Group Title: Department of Computer and Information Science and Engineering Technical Reports
Title: BEWARE : Background traffic-aware rate adaptation for IEEE 802.11 MAC
Full Citation
Permanent Link:
 Material Information
Title: BEWARE : Background traffic-aware rate adaptation for IEEE 802.11 MAC
Alternate Title: Department of Computer and Information Science and Engineering Technical Report
Physical Description: Book
Language: English
Creator: Wang, Shao-Cheng
Helmy, Ahmed
Publisher: Department of Computer and Information Science and Engineering, University of Florida
Place of Publication: Gainesville, Fla.
Copyright Date: 2007
 Record Information
Bibliographic ID: UF00095655
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.


This item has the following downloads:

2007420 ( PDF )

Full Text

BEWARE: Background Traffic-Aware Rate

Adaptation for IEEE 802.11 MAC

Shao-Cheng Wang and Ahmed Helmy
Department of Computer and Information Science and Engineering
University of Florida
Gainesville, FL., U.S.A.

Abstract- IEEE 802.11-based devices employ rate adaptation
algorithms to dynamically switch data rates to accommodate
the fluctuating wireless channel conditions. Many studies
observed that, when there are other stations transmitting in the
network, existing rate adaptation performance degrades
significantly due to the inability of differentiating losses
between wireless noise and contention collisions. They
proposed to exploit optional RTS frames to isolate the wireless
losses from collision losses, and thus improve rate adaptation
performance. In this paper, we conduct a systematic evaluation
on the effectiveness of various existing rate adaptation
algorithms and related proposals for loss differentiations, with
multiple stations transmitting background traffic in the
network. Our main contributions are two-fold. Firstly, we
observe that most existing rate adaptations do not perform
well in background traffic scenarios. In addition, our study
reveals that RTS-based loss differentiation schemes can
mislead the rate adaptation algorithms to persist on using
similar data rate combinations regardless of background
traffic level, thus result in performance penalty in certain
scenarios. The fundamental challenge is that rate adaptation
must dynamically adjust the rate selection decision objectives
with respect to different background traffic levels. Secondly,
we design a new Background traffic aware rate adaptation
algorithm (BEWARE) that addresses the above challenge.
BEWARE uses a mathematical model to calculate on-the-fly
the expected packet transmission time based on current
wireless channel and background traffic conditions. Our
simulation results show that BEWARE outperforms other rate
adaptation algorithms without RTS loss differentiation by up
to 250% and with RTS by up to 25% in throughput.

Keywords- Rate adaptation, 802.11, MAC

With the large-scale deployments of wireless local area
networks (WLANs) in homes, offices, and public areas, the
IEEE 802.11 standard has become the dominant technology
in providing low-cost high-bandwidth wireless connections.
A large part of the success of WLANs can be attributed to
the implementation of several simple yet fully distributed
algorithms in dealing with the fundamental challenges for
wireless communications, including transmissions on a
shared medium and lossy wireless channel conditions. For
example, the Distributed Coordination Function (DCF), a
Carrier-Sensed Multiple Access with Collision Avoidance

(CSMA/CA) based Medium Access Control (MAC),
mandates the stations to first check if the medium is idle
before transmit packet. The Collision Avoidance mechanism
further regulates stations to backoff for a random amount of
time before transmission attempts, such that the chance of
collisions is probabilistically low [1]. In addition, enabled by
different levels of complexity and redundancy in signal
modulation and coding schemes, the IEEE 802.11 standard
employs multiple data rates to combat the volatile nature of
wireless channel. IEEE 802.11-based stations implement rate
adaptation algorithm (RAA) to dynamically select the best
transmission rate that yields the highest performance in any
given wireless channel conditions.
The key challenges are that RAA must not only
accurately estimate the channel condition in order to infer the
most suitable data rate, but also be very responsive to the
rapidly fluctuating wireless channel dynamics. Several
approaches have been proposed [2]-[9], including the use of
received signal strength, local Acks, and packet statistics to
design a RAA that addresses the above challenges. The
effectiveness of RAAs has been extensively evaluated under
various wireless channel conditions, when there is only one
station in the network. On the other hand, in multiple-user
environment, several studies [10][11] reported that some
types of RAAs' performance, e.g. Automatic Rate Fallback
(ARF)[2], can degrade drastically. It is because, as ARF
lowers its data rate whenever consecutive frame losses occur,
collision losses introduced by contention-based IEEE-802.11
DCF can mislead ARF to think the wireless channel has
deteriorated causing it to unnecessarily lower its rate and
resulting in performance degradation.
Based on this observation, there have been a few
attempts to aid rate adaptation algorithms in dealing with the
collision effect. The key idea is to provide RAAs the ability
to differentiate between wireless losses and collision losses
such that RAAs can resume their normal functionality by
filtering only wireless losses into rate decision process. On
the one hand, [11] suggests exploiting the Request-To-
Send/Clear-To-Send (RTS/CTS) exchange to differentiate
collision and channel errors. With RTS/CTS exchanges
preceding data transmissions, the RAAs are no longer
affected by the collision effect by assuming the only cause
for the data frame transmission failure after a successful
RTS/CTS exchange is due to channel error not collision.
CARA [10] further proposes to selectively turn on RTS/CTS
in order to save the extra RTS/CTS overhead. On the other

hand, without RTS/CTS frames, [11][12] add extra frames
and fields to explicitly notify the sending station whether the
transmission failure is due to collision or channel errors.
While these proposals provide significant improvements
compared to RAAs without loss differentiation capability, it
is unclear whether loss differentiation is good enough to deal
with all kinds of mixed wireless and collision loss scenarios.
The fundamental problem is, as we will show later in this
paper, that background :i,, i;. from other contending stations
changes the throughput ranking of the operating data rates.
In other words, under the same wireless condition, the data
rate yields the highest throughput in no background traffic
scenarios is not necessarily the best one in background traffic
scenarios. As loss differentiation schemes filter out all
collision losses for RAA, the RAAs become insensitive to
the throughput ranking changes caused jointly by wireless
losses and collision losses, causing performance degradation.
In this paper, we design a new Background traffic aware
Rate Adaptation Algorithm (BEWARE) that explicitly
addresses the mixed effects from wireless and collision
losses. Our contributions of this paper are: i. we
systematically evaluate the performance of RTS-based loss
differentiation in different mixed wireless and collision
losses scenarios. We identify when and why RTS loss
differentiation does not work in certain scenarios, ii. we use
the insight in these systematic evaluations to identify a key
parameter the expected packet transmission time to
explicitly address the mixed effects from wireless and
collision losses on all available data rates. We propose an
online algorithm to estimate this parameter of all data rates
and embed this information into the RAA design as the key
rate decision maker, and iii. we compare the performance of
BEWARE with other RAAs with and without loss
differentiation, and observe up to 250% and 25%
performance improvement, respectively.
The rest of the paper is organized as follows. Section II
reviews the existing RAAs and related loss differentiation
approaches. Section III evaluates the performance of existing
RAAs and loss differentiation schemes. Section IV presents
the design of our background traffic aware rate adaptation
algorithm, and Section V evaluates its performance under
various background traffic scenarios. Section VI concludes.


In this section, we briefly review the existing rate
adaptation algorithms (RAAs) and related loss differentiation
schemes that help RAAs deal with collisions in multiple-user
environment. We discuss pros and cons of each approach.
A. Existing Rate Adaptation Algorithms

As the 802.11 standard intentionally leaves the rate
adaptation algorithms open to vendors' implementation,
there have been quite a few RAAs proposed by academia
and industry. They can be broadly classified into three
categories based on the information they collect for rate
selection decisions: 1) statistics based RAAs, 2) received
signal strength (RSS) based RAAs, and 3) hybrid RAAs.
1) Statistics-based rate adaptation algorithms

Statistics-based RAAs collects frame transmission
statistics such as number of retries, number of frame success
and failures. These statistics are further processed and
compared for different rates or pre-set thresholds to infer for
current wireless channel conditions. Based on the statistics
the RAA uses for rate decisions, we can further categorize
this class of RAAs into three different approaches. i) Retry-
based rate adaptation: This approach [2][3] uses number of
transmission successes/losses as the indicator of good/bad
wireless condition, and increase/decrease data rate
accordingly. For example, ARF [2] decreases the data rate
upon two consecutive transmission losses and increases data
rate after ten consecutive transmission successes. However,
despite its easy design, previous study [5] has shown that,
due to randomness of the wireless loss behavior, there is very
weak correlation between past consecutive transmission
successes/losses and future channel condition, and
consequently this approach tends to yield pessimistic rate
estimations. ii) Frame-Error-Rate(FER)-based rate
adaptation: this approach [4][5] calculates FER by the ratio
of the number of received ACK frames to the number of
transmitted frames. The RAA decreases and increases the
operation data rate if FER exceeds some pre-determined
thresholds. The major drawback is the pre-determined FER
thresholds. As wireless channels are so vulnerable to many
factors such as multipath, channel fading, and obstructions, it
is difficult for one set of pre-determined FER thresholds to
fit in all circumstances. iii) Throughput-based rate
adaptation: This approach [6] calculates each data-rate's
throughput based on the packet length, bit-rate, and the
number of retries collected during a predefined decision
window (-1 sec). The major drawback of this approach is the
excessive length of the decision window. As the decision
window has to be large enough to collect meaningful
statistics, it causes the rate adaptation algorithm to be less
responsive to sudden wireless condition changes.
2) Signal-strength-based rate adaptation algorithms

This class of RAAs [7][8] relies on wireless signal
strength information, such as Received Signal Strength
Indicator (RSSI) or Signal-to-Noise Ratio (SNR), to make
the rate adjustment decisions. They assume a strong
correlation between received signal information and the
delivery probability of a data rate. The RAAs pick the data
rate based on a pre-determined mapping between the
received signal strength and throughput. Meanwhile, there
are two approaches to overcome the communication issue of
piggybacking the signal strength measurement taken at the
receiver side to sender so that sender can adjust the data rate
accordingly. One has to either use explicit signaling [8],
which is incompatible to the IEEE 802.11 standard, or
assume the channel is symmetry [7], which is clearly not the
case in real-world scenarios, and thus of little practical value.
In addition, this class of RAAs suffers from other
drawbacks. Firstly, the rate adjustment mechanism requires a
priori channel model to map the received signal information
to corresponding data rate throughput. In reality, such
mapping is highly variable and a model established before-
hand may not be applicable to any environments later.

Secondly, it is not trivial to obtain reliable signal strength
estimation from the radio interfaces.
3) Hybrid rate adaptation algorithms.

The hybrid RAA [9] collects both frame transmission
statistics and received signal strength, and use statistics-
based controller as the core rate adaptation engine. The rate
decision can be overridden by signal strength based
controller if it detects a sudden changes in received wireless
signal strength. As hybrid RAA design still assumes
symmetric wireless channel and pre-established RSSI-to-rate
thresholds, this approach is not immune from the drawbacks
we discussed in signal-strength-based RAAs section.
In summary, all types of RAAs strive to obtain accurate
channel estimations from different kinds of loss
characteristics and decide when to decrease and when to
increase the rate. However, in multiple-user environment,
packet collisions incur new sources of frame losses. None of
these RAAs explicitly address this issue. In the next section,
we review several proposals that try to aid RAAs in dealing
with collision effects.
B. Loss 1;7i i. ,ii,,i;. i, for rate adaptation

Previous studies reported that ARF's performance
degrades drastically when operating with mixed frame losses
from wireless noise and contention collisions. Because ARF
treats collision losses no different than wireless losses, ARF
excessively decreases its rate upon contention collisions,
even when wireless channel is close to perfect. This "rate
poisoning" effect results in severe performance degradation.
There have been two approaches to aid rate adaptation
algorithms in differentiating wireless losses from collision
losses. i) Loss 1;f I,'.iI. ',I by RTS/CTS: [10] and [11]
suggest to exploit the RTS/CTS exchange to differentiate
collision and channel errors. With RTS/CTS exchanges
preceding data transmissions, RTS-based loss differentiation
assumes the only cause for the data frame transmission
failure after a successful RTS/CTS exchange is due to
channel error not collision. Therefore, RAA rate decision
process reacts only on wireless losses filtered by RTS/CTS,
and RAAs are no longer affected by the collision effect. Kim
et al. propose Collision-Aware Rate Adaptation (CARA), to
reduce the extra RTS/CTS overhead by selectively turning
on RTS/CTS after data frame transmissions fail at least once
without RTS/CTS. ii) Loss 1;fi( .. ',,; ii. by explicit
notification: [11] and [12] propose to add extra frames and
fields to explicitly notify the sending station of the source of
losses. However, both proposals require changes to the IEEE
802.11 standard and are not compatible with existing 802.11
compliant devices, thus they are not favorable for real-world
In summary, loss differentiation is the dominating
approach for RAAs dealing with collision effects when there
are other stations transmitting traffic in the network.
However, it is not clear whether loss differentiation is
sufficient to guide RAAs to perform well in various
multiple-user environments with mixed wireless and
contention conditions. As we will show later in the paper,
while RTS-based loss differentiation works in certain
circumstances, we also found other scenarios that RTS-based

loss differentiation performs poorly, especially when it
operates independently with other RAAs or fixed rate
background traffic.


In this section, we first explain briefly how does IEEE
802.11 rate adaptation work. In particular, we analyze how
rate selection objective varies with the level of background
traffic. Furthermore, we systematically evaluate the
performance of various RAAs with RTS loss differentiation
schemes under different scenarios. We explore the
interactions between RAAs and mixed wireless and collision
effects by varying the number of stations in the network and
the distance between stations and access point. Although
there have been a few studies [10][11] evaluated ARF's
performance in multiple-user environment, to our best
knowledge, there is no comprehensive study on how other
popular RAAs perform when there is background traffic
present in the network. As we will show in this section, it is
critical to examine how and why these RAAs do not perform
well with background traffic. By such investigation, we not
only better understand the necessity for a RAA that does take
background traffic into consideration, but also gain insight
into how to design such a RAA.
A. IEEE 802.11 Rate Adaptation and Background T,,,i(i.

We use simulations to illustrate how multiple data rates
employed by IEEE 802.11 standard address the tradeoff
between throughput and transmission range. In Fig. 1, we
simulate the throughput of 8 data rates of IEEE 802.1 la in an
indoor fading channel [13] as the distance between the
mobile station and access point increases. We can see that
higher data rates can achieve higher throughput, but their
transmission ranges are shorter. With such different
characteristics among different available data rates, the
standard then relies on rate adaptation mechanism to
dynamically select the best data rate according to the current
wireless channel condition experienced by the link. Ideally,
the data rates selected by a rate adaptation mechanism that
has perfect knowledge of the current network condition,
follow closely with the outer envelop (plotted as thick solid
line) of Fig 1. In this way, the throughput yielded by the rate
adaptation mechanism is always maximized given a
particular channel condition. We will refer to this outer
envelope concept as the best available strategy throughout
the paper.
On the other hand, Fig. 2 plots the performance of the
same data rate set under the same wireless channel condition,
but with 12 other stations transmitting saturated background
traffic in the network. Note that, not only the shape of
staircase like throughput-distance curves changes, but the
rates selected by the best available strategy also change for
the same location. It is because the data frames transmitted
by any data rate are subject to not only wireless losses but
also collision losses caused by medium contentions with
other stations. This combined effect changes the
performance ranking of data rates for a given location. Fig. 3
further illustrates this effect by plotting the rate selections by

... -.- FIX-54M
x FX54M -.. ... FIX-48M
25 FIX-36M
a / /x Best available strategy FIX-24M
a 20 FIXM FIX-18M
.... '. ... FIX-12M
j 15 ---+--- FIX-9M
SA--*-.---x- *- .'.. -.. FIX-6M
". -; o i"o Best
S10 -. .--9... ..... . .
I- +-.-+..-+ -.-.'-.'+. .- .. -.+.. 4...+.. + + .
FiX X'M 6M "''*tT
o- "- :::-- -
0 5 10 15 20 25 30 35 40 45
Distance (m)

Figure 1. Throughput versus distance for IEEE 802.11a data rates, no
background traffic
2 ----- FIX-54M
18 ...-- --. FIX-48M
1 6---.--- FIX-36M
6 Best available strategy FIX-24M
214 --.-- FIX-18M
1 F 2M "-- ...--. FIX-12M
2 P.'/ 1 4. .+. -k FIX-9M
S0 -- -. F-Best
o FIX 48M *
0- M, X .
Fix rs
02 *

5 10 15 20

25 30 35 40 45

Figure 2. Throughput versus distance for IEEE 802.11a data rates, with
12 background traffic stations

---# of BK STA=0

4 48

3 36

zo 24


0 5 10 15 20 25 30 35 40 45
Distance (m)

Figure 3. Best available data rate under different background traffic

best available strategy when operating with different number
of saturated background traffic stations and unsaturated
residential traffic benchmark scenario as specified in [14].
As we can see from Fig 3, the rate selected by best available
strategy varies widely with background traffic intensity. In
other words, the rate adaptation strategy that works well in
one background traffic scenario may not work in other
background traffic scenario, hence the rate adaptation
mechanism needs to explicitly address this phenomenon.
Previous study [15] identified that the core of RAAs is on
Ihei to decrease" and "when to increase" the transmission
rate. Here, we argue that the rate selection objectives in
terms of "where to decrease/increase to", which change with
background traffic intensity, are essential to RAA design in
multi-user environment. It is very criticalfor rate adaptation
designs to be aware of such changes and adjust its rate
selection strategy to accommodate such changes; otherwise
it will suffer from serious performance degradation.

B. Performance of RAAs in RTS Access Mode

Previous studies identified that the lack of ability in
differentiating between wireless losses and collision losses is
the main problem for ARF to suffer from rate poisoning in
background traffic scenarios. They reported the superior
performance of ARF with RTS on over that with RTS off.
However, those studies did not provide systematic
investigation into whether RAA with RTS really achieves
the optimal throughput and why it does or does not. Besides,
we convey detailed comparisons among different RAAs with
RTS on, which are also not offered by previous studies. We
include representable RAAs from all three classes of
statistics-based RAAs, i.e., ARF, ONOE [16], Sample-Rate
(SMPL) [6], and RRAA-basic [5], in addition to signal
strength based RBAR [8].
We simulate an infrastructure-based 802.11a network
with Ricean fading model [13]. We first place all stations at
2.5m away from the access point and turn on RTS for all
stations. We isolate the effects of RTS loss differentiation in
performance comparisons by enabling only one station with
RAA on, and other background traffic stations with fixed
data rate. When there is little wireless loss for the RAA-
enabled station, we observe from Fig 4 that all RAAs
perform almost the same as the best available strategy,
regardless of how many stations transmitting background
traffic in the network. We then move the RAA-enabled
station to 12.5m away from the access point, we can see
from Fig 5 that RAAs start to lose track from the best
available rate and even drop their throughput lower than that
is offered by the lowest data rate. To further explain such
scenario, we plot Fig 6 to illustrate rate selection breakdowns
of ARF, as an example, as distance to access point increases.
We can see that the rate selection of ARF remains almost the
same as number of background traffic stations increases.
This is because RTS isolates the wireless losses from
collision losses. As a result, RAA makes the rate decisions
solely on wireless losses, and RAAs become insensitive to
the throughput ranking changes, which are illustrated as the
dotted lines in Fig 6, caused jointly by wireless losses and
collision losses.
We further examine the rate selection of all statistics-
based RAAs with RTS-on, and find the same phenomenon
exists. It follows that turning on RTS misleads RAAs into
using rates only suitable for no-background-traffic in
scenarios with background traffic, where these rates are
not always suitable. As a result, RTS loss differentiation
only works well when the rate selections are similar for all
other background traffic scenarios. On the other hand, since
SNR-based RBAR makes rate decisions by signal strength,
the data rate selected by RBAR is always the same
regardless background traffic intensity. In other words,
although the exact cause of bad rate selections is different,
RBAR also performs badly in background traffic scenarios.

C. Performance of Collision-Aware Rate Adaptation

Kim et. al.[10] propose to adaptively turn on RTS-CTS
exchanges to reduce the extra overhead introduced by RTS
loss differentiation. The mechanism, as called CARA1 in

20 --FK6M
2( ----------- A RF

15 -- R-RAA

10 -RBAR

NO N2 N=5 N=8 N=12
Number of background traffic stations

Figure 4. Throughput comparison for RAA-
enabled station with RTS loss differentiation at
2.5m away from access point, with various
number of background traffic stations in RTS
access mode

N=0 N2 N=5 N=8 N=12
Number of background traffic stations

Figure 5. Throughput comparison for RAA-
enabled station with RTS loss differentiation at
12.5m away from access point, with various
number of background traffic stations in RTS
access mode

10 40 70 100 130 160
Distance (m)

Figure 6. Data rate selection for ARF with RTS/CTS
and Best available strategy, with various number of
background traffic stations in RTS access mode

[10], works in the following manner. By default, the data
frames are transmitted without RTS. When the consecutive
failure count reaches probe activation threshold (Pth), the
RTS-CTS exchange is activated. If the consecutive failure
count further reaches consecutive failure threshold (Nth), the
transmission data rate is decreased. The default values of Pth
and Nth are set as 1 and 2 in [10], respectively. The data rate
is increased as the consecutive success count reaches 10,
similar to ARF.1
In this section, we compare the performance of CARA
with ARF in basic access mode and the best available
strategy. As shown in Fig. 7, we vary the distance between
the RAA-enabled station and access point from 2.5m-45m,
and show the RAA station throughput when operating with
12 other stations transmitting background traffic at fixed
54Mbps data rate. We can see that, deviate from what
reported in [10], CARA does not always offer superior
performance over rate-poisoned ARF. In particular, when the
station is far away from AP (>25m), the most suitable rates
turn to be lower rates. In these cases, CARA's adaptive
RTS/CTS mechanism only adds overhead to packet
transmissions, and no longer functions as loss differentiator
for underlying RAA. On the other hand, when CARA does
outperform ARF, its performance is 15%-25% less than the
best available strategy. Itfollows that, while RTS-based loss
differentiation schemes help RAAs distinguish between
wireless and collision losses in some scenarios, they do not
perform well in some other scenarios.



2 ..... FIX54M
..---.- FIX-48M
8 FIX-36M
...-- FIX-12M

6 -
4, ',
2 ------
0 5 10 .15 20 25 30 35. 4 4 -
0 5 10 15 20 25 30 35 40 45

Distance (m)
Figure 7. Throughput comparison for ARF, CARA1, and Best (best
available strategy) with 12 background traffic stations in basic access mode

We do not consider the optional Channel Collision Assessment (CCA)
detection, which is called CARA-2 in [10], as CARA-2 only provides
marginal performance gain over CARA-1

In summary, by systematic evaluations on how different
RAAs perform with different operating modes in mixed
wireless and collision environment, we made the following
observations and conclusions: i) The best available strategy
varies ';. ,,.ni. ,mi\ with the level of background traffic. We
argue that any rate adaptation mechanism should be aware
of such change at the presence of background traffic, or it
will suffer from serious performance degradation, ii) We
show that none of the existing RAAs we investigated perform
well in every background :i,,.ii; scenario. iii) We see that,
even with RTS loss differentiation or CARA, there are also
situations where these mechanisms perform poorly. In fact,
in those cases, RTS loss differentiation or CARA hurt the
performance. With these valuable observations, we present a
new background traffic aware RAA design in the next


In this section, we present the design of BEWARE, a
Background traffic aWaAre RatE adaptation algorithm for
IEEE 802.11-based MAC. The goals to design such rate
selection strategy are two-fold: it has to be robust against any
degree of background traffic; meanwhile, it is also
responsive to random and even drastic wireless channel
The key idea for BEWARE is to incorporate both
wireless channel statistics and background traffic condition
indicators in accessing the effectiveness of each available
data rate. We use a mathematical model to calculate the
expected packet transmission time of each data rate, based on
wireless channel and background traffic feedback collected
from Physical (PHY) layer. The rate selection engine then
uses this metric to find the data rate that yields the highest
throughput in the given wireless channel and background
traffic condition. Furthermore, we tailor the rate selection
engine to constantly explore other data rates' performance
without hurting the overall performance. We also design a
fallback mechanism in case of sudden channel changes.
Next, we describe the mathematical model for expected
packet transmission time calculations in Section 4.1, then the
rate selection engine in Section 4.2.

selected from 0 to Wj (maximum number of backoff slots in
stage j), there are exactly k busy time slots and (n-k) idle
slots is,

T, .

by other nodes

with other nodes

Of the tagged nodes

Figure 8. Packet transmission and collision events during IEEE 802.11
MAC backoff

A. Packet transmission time estimation

The core of BEWARE design is the estimation for
expected packet transmission time of each data rate, with the
consideration of mixed effects from wireless channel
condition and collisions. In CSMA/CA-based 802.11 MAC,
the overall time duration required to complete a packet
transmission starts from the instant that a packet becomes the
head of the transmission queue and MAC layer starts
contention backoff process, to the instant that the packet
completes the backoff process by being either successfully
received or dropped because of maximum retry limit has
reached. Therefore, as shown in Fig. 8, we calculate
expected packet transmission time by carefully analyzing the
duration and occurring probability of different events take
place at backoff stages, as follows.
a) When the backoff timer decrements, the time slot is
either sensed as idle (for Ts,0o, the length of one time slot) or
as busy occupied by background traffic transmission (for
Tbusy, the average medium occupation time used by
background traffic transmissions). We define Pbusy be the
probability that, at a given time slot, the backoff timer is
frozen due to busy medium in carrier sensing. It follows that
the occurring probability of idle slot and busy slot is (1-
Pbusy) and Pbusy, respectively.
b) When the backoff timer expires (i.e. decrements to
zero), the attempt of packet transmission either fails (after
Tfai) or succeeds (after T,,,). We define Pfai to be the frame
error probability. It follows that the occurring probability of
packet failure and success is Pfai and (1- Pfai), respectively.
For the parameters required to model the above process,
we acquire all except T,10, which is specified in different
version of IEEE 802.11 standard, directly from monitoring
channel activity. Specifically, we determine Pfai by counting
the ratio of failed packet transmission attempts and total
packet transmission attempts. We also obtain Pbusy and Tbusy
by keeping track of the number and duration of experienced
collisions, respectively. On the other hand, Tail and T,,u are
directly determined by the operating data rate. Note that, in
practice, it may be difficult to obtain some of these
parameters accurately due to implementation complexity in
real devices. We can consider alternative approaches[17][18]
by using number of consecutive idle slots between two busy
slots to estimate Pbusy and Pfail.
Once these parameters are collected, we can construct a
mathematical model calculating the occurring probability for
combinations of all different backoff events throughout all
backoff stages. We first define the occurring probability
F _n that, in any single backoff stage j with backoff timer

Fkn-k = W +kPbusyk pbusn-k

, 0

Moreover, we know that any combination of number of
busy and idle slots can be a cumulative effect from
successive backoff stages. Therefore, we then define
k,n-k for probability of backoff counter being frozen (k-j)
times and idle (n-k) times that up to back off stage j (which
implies packet transmission failed times),
Snk= Fk ,nk, O k- n-k '

Si P S x Fi (2)
Sk,nk = Pfail mS,iX k-j-m,nk (2)

,for 1 i-O
,where m is the number of backoff stages specified in the
For stage 0, this term equals Equation 1. For stage greater
than zero (i.e. j=1,2,..,m), this term includes all possible
cases, from combination of previous stages) to the current
stage, which result in (n-k) idle slots, (k-j) busy slots, and j
failed transmission periods. In other words, the packet
transmission time when such combination happens can be
characterized by,

Tkn-k = (k j)*T T,u + j *Tf +(n k) To + T,,. (3)

We then use an intermediate term to consolidate the effects
from different backoff stages,

Tkf k (1 Pfi)*E(S( k *TJ Ik Sm-l1 *'m-1 (4)
k-k ,fail (S ,n-k kn-k ) + k,n-k k,n-k (4)

As a result, the expected packet transmission time are
derived as,

Tavg ZiTik,n-k
k=0 n=k

where N= (Wi-1).

Once the expected packet transmission time is obtained, it is
sent to the rate selection module for rate selection decisions.
While the accuracy of this model has been evaluated in our
previous work [19], as we will show in the next section,
such model can be very useful when integrated into a RAA
design in estimating the efficacy of data rates.
As we can see from the derivation in this section, average
packet transmission time is a function of several parameters
from the environment, i.e. Pbusy Pfail, and Tbusy. In the
presence of background traffic, these parameters should all
be considered when making rate selection decisions, as
opposed to concentrate on just wireless losses or the
differentiation of wireless losses and collision losses as it is

Backoff stage




Pbusy Tbusy(ms) p36
No background traffic 0 0 0.24
Background traffic scenario #1 0.1 0.25 0.14
Background traffic scenario #2 0.2 0.25 0.1
Background traffic scenario #3 0.1 0.5 0.105
Background traffic scenario #4 0.2 0.5 0.07

believed in previous studies. We further illustrate this point
by the following example. Consider two adjacent data rates,
36Mbps and 24Mbps, available in IEEE 802.11a standard.
When switching from the higher rate to the lower rate, we
expect a turning point that the frame error rate change
from P, to P,, resulting in T,1 < T3 For the ease of
discussion, let us assume Pf, =0 for a given wireless
environment, and the rate selection algorithm search for the
right turning point P,, to switch from 36Mbps to 24Mbps.
Table I lists the turning point in different background traffic
As we can see from Table I, the turning points of
background traffic scenarios differ significantly from that of
no background traffic scenario. The turning point shifts
toward less lossy environment (smaller packet error rate)
when background traffic level, Pbusy and Tbuy, increases. This
observation has an implication in RAA's performance: if the
RAA does not act upon such turning point differences in
different rate selections for no background traffic scenario
and background traffic scenarios, just like the persistent rate
selection tendency of RTS-based loss differentiation
mechanism in Section 3, the rate decision most likely lead to
performance degradation when dealing with mixed loss
effects from background traffic and wireless channel losses.
Therefore, as we propose in this subsection, using a model
that considers both effects from background traffic and
wireless losses in rate selections is a better way to assist
RAA in making correct rate decisions.
B. Rate selection algorithm

In this section, we describe how BEWARE makes rate
selection decisions by using the average expected
transmission time derived from previous subsection.
BEWARE adopts a rate selection approach similar to [6].
Moreover, BEWARE adopts more careful measures in
probing data rates, and implements various schemes in
dealing with more dynamic packet transmission statistics
from the mixed wireless channel and background traffic
The rate selection module in BEWARE design can be
broken down with the following three tasks:
1) Rate probing: Periodically, BEWARE sends packets
at a data rate other than the current one to update the
expected transmission time of other data rates. BEWARE
adopts various measures to ensure probing other data rates is
not done very often and the cost is not too high. BEWARE
limits the frequency of packet probing to a fraction (-5%) of
the total transmission time. BEWARE also limits the

number of retries allowed for probing packets to 2 to save
costly waiting time for unsuccessful probing. In addition,
BEWARE does not probe data rates that suffer from
excessive failures for most recent packet attempts, and those
whose expected transmission time with no background
traffic already exceed the expected transmission time of
current operating data rate.
2) Information processing: After the packet
transmission completes, the packet transmission statistics
and expected packet transmission time are fed into rate
selection module for processing. In order to keep track the
dynamics of expected packet transmission time for different
data rate, BEWARE uses exponentially weighted moving
average (EWMA) to update the expected packet
transmission time with previous samples under same
background traffic level. For transmission environment
statistics, including Tbusy, Pbusy, and Pfai, BEWARE also uses
EWMA to smooth out the biases to the sudden changes in
current wireless channel and collision conditions. In
addition, BEWARE keeps track other statistics such as
number of successful/failed packets of different data rates.
3) Rate selection decisions: The rate selection module
constantly compares the expected packet transmission time
of current data rate and that of others, and decides to change
operating data rate whenever it finds a data rate yields the
shorter transmission time (and thus highest throughput).
BEWARE also implements a short-term frame loss reaction
mechanism in case wireless channel conditions change too
rapidly. The rate selection module forces data rate changes
when the packets exhaust all retries for three times


In this section, we use ns-2 [20] to evaluate the
performance of BEWARE and other RTS-based loss
differentiation RAAs, including ARF with RTS/CTS (as
referred to ARF-RTS) and CARA-1 under various mixed
wireless and background traffic scenarios.
A. Simulation setup

We enhance the ns-2 simulator to support 802.11a
Physical layer (PHY) and Ricean fading model [13]. By
default, we set the Ricean distribution parameter, K=6, and
Doppler spread fm = 17Hz (resulted from environment
maximum velocity v=lm/s) for performance comparisons.
Later in this section, we vary K and f, to investigate the
fading effects on the RAA performance. We simulate
scenarios in an infrastructure-based network, which contains
one Access Point (AP) and a number of static wireless
stations spreading in the network. The traffic sources are
UDP flows, and we use saturated traffic as recent IETF
measurement studies [21] has shown that highly congested
environments represent realistic scenarios.
B. Performance of single station with varying distance

We first focus on RAAs' performance with varying
distance under background traffic scenarios. We place 2-12
stations on a circle around the AP within 2 meter radius, and
all stations transmit UDP background traffic with RTS

? 1 5 .---CARA1
\ -, ARF-RTS

F0 5

0 5 10 15 20 25 30 35 40 45
Distance (m)

Figure 9. Throughput comparison for Best (best available strategy),
BEWARE, CARA1, and ARF with RTS/CTS, with 12 background traffic
stations in RTS access mode

5 10
Number of contending stations

Figure 10. Aggregate throughput comparison for BEWARE, CARA1, ARF
with RTS, and ARF in close-by topology with various number of
contending stations

--e- CARA-1

$- - 5 --... .... ...-- -- ----------:: --

Number of contending stations

Figure 11. Aggregate throughput comparison for BEWARE, CARA1, ARF
with RTS, and ARF in random topology with various number of
contending stations

access mode. The transmission data rate of background
traffic stations is locked at 54Mbps because of very close
proximity to AP. We then add one RAA-enable station in the
network and measure the RAA's performance by varying the
distance between RAA-enable station and AP. We show
results with 12 stations transmitting background traffic as an
example in Fig. 9, while results with other number of
background traffic stations show similar trend. In all cases,
the performance of BEWARE follows closely to what is
offered by best available strategy by only 10% less in
throughput, and the performance of CARA-1 trails behind
BEWARE by another 10%-15%. On the other hand, the
performance of ARF-RTS significantly derails from the best

available strategy when the distance from station to AP is
close by to moderate (2.5m-35m). It is because, in this
range, the rate selections for no background traffic deviate
significantly from the rate selections for this background
traffic scenario. As we discussed in Section 3, ARF with
RTS loss differentiation suffers from performance
degradations by continuing to use the rate selections only
suitable for no background traffic.

C. Aggregated performance with varying number of
contending stations

We now evaluate aggregate performance when all
stations turn on RAA and operate with the same RAA
homogeneously. We first simulate a topology with minimum
wireless losses, in which various numbers of stations are
uniformly placed at 2.5m away from AP and each station
transmits fixed size 1500-byte long UDP traffic. As shown in
Fig. 10, ARF's aggregate performance degrades severely due
to the "rate poisoning" effect we discussed in Section 3. On
the other hand, with the help from RTS loss differentiation,
ARF-RTS performs well for any number of contending
stations. Furthermore, BEWARE and CARA-1 perform
closely and both outperform ARF-RTS in most cases, thanks
to the overhead reduction design in CARA-1 and accurate
background traffic effect estimation in BEWARE.
Secondly, we simulate a random topology with various
numbers of stations randomly scattering in the network with
maximum distance 45m away from AP to guarantee no
hidden terminals. Each station transmits UDP traffic with
random size. As shown in Fig. 11, the performance ranking
differs from what we observe in Fig. 10. While ARF still
suffers from rate poisoning and performs the worst, CARA-1
no longer outperforms ARF-RTS and ranks second from the
worst. It is because, as nodes spreading at different distance
to AP, both wireless loss and contention losses affect the
default without-RTS data frame transmissions, which cause
CARA stations decrease data rate over aggressively. On the
other hand, BEWARE still performs the best in random
topology. On average, BEWARE outperforms ARF by
200%-250% and ARF-RTS, the best proposed by previous
studies, by 20%-25% in aggregate performance.

D. Aggregated performance under various channelfading

We now compare the performance of different RAAs
under various channel fading conditions. We vary the Ricean
parameter K and Doppler spread f,i. Note that, as K
increases, the line-of-sight component is stronger and the
overall channel SNR increases. On the other hand, as f,
increases, the channel condition changes more rapidly. Fig
12 plots the aggregate performance of different RAAs under
different K in a random topology similar to what we used in
previous sub-section. We can see that, as K increases, the
overall throughput all RAAs increases as expected.
However, the ranking of RAA performance remains
unchanged. BEWARE outperforms ARF-RTS, CARA-1,
and ARF under all different K parameters we studied. We
then plot Fig 13 with the aggregate performance of different
RAAs under different Doppler spread. We can see that, as f,
decreases, BEWARE still outperforms ARF-RTS in most

5 -'-- BEWARE
--"-i-- ARF
5 -----.......------------ .............. .....

(a) K=O


Number of contending statons

(b) K=6

Number of contending stalons

(c) K=12

Figure 12. Aggregate throughput comparison for BEWARE, CARA1, ARF with RTS, and ARF in random topology under different Ricean Parameter K


o 5 10o
Num bet of contendn- g station

CA & 1
Num contedngsaton ARF

...... ....

0 5 10 15
Num ber f contending stations

(a)fm=3.5HZ (v= 0.2 m/s)

(b)fm=10HZ (v= 0.6 m/s)

(c)fm=17HZ (v= 1 m/s)

Figure 13. Aggregate throughput comparison for BEWARE, CARA1, ARF with RTS, and ARF in random topology under different Doppler Spread f,

cases, but the performance gap between BEWARE and
ARF-RTS closes. To be more specific, as BEWARE
outperforms ARF-RTS by 25% when f, =17Hz, this
advantage decreases to 5% when f, decreases to 3.5Hz.
Previous studies [12][15] reported that, as ARF is designed
to increase its rate after several consecutive packet successes,
ARF-based RAA tends to yield higher throughput by taking
advantage of the slower changing channel environment.
However, the performance of ARF degrades when the
wireless channel condition changes rapidly. On the other
hand, we can see that, as BEWARE yields comparable
performance in different f, environments, BEWARE is
robust to both fast-changing and slow-changing wireless
channel conditions.

E. Performance with heterogeneous RAA deployments

As rate adaptation is an option that is left open for
wireless card vendors to implement, it is not uncommon that
there are stations equipped with different RAAs in real world
scenarios. Therefore, it is essential to evaluate the
performance of different RAAs in heterogeneous scenarios.
In this experiment, we evaluate how different RAAs improve
the individual and aggregate performance with a gradual
upgrade deployment. We consider a network with 12 stations
randomly placed within the transmission range of the AP,
and transmit UDP traffic with random size. By default, all
stations operate with ARE without RTS/CTS, which is
considered the baseline scenario. We then gradually upgrade
a number of stations with BEWARE or ARF-RTS, and
evaluate the aggregate performance improvement over
baseline scenario and individual performance improvement

of the same station after upgrade. We can see from Fig. 12
that, as the aggregate performance of ARF-RTS improves
when upgraded stations added to the network, the individual
performance of ARF-RTS actually decreases when less than
half of the stations in the network are upgraded. When there
are just a few stations upgraded with ARF-RTS, individual
performance of upgraded stations decrease due to excessive
use of higher data rates as we discuss in Section V-B.
Meanwhile, aggregate performance increases as other
stations take advantage of the excess loss transmission
opportunities incurred by upgraded stations. On the other
hand, when there are more and more stations upgraded with
ARF-RTS, ARF-RTS stations mutually take advantage of
other upgraded stations' loss transmission opportunities, and
collectively result in higher aggregate throughput even the
rate selections made by these stations are not the most
suitable ones for the corresponding scenario. By contrast,
both individual and aggregate performance of BEWARE
start to improve when just 1 station is upgraded. In addition,
as the stations upgraded with BEWARE start to use data
rates higher than what is used before upgrade, other stations
benefit from the extra free transmission time spared by
BEWARE stations, and thus yields higher throughput even
they are not upgraded with BEWARE. Note that this is an
essential feature that, when incorporating any new algorithm
to interoperate with other existing algorithms, the new
algorithm should not hurt the performance of other existing
In summary, with the homogeneous and heterogeneous
background traffic scenarios we evaluate in this section, we
observe that, while the effectiveness of RTS-based loss

-- -CARA-

... ARF

Number of contending stations

Number of contending stations




0 o/


-- ARFRTS, individia
- -AFFRTS, aggregate
BEWARE, individual
-- BEWARE, aggregate

\4 6 8 10

Number of upgraded stations

Figure 14. Individual and Aggregate throughput improvement of
BEWARE and ARF-RTS with various number of contending stations
in heterogeneous deployments

differentiation RAAs differ in different scenarios, BEWARE
always yields the best performance for most cases. In
addition, even with only one station equipped with
BEWARE in the network, both individual performance of
BEWARE and aggregate network performance improve over
the rate-poisoned all-ARF network.


In this paper, we first identify that data rate selection
strategies of 802.11-based stations should accommodate the
different rate selection criterions in different background
traffic scenarios. This observation further helps us explain
why RTS-based loss differentiation schemes, which are
proposed by previous studies to aid rate adaptation
algorithms in dealing with collision effects, do not perform
well in certain scenarios. In particular, RTS-based loss
differentiation hurts the performance by persistently using
the same rate selections regardless of background traffic
level. Therefore, these observations motivate us to design a
rate adaptation algorithm that explicitly addresses wireless
and contention factors in its design.
We propose a novel background traffic-aware rate
adaptation, BEWARE, that uses an accurate mathematical
model to estimate the effectiveness of the data rates in given
wireless and contention conditions. We show that the rate
selections of BEWARE are close to what are selected by the
best available strategy that has global knowledge of network
conditions. We also show that, compare to other RTS-based
loss differentiation schemes, BEWARE yields the best
performance in scenarios we investigated in the paper.
As a work-in-progress, we are working on implementing
BEWARE into the real 802.11a wireless card driver. The
results of real-world experimentation and related materials
will be updated in authors' website [22]. Meanwhile, we also
plan to investigate the interactions between rate adaptation
algorithms and upper-layer protocols such as TCP. We
believe that, as the design of BEWARE fully addresses the
wireless and contention factors in MAC layer, it should
render the best performance when integrated with upper-
layer protocols.


[1] G. Bianchi, "Performance Analysis of the IEEE 802.11 Distributed
Coordination Function," IEEE JSAC, vol. 18, no. 3, March 2000.
[2] A. Kamerman and L. Monteban, "WaveLAN-II: A High-Performance
Wireless LAN for the Unlicensed Band," Bell Labs Technical
Journal, pp. 118-133, Summer 1997
[3] M. Lacage, M. H. Manshaei, T. Turletti, "IEEE 802.11 rate
adaptation: a practical approach," ACM MSWiM '04
[4] B.E. Braswell and J.C. McEachen, "Modeling Data Rate Agility in
the IEEE 802.1 a WLAN Protocol," In OPNETWORK 2001.
[5] S. H.Y. Wong, H. Yang, S. Lu, V. Bharghavan, "Robust Rate
Adaptation in 802.11 Wireless Networks," ACM MOBICOM 2006.
[6] J. Bicket. Bit-rate Selection in Wireless Networks. MIT Master's
Thesis, 2005.
[7] J. P. Pavon and S. Choi, "Link Adaptation Strategy for IEEE 802.11
WLAN via Received Signal Strength Measurement," in IEEE ICC'03
[8] Gavin Holland, Nitin Vaidya, and Paramvir Bahl, "A Rate-Adaptive
MAC Protocol for Multi-Hop Wireless Networks," Acm MobiComOl
[9] I. Hatatcherev, K. Langendoen, R. Lagendijk, H. Sips, "Hybrid rate
control for IEEE 802.11," ACM MobiWac '04
[10] J.S. Kim, S.K. Kim, S.H. Choi, D. Qiao, "CARA: collision-aware rate
adaptation for IEEE 802.11 WLANs," in: IEEE INFOCOM'06.
[11] Q. Pang, V.C.M. Leung, and S.C. Liew, "A Rate Adaptation
Algorithm for IEEE 802.11 WLANs Based on MAC-Layer Loss
Differentiation," in Proc. IEEE BRAODNETS '05.
[12] J.H. Yun and S.W. Seo, "Novel collision detection scheme and its
applications for IEEE 802.11 wireless LANs," Computer
Communications Volume 30, Issue 6, March 2007.
[13] R. J. Punnoose, P. V. Nikitin, and D. D. Stancil, "Efficient Simulation
of Ricean Fading within a Packet Simulator," IEEE VTC'00-Fall.
[14] A. P. Stephens et al., "IEEE P802.11 Wireless LANs Usage Models,"
IEEE 802.11-03/802r23, May 2004
[15] D. Qiao and S. Choi, "Fast-Responsive Link Adaptation for IEEE
802.11 WLANs," in Proc. IEEE ICC'05.
[16] MADWIFI,, online link.
[17] M. Heusse, F. Rousseau, R. Guillier, and A. Duda, "Idle Sense: An
Optimal Access Method for High Throughput and Fairness in Rate
Diverse Wireless LANs," in Proc. ACM SIGCOMM'05, Aug. 2005
[18] C. Hu and J. Hou, "A Novel Approach to Contention Control in IEEE
802.11 e-Operated WLANs", in Proc. IEEE INFOCOM '07.
[19] S.C. Wang and A. Helmy, "Performance Limits and Analysis of
Contention-based IEEE 802.11 MAC," in Proc. IEEE LCN '06
[20] "The Network Simulator -ns-2,"
[21] A. P. Jardosh, K. N. Ramachandran, K. C. Almeroth, and E. M.
Belding-Royer "Understanding congestion in IEEE 802.11b wireless
networks," IMC '05
[22] ,online link.


University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - - mvs