AbstractThe IEEE 802.11 medium-access control (MAC) protocol is usually considered to be a default standard in multihop
wireless networks. However, in a multihop network with a large
interference range, the request-to-send/clear-to-send (RTS/CTS)
handshake and virtual carrier sensing mechanism may not be able
to eliminate interference or solve hidden- and exposed-terminal
problems. This paper proposes two new MAC protocols, i.e., the
power-fixed dual (PFD) and power-aware dual (PAD) busy-tone
schemes, both of which are able to effectively prevent collision
of data/acknowledgment (ACK) packets and are applicable in
various open-space environments with different path-loss characteristics. Analytical models are developed to evaluate their performance in terms of the blocking area, saturation throughput, and
capability of mitigating aggregate interference of simultaneous
transmissions. Numerical examples are presented to show the
effectiveness of the proposed MAC protocols and the interaction
between performance metrics and key parameters. Analysis and
simulation results indicate that both PFD and PAD schemes can
achieve a much higher throughput and a lower packet-collision
ratio than IEEE 802.11 distributed coordination function (DCF),
conservative CTS reply (CCR), and dual busy-tone multiple-access (DBTMA) schemes.
Index TermsAd hoc network, IEEE 802.11, large interference
area, medium-access control (MAC), multihop wireless network,
power control.
I. I NTRODUCTION
ever, randomly distributed MAC protocols generate hiddenand exposed-terminal problems [5]. IEEE 802.11 is the standard for wireless local area networks (wireless LANs) based
mainly on the carrier-sense multiple-access/collision-avoidance
(CSMA/CA) technique. Most mobile ad hoc networks utilize
the standardized IEEE 802.11 distributed coordination function
(DCF) as their MAC protocol. Since IEEE 802.11 DCF was not
originally designed for multihop wireless networks, its requestto-send/clear-to-send (RTS/CTS) handshaking is not able to
effectively solve the hidden-terminal problem [2]. This problem
becomes more serious when the interference range increases in
a multihop wireless network.
Recently, a lot of effort has been made to study the effect
of a large interference range on multihop network performance
(e.g., [1], [2], and [5]). Here, the term large interference
range denotes the situation that the interference range of a
host is larger than its transmission range. The design of these
traditional MAC protocols was usually based on the assumption
that the interference range of a mobile host is the same as its
transmission range. This ideal assumption may not match the
realistic environment. As a consequence, few MAC protocols
are able to achieve expected performance when the largeinterference-area problem is present [6][14].
A. Related Works
In a multihop wireless network, the RTS/CTS handshake
alone is not capable of resolving the large-interference-area
problem. The interference range I of a host significantly
increases with the transmitterreceiver distance x. When the
distance x is larger than a threshold th , I is greater than
the hosts transmission range . Consequently, the hosts within
the interference range may not be able to receive either RTS or
CTS if using the CSMA/CA protocol. Thereafter, due to the unavailability of duration information of ongoing transmissions,
these hidden hosts are unable to correctly set the field network
allocation vector (NAV), which is used to indicate whether the
channel is busy or not. If there is an interfering host that is
transmitting packets, the ongoing transmission is interrupted.
This type of communication interruption demonstrates that
the RTS/CTS handshake alone is ineffective to address the
large-interference-area problem when the transmitterreceiver
distance x is larger than th .
In the literature, the MAC protocols were designed based on
the assumption that the interference range of a mobile host is
Authorized licensed use limited to: UNIVERSITY OF PLYMOUTH. Downloaded on October 11, 2009 at 02:27 from IEEE Xplore. Restrictions apply.
LENG et al.: MAC PROTOCOLS FOR MULTIHOP WIRELESS NETWORKS WITH A LARGE INTERFERENCE AREA
2967
Fig. 1.
System model.
Authorized licensed use limited to: UNIVERSITY OF PLYMOUTH. Downloaded on October 11, 2009 at 02:27 from IEEE Xplore. Restrictions apply.
2968
the PAD scheme, in which the busy tones are transmitted with
an adaptive power level such that they can exactly indicate the
instantaneous interference range.
In the proposed MAC protocols, data and control packets
share a single physical channel. Two out-of-band busy tones
(sine waves), i.e., the transmitting busy tone BTt and the
receiving busy tone BTr , are placed at different frequencies
with a sufficient frequency guard band in the signal channel.
Unlike RTS and CTS packets, busy tones do not collide. Since
busy tones can always stay on to avoid any potential interfering
host within the entire transmission duration of a packet, the
busy tones provide protection for data packets and a complete
solution to both hidden- and exposed-terminal problems. On
the other hand, since busy tones change the single-channel
assumption of IEEE 802.11, it may be difficult to directly use
busy tones in the current IEEE 802.11 MAC. However, some
latest international (draft) standards, such as IEEE 802.11p
together with IEEE 1609.4 for vehicular wireless networks
[11], apply the multichannel media-access control mechanism
to improve the transmission efficiency. It is likely that busytone-aided MAC schemes can easily be compatible with new
IEEE 802.11 standards in the near future.
Fig. 1 shows the system model, where we presume that host i
transmits data packets to host j with a distance x meters. When
there is an ongoing communication between hosts i and j, other
hosts in the neighborhood may transmit packets and become the
potential interfering hosts. Let r denote the radio transmission
range of hosts in a multihop wireless network. In the openspace environment, if the transmission power of a packet at the
transmitter is Pt , the received signal power at the distance x is
given by [3]
Pr (x) =
Pt Gt Gr h2t h2r
xk
(1)
PBTth TSIR rk
.
Gt Gr h2t h2r
(3)
1/k
(4)
Authorized licensed use limited to: UNIVERSITY OF PLYMOUTH. Downloaded on October 11, 2009 at 02:27 from IEEE Xplore. Restrictions apply.
LENG et al.: MAC PROTOCOLS FOR MULTIHOP WIRELESS NETWORKS WITH A LARGE INTERFERENCE AREA
2969
Fig. 2.
1/k
Compared with IEEE 802.11 DCF, PFD only requires additional dual busy tones, which are transmitted with the maximum
power PBTmax if the interference range is larger than the transmission range. A PFD-enabled host should sense a busy tone
around before it intends to send out a packet. Accordingly, those
hosts that do not hear RTS/CTS can sense busy tones and defer
their transmissions during an ongoing communication. As a
result, PFD is able to solve the hidden-terminal problem caused
by the large interference range. For instance, as shown in Fig. 1,
during the data transmission from host i to host j, the dual busy
tones in the PFD scheme are able to prohibit all transmissions
from the potential interfering host m or n. Since the dual busy
tones are always transmitted with the highest power level, the
spatial region reserved for the ongoing traffic is larger than the
interference range. In this case, the exposed-terminal problem
in PFD may be worse than that in IEEE 802.11.
B. PAD Busy-Tone Scheme
To improve spatial reuse, we propose the PAD protocol with
a power control strategy by adaptively adjusting the transmission power of the busy tones. The power level is able to exactly
indicate the interference range. Fig. 1 shows that within PAD,
during the transmission of a data packet from host i to host j,
the transmitting busy tone BTr from host i covers host js
interference range, and the transmitting busy tone BTt from
host j covers host is interference range. Furthermore, the VCS
and physical channel sensing implemented in IEEE 802.11
DCF are replaced by the component of busy-tone detection.
A PAD-compatible host can transmit a packet only if no busy
tones are detected. Fig. 3 shows the time sequence in PAD, as
explained in the list that follows.
1) Before host i intends to send RTS to host j, host i first
senses busy tones around. If no busy tones are heard
during the DCF interframe space (DIFS) period, host i
sends out RTS and initializes BTt transmission with the
maximum power level PBTmax .
2) Upon receiving RTS, host j senses the receiving busy
tone BTr around. If BTr is detected during the short
interframe space (SIFS) period, host j ignores RTS.
Otherwise, host j replies with CTS and transmits the
receiving busy tone BTr with the transmission power
level PBT (x).
(5)
TSIR PBTth Pt
Pr (x)
Authorized licensed use limited to: UNIVERSITY OF PLYMOUTH. Downloaded on October 11, 2009 at 02:27 from IEEE Xplore. Restrictions apply.
(6)
2970
Fig. 4.
reuse factor. On the other hand, if x > 0.56r, the blocking area
of PAD is larger than that of IEEE 802.11. However, the tight
blocking area reserved by IEEE 802.11 cannot efficiently avoid
collisions (the analysis can be found in the next section). Fig. 4
also shows that if x 0.56r, PFD has the same blocking area
as IEEE 802.11. When x > 0.56r, PFD has the largest blocking
area among all the three protocols.
B. Saturation Throughput
The saturation throughput represents the throughput in a
heavy traffic situation, i.e., host i always has a packet in its
buffer to send, and the receiver host j is randomly chosen
from one of its neighboring hosts. This performance metric is
very important, and an approximating model was developed
by Takagi and Kleinrock [15] and used to study carrier-sense
multiple access and busy-tone multiple access [16]. In this
paper, we will develop an analytical model to evaluate the
saturation throughput in PFD and PAD under a general pathloss model for open-space environments.
Assume that mobile hosts are distributed in a multihop
wireless network as a 2-D Poisson point process with its density
[15], [16]. Let Pr (i, S) denote the probability that i hosts are
found within area S. Pr (i, S) is given by
Pr (i, S) =
(S)i S
e
.
i!
(7)
Authorized licensed use limited to: UNIVERSITY OF PLYMOUTH. Downloaded on October 11, 2009 at 02:27 from IEEE Xplore. Restrictions apply.
LENG et al.: MAC PROTOCOLS FOR MULTIHOP WIRELESS NETWORKS WITH A LARGE INTERFERENCE AREA
2971
TABLE I
PROBABILITY COMPUTATION FOR IEEE 802.11
Fig. 5.
(1 p )i
i=0
di d
e = ep d .
i!
(8)
PrWW = (1 p )ep d .
(9)
T
=
Wait
TSucceed = tRTS +tCTS +tDATA +tACK +3 +DIFS
(10)
(rA , rB , x) =
0,
1
2
that RTS, CTS, data, and ACK packets are successfully received, respectively. PrWS (x) is given by
PrWS (x) = p (1 p )PrRTS (x)PrCTS (x)PrDATA (x)PrACK (x).
(11)
Let PrWD (x) denote the probability that the data/ACK dialogue fails due to collision after a successful RTS/CTS handshake. Then, we have
PrWD (x) = p (1 p )PrRTS (x)PrCTS (x)
{1 PrDATA (x) + PrDATA (x) [1 PrACK (x)]}
(12)
2 2 2
2 2 2 x rA rB
2
x +rA rB
x rA +rB
2
2
2
+ rB arctan
, rA rB < x rA
rB rA + + rA arctan
Authorized licensed use limited to: UNIVERSITY OF PLYMOUTH. Downloaded on October 11, 2009 at 02:27 from IEEE Xplore. Restrictions apply.
(13)
2972
2x
2p (1 p )
P
(x)dx
=
r
r2 WS
r2
PrWS =
0
THPFD =
r
(14)
From (12), the transition probability PrWD from the W ait state
to the Df ail state is derived as
PrWD
2p (1 p )
=
r2
r
W PrWW + S + D + C = W .
(15)
(16)
(23)
1
1
=
.
2 PrWW
2 (1 p )ep d
1/k
ep rI2 , x r
1 + TSIR
(24)
PrRTS (x) =
1/k
ep ,
<xr
r
1 + TSIR
where = rI2 + 2tRTS (rImax , rI , x). Based on (23), the
transition probability PrWS is given by
PrWS
(18)
(22)
TH802.11 =
S tDATA
.
W TWait + S TSucceed + C TCfail
W =
2p (1 p )
r2
r
xPrRTS (x)PrCTS (x)dx.
(20)
(21)
2p (1 p )
=
r2
r
xPrRTS (x)dx.
(25)
1
1
=
.
2 PrWW
2 (1 p )e 10p r2
(26)
(27)
S tDATA
.
W TWait + S TSucceed + C TCfail
(28)
Authorized licensed use limited to: UNIVERSITY OF PLYMOUTH. Downloaded on October 11, 2009 at 02:27 from IEEE Xplore. Restrictions apply.
LENG et al.: MAC PROTOCOLS FOR MULTIHOP WIRELESS NETWORKS WITH A LARGE INTERFERENCE AREA
(30)
Nt , =i
rT
2973
TABLE II
FHSS SYSTEM PARAMETERS
Authorized licensed use limited to: UNIVERSITY OF PLYMOUTH. Downloaded on October 11, 2009 at 02:27 from IEEE Xplore. Restrictions apply.
2974
Fig. 7.
B. Simulation Results
Next, simulation is carried out by using OPNET [21] to
evaluate the performance in a more comprehensive way. We
compare the performance of five MAC schemes, i.e., IEEE
802.11, PFD, PAD, DBTMA, and CCR [1]. In each simulation
Authorized licensed use limited to: UNIVERSITY OF PLYMOUTH. Downloaded on October 11, 2009 at 02:27 from IEEE Xplore. Restrictions apply.
LENG et al.: MAC PROTOCOLS FOR MULTIHOP WIRELESS NETWORKS WITH A LARGE INTERFERENCE AREA
Fig. 8.
V. C ONCLUSION
This paper has proposed two MAC protocols, namely, PFD
and PAD, for multihop wireless networks covering large interference areas. PFD introduces dual busy tones with fixed
transmission power to identify the maximum possible interference range when the interference range is larger than the
transmission range. PAD uses busy tone sensing to replace the
VCS and physical channel sensing, as used in IEEE 802.11.
In PAD, the dual busy tones transmit with an adaptive power
level to exactly match the instantaneous interference range. It
is shown that PFD and PAD can efficiently avoid collisions
during data and ACK packet transmissions. PAD is able to
solve the hidden- and exposed-terminal problems. An analytical
model has been developed to derive the performance metrics
with respect to the blocking area and saturation throughput in
IEEE 802.11, PFD, and PAD schemes. Analysis and simulation
results indicate that both PFD and PAD can achieve a much
higher throughput and a lower packet-collision ratio than IEEE
802.11 DCF, CCR and DBTMA. Moreover, it is found from
the analysis and simulation that the dual busy tones of both
PFD and PAD can partially indicate the aggregate interference
introduced by simultaneous transmissions.
R EFERENCES
[1] K. Xu, M. Gerla, and S. Bae, Effectiveness of RTS/CTS handshake
in IEEE 802.11 based ad hoc networks, Ad Hoc Netw., vol. 1, no. 1,
pp. 107123, Jul. 2003.
[2] S. Xu and T. Saadawi, Does the IEEE 802.11 MAC protocol work well
in multihop wireless ad hoc networks? IEEE Commun. Mag., vol. 39, no.
6, pp. 130137, Jun. 2001.
[3] T. Rappaport, Wireless Communications: Principles and Practice.
Englewood Cliffs, NJ: PrenticeHall, 1996.
2975
Supeng Leng (S01M06) received the B.Eng. degree from the University of Electronic Science and
Technology of China (UESTC), Chengdu, China, in
1996 and the Ph.D. degree from Nanyang Technological University (NTU), Singapore, in 2005.
He is an Associate Professor with the National
Communication Laboratory, UESTC. He has experience as an R&D Engineer in the field of computer
communications and as a Research Fellow with the
Network Technology Research Center, NTU. His
research focuses on ad hoc/sensor networks, wireless
mesh networks, ultrawideband networks, and cognitive radio networks.
Authorized licensed use limited to: UNIVERSITY OF PLYMOUTH. Downloaded on October 11, 2009 at 02:27 from IEEE Xplore. Restrictions apply.
2976
Jie Xiang received the B.Eng. degree in communication engineering from Wuhan University, Wuhan,
China, and the M.Eng. degree in communication
and information systems from Huazhong University of Science and Technology, Wuhan. He is
currently working toward the Ph.D. degree with
the Simula School of Research and Innovation,
Lysaker, Norway, and the University of Oslo, Oslo,
Norway.
His current research focuses on radio resource
management, dynamic spectrum access schemes,
and networking in cognitive radio wireless networks.
Authorized licensed use limited to: UNIVERSITY OF PLYMOUTH. Downloaded on October 11, 2009 at 02:27 from IEEE Xplore. Restrictions apply.