Академический Документы
Профессиональный Документы
Культура Документы
1
1. Introduction
wireless networks. The general idea of a multiple access MAC protocol is that a station
senses the medium before a transmission. Only if the medium is idle, does the station
sends data. But since the interference at the receiver is what we are trying to avoid, the
going on when C has a frame to send to B, since C is outside the transmission range of A
(hidden from A), C senses the medium as idle, then starts the transmission, which will
corrupt the data from A received at B. Such kind of collision caused by stations hidden
from each other are called hidden station problem, also known as hidden terminal
problem, hidden node problem, etc. This problem is originally defined in [1]. Since then,
e on
A ’ i ss
ns C’s
ns ang
ng si
s ion
ra mis
m e
r
tra
A B C
2
2. Solutions
BTMA assumes that all stations are within the range of transmission of the base
station or access point. And the idea of BTMA is to divide the bandwidth of a channel
into data channel and busy-tone channel. If the base station of access point senses a busy
data channel, it will transmit a busy-tone signal on the busy-tone channel. Each station
listens to the busy-tone channel to find out whether the data channel is free before a
transmission. The constraints of this approach include “the use of a separate channel to
convey the state of the data channel, the need for the base station to transmit the busy
tone while detecting carrier in the data channel, and the difficulty of detecting the busy-
In [3], the “exposed station” problem that is closely related to the hidden station
problem is discussed. As shown in Fig. 2, the transmission from C to D has been going
on when B wants to send data to A. Since B senses that the medium is busy, it will defer
this transmission. But in fact, the transmission from B to A will not affect the
ÆA) is prohibited because B is exposed to the ongoing CÆD transmission. Although the
exposed station problem does not cause collision, it degrades the network throughput.
3
X
A B C D
CSMA/CA method (used by the Apple Localtalk network for somewhat different
sends out the data frame to B, it transmits a Request-To-Send (RTS) to B. Upon correctly
receiving the RTS, B replies a Clear-To-Send (CTS) frame. After A receives the CTS, it
starts the data frame transmission. During this procedure, stations that overhear the RTS
frame will defer all transmissions long enough for A to receive the CTS. While stations
(including C) that hear the CTS will keep quiet (because they are in B’s range, so their
Using such a protocol, the hidden station problem is resolved with the help of
RTS/CTS handshake, i.e., C will not interrupt the transmission from A to B even though
it cannot sense the ongoing transmission. Meanwhile, the exposed station problem is also
addressed to some extent. Consider the scenario described at the beginning of this section.
Assume that the CÆD transmission used RTS/CTS/DATA handshake, B must have
heard C’s RTS, but B could not hear D’s CTS. According to MACA, B only has to wait
4
long enough for C to finish receiving the CTS from D. After that, B can go ahead to
initiate the transmission to A instead of wait until CÆD data transmission completes.
MACAW is a protocol based on MACA after the study of a single channel wireless
LAN at Xerox Corporation’s Palo Alto Research Center [4]. In [4], implementation
details such as the backoff algorithm and the frame queues are analyzed and revised
accordingly. Moreover, the needs to improve higher layer (TCP/UDP) throughput leads
to the introduction of ACK for each DATA frame, which brings necessary modifications
on MACA.
collision on C with D’s ACK. On the other hand, since B did not hear D’s CTS, there is
no way for B to know when the DATA/ACK handshake between C and D finish. To
Sending) control frame is added. It works as follows. After the sender receives the CTS
and before it starts the DATA transmission, it sends out a DS frame to let all stations in
its range to know that the RTS/CTS handshake has succeeded and the DATA/ACK
handshake is about to occur. The stations that overhear the DS frame will defer all
transmissions until the ACK frame time slot has passed [4].
MACA, MACAW, as well as their variants share the following feature, which is
“stations do not sense the channel before transmissions” [5]. “A station defers its
5
transmission only after it has received and understood a complete RTS or CTS” [5].
• RTS/CTS exchange with no carrier sensing, such as MACA and its variants, FAMA-
Fig. 3 At t1, S sends RTS to R. At t2 > t1, R replies a CTS. At the mean time,
H sends an RTS to R. At t3 > t2, DATA sent from S collides with the RTS
from H at R [5].
Fig. 3 shows the scenario when collision happens using MACA. FAMA solve this
problem by making CTS last long enough or be repeated enough times to avoid such kind
of collisions. It is proved in [2] that for FAMA-NCS, correct floor acquisition can be
achieved as long as γ > τ and γ + 2τ + ε < γ ′ < ∞ with the presence of hidden stations.
and ε is the hardware transit-to-receive transition time. Meanwhile, for FAMA-NPS (no
carrier sensing), with the presence of hidden stations, correct floor acquisition can be
achieved if the receive repeats at least 2N+1 CTS in response to an RTS and the
6
where N is the total number of neighbors any node in the network may have, plus the
maximum number of neighbors any one of those neighbors may have, and γ is the
transmission time of RTS or CTS. It can be seen that such a restrictive requirement
hidden station problem. They still have limitations. And there is still space to improve the
performance.
As shown in Fig. 4, the CTS from B to A collided with D’s data frame at node C so
that C cannot understand the CTS and find out the updated NAV value. Later, after the
DÆE data transmission finish, C thinks (and senses) that the channel is free to use. It
initiates a transmission to B by an RTS, which collides with the data frame sent from A at
node B. The reason behind this situation is that node C is “masked” by ongoing DÆE
transmission so that it cannot correctly hear any CTS sent by other stations. [6] evaluates
the impact of this problem on IEEE 802.11 ad hoc network by mathematical analysis and
experiments and concludes that the masked node problem significantly affect the delay
and throughput performance [6]. Unfortunately, up to this point, there is no positive result
7
Fig. 4 Masked Node Problem [6]
be inhibited. As shown in Fig. 5, in (1), the transmissions AÆB and PÆQ can run in
must wait for the other to fully complete. Similar limitation also exists for (2). The
way handshake [7]. Look at Fig. 5 (1), suppose concurrent data transmissions AÆB and
PÆQ are going on simultaneously. Assume AÆB data transmission finishes before
PÆQ data transmission. Since B has to send an ACK back to A (A & B switch their
sender/receiver roles), this ACK will collide with the data transmission from P at node Q.
8
Fig. 5 Using RTS/CTS, these traffic patterns are prohibited. But (1) and (2)
delineate the intervals for both the DATA and ACK transmissions, instead of only to
specify the total (DATA+ACK) transmission duration. Fig. 6 illustrates this scheme. B
has overheard Q’s CTS when it receives A’s RTS. If A allows its transmission schedule
to be modified, B will send back a CTS with modified TDATA and TACK that synchronize
In the real world, a lot of factors may affect the effectiveness of RTS/CTS scheme. In
[8], the impacts of the difference between the transmission range and carrier sensing
9
range is discussed. The motivation is that the assumption that hidden nodes are within
transmission range of receivers cannot always hold due to the fact that power needed for
interrupting a frame reception is much lower than that of delivering a frame successfully
[8]. In [9], node mobility is considered. And [10] takes different transmission rates for
control frames and data frames into consideration. Interested readers can refer to these
4. Performance Analysis
RTS/CTS handshake can reduce the chance of collisions resulted from the hidden
stations with the price of control frames overhead. An interesting issue is the impact of
RTS/CTS on the throughput performance. It is shown in [11] that only when the number
of stations is very few (2 or 3), does RTS/CTS mechanism perform a little worse. When
the number of stations is greater than or equal to 5, RTS/CTS always has better
transmission probability) is less when RTS/CTS is used. Hence, RTS/CTS method should
trigger RTS/CTS. Namely, when the data frame size is smaller than the threshold, do not
use RTS/CTS; otherwise use it. The impact of RTS threshold on the throughput
contending stations is the main factor that influences the optimal threshold value while
the frames arrival rate and the length of frames are ignorable. Since the best throughput
performance is obtained with the RTS threshold set to a small value, [12] suggest always
10
5. Conclusions
In this survey, I make a brief summary on the hidden station problem. Various
solutions are discussed. BTMA [1] uses a separate busy-tone channel to make all in range
stations aware of ongoing transmission. MACA [3] eliminates the need for a separate
busy-tone channel and provides the RTS/CTS/DATA handshake solution. From then on,
the RTS/CTS mechanism became the basic building block for the major solutions for the
hidden station problem. The idea is to have the sender initiate an RTS frame to the
receiver, and the receiver replies a CTS frame to the sender so that all stations within the
transmission range of the receiver defer their transmission until the data transmission
completes. Because of the need for ACK for DATA, major revision is made in MACAW
that stations in sender’s transmission range defer their transmissions until the sender
successfully receives the ACK from the receiver. Noticing the problem caused by the
absence of carrier sensing in MACA as well as its variants, FAMA [5][2] introduced a
new protocol (FAMA-NTR [5], FAMA-NCS [2]) in which a station senses the medium
before transmitting RTS. And RTS/CTS-based no carrier sensing protocols like MACA
and its variants becomes FAMA-NPS [2]. In addition, FAMA take the propagation delay
into account and gives RTS/CTS length restriction to guarantee the correct floor
acquisition.
Even after these efforts, there are situations when collision caused by hidden stations
may still happen. The masked node problem [6] is an example. This problem happens
11
that this station may start a transmission later to corrupt the transmission specified by that
CTS. On the other hand, there is room to improve the throughput performance by
RTS/CTS frames. Furthermore, physical factors such as the difference between the
transmission range and carrier sensing range [8], node mobility [9], and different
transmission rates for control frames and data frames [10] also play important roles in the
between reducing the retransmission caused by hidden station problem and adding
control overhead. Throughput performance analysis in [11] and simulation results in [12]
6. Future Work
Here I list some future work on the hidden station problem after the survey of its
current development.
Recall that [4] digs into the backoff algorithms in MACA and makes revisions to
(MACA and its variants do not have carrier sensing for RTS transmission), I believe that
analysis on the control frame collision problem should also be done on current RTS/CTS
12
(b). Hidden-station-aware protocols
Suppose stations are stationary, if all data transmissions use RTS/CTS mechanism, it
is not hard for a station to be aware of stations that are hidden from it. Consider Fig. 1.
Assume that A sends an RTS to B, then B replies with a CTS which will be heard by C.
By looking into the CTS frame, C can find out that the CTS is for station A. But since C
did not hear A’s RTS, C can realize that A is hidden from itself. The impact of such
awareness on the hidden station problem is yet to determine, but I believe that this is
helpful information that could be taken advantage of. Moreover, if this approach could be
taken, the node mobility issue needs to be addressed to guarantee that each node has up-
Throughput has been the major performance metrics. It is reasonable that we want to
maximize the utilization of the network. But for a station, its concerns might be selfish,
e.g., to maximize its own throughput. Specific to the hidden station problem, although a
formal proof is not there yet, I believe that in current IEEE 802.11, using RTS/CTS
mechanism gains a station better chance to get a hold of the medium since the IFS for
RTS (SIFS) is the shortest among all IFS’s. Of course, the tradeoff is that RTS/CTS
introduces the control overhead. The game theory point of view is to regard this situation
as a non-cooperative game in which each station is trying to maximize its own throughput
by action of whether to use RTS/CTS, when to trigger it (by configuring its own
dot11RTSThreshold), etc. I hope that the tools in game theory could help to give us a
13
Reference
[1] F. Tobagi and L. Kleinrock, “Packet Switching in Radio Channels: Part II – The
Hidden Terminal Problem in Carrier Sense Multiple-Access and the Busy-Tone
Solution”, IEEE Trans. Comm., vol. 23, no. 12, pp. 1400-1416, 1975.
[3] P. Karn, “MACA – A New Channel Access Method for Packet Radio”, Proc.
ARRL/CRRL Amateur Radio Ninth Computer Networking Conf., pp. 134-140, 1990.
[6] S. Ray, J. B. Carruthers, and D. Starobinski, “Evaluation of the masked node problem
in ad-hoc wireless LANs”, IEEE Trans. on Mobile Computing, vol. 4, no. 5, 2005.
[8] K. Xu, M. Gerla, and S. Bae, “How effective is the IEEE 802.11 RTS/CTS handshake
in ad hoc networks?”, Proc. IEEE GlobeCom ’02, 2002.
[9] Z. Haas, “On the performance of a medium access control scheme for the re-
configurable wireless networks”, Proc. MILCOM ’97, 1997.
[10] M. Borgo, A. Zanella, P. Bisaglia, and S. Merlin, “Analysis of the hidden terminal
effect in multi-rate IEEE 802.11b networks”, International Symposium of Wireless
Personal Multimedia Communication WPMC 2004, Spet. 2004.
[12] S. Sheu, T. Chen, J. Chen, and F. Ye, “The Impact of RTS Threshold on IEEE
802.11 MAC Protocol”, Proc. IEEE ICPADS ’02, 2002.
14