Generic Coexistence Method between IEEE 802.16 and IEEE 802.11 based Wireless Networks

Full text

6th Karlsruhe Workshop on Software Radios
193
Generic Coexistence Method between IEEE 802.16
and IEEE 802.11 based Wireless Networks
Mohammad M. Siddique,
Bernd-Ludwig Wenning,
Carmelita G¨
org
Department of Communication Networks
University of Bremen
Germany
Email: [mms, wenn, cg]@comnets.uni-bremen.de
Maciej Muehleisen
Chair of Communication Networks
RWTH Aachen University
Germany
Email: mue@comnets.rwth-aachen.de
Abstract—Due to the high scarcity and high costs of radio
spectrum, more and more radio services are occupying unlicensed
bands for their operation. Due to this, there is a high risk of
destructive interference which degrades the performance and
fails to support Quality of Service (QoS) for systems operat-
ing in these bands. IEEE 802.11 based wireless networks are
already operating in unlicensed band. A new competitor for
unlicensed bands is the IEEE 802.16 based wireless metropolitan
area network. Therefore, spectrum sharing between coexisting
competing wireless systems like 802.11 and 802.16 is an upcoming
challenge. To understand the characteristics of interference in
such a heterogeneous scenario, an analysis of possible interference
is presented and the performance of the legacy systems is
evaluated. Then a spectrum sharing concept is proposed which
can generally be applied to both systems. In this paper, the
proposed concept is adapted for coexisting 802.16 and 802.11e
based systems, which is an extension of 802.11. In this case,
the 802.11e Hybrid Coordination Function (HCF) Controlled
Channel Access (HCCA) is extended to provide a protocol for
airtime sharing. Simulation results are presented showing that
the proposed algorithm provides excellent improvement of system
performance in the context of capacity and channel utilization
compared to the case without applying any spectrum sharing
method.
Index Terms—IEEE 802.11, IEEE 802.16, Coexistence.
I. INTRODUCTION
Wireless Local Area Networks (WLANs) are one of the
most popular and commercially successful wireless technolo-
gies that provide wireless connectivity for fixed, portable and
moving stations within a local area. The Institute of Electrical
and Electronics Engineers (IEEE) specified a standard for
WLAN which is known as IEEE 802.11 [1]. Various amend-
ments have been done on the base standard IEEE 802.11 by
extending the protocol to improve the performance in several
contexts. For example, to provide QoS Support, IEEE 802.11e
has been defined [3]. WLAN operates in unlicensed bands
like the Industrial, Scientific and Medical (ISM) band (in 2.4
GHz) and the Unlicensed National Information Infrastructure
(U-NII) band (in 5.0 GHz) depending on its Physical (PHY)
layer protocol. Another emerging wireless technology for
Broadband Wireless Metropolitan Area Networks (BWMAN)
is IEEE 802.16 [4], commercially known as Worldwide Inter-
operability for Microwave Access (WiMAX). During its early
stage, 802.16 technology is being seen as an alternative to the
wired Internet access like DSL, however, today it also supports
mobile nodes. The IEEE 802.16 systems can operate between
2 GHz and 11 GHz where licensed and unlicensed bands are
located.
Licensed spectrum is becoming more and more expensive.
So from the economical point of view, there is a high probabil-
ity in the future that 802.16 will operate in the same unlicensed
bands (e.g. ISM and U-NII bands) where 802.11 is operating
[5]. Aside from economical consideration, technical recom-
mendations considering the evolution of technologies are also
going to that direction. According to the recommendation
from e.g. the 4G draft [6], International Telecommunication
Union - Radiocommunication (ITU-R), the next generation
wireless networks will be an integration of different wireless
standards like WLANs, WiMAX and cellular networks. This
increases the probability that multiple different radio access
technologies will be present in the future, all or some of which
have to operate in the same (unlicensed) frequency band. In
such a case several possible coexistence scenarios could occur.
One example scenario which appears in apartments or office
buildings in dense urban areas can be defined as follows. An
IEEE 802.11 system starts using the same unlicensed channel
which is used by an IEEE 802.16 system or vice versa, because
an alternative channel is not available. This is denoted for
further reference as a heterogeneous networks coexistence
scenario. One reason of having no alternative channel is that
other channels may be used by other IEEE 802.11 systems or
by different technologies like Bluetooth or ZigBee which are
also operating in the same unlicensed band.
One of the main drawbacks of an unlicensed band is
unpredictable interference. If the systems are not managed
to use the spectrum properly, this interference leads to poor
performance. Destructive mutual interference between unco-
ordinated wireless systems severely decreases the spectral
efficiency and performance. Performance can be analyzed from
different perspectives, for example, in the overall scenario, on
a per system basis or on a per user basis. However, from the
user’s perspective, fair spectrum sharing and assured QoS (e.g.
throughput, delay, etc.) are required. So there is an increased

6th Karlsruhe Workshop on Software Radios
194
requirement to efficiently utilize the unlicensed spectrum
bands by means of spectrum sharing or coexistence methods.
The more systems operate within a mutual range, the more
they require methods for coexistence or even cooperation.
The objective of this paper is therefore twofold. First, an
analysis of possible interference occurring in a heterogeneous
coexistence scenario is shown and the system performance
considering the capacity in such a scenario is evaluated.
Secondly, a generic spectrum sharing algorithm by means of
Medium Access Control (MAC) layer scheduling which is de-
veloped in the framework of the ”Policy-based Spectrum Shar-
ing for unlicensed Mesh (PoSSuM)” project [7] is described
and applied to the same heterogeneous scenario and evaluated.
The evaluation has been done by a simulation environment
called Open Source Wireless Network Simulator (openWNS)
[8], [9]. In the framework of this paper a combined simulation
platform for simulating the heterogeneous scenario is modeled
and published as open source under Lesser General Public
License (LGPL) in [10].
For the rest of the paper, the unlicensed band in 5 GHz
and IEEE 802.11a [2], which is a supplement of 802.11
considering the operation in 5 GHz band are considered. It
is important to note that in the standard IEEE 802.11-2007,
all the amendmends like 802.11a, 802.11e are included.
A. Related Work
The IEEE standard draft 802.16h [11] proposes methods for
802.16 system coexistence. One approach is to use a specified
protocol allowing multiple systems to negotiate their resource
usage. In [12] Rapp evaluates the coexistence of HiperLAN/2
[13] systems. Scheduling policies creating Silent Periods as
transmission opportunities for other systems are introduced.
Results are derived by simulation. A similar coexistence
scheme is presented in [14] and [15]. In [14] a scheme is
presented and evaluated analytically on how this idle period
can be exploited by letting a second system fill the subframes
from the other time direction. A scheme that reschedules data
and allows multiple systems to coexist by shifting their frame
start is described and analyzed in [15]. In [16], a concept of
using a busy tone signal for protecting 802.16 transmission
in heterogeneous coexistence scenario is presented, but not
evaluated. The IEEE Standards Coordinating Committee 41
(SCC41) is currently working on enabling network coexistence
through dynamic spectrum access and cognitive approach.
B. Outline
The rest of the paper is structured as follows. Section II
provides background information about different channel ac-
cess methods in the case of 802.11 and 802.16, Section III
gives an analysis of possible interference in the heterogeneous
scenario. Section IV is about the spectrum sharing method and
its adaptation to coexisting 802.11 and 802.16 systems. The
simulation setup and results are provided in Section V. The
paper is concluded in Section VI.
Fig. 1. IEEE 802.11 channel access method: Distributed Coordination
Function (DCF) [1]
II. MEDIUM ACCESS CONTROL
A. IEEE 802.11
The IEEE 802.11 standard defines two channel access
mechanisms for WLAN namely the Distributed Coordination
Function (DCF) and the Point Coordination Function (PCF).
1) Distributed Coordination Function: The DCF is a con-
tention based random channel access scheme based on the Car-
rier Sense Multiple access/Collision Avoidance (CSMA/CA)
protocol. Fig. 1 shows the channel access method of 802.11
in DCF mode. Details of the DCF are given in the standard
[1] which describes the rules to follow when a MAC layer
Service Data Unit (MSDU) is arriving at the MAC layer from
a higher layer. The basic rules are as follows:
•(a) A Station which desires to initiate a transmission,
invokes the Carrier Sense mechanism to determine the
idle/busy state of the channel.
•(b) If the medium is busy, the station must wait for
the channel to become idle. It is referred as ’access
deferral’ according to the standard. In this case, the access
is deferred until the medium is determined to be idle
without interruption for a period of DCF Interframe Space
(DIFS).
•(c) If the channel is idle without interruption for a period
of DIFS, the station generates a random backoff period
for an additional deferral time before transmitting, unless
the backoff timer already contains a nonzero value. Only
if the channel remains idle for this additional random
time, the station initiates transmission.
2) Point Coordination Function: The PCF is a centrally
controlled channel access scheme based on a polling mech-
anism. PCF is an optional function and only working in
Infrastructure Mode. When the WLAN system is set up to
work in PCF, the channel access time is divided in periodic
intervals called superframes. The PCF operates in the first part
of the superframe, called the Contention Free Period (CFP).
DCF operates in the remaining part of the superframe called
Contention Period (CP).
It has been shown in literature that the DCF and the PCF
have limitations concerning the support of QoS [17]. This
motivates the development of 802.11e to provide user level
QoS. In 802.11e traffic flows are differentiated into categories
like voice, video, best effort, and background and they are

6th Karlsruhe Workshop on Software Radios
195
Fig. 2. IEEE 802.11e superframe structure
served according to their access priority. Wireless multimedia
extension (WME) [18] based on IEEE 802.11e is the commer-
cial version of WLANs with QoS support.
B. IEEE 802.11e and Hybrid Coordination Function
IEEE 802.11e [3] defines a coordination function called
Hybrid Coordination Function (HCF) to provide a QoS guar-
antee. The HCF includes two channel access mechanisms: the
Enhanced Distributed Channel Access (EDCA) and the HCF
Controlled Channel Access (HCCA). The EDCA is generally
an enhancement to DCF by introducing access categories and
priorities. It will not be further described as it is out of the
scope of this paper. The HCCA is an enhancement to the PCF.
Fig. 2 gives an overview of the IEEE 802.11e superframe
structure in the time domain. Each superframe starts with a
beacon frame.
1) HCF Controlled Channel Access: In HCCA the Hybrid
Coordinator (HC), which is located in the access point (AP),
has control over the channel. One main feature introduced
in HCF is the Transmission Opportunity (TXOP). A TXOP
specifies the duration of time in which a station can occupy
the medium uninterrupted and exchange multiple consecutive
frames with only Short Interframe Space (SIFS) spacing
between an acknowledgement (ACK) and the next data frame.
TXOP is defined by a starting time and a specified maximum
length. A station is granted a TXOP (called polled TXOP) by
the HC. The polled station (in the uplink case) is informed
about the allocated TXOP by a CF-Poll frame. Other stations
in the network set their network allocation vector (NAV)
according to the duration field of the CF-Poll frame. Another
special improvement in the HCCA is the contention free burst,
known as Contention Access Phase (CAP), which is initiated
during a Contention Period or a Contention Free Period. It
may span across multiple consecutive polled TXOPs. The HC
can start a CAP by sending a CF-Poll or a data frame (in the
case of uplink or downlink respectively) when the medium is
idle for more than a PCF Interframe space (PIFS) period.
C. IEEE 802.16
The IEEE 802.16 standard defines a centrally controlled
wireless communication protocol where the channel occu-
pation of IEEE 802.16 systems is fully controlled by the
scheduler in the Base Station (BS). Subscriber Stations (SSs)
associate with the BS forming a cell.
IEEE 802.16 supports Frequency Division Duplex (FDD)
and Time Division Duplex (TDD) operation but TDD is
mandatory for unlicensed operation [15]. The IEEE 802.16
system follows a periodic MAC frame as shown in Fig. 3.
If Time Division Duplex is used, each frame consists of a
downlink (DL) and an uplink (UL) subframe. Each frame
starts with a preamble followed by the Frame Control Header
(FCH). Besides general information about the system, the FCH
provides the first part of the so called Medium Access Pointer
(MAP). The MAP is formed by the scheduler at the beginning
of each frame deciding the exact structure of the current frame.
It therefore contains the information describing which node
should transmit or receive at which offset from frame start
and which Modulation and Coding Scheme (MCS) should be
used. For the downlink it inspects the queue and possible MCS
for each Subscriber Station and grants it an appropriate share
of the frame if possible. For the uplink the scheduler relies
on information from the SSs to estimate their demands. A SS
can register its traffic demands through bandwidth requests in
the Random Access phase. This Random Access phase is also
used for initial access of SSs. The scheduling algorithm is not
defined by the standard. It is common to fill the subframes
by Protocol Data Units (PDUs) in ascending time order. Idle
periods occur at the end of the downlink and uplink subframe
if they are not fully utilized. The Transmit Transition Gap
(TTG) is located between downlink and uplink subframes to
allow time for the BS to switch from transmit (Tx) to receive
(Rx) mode and SSs to switch from Rx to Tx mode; and the
Receive Transition Gap (RTG) is located between uplink and
downlink subframes to allow the other way round.
The Current standard version 802.16-2009 also supports
Orthogonal Frequency Division Multiple Access (OFDMA),
however in the following we only focus on Orthogonal Fre-
quency Division Multiplexing (OFDM) based systems.
III. INTERFERENCE IN HETEROGENEOUS SCENARIOS
In this section an analysis is presented which shows the
interference and collision events that could occur when IEEE
802.11 and IEEE 802.16 systems are collocated and operate
in the same unlicensed channel. It is assumed that all stations
are within mutual interference range. Fig. 4 shows the events
which occur during the channel access of 802.16 system and

6th Karlsruhe Workshop on Software Radios
196
   
    





   


 

        




















Fig. 3. IEEE 802.16 Time Division Duplex MAC Frame [14]
Fig. 4. Timing diagram for channel access by collocated IEEE 802.11 and 802.16 systems; possible interference and collisions events
legacy 802.11 DCF systems. The upper part of the figure
shows the 802.16 channel access which follows the rules
mentioned in section II-C and Fig. 3. The lower part shows
the 802.11 channel access which follows the rules mentioned
in section II-A1 and shown in Fig. 1.
The IEEE 802.16 system provides three idle periods: At
the end of the downlink subframe, the end of the uplink
subframe and during the Random Access Phase at the end
of the frame. The last two physically create one. Some major
events including collision events are identified in the figure and
numbered for further reference and explanations as follows.
Events like (4) and (5) are described in [16].
1) When the channel is busy due to transmissions of the
802.16 system, 802.11 channel access is deferred up to
when the channel is idle. The 802.11 system can take
control of the channel according to rule (b) mentioned
in section II-A1.
2) When the channel is idle and a MAC layer Service
Data Unit (MSDU) is available from a higher layer, the
802.11 station starts the frame exchange according to
rule (a) mentioned in section II-A1. As in the case of the
802.16 system, channel access is centrally scheduled, the
802.16 system starts its uplink subframe and occupies
the channel if a MSDU is available. Transmission is
not deferred if the channel is being used by the 802.11
system. This results in interference, and the probability
of losing interfered packets in both systems is very high.
3) The same as (1) happens during the uplink subframe of
802.16.
4) The same as (2) can happen when the next 802.16 frame
starts. It is even more critical because in this case the
preamble, FCH and MAP of the 802.16 frame would be
lost. Eventually, the whole frame will be lost.
5) When Subscriber Stations with no queued PDUs are not
accessing the channel though they are scheduled to do so
and the MAP is successfully transmitted, this generates
an idle period. The 802.11 system can take over the
control of the channel following the rule (b) which
would cause interference to following 802.16 PDUs.
It is important to note that the 802.11 system can take over the
control if required when the idle period duration is equal or
more than DIFS, depending on 802.11 system’s backoff state.
This characteristic, which is sometimes realized as ’selfish’
behavior of 802.11 DCF channel access method, is one of the
main problems in the context of collocated wireless networks,
when the other system is following tightly scheduled medium
access control like Time Division Multiple Access (TDMA)
and Time Division Duplex (TDD) as in IEEE 802.16. In such
a case, coexistence performance can be improved by one of
the following methods.

6th Karlsruhe Workshop on Software Radios
197
Fig. 5. Systems provide idle periods and shift their frame start to enable
coexistence
•Protecting the starting time of the scheduled frame ex-
change duration of 802.16 [16]: Some percentage of
capacity will be lost to send a protecting signal which
depends on the busy signal duration. It can be adapted
according to the estimated maximum length of 802.11
frames, based on measurements.
•Applying the Listen-Before-Talk (LBT) mechanism in the
case of the 802.16 system: In the 802.16h [11] standard
the LBT concept is introduced and applied before the
beginning of the superframe and if the channel is found
busy, no transmission will take place in the succeeding
frame. This enables the systems to reduce the number of
collisions, however, there is a challenge to keep the QoS
values like delay and jitter within the required levels. The
802.16h protocol is not further evaluated in this paper.
•Making the channel access of 802.11 more regular like
802.16 and adapting the periodicity, service starting point,
and service period length of both systems: This is the
main focus of this paper and introduced in the following
section IV.
IV. COEXISTENCE METHOD
We have taken into account three following assumptions for
this paper:
1) The systems have a method to estimate their own traffic
demand and the traffic demands of other systems,
2) the systems have a method to detect the beacons/FCHs
from other systems and shift their own frame starting
time referring to the beacons/FCHs and
3) systems follow a common periodic time interval to serve
their stations.
The first two methods, in the following called ’detection
methods’, can be developed by using the radio resource
measurement. For example in the case of 802.11, measurement
techniques on the basis of IEEE 802.11k can be applied to
identify the idle/busy periods and to develop the detection
methods. Measurement of spectrum utilization helps to esti-
mate the traffic demand of other systems [5]. A method of
estimating the number of systems in the vicinity is given
in [19]. The outcome of these detection methods can be
considered as input to the spectrum sharing algorithms for
decision making and scheduling. Detection methods and traffic
demand estimation are considered separate research topics
which are not in the focus of this paper, however, they are
ongoing work.
The following coexistence method allows systems to coexist
by multiplexing their channel access in the time domain, which
can be seen as ’TDMA between systems’ as shown in Fig. 5.
Each system leaves some capacity which can be used as
’spectrum opportunities’ or ’idle periods’ for other systems.
Distributing idle periods randomly results in collisions. Hence
one of the main features of this scheme is that it occupies the
channel and keeps idle periods in a regular pattern. Therefore,
we refer to the scheme as ’Regular Channel Access (RCA)’. It
helps, on the one hand, systems to reliably predict the length
of the idle periods and their offset in the superframe during
the detection phase; on the other hand, it helps systems to
utilize the idle periods for own transmissions causing less or
no collisions with each other due to orthogonality in time.
It is worth to mention that it is often experienced that the
capacity or bandwidth requirements of the applications used
in the systems are less than the channel capacity. In the case
of high bandwidth requirements, the system reserves some
capacity for the admission of other systems. The duration of
occupying the channel by the own system and the duration of
idle periods for other systems can be adapted by estimating
own and traffic demands of other systems. The time period or
the air time allocated to the own system is calculated to
Tallocown =T Down
T Down +T Dothers ×RI (1)
Tallocothers =T Dothers
T Down +T Dothers ×RI (2)
Here, TD (traffic demand) is defined as a ratio which is within
the range [0,1], where 1 means the demand is equal to the
channel capacity and RI means RCA Interval described below.
A. RCA in Heterogeneous Networks
Fig. 6 shows the RCA in the case of collocated 802.11 and
802.16 systems. Here, the 802.16 frame length is considered
as RCA Interval. The 802.16 system schedules downlink
(DL) and uplink (UL) subframes at the beginning of the
RCA Interval in such a way that there is only a TTG duration
gap between downlink and uplink. In such a case transmissions
during the downlink and uplink subframes can be viewed
logically as one continuous busy period. TTG is shorter than
DCF Interframe Space (DIFS). The rest of the airtime in the
superframe is kept as idle period for other systems. The air
time (resource) allocation for 802.16 (DL+UL) transmissions
inside the 802.16 frame can be dynamically adjusted consid-
ering the traffic load of the own system and others, using
equation (1) and (2).
To enable RCA in 802.11, an 802.11e based system, like
Wireless multimedia extension is considered. By introducing
a MAC scheduler in the HC of the 802.11 system, a regular
channel occupation and provision of idle periods is possible as
shown in the lower part of Fig. 6. The basic functions are as

6th Karlsruhe Workshop on Software Radios
198
Fig. 6. Timing diagram for channel access by collocated IEEE 802.11 and 802.16 system with Regular Channel Access (RCA) method; mitigate the
interference and collisions
follows: The interval between the start of two successive chan-
nel occupations by 802.11 system is realized as RCA Interval
which is configured to be equal to the 802.16 frame length. The
required air time allocation for 802.11 transmissions inside
RCA Interval can be calculated considering the equation (1)
and (2). To fill up the idle period left by the 802.16 system,
the 802.11e system schedules its PDUs with an offset from
the beginning of 802.16’s FCH. The offset is calculated to
T ime Shif tow n =T allocothers (3)
By this proposed approach, the operation of coexisting
systems is coordinated and synchronized indirectly with the
help of regular channel occupation on the one hand and with
the use of measurement techniques on the other hand. In the
following, evaluation results of 802.11 and 802.16 system
performance in the coexistence scenario are shown.
V. SIMULATION SETUP AND RESULTS
To understand the characteristics of the effects of inter-
ference and collisions in the systems, a scenario with one
Base station and one Subscriber Station for the 802.16 system
and one Access Point and one station for the 802.11 system
is considered. This also resembles the apartment scenario
mentioned before, where other orthogonal frequency channels
in the unlicensed band are occupied by other systems. The
system parameters are listed in Table I. Both systems are
using the most robust modulation scheme of binary phase shift
keying (BPSK) with a coding rate of 1/2, which can provide
6 Mbit/s of Data rate at the physical layer. This is considered
during simulation to show the fundamental operation of the
algorithm. Each Simulation is run for 500 seconds. In each
simulation run, the static offered traffic for both systems during
the span of simulation duration is considered.
The results are shown for three cases.
1) Case1: Legacy 802.11 and 802.16 systems.
2) Case2: Legacy 802.11 and RCA enabled 802.16 systems
3) Case3: RCA enabled 802.11 and 802.16 systems.
Throughput and delay are evaluated as performance metrics.
TABLE I
SYSTEM PARAMETERS
Common
Carrier Frequency 5.470 GHz
MCS BPSK 1/2
Bandwidth 20 MHz
802.16
Frame length 10 ms (720 Symbols)
1 Symbol duration 1/72 ms
Preamble+FCH 3 Symbols
MAP 4 Symbols
DL subframe 355 Symbols
TTG 2 Symbols
UL subframe 328 Symbols
Random Access 26 Symbols
RTG 2 Symbols
802.11
Slot Duration 9 μs
SIFS 16 μs
PIFS 25 μs
DIFS 36 μs
CWMin 15
CWMax 1023
ACK Duration 44 μs
802.11e
RCA Interval 10 ms
CF-Poll Duration 56 μs
QoS-Null Duration 56 μs
A. Throughput Evaluation
In this context throughput measured on top of the MAC
layer is considered.
1) Case1: Legacy 802.11 and 802.16 systems: Fig. 7 shows
the throughput averaged over 10 ms over time between 6 s to
7 s. The throughput averaged over 1 s is also shown. Note
that the throughput of 802.16 in the figure is the summation
of downlink and uplink throughput. Here, 2 Mbps (1 Mbps DL
+ 1 Mbps UL) traffic load and Protocol Data Units (PDUs)
of 375 bytes in the case of 802.16 system and 2 Mbps traffic
load and PDUs of 1480 bytes in the case of 802.11 system
are configured as offered traffic load. PDU Inter arrival times

6th Karlsruhe Workshop on Software Radios
199
6 6.1 6.2 6.3 6.4 6.5 6.6 6.7 6.8 6.9 7
0
1
2
3
4
5
6x 106
Simulation time / s
Throughput / bps
802.11 (10 ms average)
802.11 (1 s average)
802.16 (10 ms average)
802.16 (1 s average)
b
h
c
a
f
d
g
e
− legacy (DCF) in 802.11 and
legacy 802.16
− 2 Mbps 802.11 Traffic
− 2 Mbps 802.16 Traffic
Fig. 7. Throughput of legacy 802.11 and 802.16 against time
0 0.5 1 1.5 2 2.5 3 3.5 4
x 106
0
1
2
3
4
5
6x 106
Offered IEEE 802.11 Traffic / bps
Throughput / bps
802.16
802.11
− legacy (DCF) in 802.11 and
legacy 802.16
− 2 Mbps 802.16 Traffic
Fig. 8. Mean throughput of both systems in Case1 against 802.11 traffic
follow an exponential distribution. The 802.16 system uses a
periodic frame of 10 ms length as shown in Fig. 3.
IEEE 802.11 follows DCF, so it does not access the channel
when it senses it busy. Each simultaneous transmission causes
data loss. When all uplink and downlink PDUs of 802.16
collide with 802.11 PDUs, this results in zero throughput
(points (a) in the figure) for 802.16. The rest of the frame
airtime, if any left, is occupied by 802.11 for retransmission
of the lost PDUs first and more PDUs depending on traffic
load secondly; resulting in throughput like (b) or (c) for
802.11. Throughput like (d) or (e) for 802.16 and (f) or (g) for
802.11 correspond to the situation when some PDUs in either
downlink or uplink are not lost in the case of 802.16. Points
like (h) correspond to the situation when both downlink and
uplink are not lost. Due to the Poisson traffic source, different
random events are observed. However, the main noticeable
observation is the long time average which clearly shows the
huge difference between the offered and received traffic for
both systems. This is due to the high number of collisions.
Fig. 8 shows the mean throughput over offered 802.11
traffic. A stacked graph is used showing the throughput of each
system and each traffic direction as well as the total throughput
6 6.1 6.2 6.3 6.4 6.5 6.6 6.7 6.8 6.9 7
0
1
2
3
4
5
6x 106
Simulation time / s
Throughput / bps
802.11 (10 ms average)
802.11 (1 s average)
802.16 (10 ms average)
802.16 (1 s average)
− legacy (DCF) in 802.11 and
RCA model in 802.16
− 2 Mbps 802.11 Traffic
− 2 Mbps 802.16 Traffic
Fig. 9. Throughput of both systems in Case2 against time
of both systems. The 802.16 uplink and downlink throughputs
are decreasing significantly with increasing 802.11 traffic up
to 2 Mbps; because it increases the probability of collision
events like (2) and (4) mentioned in III and shown in Fig. 4.
However, the 802.11 throughput is not affected as lost packets
are retransmitted in the idle time period inside the current
frame. For offered IEEE 802.11 traffic of more than 2 Mbps,
the 802.11 throughput reaches saturation. The trend of the
curves of 802.11 and 802.16 shows one of the characteristics
of 802.11 which is ’selfishness’ during channel access. This
is harmful for other systems in the context of coexistence.
Overall, it has been found that around 2.5 Mbps capacity can
be achieved, in other words, 40% of the channel capacity can
be utilized.
These throughputs can be improved by physically joining
the downlink and uplink together as discussed in IV-A and
verified in the next results.
2) Case2: Legacy 802.11 and RCA enabled 802.16 systems:
Fig. 9 shows the same kind of performance metric like before;
however, in this Case only downlink traffic is configured in
the 802.16 system equal to the sum of downlink and uplink
(2Mbps) in Case1. The 1 s average shows clearly that the
difference between the offered and received traffic is reduced
and throughput is increased for both systems; significantly for
802.16. The reason is that the number of collisions is reduced
as the number of events like (3) mentioned in III and shown
in Fig. 4 are avoided.
Fig. 10 shows the individual and overall system throughput
against offered 802.11 traffic. As expected, the impact of
interference is less severe if only downlink traffic is present. In
this case, the channel is idle for about half of the time. At low
loads of 1 Mbps and 2 Mbps, no data loss occurs in the IEEE
802.11 system. A trend of decreasing 802.16 throughput like in
Case1 is observed, the level of throughput is higher than in the
case of Case1 as the number of collisions between the systems
is less. The overall channel utilization is improved up to 60%
which resembles better throughput than Case1. The downlink
only results give a hint towards the achievable performance

6th Karlsruhe Workshop on Software Radios
200
0 0.5 1 1.5 2 2.5 3 3.5 4
x 106
0
1
2
3
4
5
6x 106
Offered IEEE 802.11 Traffic / bps
Throughput / bps
802.16
802.11
− legacy (DCF) in 802.11 and
RCA model in 802.16
− 2 Mbps 802.16 Traffic
Fig. 10. Mean throughput of both systems in Case2 against 802.11 traffic
0 0.5 1 1.5 2 2.5 3 3.5 4
x 106
0
1
2
3
4
5
6x 106
Offered IEEE 802.11 Traffic / bps
Throughput / bps
0.5 Mbps 802.16 Traffic
1.0 Mbps 802.16 Traffic
1.5 Mbps 802.16 Traffic
2.0 Mbps 802.16 Traffic
2.5 Mbps 802.16 Traffic
3.0 Mbps 802.16 Traffic
− legacy (DCF) in 802.11 and
RCA model in 802.16
Fig. 11. Overall mean throughput of both systems in Case2
if the 802.16 downlink and uplink traffic flows are scheduled
directly after each other in the beginning of the superframe
or the uplink subframe is filled from the back as presented
in [15]. It can be concluded from these results that such a
modified scheduling is able to increase the overall and system
wide throughput performance. We can consider this as a model
of regular channel access in an 802.16 system. In this figure,
802.16 offered traffic is fixed. To see the impact of offered
802.16 traffic in a coexistence scenario, Fig. 11 is depicted.
Increasing the offered load of both systems results in more
collisions and overheads. After some thresholds of overall
offered traffic, the maximum achievable overall throughput
which is around 3.5 Mbps is even degraded.
From the above results it is found that unrestricted and
uncoordinated channel access of 802.11 severely degrades the
performance.
3) Case3: RCA enabled 802.11 and 802.16 systems: Initial
simulation results verify the benefit of regular channel access.
In the simulated scenario, only the downlink is active for
the 802.16 system, which as discussed above, models the
scheduling of downlink and uplink directly after each other.
The IEEE 802.11 system uses RCA with an RCA Interval of
0 0.5 1 1.5 2 2.5 3 3.5 4
x 1
0
6
0
1
2
3
4
5
6x 106
Offered IEEE 802.11 Traffic / bps
Throughput / bps
802.16
802.11
− RCA in both 802.11 and 802.16
− 2 Mbps 802.16 Traffic
Fig. 12. Mean throughput of both systems in Case3 against 802.11 traffic
10 ms (which is the superframe length of 802.16).
The results in Fig. 12 show individual and overall through-
put against offered 802.11 traffic when the 802.16 offered
traffic is 2 Mbps. In this case the airtime is equally divided
between the systems. In this RCA case, the 802.11 system
utilizes the idle periods by shifting its channel access starting
time and limiting its own channel occupation to fit into those
idle periods. This process decreases interference, resulting in
lower probability of losing packets, which eventually increases
the throughput performance of the 802.16 system. Due to fixed
allocation of airtime, the 802.11 system does not interfere at
all. The 802.11 throughput goes to saturation at less offered
load due to fixed allocation, however better fairness between
systems is achieved.
Fig. 13 shows the overall throughput. Here the airtime
in the superframe is equally divided between the systems.
Due to the fixed capacity allocation of 50%, the 802.11
throughput is not varied much for different 802.16 offered
load and the 802.16 throughput is almost equal to the 802.16
traffic up to 3 Mbps. The dark line in the figure shows
the coexistence performance when both systems have equal
traffic load and airtime is allocated accordingly. Maximum
achievable throughput is almost 5.5 Mbps which shows that
proper sharing can improve the spectrum utilization up to 90
percent. The same is true for differing traffic demands of the
systems.
B. Delay Evaluation
The end-to-end MAC delay is generally defined as the
summation of delay components like queuing delay, channel
access delay, transmission delay and retransmission delay (if
Automatic Repeat Request (ARQ) applies), propagation delay
and processing delay. The last two delay components are not
considered (set to zero) in the following delay evaluation.
Fig. 14 shows the mean delays experienced by the 802.11
and 802.16 systems in considered Cases from 1 to 3. Delays
are shown against the 802.11 offered traffic and for 2 Mbps
802.16 offered traffic. Applying the ARQ in the MAC layer for
the 802.16 system is optional according to the 802.16 standard

6th Karlsruhe Workshop on Software Radios
201
0 0.5 1 1.5 2 2.5 3 3.5 4
x 106
0
1
2
3
4
5
6x 106
Offered IEEE 802.11 Traffic / bps
Throughput / bps
0.5 Mbps 802.16 Traffic
1.0 Mbps 802.16 Traffic
1.5 Mbps 802.16 Traffic
2.0 Mbps 802.16 Traffic
2.5 Mbps 802.16 Traffic
3.0 Mbps 802.16 Traffic
− RCA in both 802.11 and 802.16
Airtime allocated according to
traffic demand estimation
Fig. 13. Overall mean throughput of both systems in Case3
0 0.5 1 1.5 2 2.5 3 3.5 4
x 106
10−3
10−2
10−1
Offered IEEE 802.11 Traffic [bps]
Mean delay [s]
Case1: 802.11
Case1: 802.16
Case2: 802.11
Case2: 802.16
Case3: 802.11
Case3: 802.16
Fig. 14. Mean delays experienced by both systems in Case 1-3 with 2 Mbps
802.16 Traffic
[4]. Retransmission delays are not counted for 802.16 as ARQ
is not applied to 802.16 in the simulation presented here. The
802.16 delays almost remain constant with increasing 802.11
offered traffic.
The 802.11 system has ARQ in MAC layer. The higher
the 802.11 traffic, the higher is the collision rate in Case1
and Case2 which causes higher retransmission delays. As
stated before the collision rate is higher in Case1 than Case2
and that increased the retransmission delays and queueing
delays in Case1 more than in Case2. The delays in Case3 are
experienced mainly due to queueing and access delays. Due
to the traffic characteristics, in some points of simulation time
more packets are generated than expected and queued in the
buffer. Those queued packets are served in later subsequent
RCA Intervals which causes higher delays. This phenomenon
is more visible with increased 802.11 offered traffic.
Fig. 15 shows the mean delays against the offered traffic
of both systems and for Case2 and Case3. In Case2, mean
802.11 delays are increased with increasing the offered traffic
of both systems due to the reason mentioned above. When
the overall traffic exceeds the channel capacity, the upper
limit of mean delay is reached. This is due to the limited
buffer size, however taking into account the buffer loss. In the
simulation the strategy of dropping the new arrived packets
when the buffer becomes full is considered. In Case3 mean
802.11 delays however do not vary significantly with 802.16
offered traffic load. This is due to two reasons: The 802.16
system does not interfere the 802.11 system and according
to the simulation configuration systems have fixed resource
allocation. The 802.16 system experienced higher mean delay
with increased 802.11 traffic for both cases.
The delay results could be improved on the one hand by
allocating the resource (i.e. the airtime in the RCA Interval)
according to the traffic demand of the systems and on the
other hand by applying optimized buffer management, which
are ongoing work.
VI. CONCLUSION AND OUTLOOK
In this paper, an analysis of possible interference that can
occur in the heterogeneous coexistence scenario with legacy
802.11 and 802.16 systems is identified and shown. Simulation
results for throughput and delay are evaluated considering the
mentioned ’apartment scenario’ that can happen in near future.
It shows that only 40% of the avaiable capacity could be
utilized. A generic spectrum sharing method is developed and
described enabling the systems to operate in harmony and to
mitigate interference, resulting in increased spectral efficiency.
Applying the method in a first step only to the 802.16 system
shows the channel utilization is increased up to 60% and delay
is decreased. When both systems are following the method, the
throughput improvement is even higher and fairness between
the systems is observed. In this case the 802.11 delays are
decreased compared to without using the proposed method,
however delays are not varied against 802.16 offered traffic
as airtime allocation is not changed during simulation setup.
A heuristic method of airtime allocation based on traffic
demand estimation could be applied to even increase the
performance. From the implementation point of view the main
advantages of the adapted scheme are as follows: The method
does not change or violate the standards, the implementation
complexity of this algorithm is rather low, and according to
our knowledge, there is no impact to higher network layers.
As a requirement the systems have to detect the MAC-frame
period of the other system. They can then either use the same
or a fraction of the period.
For future work, the performance of coexistence of legacy
802.11 and 802.16h based systems where Listen-Before-Talk
is applied will be evaluated. Application of the proposed
regular channel access based coexistence method in 802.11
and 802.16h based systems and a performance comparison
will be done.
ACKNOWLEDGMENT
The authors would like to thank the German Research
Foundation (DFG) project ”Policy-based Spectrum Sharing

6th Karlsruhe Workshop on Software Radios
202
0 0.5 1 1.5 2 2.5 3 3.5 4
x 106
10−3
10−2
10−1
Offered IEEE 802.11 Traffic [bps]
Mean delay [s]
0.5 Mbps 802.16 Traffic: 802.11
0.5 Mbps 802.16 Traffic: 802.16
1.0 Mbps 802.16 Traffic: 802.11
1.0 Mbps 802.16 Traffic: 802.16
1.5 Mbps 802.16 Traffic: 802.11
1.5 Mbps 802.16 Traffic: 802.16
2.0 Mbps 802.16 Traffic: 802.11
2.0 Mbps 802.16 Traffic: 802.16
2.5 Mbps 802.16 Traffic: 802.11
2.5 Mbps 802.16 Traffic: 802.16
3.0 Mbps 802.16 Traffic: 802.11
3.0 Mbps 802.16 Traffic: 802.16
0 1 2 3 4
x 106
10−3
10−2
10−1
Offered IEEE 802.11 Traffic [bps]
Case2 Case3
Fig. 15. Mean delays experienced by both systems in Case 2 and 3 against offered traffic
for unlicensed Mesh” (PoSSuM) which has funded the work
presented.
REFERENCES
[1] “Information Technology – Telecommunications and information ex-
change between systems – Local and metropolitan area networks –
Specific requirements – Part 11: Wireless LAN Medium Access Control
(MAC) and Physical Layer (PHY) Specifications,” IEEE Std., 1999.
[2] “Supplement to IEEE Standard for Information technology – Telecom-
munications and information exchange between systems – Local and
metropolitan area networks – Specific requirements, Part 11: Wireless
LAN Medium Access Control (MAC) and Physical Layer (PHY) spec-
ifications, High-speed Physical Layer in the 5 GHz Band,” IEEE Std.,
1999.
[3] “Information Technology – Telecommunications and information ex-
change between systems – Local and metropolitan area networks –
Specific requirements Part 11: Wireless LAN Medium Access Control
(MAC) and Physical Layer (PHY) specifications Amendment 8: Medium
Access Control (MAC) Quality of Service Enhancements,” IEEE Std.,
November 2005.
[4] ”IEEE 802.16, IEEE Standard for Local and metropolitan area networks,
Part 16: Air Interface for Fixed Broadband Wireless Access Systems,”
IEEE Std., October 2004.
[5] L. Berlemann, ”Distributed Quality-of-Service Support in Cognitive
Radio Networks,” Thesis (PhD), Chair of Communication Networks,
RWTH Aachen University, Aachen Germany: Wissenschaftsverlag
Mainz, 2006.
[6] D. Rouffet, S. Kerboeuf, L. Cai, and V. Capdevielle, ”4G Mobile,” Tech.
Rep., 2005.
[7] A. K ¨
onsgen, M. Siddique, C. G¨
org, L. Berlemann, G. Hiertz,
S. Mangold, S. Max, and M. M¨
uhleisen, ”Coexistence and Radio
Resource Optimization of Wireless Networking Technologies,” Aachen,
Wissenschaftsverlag Mainz, 2008, no. 1000, pp. 43–55.
[8] M. Muehleisen, D. B¨
ultmann, R. Jennen, S. Max, J. Mirkovic, R. Pabst,
and M. Schinnenburg, ”openWNS: Open Source Wireless Network
Simulator,” Aachen, Wissenschaftsverlag Mainz, 2008, no. 1000, pp.
121–140.
[9] “OpenWNS - Open Wireless Network Simulator: A Simulation Platform
for Wireless and Multi-cellular Mobile Communication Systems,” 2009.
[Online]. Available: http://www.openwns.org
[10] ”OpenWNS SystemTest Wifi and Wimac Coexistence,”
November 2009. [Online]. Available: http://launchpad.net/
openwns-systemtest- wifi-wimac- coexist
[11] ”IEEE Std 802.16h/D4, IEEE Standard for Local and Metropolitan Area
Networks. Part 16: Air Interface for Fixed Broadband Wireless Access
Systems. Improved Coexistence Mechanisms for License-Exempt Oper-
ation,” IEEE Std., February 2008.
[12] J. Rapp, ”Hiperlan/2 System Throughput and QoS with Interference Im-
proving Strategies,” in IEEE VTS 53rd Vehicular Technology Conference,
vol. 4, 2001.
[13] M. Johnsson, ”HiperLAN/2: The Broadband Radio Transmission Tech-
nology operating in the 5 GHz Frequency Band,” 1999.
[14] M. M¨
uhleisen, R. Jennen, M. Siddique, and C. G¨
org, ”Analysis of a
TDMA Coexistence Approach for IEEE 802.16 Systems,” in Leistungs-
, Zuverl¨
assigkeits- und Verl¨
asslichkeitsbewertung von Kommunikation-
snetzen und verteilten Systemen, MMBnet 2009, Hamburg, Germany,
2009, pp. 65–69.
[15] M. Muehleisen, R. Jennen, M. Siddique, and C. G¨
org, ”IEEE 802.16
Coexis tence through Regular Channel Occupation,” in Proceedings of
the European Wireless, European Wireless, 2009, pp. 211–215.
[16] L. Berlemann, C. Hoymann, G. Hiertz, and B. Walke, ”Unlicensed
Operation of IEEE 802.16: Coexistence with 802.11(a) in
Shared Frequency Bands,” in Proceedings of the 17th Annual
IEEE International Symposium onPersonal, Indoor and Mobile Radio
Communications, Helsinki, Finland, 2006, p. 5.
[17] Q. Ni, ”Performance Aanalysis and Enhancements for IEEE 802.11e
Wireless Networks,” Network, IEEE, vol. 19, no. 4, pp. 21–27, 2005.
[18] ”Wi-fi MultimediaTM(WMM R
)” [Online]. Available: http://www.wi-fi.
org/knowledge center/wmm
[19] Yue Fang, Daqing Gu, McDonald, A.B., Jinyun Zhang, ”A two-level
carrier sensing mechanism for overlapping BSS problem in WLAN,”
Local and Metropolitan Area Networks, 2005. LANMAN 2005. The
14th IEEE Workshop on , vol., no., pp.6 pp.-6, 18-18 Sept. 2005