1. A brief review of an application of queuing theory for
the performance and QoS analysis of IEEE 802.11
Panth Shah1
Graduate Student of Computer Engineering
Illinois Institute of Technology
Chicago, Illinois, U.S.A.
pshah80@hawk.iit.edu
Abstract— IEEE 802.11 is a standardized protocol for the
development of Wireless LAN with MAC and PHY specification
which was a major breakthrough to develop a best effort service
model for wireless technology. Distributed Coordination function
(DCF) is the primary technique for medium access control
(MAC) of 802.11. For IEEE 802.11 to meet the QoS
requirements, it is very much important to analyze the
performance metrics such as throughput or delay in both
saturated and non-saturated case. A required improvisation in
performance can be obtained by reserving bandwidth, analyzing
number of nodes and packet size distribution, and packet arrival
pattern for IEEE 802.11 based WLAN. A comprehensive study is
been conducted in this research paper regarding the
performance analysis for IEEE 802.11 protocol by measuring the
throughput, mean frame delay, packet loss rate, queuing delay
and channel busyness. In this paper, I have also studied different
probability and queuing models and conducted a generalized
survey to compare those models to analyze the performance of
IEEE 802.11 WLAN for different cases. An analytical study is
been conducted to check the results supporting important QoS
requirements for IEEE 802.11 MAC and Physical layer protocol
for WLAN supporting multimedia application.
Index Terms—IEEE 802.11, queuing analysis, service time,
QoS, throughput analysis, DCF, RTS/CTS, wireless local area
network (WLAN)
INTRODUCTION:
In recent years, communication on wireless medium has
always been a field of research for various researchers to
develop a sustainable and compatible infrastructure which can
support a multi user access service using broadband
bandwidth capability with layered architecture. Cellular
phones, wireless internet access and home network based on
wireless technology are the practical development of proposed
Wireless communication model. By the time, a standard has
been set up by IEEE 802.11 for the development of Wireless
LAN which is IEEE 802.11. IEEE 802.11 protocol is
developed with the support specification of Multiple Access
Control (MAC) and Physical layer Infrastructure. For WLAN,
it is architectures conducting an Access Point (AP) in the
middle and multiple users to access the same AP. In IEEE
802.11, the mechanism used for multiple access to the medium
by multiple users is Distributed Coordinated Function (DCF)
which is based on CSMA/CA protocol.
As a major technique for multiple access in IEEE 802.11,
DCF is has been researched and analyzed by many researchers
to obtain the optimum performance in accessing channel by
avoiding packet collision and providing sufficient amount of
random back off time for the retransmission. Two way hand
shacking and four way hand shacking are the two techniques
for basic access mechanisms associated with DCF. In two way
hand shacking mechanism, an ACK is sent by the receiver to
the sender to give feedback for the received packet. While in
four way hand shacking mechanism, RTS/CTS method is
used. There is a reserved slot for the transmission of RTS/CTS
before packet transmission and after packet reception. But, for
this RTS/CTS mechanism Hidden Terminal problem is one
major performance issue.
For the analysis of the performance of IEEE 802.11 DCF
mechanism, Binachi [1] proposes a model based on Markov
chain model with the assumption of ideal channel and having
a finite number of terminals in the system. In addition to this,
analyze the behavior of IEEE 802.11 in saturation and non-
saturation condition is also an extensive study topic for many
researchers. Researchers Zhai, Chen and Fang [2] came up
with the study of maximum IEEE 802.11 protocol capacity for
DCF based WLAN under non-saturated case and resulted that
protocol supports strict QoS requirements under unsaturated
traffic condition. In [3], researchers have developed an
analytical model based on discrete time G/G/1 queuing to
check the performance of the IEEE 802.11 MAC protocol with
respect to delay and queue length. Here, each node in the
network is been studied and modeled as a G/G/1 queue to
determine number of connections in the case of delay and to
evaluate the efficiency of delay reduction techniques for the
given network in multiple access protocols. While evaluating
the performance of an IEEE 802.11 MAC layer, position of
nodes in reference to Access Point (AP) is also a metric to
calculate the saturation throughput of the medium. In [4],
researcher had developed a model to calculate the saturation
throughput conditioning the station position with respect to the
other stations. In this paper, assumption that all stations are to
be placed at the same distance taken by Bianchi in [1] to
compute the performance of IEEE 802.11 MAC layer using
discrete time Markov chain is extended.
In this paper, various IEEE 802.11 QoS schemes are studied
and demonstrated with their respective techniques used for the
analysis of the performance metrics of DCF based WLAN
developed based on IEEE 802.11 protocol. Main aim of this

2. paper is to provide a survey of the different performance
issues examined so far, and possible solution for those
problems. The intended purpose of this brief review is to find
out the most accurate probability and/or queuing model for
designing Wireless LAN system with IEEE 802.11 protocol by
taking all the performance and design issues into
consideration. Also, the new features added into IEEE
802.11e standard is also described and evaluated for its
performance [5].
REVIEW OF IEEE 802.11 STANDARD:
IEEE 802.11 is a protocol developed as a Wireless
Communication standard with optimized Physical Layer
(PHY) and Medium Access Control (MAC) sublayers. For
multiple channel access support, IEEE 802.11 protocol uses
DCF mechanism with CSMA/CA with complexity
consideration in wireless environment. While developing
wireless LAN using this protocol, station can't listen the
collision while transmitting. In this standard, two types of
network configuration modes [5] are provided: one is
infrastructural mode and the other on is ad-hoc mode. In
infrastructural mode, stations are connected with access point
for transmission and reception. Whereas, in ad-hoc mode
stations can communicate with each other without any Access
Point.
A. Distribution Coordination Function(DCF) in IEEE
802.11 standard:
In wireless communication system developed on this
standard, when a station is ready to transmit a new packet,
station checks whether the channel is idle or busy. This is very
much important to avoid the collision of packets after
transmission. DCF uses Carrier sense multiple access with
collision avoidance (CSMA/CA) technique with binary
exponential back off for collision free packet transmission
between stations. DCF defines two methods to access the
channel: two way hand shacking and four way hand shacking.
Before initiating a packet transmission, each station senses the
medium/channel and generates a binary exponential back off.
DCF interface space is a time interval defined as an interval
for which the medium is idle. A slotted back off time is
generated randomly from a contention window (CW) size
which is defined as Back off time = rand [0; CW] ⋅ slot time
[5]
. Contention window size changes according to each
successful or unsuccessful transmission of packets. Initially,
for the first transmission CW is set to minimum value and the
size doubled after each unsuccessful transmission. After
successful transmission, CW sets to minimum value.
By checking the current channel status, back off time is
been set up and only when back off time reaches to zero,
station can access the channel. To notify the successful
transmission, a positive acknowledgement (ACK) signal is
sent back from receiver to sender station. Shortest Interface
Space (SIFS) is defined as a time duration between data frame
and ACK received for that frame. This two way handshaking
technique for the packet transmission can be stated as a basic
access mechanism. Another method for reliable transmission
is to use four way handshaking scheme in which RTS/CTS
signals are used to send a transmission request from sender to
receiver and a clear transmission request from receiver to
sender. This is the way where a system performance can
significantly be improvised when there is a hidden terminal
issue. RTS and CTS are the frames which carries the
information of length of the packet which is transmitted.
Receiver than using this information about packet updates the
Network Allocation Vector (NAV) to determine the time for
which channel will be busy. RTS/CTS mechanism is efficient
for system containing large packets, as it reduces the length of
frames in the contention process. Also, in RTS/CTS collision
can only occur on the RTS frame and is detected when CTS
response lacks.
Figure 1. Operational diagram of IEEE 802.11[7]
Figure 2. Functionality of CSMA/CA and RTC/CT[8]
B. IEEE 802.11e EDCA:
The IEEE 802.11e uses the EDCA mechanism which is an
extension of mechanism DCF used by IEEE 802.11. The
extension is required to add QoS requirement into
conventional IEEE 802.11 standard. In this mechanism, a
concept of Access Category (AC) with parameters such as
inter frame space duration, minimum contention window size,
maximum contention window size and so on is presented
which is the category in which stations separates it's arrival
traffic into. Each Access Category is numbered from 0 to 3
and having its predefined parameters mentioned. Arbitration
Inter-Frame Space in EDCA is defined as SIFS + AIFSN(AC)
x Tslot
[6]
.
A trace of backoff counter is important for the packet
transmission as when backoff counter reaches to zero, a packet
is transmitted until and unless there no other packet from a
higher priority category is ready for the transmission. In this
case, if there is any lower priority packet ready to be
transmitted, it experiences a virtual collision with the higher

3. priority packet intended to get transmitted.
LITERATURE REVIEW:
Examination of the properties and performance of IEEE
802.11 has been researched and analyzed by different
researchers based on various performance metrics. There has
been a various models proposed to analyze the saturation
throughput and packet delay or packet loss for IEEE 802.11
DCF protocol.Authors and Affiliations
Performance analysis: To characterize the behavior of
802.11and analyzing the performance of the system based on
this protocol, Binchi [1] proposed a Markov chain based model
for the back off window size analysis. Binchi's model is very
simple and easily understandable using the concepts of
Queuing theorem. Packet transmission probability and
throughput is been observed and formulated by using the
stochastic process representing the back off time counter for
the given node. Back off time counter decreases at the
beginning of the each time slot and this decrement stopes when
the channel is sensed busy. The value of the back off time
counter for each node in the system depends on its transmission
history, the stochastic process is non-Markovian. Let s(t) be
the stochastic process representing the back off stage (0,….,m)
of the station at time t. But, the major approximation in the
given model is that, at each transmission regardless of the
number of retransmission suffered, each packet may collide
with constant probability of p which is also independent to the
probability of the other node. The assumption also made up
here is that every node/station in the system is at the same
distance from the Access Point (AP) in the wireless system.
In [4], a Distance Aware Model is proposed where the
assumption taken by Binchi[1] is been eliminated. The model
proposed here considers the interference from the other station
in WLAN and also packet loss is computed by the background
noise. This model is called as DAW (Distance Aware) model.
In this model, the distance dk is been taken into consideration
which is the distance of Station to Access Point. Packet loss for
this model is a subject to this distance denoted by pk(dk). For
this model, fixed and random topologies are taken into
consideration for the analytical purpose and to calculate the
throughput of the system. By this model, it is concluded that
the model achieved more realistic results compare to the model
which doesn't consider the condition where the station is
moving with reference to Access Point.
In [2], the authors have conducted a comprehensive study
to analyze the performance of the network with respect to
maximum protocol capacity, throughput, and delay and packet
loss rate. But in this paper, researchers have taken non-
saturated case into consideration apart from just a saturated
case for the network a maximum protocol capacity according
to this paper can only be achieved in non-saturated case. The
author in this paper has tried to simulate their study around the
non-saturated case and tune the 802.11 to work on the
maximum throughput, minimum loss rate and packet delay
rate. In this paper, authors assumed that the traffic is uniformly
distributed among the stations where the total number of
nodes taken here is n and also the transmission probability for
each node in any time slot is also given as pt. The probability
of back off time slot to be idle, probability of one successful
transmission and collision probability is explained as follows:
Figure 3. Markov Chain Model proposed for the back off
window size [1]
Here, the impact of fading channel is also discussed with
the assumption that the channel is perfect. The condition
where the faded channel can affect the performance of the
system can be observed from packet losses. In here, it is
considered that channel fading is not a serious problem in
WLAN which takes low node movement and stable channel
into consideration. If the channel is faded, than packet loss in
not only a subject of collision, but packet loss can also be
done because of channel fading. From the analytical model
presented here, the normalized throughput decreases with the
service time increases and mean and variation of delay
increases along with the service time increment. Also packet
loss rate is also increased. From the results simulated and
analyzed in this paper, network QoS parameters such as
throughput, delay and packet loss can be tuned to get the
optimum performance of the network. Also, using channel
busyness ratio network utilization can accurately be
monitored.
In [7], the author has discussed about different scenario
where IEEE 802.11 protocol shows serious performance
issues. The author has summarized different performance
issues extracted and possible solution for those issues. Also,
the classification of issues has been extended to help
designing MAC protocol for ad-hoc network apart from
WLAN. In this paper, author has classified the issues into
three main categories which are: Configuration with long term
fairness issue. Configuration with short-term fairness issue
and Configuration that results in overall throughput

4. decrement. At the end it is concluded that cross layer protocol
design is one of the best and effective solution for the
problems encountered into complex systems which isolated
MAC layer into its design.
ANALYTICAL MODEL PROPOSED FOR IEEE 802.11
DCF:
Back off counter value is a fair representation of a state of
IEEE 802.11 DCF based wireless network and the back off
stage the station is situated in currently. While using the
concept of queuing theory and properties of probability model
into consideration to observe the performance of DCF
mechanism, MAC buffer is taken as a queue and number of
packets currently present in the queue is taken as a
performance metric for the consideration of queuing theorem.
Here, we have modeled a behavior of typical wireless station
and also includes the effect of other stations on the
performance of that station at the time of packet transmission.
Here by comparing the entire WLAN system with customer-
server model to analyze the behavior of the system using
queuing theory. In this case, we are modeling the behavior of
every wireless station present into a wireless system where
back off stage is mapped on a queue and every single wireless
node is articulated as a single customer. This is the reason, the
proposed system is closed queuing network. While designing
this model on the current network system, two dimensional
array of queues are taken where each queue is modeled on
(k, i)th position in which i represents current back off stage
and k represents the number of packets currently in the MAC
buffer.
Figure 4. IEEE 802.11 DCF modeled on Queuing network
modeling [6]
As, there is only one customer in the system modeled on
queuing network, Queuing system is considered M/G/inf.
According to the model proposed, traffic equation for the
given queuing network is given as follows:
A. Model for Saturation Throughput and Packet Delay
analysis for IEEE 802.11 DCF with channel fading:
For WLAN developed using this protocol, station's
transmission probability for packet transmission can be
evaluated using the modified Markov chain model presented
here. This modified Markov chain includes the analysis of
back off window size for each station that takes frame-error
into consideration and maximum allowable number of
retransmission attempts are also be analyzed by taking each
station as a queue. The frame error rate has a significant
impact on the evaluation of throughput, mean frame delay and
discarding probability. For the analysis purpose, transmission
probability of each station is taken in a randomly chosen time
slot and so the Modified Markovian chain is analyzed as in
fig. 5. Moreover, this model is extended by taking error free
probability Pf into consideration. Finite number of
retransmission attempts (m + f + 1) which is the number after
which the frame is discarded from the transmission queue with
an addition of a new frame in the queue. A current state (i, k)
of a station is determined by the current value of the back off
timer k after which it will start suffering unsuccessful
transmission attempts.
From the analysis, it is observed that the saturation
throughput decreases with increment in number of stations in
basic access mode while others are stable with RTS/CTS
access mode. While comparing the ratio between the length of
the useful frames for transmission and RTS frames, the
difference is not major. This is because of the frame delay in
each station which is originated from the back off defer period
in the station queues significantly. Impact of frame error rate
over the performance of the protocol is observed using the
following result shown in fig. 6. The transmission probability,
saturation throughput, frame discard probability and mean
delay after analyzing this model is obtained as follows:
Transmission probability:

5. Saturation Throughput:
Frame Discard Probability:
Mean Delay:
Average duration of a renewal cycle Trc:
Figure 5. Fading channel analysis for IEEE 802.11 DCF using
finite-state station model [9]
While considering the frame-error rate Pf on the overall
performance measure of the system under fading channel
condition, this parameter will effect on saturation throughput,
delay rate and discard probability of the system/model
proposed here. In the result shown below, increasing frame-
error rate from 0.01 to 1, throughput degrades towards zero
and by increasing discard probability towards 100%.
Saturation mean delay also increases with the increment in
frame-error rate.
(a)
(b)
(c)
Figure 6. Frame-error rate impact over the performance: (a)
saturation throughput, (b) Delay, (c) Discard probability [9]

6. QOS ANALYSIS BASED ON PERFORMANCE PARAMETERS AND
MECHANISM:
While designing wireless links using given protocol, a
major challenge associated with the QoS of the system is the
performance of upper layer application. In conventional IEEE
802.11 protocol, there is an isolated MAC layer which can't
support the QoS specification of upper layer applications as
the analysis of upper layer application performance won't help
evaluating the QoS of MAC layer. The specified
characteristics observed for all the layers of this protocol are
high loss rates, burst or frame loss, high latency rate and jitter.
QoS mechanism for 802.11 can be classified into three
categories given as: Service differentiation, Admission control
and bandwidth reservation and Link adaption. In WLAN
environment, except relatively low data rate and higher error
caused due to RF characteristics like multi-path fading due to
channel fading, CSMA/CA proposed in IEEE 802.11 DCF
also trapped into a problem with different layers.
Though, there are different QoS mechanism applied to
every layer of the network, various ideas are applied based on
the queuing models to analyze the 4 major performance
parameters of the system for QoS.
A. Differentiation Serving Model System:
Different requirements of different application are served
using different layers from the proposed model. For IEEE
802.11, there are two methodologies used. 1. Priority Based
and 2. Fair Scheduling based service. This classification
mechanism replicates the traffics into various flows. So, each
flow can be handled according to specific requirement they
should be handled. Priority based service always serves those
traffics with the highest priority out of the total traffic.
Whereas, second service model schedules the bandwidth on
the basis of weight of each traffic flow. The specified service
differentiation mechanisms are [8]
:
1. Enhanced DCF (EDCF)
2. Persistent Factor DCF (P-DCF)
3. Distributed Weighted Fair Queue (DWFQ)
4. Distributed Fair Scheduling (DFS)
5. Distributed Deficit Round Robin (DDRR)
Figure 7. Differentiation Serving Model System for IEEE
802.11
B. QoS Mechanism for Admission Control (CAC):
To obtain optimized output under heavy traffic load
condition, Differentiation Serving Model is not a correct
choice. For any multimedia application in IEEE 802.11 to be
analyzed for its performance, saturation delay is a critical
parameter to be considered. When the saturation delay is
large, it leads to a failure to support multimedia application.
Also, bandwidth provisioning in contention based CSMA/CA
channel access mechanism is an impossible case. So, for this
case, admission control and bandwidth reservation is an
important considerations for guaranteed QoS By considering
broader approach of wireless communication, admission
control scheme is extended into measurement based and
calculation based scheme. A concept of delay sensitive traffic
is a critical methodology of observing WLAN traffic using
Queuing models developed. This option assumes that the
traffic entering into WLAN is pre shaped and is given a
deterministic bound because of the limited capacity of a
wireless medium. So to maintain QoS of a wireless system
based on IEEE 802.11 it is very much important to analyze the
traffic type and whether it may violate the QoS of the system
or not. For this decision to be taken, CAC algorithm [10] is
developed.
CONCLUSION AND FUTURE WORK ON AN OPEN ISSUE
So far, it has been surveyed in this paper about the models
researchers had proposed by the time for the performance
analysis and relational issues related to the performance of
IEEE 802.11 protocol under different considerations. In this
article, I have represented a Queuing theory based probability
analysis model for packet transmission and packet lost analysis
to find the maximum throughput of the system. Also, for both
saturated and non-saturated environment, a model based on
Markovian chain is been studied in this paper to find the
effectiveness and essentiality of packet loss rate to measure the
performance of the system using performance metrics such as
throughput, packet loss rate, packet arrival timing and so on.
Also, from the research obtained it has been devised that
RTS/CTS mechanism in channel access is superior in most of
the cases and while using this mechanism, performance is only
partially dependent on the system parameters.
Also, a hidden terminal problem associated with
CSMA/CA is also discussed in this article and how QoS
parameters are added into IEEE 802.11e, which is an
improvised version of IEEE 802.11. Discussion about the
choice of EDCF over DCF is also obtained to check how QoS
is important to analyze while working with multimedia
applications in IEEE 802.11 protocol.
It is also concluded from the study that isolated MAC
layer is not a correct choice as information from the lower
layer is important to analyze the performance of the upper
layers.
Also, it has been discussed in the paper that QoS is been
guaranteed in IEEE 802.11 WLANs by classifying QoS
approaches into two schemes: One is Differentiation Serving

7. Model System and the other one is QoS mechanism for
Admission Control (AC).
REFERENCES
[1] G. Bianchi, “Performance analysis of the IEEE 802.11
distributed coordination function,” IEEE J. Select. Areas
Commun., vol. 18, pp. 535-547,
Mar. 2000.
[2] H. Zhai, X. Chen, and Y. Fang, “How well can IEEE
802.11 wireless LAN support quality of service?”, IEEE
Trans. on Wireless Communications, Vol. 4, No. 6,
pp.3084-3094, Nov. 2005.
[3] O. Tickoo and B. Sikdar, “Queuing analysis and delay
mitigation in IEEE 802.11 random access MAC based
wireless networks,” in Proceedings of IEEE INFOCOM,
Hong Kong, China, March 2004.
[4] M. H. Mamshaei, G. R. Cantieni , C. Barakat, and T.
Turletti, “Performance Analysis of the IEEE 802.11 MAC
and Physical Layer Protocol,” in World of Wireless
Mobile and Multimedia Networks, 2005.
[5] Q. Ni, “Performance Analysis and Enhancements for
IEEE 802.11e Wireless Networks,” IEEE Network, vol.
19, no. 4, July 2005.
[6] E. Karamad, F. Ashtiani, “Performance analysis of IEEE
802.11 DCF and 802.11e EDCA based on queueing
networks,” in Communications, IET, vol. 3, issue 5, Feb.
2009.
[7] C. Chaudet, D. Dhoutant, I.G. Lassous, “Performance
Issues with IEEE 802.11 in Ad Hoc Networking,” in
Communication Magazine, IEEE, vol. 43, issue 7, July
2005.
[8] H. Zhu, M. Li, I. Chlamtac, B. Prabhakaran, “A Survey of
Quality of Service in IEEE 802.11 Network,” in Wireless
Communication, IEEE, vol. 11, issue 4, Aug. 2004.
[9] Z. Hadzi-Velkov, B. Spasenovski, “Saturation
Throughput - Delay Analysis of IEEE 802.11 DCF in
Fading Channel,” in Communications, 2003. ICC '03.
IEEE International Conference, pg. 11-15, vol. 1, May
2003
[10]J. R. Gallardo, P. Medina, W. Zhuang, “QoS Mechanisms
for the MAC Protocol of IEEE 802.11 WLANs,” in 2nd
International Conference of Quality of Service in
Heterogeneous Wired/Wireless Networks, 2005
.