to transmission k!
k=0
mode and transmits a beacon packet with P tpc to P tmax . (1)
After beacon period, it again switches back to receiving mode. where, k is number of interferers in the vulnerablezone. After
This process is repeated till the reception of the DATA packet simplifying (1), we obtain
completes (see Fig. 2). Although, there will be some overhead
= Ptr (1 Ptr ) eAu (d)Ptr .
v
in terms of the switching over and beacon signalling, ARIMA Prts succ (d) (2)

295
 2) CTS Success Probability Pcts succ : Successful reception 1

of RTS packet at the receiver ensures that nodes in the carrier

sensing range of the receiver are not transmitting so receiver 0.8

Packet success probability

nds the medium idle. Also, the nodes in carrier sensing
range of the transmitter sensed RTS transmission and they 0.6
will differ their transmission by EIFS period which cover the
CTS transmission period. Hence, CTS does not get dropped
0.4
at the transmitter. Thus, ana_WPC
ana_BASIC_PC
Pcts succ = 1. (3) 0.2
ana_ASIMA
ana_ARIMA
3) DATA Success Probability Pdata succ : With data dura- 0
10 20 30 40 50 60
tion, Tdata and slot duration Ts , the average number of slots Average number of neighbors
when hidden nodes could initiate their transmission is
Fig. 3. Average Packet Success Probability, R = 30m.
 
Tdata
Ndata = . (4)
Ts
C. Simulation Study
Therefore, the probability associated with successful DATA
packet after RTS success is Pdata succ and can be written as The network performance evaluation has been carried out
via discrete event simulations using network simulator, NS-
 Ndata 2. We have developed NS-2 patch for the proposed protocol

 v k
k (Au ) Av
by modifying the existing NS-2 properties for IEEE Wireless
Pdata succ (d) = (1 Ptr ) ue , 802.11. In a 300 300 m2 area we deployed nodes uniformly
k! (5)
k=0
Ptr Av
randomly with a network density = 0.02. With a BER of
=e u Ndata . at most 103 , we use a two ray ground propagation model
Note that, using similar techniques as in (4) we can com- with Log-normal channel fading was simulated with a 4 dB
pute Nrts , Ncts , and Nack , where Tdata = Ts , and hence standard deviation. We assume the basic Binary Phase Shift
Nrts = Ncts = Nack = 1. Since the ACK packet transmission Keying (BPSK) modulation with Non-Return to Zero (NRZ)
time is negligible as compared to DATA transmission time, we signaling technique. Fixed packet size of L = 2346 bytes
assume that ACK packets are always received successfully. with an MPDU of 2316 bytes was taken with a data rate of
Total probability of complete successful packet transmission 20 Mbps. These values correspond to the existing standard
process can be expressed as: coding mechanism of IEEE 802.11 standard. The initial energy
of each node was taken as 30 J. Sufcient simulation runs were
conducted with 100 different seed values to have a condence
Ptot succ (d) = Prts succ (d) Pcts succ (d) Pdata succ (d).
level of 95% within the range of 2% of the results obtained.
(6) 1) Simulation Results - Effect of network density, for a
As, Ptot succ (d) is a function of inter-nodal distance d and xed nodal coverage range: Fig. 4 and Fig. 5 capture relative
we need to nd the expectation of Ptots ucc (d) as follows: performance of different MAC protocols at different node
densities with a xed transmitter-receiver distance. From these
 R gures, we have observed that as the number of neighbors
Ptot succ = f (d)Ptot succ (d)dd, (7) increase, the probability of success decreases. We further
0
observe that ARIMA performs signicantly better than other
where, f (d) is the PDF of d. As, we have assumed Uniform discussed protocols WPC, BASIC-PC and ASIMA.
random distribution of nodes, f (d) is given by: 2) Simulation Results - Effect of source-destination dis-
tance, for xed transmission range and network density: In-
2d
f (d) = . (8) crease/decrease in the distance between source and destination
R2 have the impact of increase/decrease in the interference zone
B. Analytic Results during the packet transmission. Therefore, we have conducted
simulations by varying the distance between source and desti-
Using the mathematical analysis presented in Section V
nation, keeping a xed average number of neighbors. In Fig.
we attempted to plot (7) using Matlab. Fig. 3 explains the
6 and Fig. 7 we have compared the performance of ARIMA
probability of success at different node densities. We also plot
with the existing protocols with variable source-destination
the packet success probability of three existing protocols such
distance. From these gures, we observe that ARIMA per-
as WPC, BASIC-PC and ASIMA. From Fig. 3, we observe
forms better even with variable distance, as compared to that
that ARIMA protocol has higher packet success probability
of the existing protocols in terms of success probability and
than other protocols.

296
 1

1.6

0.8
Packet success probability

1.2 WPC

Throughput
0.6 WPC
Basic_PC
Basic_PC
ASIMA
ASIMA 0.8
0.4 ARIMA
ARIMA

0.4
0.2

0 0
10 30 50 70 90 5 10 15 20 25 30
Average number of neighbors Transmitterreceiver distance (m)

Fig. 4. Average Packet Success Probability, R=30 Fig. 7. Successful information bits per second.

1.8
Initiated Multiple Access (ARIMA), avoids interference at the
receiver and ensures successful reception of the data packet.
1.4
The receiver initiated repeated power transition method of
WPC
ARIMA reduces the interference and hence increases the
success probability of the DATA packet and is able to
Throughput

Basic_PC
1 ASIMA achieve improved QoS performance of the network in terms of
ARIMA throughput, which is veried through analytical and simulation
results. We have further observed that ARIMA outperforms the
0.6 existing protocols such as WPC, BASIC-PC and ASIMA in
terms of throughput and success probability. Since, the success
probability is improved, the number of re-transmissions are
0.2
10 30 50 70 90
also reduced resulting in improve in the energy efciency
Average number of neighbors of ARIMA, which requires further investigation. We plan to
Fig. 5. Successful information bits per second, R=30 investigate the energy efciency of ARIMA and extend the
model to an interference aware energy model and analyze the
effect of power control in such scenario as future work.
throughput.
R EFERENCES
[1] Wireless lan medium access control (mac) and physical layer (phy)
1
specications, ieee std 802.11, in 1999 Ed. (R2003).
[2] K. Xu, M. Gerla, and S. Bae, How effective is the IEEE 802.11 RTS/CTS
handshake in ad hoc networks, in Proc. IEEE GLOBECOM, vol. 1,
0.8
(Taipei, Taiwan), pp. 7276, Nov. 2002.
Packet success probability

WPC
[3] F. Ye, H. Yang, H. Yang, and B. Sikdar, A distributed coordination
Basic_PC scheme to improve the performance of IEEE 802.11 in multi-hop net-
0.6
ASIMA works, IEEE Trans. Commun., vol. 57, pp. 29032908, Oct. 2009.
ARIMA [4] J. Deng, B. Liang, and P. Varshney, Tuning the carrier sensing range of
IEEE 802.11 MAC, in Proc. IEEE GLOBECOM, (Dallas, TX, USA),
0.4
Dec. 2004.
[5] M. M. Vegad, S. De, and B. Lall, A liberal carrier sensing for increased
spatial reuse in multi-hop wireless ad hoc networks, in Proc. IEEE Intl.
0.2
Conf. Commun. (ICC), (Cape Town, South Africa), May 2010.
[6] Y. Zhu and H. Zheng, Understanding the impact of interference on
collaborative relay, IEEE Trans. Mobile Computing, vol. 7, pp. 724736,
0
5 10 15 20 25 30 June 2008.
Transmitterreceiver distance (m) [7] Y. Yang, J. C. Hou, and L.-C. Kung, Modeling the effect of transmit
power and physical carrier sense in multi-hop wireless networks, in Proc.
Fig. 6. Average Packet Success Probability. IEEE INFOCOM, (Anchorage, AK, USA), May 2007.
[8] A. Sharma, B. Panigrahi, and S. De, Impact of interference on nodal
communication range in wireless ad hoc networks, in Proc. Nat. Conf.
VI. C ONCLUSION AND F UTURE W ORK Commun. (NCC), (Bangalore, India), Jan. 2011.
[9] E.-S. Jung and N. H. Vaidya, A power control MAC protocol for ad hoc
In this paper we have conducted an extensive study on the networks, in Proc. ACM MOBICOM, (Atlanta, GA, USA), pp. 3647,
variable transmission power control protocols and propose a Sept. 2002.
novel receiver initiated power control protocol for wireless ad-
hoc networks. Our proposed protocol Asymmetric Receiver

297