A „Wireless LAN‟ is a collection of nodes in the Local Area Network that
communicate by using high frequency radio waves. There has been an increasing trend
to deploy Wireless LAN as an enterprise solution for mobility, seamless, anywhere or
anytime connectivity. Demand to provide Voice and Data services over the Wireless
Network are gaining momentum rapidly. Moreover, „Wireless LAN‟ (WLAN) and „Voice
over IP‟ (VoIP), both the technologies are maturing very fast. These two technologies
are hardly meant for each other but with the advent of new technology and standards
they can work well together. However, there are various challenges and obstacles while
providing VoIP over Wireless Networks. One of the major issues is to provide Quality of
Service (QoS). Due to the unpredictable nature of the wireless environment, providing
QoS service is a challenging task. Voice traffic is more susceptible to delay and noise in
the wireless domain than data traffic. However, other issues with providing Voice
services over Wireless LAN are Number of simultaneous users being supported, power
management of the handset and security issues etc. This research paper investigates
some of the problems and their solutions, to enable Voice over Wireless LAN
(VoWLAN). In order to ensure Quality of Service (QoS), Voice application requires a
dedicated bandwidth with several time constraints to avoid jitter and delay in the
received signal. This research paper will discuss the feasibility of using Legacy IEEE
802.11 Protocol as wireless standard for voice application. IEEE just proposed an
extension to the established Wireless 802.11x standards, as a draft IEEE 802.11e to
ensure QoS over Wireless LAN. This paper covers the new MAC protocol and how it
fulfills the goal of better Quality of Service, a mandatory requirement for voice
Introduction: Voice over IP and Wireless LAN
Wireless Networking has brought some phenomenal changes to the
communications world. It has changed the way we interact, do business and even the
way we live our lives. It has provided users with the ability to roam freely and still share
useful data and information among themselves. Wireless Local Area Network, also
called Wireless LAN or WLAN is a technology in which a mobile user can connect to the
local area network (LAN) through a wireless connection. IEEE introduced standards
IEEE 802.11 or Wi-Fi, which defines specific technologies for Wireless LANs. IEEE
802.11b is one of the most popular amongst those. There has been an increasing trend
in deploying wireless LAN as a commercial and enterprise solution. It‟s been predicted
that the market for Wireless LAN will keep on growing on a larger scale and it will be
worth billion of dollars by the end of year 2005. Wireless Networks (IEEE 802.11) are
broadly classified into two categories: Infrastructure Networks and Independent
Networks . In an infrastructure network, the mobile stations communicate with a
centralized access point (AP) to connect to the outside network. Also APs are used as
channel to communicate between the nodes in the same network. On the other hand,
Independent (Ad-Hoc) Network is different; it does not require any centralized access
point (AP) to communicate with other nodes. All nodes form a mesh network and
communicate with each other. Wireless LAN has some obvious advantages, which are
the driving forces for the deployment of the wireless equipments on such a large scale.
The most obvious advantage of wireless networking is mobility. Wireless Users can
connect to an existing network through an access point and are allowed to roam freely.
Other advantages are:
Cost: Wireless LANs can cost less to implement than Wired LANs in some
situation where the cost of laying down wires is relatively high.
Portability: A Wireless LAN is portable to any physical location. It is now a more
preferable communication network for enterprise.
Scalability: A Wireless LAN can be configured to small and large offices, with a
peer-to-peer network or with a backbone wired LAN.
Voice IP is a revolutionary technology; this technology is used to send voice packets
over the packet switching network rather than dedicated voice transmission lines. Some
advantages are the low cost per call, the lower infrastructure cost, and ability to
implement various new features. VoIP running over the Wired Local Area Network (LAN)
has very well established itself in the enterprise. The obvious benefit of using the same
medium for both data and voice is widely accepted. Enterprises are excited about the
idea of using their established wireless network for voice services just as wireless
network is used for other mobile applications. 
The VoWLAN has many applications and advantages over the Voice over Wired
Network. One is mobility again, the ability for an employee of an enterprise to remain in
touch while moving within the enterprise Local Area Network. Also using an existing
Institute of Electrical and Electronics Engineers (IEEE) 802.11 standard for wireless
network for both voice and data traffic can reduce overall cost of implementation and
maintenance. Calls from the VoWLAN private enterprise network can be connected to
outside world by using a public switched telephone network (PSTN) providing an
excellent solution for connectivity.
Other Applications like medical facilities are beginning to implement Voice over
IP as a possible solution to communicate with their mobile employees and to keep them
connected with each other. The use of mobile phones is prohibited in some of the
hospital areas as they have the potential to interfere with the medical equipments. The
use of 802.11 WLAN provides excellent way of communication without interfering with
the equipments normal working. 
Rest of the paper is organized as follows. Section 2 will discusses some of the specific
requirements for Voice support over any IP Network. Section 3 presents various
problems to enable Voice over IP over Wireless Local Area Network. Section 4
overviews PHY and MAC Layer of Legacy IEEE 802.11 and new IEEE 802.11e protocol.
Section 5 shows simulation results of 802.11e MAC protocol also compares
performance with legacy 802.11.Section 6 presents some of the limitations and future
work. Section 7 concludes the paper
2: Requirements: Voice over IP
Voice Traffic is not the same as data traffic. It has some stringent requirements
regardless of whether the voice is sent over wired network or wireless network. VoIP has
already established itself as an attractive, inexpensive and reliable solution on wired
Network for enterprise model, but has some basic requirements irrespective of its
The VoIP packets are small in size, uniform and transmitted at a constant
interval for the standard G.711 encoding scheme. The data rate requirement of
64kbps for G.711 standard should be met in order to provide quality of service.
 Hence, a dedicated bandwidth is required to provide effective service for
voice traffic. As number of user‟s increases this constraint can become harder to
Quality of Service for Voice Traffic is also dependent on packet losses, common
in wireless domain. Too large a packet loss may degrade the service quality to a
possible unacceptable value.
Latency and Jitter (variation in delay) are other pertinent issues with voice traffic.
Some latency and Jitter can be accommodated, but excessive delay and jitter
may degrade the voice quality. The delay should not be more than 150 ms to get
acceptable voice quality 
VoIP is based on Real Time Transport Protocol (RTP) with User Datagram
Protocol (UDP), an unreliable service. Packet losses due to burst error or
network transmission error are not reported back. Packet loss should be below
certain level in order to get acceptable voice quality.
These VoIP requirements are usually met in case of the Wired Local Area Network. As a
TCP/IP based scheme cannot guarantee reliability which is essential in case of VoIP
traffic, it just makes “best effort” to do so. TOS field in IP protocol describes the type of
service being used. High priority field is given low value of field and vice versa. Also
Priority Queuing is done in case of Wired LAN to meet stringent requirement of QoS
(Quality of Service). 
As discussed earlier Voice over Wireless LAN (VoWLAN) does provide excellent
opportunities and advantages but it adds additional areas of concern while using
Wireless LAN as a medium for the voice traffic.
3: Issues: Voice over Wireless LAN
The current lack of QoS standard raised questions about the practicability of using Voice
traffic over Wireless LAN. The general issues with using a wireless network for voice
traffic include the following:
Bandwidth Availability for voice traffic: Prioritizing the traffic is not provided in
802.11b wireless standard. PCF (Point Coordination Function) does ensure
some priority of traffic but is not effective in providing good QoS. IEEE 802.11a
and g provides 54 Mbps while 802.11b (most common) run at maximum
throughput of 11Mbps. Wireless Medium is shared by all contending stations
which creates bottleneck on the availability of bandwidth. Also IEEE 802.11
protocol uses Link adaptation schemes and lowers its throughput to as low as
1Mbps by changing to underlying modulation techniques like QPSK, BPSK etc.
Hence the dedicated bandwidth to use for all stations further goes down.
Latency: Again, in order to ensure QoS, the latency and delay due to network
congestion or traffic should be reduced to a minimal value to get acceptable
voice quality. Usually this value should be less that 150ms . On a lightly loaded
AP, they may not be significant traffic and contention; thus this stringent
requirement may be met. But as the numbers of stations accessing through the
AP increases, this latency requirement cannot be met hence demeaning
acceptable voice quality. No priority queuing or prioritizing of traffic is done in
present IEEE 802.11x protocols   
Jitter: Another issue over Wireless medium is jitter, caused due to the delay
between packets arrival, which is caused due to contention. Different station
traffic may collide with the voice transmission resulting in unpredictable delay
between packets. If the arrival time between packets is long, the receiving device
usually tries to handle this delay by introducing packet buffering. Packets are
buffered, assembled and finally played back. This buffering of the packets
causes additional fixed latency. To make matters worse if there is loss of
packets, the voice received will be of unacceptable quality. 
Security Issues: One of the major issue while using Wireless medium as the
transmission technology is security and privacy. Sniffing of packets in wireless
domain is much easier than in wired. Security issues in VoWLAN should be
addressed effectively. Security in wireless environment can be ensured by IEEE
802.11i standard and providing WEP (Wired Equivalent Protocol). This research
paper has not investigated this issue but creating secure environment is an
important requirement for VoWLAN.
Battery Life: Battery life is a requirement for any wireless device, so that they
can operate effectively for a longer period of time. IEEE 802.11b does provide
some propriety scheme to decrease battery usage, IEEE 802.11e has provided
an enhancement to save battery life. Automated Power Save Delivery (APSD) is
enhancement to improve the battery life. The basic idea is that the mobile station
must sleep for certain beacon intervals, while the voice session is not in
progress. The Access Point (AP) should buffer all the data for station in the mean
time. Station should indicate to the AP, when it wakes up. APSD is targeted
towards small battery operated devices which can switch off their radio and
hence improve battery life. This paper as focused on QoS for wireless Networks,
did not cover details on this issue. 
Capacity: Latency, delay and jitter provide an upper limit to the number of
simultaneous voice connection in any wireless communication systems. To
provide QoS, trade off has to be made to between number of simultaneous users
and availability of bandwidth.
4: Legacy IEEE 802.11
4.1: PHY Layer for Legacy IEEE 802.11
This research paper is focused on discussing MAC Layer issues and QoS support in
proposed IEEE 802.11e MAC protocol, which has to work with the Legacy IEEE 802.11
PHY Layer. This discussion will cover some of many issues to deal with, at the PHY
Layer when designing a system to support voice over wireless LAN.
Physical Layer in wireless domain is the bottleneck to the capacity and in many
cases to the efficiency of the system. Dynamic variation in channel conditions due to
increase in noise level, interference and path loss effect cause direct impact on packet
loss and overall though put of the system.
QoS is a major requirement and the above effects can cause direct impact on it.
Also as the number of users contending for the same resource increases, the data
requirement increases which causes collision and congestion in the system, decreasing
the overall effective capacity of the system for each station.
At PHY Layer, various modulation and coding schemes are used by station to
transmit data. Modulation schemes allow more bits per symbol, helps in increasing the
data rate; however symbols gets closer to each other ( in constellation diagram) and
small errors can led to erroneous decoding of the signal. Various coding schemes can
take care of some of errors. As the number of bits per symbol is decreased, the actual
data rate decreases which effects overall throughput of the system.
Also, besides increasing data throughput, Bit Error Rate (BER) should be kept
below a certain level to have desired QoS. Appropriate modulation scheme is required to
support Voice over WLAN. BER is dependent on the distance between the symbols in
IEEE 802.11 uses various modulation schemes like BPSK, QPSK. The plot
between Bit Error Rate (BER) and Signal to Noise ratio (S/N) provides insight into PHY
Layer parameters to be considered while designing a system. IEEE 802.11b used
DBPSK for 1Mbps data rate and DQPSK for 2Mbps. 
To design the VoWLAN system for office environment, some other parameters
like fading of the channel over a small distance and path loss are also to be considered.
Indoor Propagation proves effective in predicting path loss in such environments.
As shown in Figure 1: BER (Bit Error Rate) is plotted against Signal to Noise ratio
in the presence of AWGN (Additive White Gaussian Noise) and Rayleigh fading. Clearly,
for BER of 10 -6, the DQPSK (AWGN) requires far low S/N than DQPSK (Rayleigh). So,
the type of noise and fading has effect on channel design. Also, DBPSK (AWGN)
requires 12 dB of S/N than DQPSK (AWGN) which requires 13dB of Signal to Noise
ratio. But the data rate transmitted by DQPSK is double than DBPSK. However, symbols
in DQPSK are closer (in constellation diagram), more prone to errors in transmission.
For a particular kind of wireless channel and S/N, there is always a trade off between
data rate and Bit errors in transmission. Simulation is performed in MATLAB.
Figure 1: Bit Error Rate vs S/N
Also, PER (Packet Error Rate) is another way to look at errors in the system.
So, PER = 1 – (1 - BER)N
Where, N is number of bits in the packet.
4.2: MAC Layer for Legacy IEEE 802.11
DCF (Distribution Coordination Function):
IEEE 802.11b DCF provides a standard Ethernet like contention based service. If there
are multiple stations, DCF allows them to contend for the time slot and transmit in that
allocated time slot. In IEEE 802.11b, DCF is based on CSMA/CA (Carrier Sense Multiple
Access with Collision Avoidance). Each station first detects whether the medium is free
and then transmits; it is like listening before staring to talk to avoid collision with other
stations. DCF mode specifies two types of Inter Frame Spacing (IFS), namely
Distributed Inter frame Spacing (DIFS) and Short Inter frame Spacing (SIFS). In DCF
these two different kinds are used to prioritize traffic. SIFS is used for high priority traffic
like RTS/CTS and positive acknowledgments. High priority begins when the SIFS has
elapsed. DIFS is the minimum time station waits, if the medium is free it may transmit.
Figure 2 shows the relationships between various Inter frame Spacing. RTS/CTS are
sent to avoid hidden terminal problem, but it can still occur.
If two or more stations have a packet to send, each of them first senses the
medium for duration of DIFS and if medium is idle, one station may attempt to transmit
the packet. Otherwise, if station senses medium is busy it enters the back-off counter, in
which it chooses a randomly uniform back-off value from collection of values known as
contention window. A problem arises if two or more stations intend to transmit at the
same time, both stations will then back-off for random amount of time from a collection of
values. Also mentioned in the standard, the contention window size is always a value 1
less that power of 2 (eg 31, 63,255), there is a minimum and maximum limit to
Contention window 
A station having a back-off counter continues to listen to the medium until it finds
out the idle period of DIFS, it then decrements the counter after every time slot. When
this back-off timer becomes zero a station attempts to transmit the packet. After every
unsuccessful attempt to transmit packet, the size of the contention window is doubled
and station will back-off for that time before retrying. Undoubtedly, the station with the
lowest random back-off timer will get the chance to send the packet first through the
IEEE 802.11 DCF works very well, where numbers of stations contending for the
medium are relatively less. As the number of stations attempting for the medium
increases, this method for contending the medium can cause abundant delay, not
suitable for voice applications. Also, DCF mode does not provide any mechanism for
prioritizing traffic; it treats data and voice packets all “as same.” Latency and no
prioritization, which can cause future delay significantly degrades the voice quality to an
Figure 2: DCF (Distribution Coordination Function)
PCF (Point Coordination Function)
IEEE PCF was designed to provide near real time services. PCF is a contention free
service in which stations competing for the medium are allotted by the centralized
function known as point coordinator (PC), implemented in the AP. Each station is
selected by the method of polling and not by contending. Token based service protocols
are used in Ethernet. In PCF service tokens are replaced by polling mechanism 
As mentioned earlier PCF has higher priority that DCF as it has shorter interval
than DIFS; the PCF Inter frame space (PIFS) is used during PCF period. This duration is
larger that Short Inter Frame Space (SIFS). Time is divided into repeated periods called
super frames. In PCF, Super frame is formed by the combination of Contention Free
period (CFP) and Contention period (CP) alternate over time.
PCF is used during the CFP and DCF is use during CP. It is a requirement that
super frame includes a CP of some length, which at least allows one frame delivery with
acknowledgement during DCF.
Synchronization of the contending stations is done by beacon frame. The PC,
located at the AP, generates beacon frame at regular intervals; this time interval is called
as Target beacon transition time (TBTT) . Stations are polled in the PCF by PC and
send some data with CF-Poll to this station and waits for the acknowledgement. If
acknowledgement does not arrive in PIFS duration, the PC will poll the next station in
the polling list. PC keeps polling station until CFP expires. Again, there is no contention
in PCF as stations are polled. Polling list is updated during CP.
IEEE PCF is an optional service and not widely used, as many manufactures
claim that it inhibits interoperability with other access point in the network and does not
allocate better bandwidth than DCF 
4.3: MAC Layer for IEEE 802.11e
To support QoS over the Wireless LAN, IEEE defined the draft IEEE 802.11e, which
introduces EDCF (Enhanced Distributed Coordination Function) and HCF (Hybrid
Coordination Function). Stations supporting this new protocol may be able to support
Quality of Service over Wireless LAN. Hybrid Coordinator (HC) is again a centralized
coordinator similar to PC in case of Legacy IEEE 802.11. Each super frame is again
divided into two phases, CP and CFP which alternate over time continuously. The EDCF
is used in CP and HCF is used both phases. IEEE 802.11e MAC extension can be
applied on any IEEE 802.11.
Enhanced Distributed Coordination Function (EDCF)
IEEE 802.11e realizes the idea of providing QoS by introducing Traffic Categories (TC),
there can be eight different TCs for each station. These TCs are similar to those being
provided in Wired LAN. Each TC has a different queue, with in a station, having different
QoS parameters like Contention Window, persistence factor (PF) and also each queue
maintains its own back-off counter. These different queues act as virtual stations within
one station. 
In case of CP, each TC within the stations contends for Transmission
Opportunity (TXOP) and detects the channel to be idle for AIFS (Arbitration Inter Frame
Space) then starts the back-off counter. Each AIFS should be as large as DIFS, in case
of legacy 802.11, and can be increased for each TC . After waiting for AIFS interval,
each TC within each station can start the back-off counter. The value of back off counter
is selected from a uniform set of values of CW. The minimum and maximum size of the
CWmin and CWmax are defined. Priority over legacy 802.11 MAC can be provided by
setting CWmin less than legacy 802.11 CWmin (CWmin < 15 in case of 802.11a PHY)
. One of the important modifications in IEEE 802.11e is that each TC with in station
contends for TXOP, defined as duration of time when a station can initiate transmission.
Access to the TXOP is contended in case of EDCF and granted in HCF by HC.
Persistence Factor (PF) is defined for each TC within a station; this is one of the
QoS parameters for each TC. Every unsuccessful transmission will generate a new
value of Contention Window based on the value of PF [TC]. This new value will be
uniformly distributed. Unlike legacy CW is doubled after every unsuccessful attempt [PF
= 2]. Aim is to provide different CW for each queue in order to reduce the probability of
collision within station .
Figure 3: Enhanced Distributed Coordination Function (EDCF) 
Each queue contending for TXOP may also collide as the counter of two or more
parallel TCs reach zero at same instance, this kind of collision is called as Virtual
Collision. A scheduler grants TXOP to TC with the highest priority out of the TCs that is
virtually collided within the station.
Figure 4: Virtual backoff of eight traffic categories: (1) left one: legacy
DCF, close to EDCF with AIFS=34us, CWmin=15, PF=2; (2) right one:
EDCF with AIFS[TC]>=34us, CWmin[TC]=0-255, PF[TC]=1-16. 
Hybrid Coordination Function (HCF)
Hybrid Coordinator (HC), present in the AP, coordinates the allocation of TXOP
to each contending stations by polling. To give priority of HCF over EDCF, PIFS is set to
shorter duration than DIFS and AIFS. During the CFP, stations are polled on the basis of
the priority of traffic. Stations do not contend during CFP, as they are polled by HC.
During CP, each TXOP begins when the medium is contended by the EDCF rules i.e
when back-off plus the AIFS counter expires or when the station receive a special poll
frame, the QoS CF-Poll from the HC . Unlike legacy PCF, HC can issue QoS CF-Poll
during the CFP, so only HC can grant access to the medium. CF-end indicates the end
of the CFP announced by HC. HC can also use control contention, a way to learn which
station needs access to the medium and for which duration of time . Each station is
allowed to send request to HC for the allocation of the poll TXOPs, without contending
with each other.
Figure 5: Hybrid Coordination Function (HCF)
5: Simulation Results:
5.1: Simulation 1:
These simulations consist of three types of sources, but we are interested in Voice over
the Wireless LAN and background data traffic.
The model of the bursty data traffic is based on the Poission distribution, which
generates the data messages of different sizes and respective probability. The data
traffic generated is (64, 0.6), (128, 0.06), (265, 0.04), (512, 0.02), (1024, 0.25). 
The bit rate of 24 kbps is generated by an audio source with 60 byte messages
being generated periodically with gaps of 20 ms. The speech vocoder G.729A is used
with overhead of RTP/UDP/IP . Various other Network parameters for IEEE 802.11
and IEEE 802.11e for the simulation are set according to the standards.
Scheduling algorithm selection provides significant effect on the performance of
the EDCF and PCF . In EDCF, prioritization is achieved as each TC has a different
queue, within one station, with different QoS parameters like Contention Window, PF
and each queue maintains its own back-off counter. Thus the AIFS is for the High priority
traffic has a contention window with low value than for Low priority traffic.
Simulation results show the significant advantages of using IEEE 802.11e over
legacy IEEE 802.11. These simulation results also show that EDCF performs well with
out with out HCF.
Figure 6: Average Delay vs No. of busty Data Sessions
As discussed earlier in this research paper Legacy DCF cannot be able to
support Voice application, as it does not provide any prioritizing mechanism and has
delay due to contention. EDCF and PCF provide queues for different traffic categories.
Figure 6 shows that IEEE 802.11e performs well over Legacy IEEE 802.11 by keeping
the overall delay below 2.3 ms for VoIP traffic where as in PCF+DCF where average
delay is raised to 6ms with more than 9 busty traffic data sessions.  In case of DCF,
all real time traffic (voice or video) are not assigned any priority over data traffic, which
cause them to contend with busty data traffic during the CP and can only be transmitted
with some QoS guarantee within CFP. This causes significant delay to real time traffic.
Video Traffic, not focused in this research paper, has much worse performance,
when only DCF is used; the average delay peaks to 100 ms with in 7 data sessions.
However, as figure shows EDCF has outperformed DCF by keeping delay low. 
Also as the number of data sessions increases, the voice traffic has to contend
with the bursty data traffic during CP, which increases the probability of wait for voice
traffic during the whole CP. While in EDCF+HCF, the voice traffic is prioritized over data
traffic and HCF can also able to use „QoS CF-Poll’ on need basis, which causes
decrease in the delay for high priority traffic by able to fetch TXOPs at any time during
CP  But, on the other side, it can cause the starving of lower priority traffic for the
resource and hence increases the delay for lower priority traffic.
The maximum delay of the packet is another parameter by which performance of
IEEE 802.11e and Legacy IEEE 802.11 can be compared.
Figure 7: Max. Delay Vs. No. of bursty data sessions
IEEE 802.11e is always able to keep maximum delay below 13ms for VoIP traffic
irrespective of whether HCF is being used or not. EDCF performance is very
comparable with EDCF+HCF performance. But, there can be one major problem in only
using EDCF. If the station is greedy, then it can take advantage of the IEEE 802.11e
scheme and set its traffic parameters to the highest priority (Small AIFS and small back
off Contention Window). This can cause fairness problem while contending for the same
Network resource. But this can be avoided by fine tuning of the TC parameters for every
contending station. As can be observed, EDCF does relatively well with out with the use
On the other hand, in PCF+DCF operation, some VoIP packets maximum delay
can reach over 20ms within the first three data sessions, which can cause a significant
decrease in the QoS. 
5.2: Simulation 2:
In this scenario four stations are communicating with the Access Point, which is
connected to the backbone wire line network. Each station generates one type of traffic
with a different Traffic Category (Access Category). Priority of each AC is done by
assigning different Values of AIFS, CWmin, CWmax and TXOP limit as shown in the
Figure 8: EDCF parameters 
From Figure 8, the station with AC (3) has the highest priority because of the
lowest values of back off window (CWmin and CWmax). Five cases were run by
increasing the traffic rate from 100k to 500k. Also AC (3) generates a continuous traffic
to simulate voice traffic [packet size 220 bytes, rate= 88 k, ns2 as simulation
Figure 9: Average delay Vs. Network Load (IEEE 802.11) 
Figure 10: Average delay Vs. Network Load (IEEE 802.11e) 
Figures: 9 and 10 suggest that IEEE 802.11 does not distinguish between
different kinds of traffic, no priority to real time traffic. Hence the delay of all the ACs is
the same. As Observed, performance is improved with IEEE 802.11e as delay for high
priority traffic is less than 8 ms with a source rate of 500ms. This delay comparison will
become more and more prevalent as the data rate increases. Surely, IEEE 802.11e with
prioritization has out performed legacy 802.11 MAC protocol. On the other hand, the
delay for AC0 (Lowest Priority Traffic) has been increased to significant amount
compared to that of AC 3 (Highest Priority Traffic), which introduces fairness problem
as low priority traffic is easily starved by the high priority traffic. Again this will become
more prevalent as data traffic increases.
Figure 11: Throughput Vs. Network Load (IEEE 802.11)
Figure 12: Throughput Vs. Network Load (IEEE 802.11e) 
Above Figures: 11 and 12 show that for 100 kbps and 200 kbps, the difference
between the IEEE 802.11 and IEEE 802.11e is not significant, as it can be argued that
there are still enough resources available to support that much traffic. But as the traffic
load increases, the throughput of the Highest Priority traffic goes up in case of IEEE
802.11e protocol compared to IEEE 802.11. This gap will widen up more as the amount
of traffic increases. As expected, the throughput of Low priority traffic went down. Low
priority traffic (or station originating it) is being starved of network resources. This
problem can be avoided by properly tuning EDCF and HCF parameters in accordance
with delay requirement, which is dependent on the kind of application in use and the
physical layer. Also there is a need for some kind of control scheme to avoid fairness
issues when sharing Network resources.
Every solution or system has some limitations it is worth while to focus on some of the
limitations to design better approaches:
IEEE 802.11e draft proposed new MAC Layer protocol. This can ensure Voice
Services over wireless network by keeping latency low and increasing throughput
as shown in  . The simulation environment   includes only one
Access Point and stations are at fixed distance. Multiple access points are to be
considered in these simulations as other APs can cause channel interference,
particularly effecting Polling Beacons during the CFP.
In simulations  , stations are assumed to be fixed and not moving. In order
to provide voice services over Wireless Network, the mobility of the station/user
with respect to AP and also handling of voice call from one AP to another should
be considered in simulation environment. A richer set of test scenario including
these and other situations of interest to make simulation results more general.
Link Adaptation is usually done in IEEE 802.11 protocol due to increase or
decreases in the noise floor level. Simulations in   were run at a constant
rate, assuming environment is not changing. Link adaptation must be considered
in these simulations as these adaptations will then effect the necessary changes
required in the MAC layer parameters (Contention Window, PF etc). Also Link
adaptation causes changes in the capacity of the system which affects QoS.
Simulation results   show that Access mechanisms in IEEE 802.11e has
proved effective in ensuring priority of voice traffic. But with these mechanisms
lower priority traffic is easily starved for resources by higher priority traffic, which
causes fairness problem within the system. Need is for some kind of Access
control or fairness mechanism to ensure fairness and also the QoS for high
Effective tuning of station parameters for HCF+EDCF is required, especially in
EDCF alone which can have greedy station problem. Thus, EDCF parameter for
station generating real time traffic should be assigned by centralized mechanism
considering other station requirements.
Effective scheduling algorithms should be developed to ensure effective
implementation of polling mechanism.
Demand to provide voice and data services over the same network are ever increasing.
Providing voice over Wireless Local Area Network is a very challenging task. A draft of
IEEE 802.11e has been proposed which ensures QoS over Wireless Networks, thus
provides mechanism to enable voice services over unpredictable wireless network. This
paper has presented some simulation results on QoS support in Wireless LAN using
802.11e and compared its performance with Legacy IEEE 802.11 protocol.
Generally speaking from these simulation results in :
The Access mechanism in 802.11e EDCF ensures priority, has been proved
effective. Higher priority traffic has achieved more throughput and lower MAC
Like PCF, HCF is also a centralized polling mechanism controlled by AP, which
polls stations during the Contention Free Periods (CFP). Nevertheless, HCF is
provides better services to high priority traffic as unlike PCF , HCF can issue
special QoS CF-Poll during Contention Period (CP). This will provide
Transmission Opportunity (TXOP) to high priority Traffic thus keeping lower
transmission delay for real time high priority traffic.
An apparent problem is observed, during the high traffic conditions the low
priority traffic is easily starved by the high priority traffic . This causes fairness
problem in the system, as network resources are not properly shared. This can
increase delay in low priority traffic. Some kind of control mechanism to share
network resources has to be developed, which ensures proper sharing of
network resources. This is also important to ensure QoS for high priority voice
The author would like to acknowledge Dr. Richard Wolff, Professor in Electrical and
Computer Engineering Department for useful ideas and support.
 Priyank Garg, Rushabh Doshi, Russell Greene, Mary baker, Majid Malek, Xiaoyan Cheng
Using IEEE 802.11e MAC for QoS over Wireless.
 Farooq Anjum, Moncef Elaoud, David Famolari, Abhrajit Ghosh, Ravichander Vaidyanathan,
Ashutosh Dutta, Prathima Aggarwal, Toshikazu kodama, Yasuhiro Katsube Voice Performance in
WLAN Networks – An Experimental Study.
 Antonio Grilo, Mario Nunes Performance Evaluation of IEEE 802.11E.
 M. Coupechoux, V.Kumr and L. Brignol Voice over IEEE 802.11b Capacity.
 Stefan Manglod, Sunghyun Choi, Peter May, Ole Klein, Guido Hiertz, Lothar Stibor, IEEE
802.11e Wireless LAN for Quality of Service
 Technical Document: VoIP How to: http://www.tldp.org/HOWTO/VoIP-HOWTO-4.html#ss4.6
CommDesign Article: URL:
 Cisco Technical Document: 802.11b IP Issues URL:
 Deploying WLAN in Hospitals http://www.wi-fiplanet.com/tutorials/article.php/2246471
 Technical Document: G.711 http://www.spiritcorp.com/g711.html
Implementing QoS in WLAN URL:
 802.11 Wireless Networks Book, The definitive guide
 Technical Article: 802.11b Physical Layer revealed
 Dajaing He, Charles Q.Shen Simulation Study of 802.11e EDCF
 IEEE standards website: http://standards.ieee.org/wireless/