You are on page 1of 6

Proceedings of 2009 12th International Conference on Computer and Information Technology (ICCIT 2009)

21-23 December, 2009, Dhaka, Bangladesh

A QoS Aware Route Selection Mechanism Using Analytic


Hierarchy Process for Mobile Ad Hoc Network
Abu Hamed Mohammad Misbah Uddin t , Mohammad Iftekhar Monir, Shahid Md. Asif Iqbal t
t The Royal Institute of Technology, Sweden
+Dept. of Computer Science & Engineering, Premier University, Chittagong, Bangladesh
cseminhaj@yahoo.com, monirI175@yahoo.com, asifcsep@yahoo.com

Abstract ing the results [2]. Keeping that in mind, we have de-
To truly realize potential ofMANET, multimedia servic- vised an on demand source routing protocol for
es must be provisioned with a minimum level ofQoS. To MANET which will use six important QoS attributes to
meet the QoS requirement of such services, many find out the best route by the help of Analytic Hierarchy
attributes need to be considered. To keep the routing Process.
process lightweight, standard QoS aware routing pro- Remainder of the paper is designed as follows. In sec-
tocol in MANET works with one or two such parame- tion II we discuss the theoretical background for our
ters. In this paper, we have proposed an on-demand proposal. Section III gives an overview of related re-
source routing protocol for MANET that works with six search works in the field. Section IV presents the pro-
important QoS attributes by varying priority for differ- posed mechanism. Finally, section V summarizes our
ent category oftraffic flow. We have reflected this vari- work by discussing the achievement, limitation and fu-
ation by incorporating Analytic Hierarchy Process in ture work of the proposal.
the proposal.
II. THEORETICAL REVIEW
Keywords: AHP, MANET, QoS, Traffic Categoriza-
tion, Weight.
A. QoS Parameters and Traffic Categorization
I. INTRODUCTION The ITU has defined five important QoS Parameters for
Mobile Ad Hoc Network has gained importance as a IP transport: Transfer Delay, Delay Variation, Loss Ra-
tio, Error Ratio, and Throughput [3]. Here, Loss Ratio
recognized field of research due to its decentralized,
and Error Ratio can also be described in terms of Relia-
dynamic and self configuring nature. In its early stage,
bility. Other than these, some more parameter can be
much of the efforts have been placed in providing solu- regarded in case of wireless ad hoc network, whether
tion for the best effort services. However, since the last mobile or stationary. As every mobile node acts as in-
decade, it has been observed that multimedia services termediate router, load and power condition of these
have been a major catalyst for mass recognition of a nodes should also be considered [4]. Therefore, when
technology. In failing to guarantee such services QoS aware routing is concerned above mentioned pa-
through typical best effort design, use of MANET in the rameters should be taken into concern.
real world has been limited. Therefore focus has been Some wireless technologies have defined service classes
shifted towards the provision of better defined QoS in and related QoS parameters. ITU has defined six service
this field. classes (class 0-5) with recommended values of the QoS
QoS routing plays a major role in QoS provisioning as parameters given in [3]. Enhanced Distribution Coordi-
it tries to find the best route to serve application's QoS nation Function (EDCF) has introduced prioritize four
requirement. Some QoS routing protocol works as an access categories, designated 3, 2, 0, 1 from highest to
integral part of session admission control whereas some lowest priority for IEEE 802.11 in [5]. These four
tries to improve overall performance through particular access categories are not allocated to any particular type
metrics. Majority of proposed QoS routing for MANET of traffic but allocation that is often use is: voice, video,
has included throughput and delay as routing parameter. best effort and background. QoS categorization in
However, many other metrics are also important to 802.16 is provided by means of four alternate schedul-
ing services: Unsolicited Grant, Real Time Polling, Non
quantify QoS. But working with more attributes is diffi-
Real Time Polling and Best Effort [6]. UMTS specifies
cult and therefore, researchers have avoided working
four service classes: Conversational, Streaming, Interac-
with more than two attributes to design QoS routing [1].
tive and Background [7]. As there is no specified traffic
Analytic Hierarchy Process is a powerful and flexible categorization for Ad Hoc Wireless network, one of
decision making tool to set priorities when both qualita- these categorizations can be utilized in modified form, if
tive and quantitative aspects of a decision are consi- QoS provisioning is in concern.
dered. When complex multi-criteria based decision
making problem needs to be solved, AHP is a very good
choice as it simplifies by reducing complex decisions to
a series of one-on-one comparisons and then synthesiz-

978-1-4244-6284-1/09/$26.00 ©2009 IEEE 552


B.AHP path that has the largest packet capacity by using resi-
The Analytic Hierarchy Process (AHP) is a structured dual battery power and the link error rates.
solution process to work with complex decisions of
dealing multiple criteria to choose one of the multiple IV. PROPOSED SCHEME
options. It is used globally in a wide variety of decision
situations, in fields such as government, technology, A. Process Overview
business, industry, healthcare, and education.
Our proposed system contains three important blocks:
AHP first decompose decision problem into a hierarchy
of more easily understandable sub-problems, which are Flow Classifier, Routing Information Base (RIB) and
evaluated independently. The elements of the hierarchy AHP Engine. Flow classifier assigns access class to
can relate to any aspect of the decision problem- the traffic flows according to their nature through some
tangible or intangible, carefully measured or roughly defined process. Routing Information Base runs an on
estimated, well- or poorly-understood -anything at all demand source routing protocol similar to DSR protocol
that applies to the decision at hand. and gathers values of QoS attributes for different route
Once the hierarchy is built, the decision makers syste- throughout the process. AHP pre- computes the weight
matically evaluate its various elements by comparing distribution of the QoS parameters and stores them. The
them to one another at a time. In making the compari- engine also accumulates the values of QoS attributes of
sons, the decision makers can use concrete data about different routes from RIB. Finally, it performs calcula-
the elements, or they can use their judgments about the tion using the weight and the value of QoS parameters
elements' relative meaning and importance. The AHP according to AHP process to generate routing metric.
converts these evaluations to numerical values that can Each unit is explained in the following sections.
be processed and compared over the entire range of the
problem. A numerical weight or priority is derived for B. Flow Classifier
each element of the hierarchy, allowing diverse and
Importance of the QoS attributes varies for different
often incommensurable elements to be compared to one
another in a rational and consistent way. In the final traffic flow. To reflect the appropriate importance of
step of the process, numerical priorities are calculated attributes while calculating metric for QoS routing pro-
for each of the decision alternatives. These numbers tocol, common user traffic needs to be divided into
represent the alternatives' relative ability to achieve the multiple classes so that attribute priority are easily
decision goal, so they allow a straightforward consider- attached to them. As there is no strongly defined allo-
ation of the various courses of action [2]. cation strategy to common user applications to differ-
ent classes defined in EDCF, following categoriza-
III. RELATED WORKS tion (adapted from UMTS QoS classification [7]) is
used in the design.
There have been regular efforts in designing QoS aware
Table I: Traffic flow category
routing protocols. As supporting more than one QoS Category 3 VoiceNideo Telephony
constraint make the QoS routing problem NP-complete Category 2 Streaming, AudioNideo on demand
[8], majority of the routing protocols have focused on Category 0 Web browsing, Database access, Re
providing QoS based on one or two metrics, mostly Category 1 Email,File Download
throughput and delay. In [9], the authors have designed
a QoS-aware routing protocol through the use of the
Flow classifier uses category defined in table I to assign
approximate bandwidth estimation by using two (ad- a class to a flow by one the following mechanism: IP
mission control and feedback) bandwidth estimation ToS, 802.1p, DSCP [17].
methods. Zhu and Corson, has developed an end to end
bandwidth based QoS routing protocol for mobile ad
C. RIB & The Routing Process
hoc networks employing TDMA [10]. Similarly,
MACAIPR [11] provides guaranteed bandwidth support The RIB runs an on-demand source routing protocol
via reservation for real time traffic. On the other hand, similar to DSR [18]. According to the design, route re-
Application Aware QoS Routing [12] works with both quest is generated by the source and route is selected by
transmission delay and session throughput requirement the destination. The route discovery process works as
to assure throughput, bounded delay and jitter. ODCR follows. When a node wants to send data to another
[13] uses delay to find out the best route with bounded node but it does not have the route to reach the destina-
delay. tion, the source node broadcasts a route request packet
Other QoS parameter based routing solution is also evi- directed towards the destination. When an intermediate
dent. DSARP [14] has used load condition (buffer full- node receives the broadcast it adds itself as well its cur-
ness) to provide bounded delay and jitter. GAMAN [15] rent values of QoS attributes.
uses node traversal delay and reliability to achieve ro- As discussed in the literary review, common parameters
bust QoS routing solution. MRPC [16], a power aware that attributes to the QoS of a traffic flow in MANET
routing protocol uses Min Max formulation to select the are: Throughput, Delay, Jitter, Reliability (defined by

553
Packet Loss and Bit Error Rate), Load and Battery pow- Usually the AHP method consist four steps: 1) Create
er. These QoS metrics can be classified as additive, the input values by pair-wise comparisons of decision
concave or multiplicative metrics based on their ma- elements, 2) Estimate the relative weights of the deci-
thematical properties [19]. Additive metrics are defined sion elements, 3) Check for consistency, 4) Combine
as ILi(m), over path P, where L, (m) is the value of the the relative weights to determine the ranking of the dif-
metric m over link L, and L, P. The value of concave ferent decision alternatives. While designing solution
metric Cmisdefined as the minimum value of that metric for MANET, one has to keep in mind that the protocol
over a path, Cm = mirul., (m)). Finally, multiplicative should be as lightweight as possible. That's why we
metric MIn is calculated by taking the product of the have used a variant of AHP process explained in [20].
values along a path, M m=I1 L, (m). Thus, delay is an This mechanism reduces the processing by eliminating
additive metric, as it is cumulative over the whole path step 3 and simplifying step 2.
and throughput is a concave metric. Therefore, these six The next step is creating judgments on the parameters'
metrics are calculated for each link over a route accord- importance, made in pairs, aij, relating the importance
ing to its nature and placed in route request packet. of parameter i to that of parameter j. A theoretically
When the destination receives the route request pack- justified fundamental scale according to [2], is used to
etCs), it replies the best route to the source. Upon receiv- represent the intensities between each attribute. Usually
ing the packet, RIB in destination node stores each the scale consists of nine levels but to make it easier, we
route along with its QoS values for the six parameters. use more restricted scale with three levels mentioned in
To compute the best route, the destination node uses the following table. Reason behind this simpler scale is
QoS attribute values only. While sending route request, that we have categorized the value and requirement of
source node sets category of the traffic flow for which the attributes in three levels: high, medium and low.
the route is intended. This is used by the RIB in the des- Table II: Intensity of Importance Scale
tination node, as it informs AHP engine to perform cal- Intensity of Impor- Definition
culation according to the category. 1 EquallyImportant
3 Moderately More Importan1
5 Strongly More Important
D. AHP ENGINE
The AHP engine collects the discovered routes along At first, keeping the objective of route selection in fo-
with their QoS values from the RIB. It is also informed cus, QoS attributes are pair-wise compared. A square
about the flow class for which the route is intended. matrix of order n, A = [~j], is created using the compar-
From the class definition, AHP uses the appropriate pre- isons where, ~j>O, indicating the importance of parame-
computed priority weight and multiplies these weights ter i relative to parameter j as show in equation (1).
with perceived QoS values through a process defined in
all a12 a13 ali)
Analytic Hierarchy Process. Outcome of the calculation a2l a22 a23 a2i
is the relative preference of the routes. Using relative A= (1)
... ... .. .
preference value, destination node identifies the best (.atl ail ai~ afi
route and sends route reply accordingly. AHP process is
defined as follows. Obviously, ~j = 1, when i=j, while ~j = 1/a.;i; which re-
In the first step, the problem is organized into a three flects the reciprocal importance of parameter j relative
level hierarchy as figure 2. The key objective here is to to parameter i. Then we are going to determine the
select the best route suited to QoS requirement of the weight distribution of the parameters using the compari-
traffic flow. Therefore route selection is placed at the son matrix mentioned in equation (1), in which Wi is the
root of the tree.
1,...---.-·-
. .......
------. weight of parameter i in the weight vector w = [w}, W2,
W3, .... , w n] for n attributes. This vector can by found
by the equation (2) based on [2], given as follows:

El E·Ei:3 El; El=


w.I = "t:"ft ~''; "t:"D ~ft ~"
L.Jj=l W1J .Lii=l L j=l W1J .
(2)

........ .. Ii .... ,-.. Where, aij is the (ij)th entry of the comparison matrix A.
. ".",~...' .,
: ""'_'0J- ~- . --'i" L_, ,". ~." '"
~
Thus, we find Wij, column matrix, where i = 1 to n, for a

E-~8~t::J
particular flow class j. Aligning column matrices, Wi},
Wi2, Wi3 ... to column 1, 2, 3 ... and so on, we form
another matrix B = [bij] of order n X m, where, where n
is the number of parameter and m is the number catego-
Fig. 1. AHP 3 Tier Hierarchy ries.
We perform another set of pair-wise comparison but
Factors contributing to the objective are shown in the this time for the routes and build a set of comparison
tier 2 and alternatives or route options are mentioned in square matrix for all QoS attributes, C = [Cij] of order p,
tier 3 of the hierarchy. where p is the number of routes. Using the mechanism

554
given in equation (2), we construct Rij ; where, Rij is Cat T D J R R B Row Score
2 Sum
route weight distribution of route i for parameter j. We
represent Rij in matrix D = [Dij ] p x ne where p is the T 1 5 1 5 5/3 5/3 15.3 0.28
3
number of routes and n is the number of parameter. Fi- D 1/5 1 1/5 1 1/3 1/3 3.07 0.05
nally, we perform matrix multiplication within D x B = J 1 5 1/5 5 5/3 5/3 15.3 0.28
[E], x m- This is the solution matrix. Highest value of 3
each column represents the best route for a particular R 1/5 1 1/5 1 1/3 1/3 3.07 0.05
category of flow traffic. L 3/5 3 3/5 3 1 1 9.2 0.17
B 3/5 3 3/5 3 1 1 9.2 0.17
To- 55.2 1
E. Case Study tal
To demonstrate how AHP engine works we will use an
example ad hoc network scenario. The topology of the Table V: Factor comparison category 0
scenario is given below (figure 2). S wants to send Ca T D J R L B Row Seo
t Sum re
VoIP traffic to D and it is discovering route for D. In 0
this topology, S can reach to D, using one of the three T 1 1 3 3/5 1 1 7.6 0.1
routes (Route 1: H-I-J-K, Route 2: A-B-C, Route 3: E- 7
D 1 1 3 3/5 1 1 7.6 0.1
F-G). Following section demonstrates how AHP calcu-
7
lates the matrix. J 1/ 1/3 1 1/5 1/3 1/3 2.53 0.0
3 5
R 51 5/3 5 1 5/3 5/3 12.6 0.2
3 7 7
L 1 1 3 3/5 1 1 7.6 0.1
~.,
7
."-.
1,-.. B 1 1 3 3/5 1 1 7.6 0.1
7
To- 45.6 1
tal

Table VI: Factor comparison category 1


Cat 1 T D J R L B Row Score
Sum
T 1 1 1 1/5 1/3 1/3 3.87 0.07
Fig. 2. Ad Hoc Network topology D 1 1 1 1/5 1/3 1/3 3.87 0.07
J 1 1 1 1/5 1/3 1/3 3.87 0.07
R 5 5 5 1 5/3 5/3 19.33 0.37
Using the stringency of common application with their
L 3 3 3 3/5 1 1 11.6 0.21
QoS requirement adapted from [21], following matrixes B 3 3 3 3/5 1 1 11.6 0.21
are pre-computed to all nodes which shows the pair Total 54.14 1
wise weight distribution of the QoS attributes for differ-
ent categories. Although almost all of the QoS attribute Now, different routes are pair wise compared for each
requirement varies for different access class traffic but QoS attributes. Before forming this pair wise compari-
requirement of load and battery power level is uniform sons we need to obtain the values for the QoS attributes
for all categories. That's why we have placed uniform and order them according to the intensity of importance.
intensity of importance of those two attributes through- To do that we use two threshold values: ul and a2,
out the matrices. We have used following notation for where ul > a2 > O. If perceived value of a QoS attribute
simpler usage: Throughput = T, Delay = D, Jitter =J, is beyond ul then its intensity level is considered high.
Reliability = R, Load = L, Battery Power = B. On the other hand, if the value lies between ul and a2
Table III Factor comparison catesorv 3 then intensity level is medium. Otherwise, intensity is
Cat T D J R L B Row Score considered low. We will assume that from a set of per-
3 Sum ceived values of the attributes in the network, following
T 1 3/5 3/5 3 1 1 7.2 0.132 table is formed:
D 5/3 1 1 5 5/3 5/3 15.33 0.28 Table VII: Ranking for other categories
J 5/3 1 1 5 5/3 5/3 15.33 0.28 Parameter Route 1 Route 2 Route 3
R 1/3 1/5 1/5 1 1/3 1/3 2.4 0.044 Throughput High Medium Medium
L 1 3/5 3/5 3 1 1 7.2 0.132 Delay Medium Low High
B 1 3/5 3/5 3 1 1 7.2 0.132 Jitter Medium Low High
Total 54.66 1 Reliabilitv High Medium High
Load Medium High Low
Battery Power Medium High Low
Following matrices shows the pair wise comparison and
Table IV: Factor comparison category 2
priority rating of the each route for each of the parame-

555
ters based on the information in table VIII. ble for category 1. The destination node will send a
Table VIII: Pair wise companson for Thr ou 1lPU
h t route reply to the source accordingly.
Throughput Route Route Route Row Score T able XII: Ra nkimg for ot h er catezones
1 2 3 Sum
Category 2 Category 0 Category 1
Route 1 1 5/3 5/3 4.33 0.44
0.3638 0.3649 0.3599
Route 2 3/5 1 1 2.6 0.27
Route 3 3/5 1 1 2.6 0.27 0.3088 0.2659 0.2249

Total 9.53 1 0.3218 0.3658 0.4138

Similarly, pair-wise weight distribution of all the routes v. CONCLUSION


for other attributes are calculated and following matrix Current trend in research in the field of QoS routing in
is formed. MANET shows that that, researchers are putting a great
Table IX: Priority ratin ~ of options against a lIt:actors emphasis on the session admission (QoS route finding)
Throu- Jit- Reli- Battery capability of their protocol, which is admittedly very
Delay Load
ghput ter ability Power
important. Conforming to that, we present a QoS aware
Route
0.44 0.33 0.33 0.39 0.33 0.33
route selection mechanism for mobile ad hoc network.
1 In order to make a more accurate route selection for best
Route QoS support we select the most important QoS
0.27 0.56 0.56 0.22 0.11 0.11
2
attributes from different aspects. AHP is used to calcu-
Route
0.27 0.11 0.11 0.39 0.56 0.56 late the integral value of multiple parameters and make
3
the route selection decision in which the human aspect
is involved to determine the importance between differ-
The last step is to establish the overall ranking among
ent criteria. As the calculations are neither numerous
the alternatives by combining the priority matrices - nor complicated, a use for real time route selection is
weight distribution of criteria and weight distribution of possible. It can be expanded for additional criteria and
options. The outcome from this product is the overall attributes. So, this scheme significantly advances the
ranking for the options in achieving the goal. Following system flexibility and extensibility, and provides more
matrix multiplication will determine the overall ranking accurate and effective route selection at any time.
of the routes for VolP traffic flow. As VolP traffic falls However, session completion is also as important as
into category 3, weight distribution of routes are multip-
session admission from a user perspective. This is be-
lied with the weight distribution of attributes for catego- cause the perceived QoS is better when some sessions
ry 3. Product from the above combination results the
are blocked but none are dropped in mid session. In this
final overall ranking shown in table XI, with Route 2 proposal we have not explained any QoS assurance pro-
coming out as the most favorable alternatives for cate- cedure. We will look into the aspect of session comple-
gory 3. tion in future research. Furthermore, we will also look
Table 10: Combined Matrix into fast local QoS route-repairing schemes require to
Thro- Del- Jitt- Reli- Lo- Batt-
Cat improve QoS session completion rates and protocols'
ugput ay er ability ad ery
3
Power robustness against mobility. Finally, experimental result
0.44 0.33 0.33 0.39 0.33 0.33 0.132 from simulated environment will reflect the strength of
0.27 0.56 0.56 0.22 0.11 0.11 * 0.28 our proposal. We are going to provide outcomes of si-
0.27 0.11 0.11 0.39 0.56 0.56 0.28 mulated experiments in the upcoming works.
0.044
0.132 REFERENCES
0.132
[1] Hanzo-II, L.; Tafazolli, R.,"A SURVEY OF QOS
ROUTING SOLUTIONS FOR MOBILE AD HOC
Table XI: Ranking of routes or category 3
Route
NETWORKS", Communications Surveys & Tuto-
Category 3
rials, IEEE Volume 9, Issue 2, Page(s):50 - 70,
Route 1 0.34716 Second Quarter 2007.
Route 2 0.38796 [2] Saaty, Thomas L. "How to make decision: The
Route 3 0.26224 Analytic Hierarchy Process", European Journal of
Operational Research, Vol. 48, pp 9-26, 1990.
[3] ITU-T Y.1540 AMD 1: Internet protocol data
Table XII shows the ranking of the routes if other cate- communication service - IP packet transfer and
gories of traffic are considered. Here, Route 1 is the availability performance parameters Amendment 1.
most favorable route in category 2, Route 3 is the most [4] Prasant Mohapatra, Jian Li, and Chao Gui, "QoS in
favorable for category 0 and Route 1is the most favora- Mobile Ad hoc Networks," Special Issue on QoS in

556
Next-Generation Wireless Multimedia Communica- [18]D. Johnson, Y. Hu, D. Maltz, "RFC 4728: The Dy-
tions Systems in IEEE Wireless Communications namic Source Routing Protocol (DSR) for Mobile
Magazine, June 2003. Ad Hoc Networks for IPv4", February 2007
[5] Daqing Gu Jinyun Zhang, "QoS enhancement in [19]T. B. Reddy, I. Karthigeyan, B. Manoj, and C. S. R.
IEEE 802.11 wireless local area networks", Com- Murthy, "Quality of service provisioning in ad hoc
munications Magazine, IEEE, Volume: 41, Issue: wireless networks: a survey of issues and solu-
6, page(s): 120- 124, June 2003. tions." available online:
[6] IEEE 802.16-2004, "IEEE Standard for Local and http://www.sciencedirect.com. Last accessed on
Metropolitan Area Networks - Part 16: Air Inter- 29th July, 2009.
face for Fixed Broadband Wireless Access Sys- [20]Dr S. Tom Foster, Dr. Gerald LaCava, "The Ana-
tems", October 2004. lytical Hierarchy Process: A step by step ap-
[7] UMTS: Quality of Service, proach", Available at:
http://www.umtsworld.com/technology/qos.htm. https://acc.dau.mil/GetAttachment.aspx?id=5479&
Last accessed on 29th July, 2009 pname=file&aid=18722&lang=en-US, Last ac-
[8] S. Chen," Routing Support for Providing Guaran- cessed on 29th July, 2009.
teed End-to-End Quality-of-Service", PhD thesis, [21]Andrew S. Tanenbaum, "Quality of Service: Re-
University ofIL at Urbana-Champaign, 1999. quirements", Computer Network, Chapter 5, page:
[9] Lei Chen, Wendi B. Heinzelman, "QoS-Aware 397, Fourth Edition, Prentice Hall, October 2002.
Routing Based on Bandwidth Estimation for Mo-
bile Ad Hoc Networks", Selected Areas in Com-
munications, IEEE Journal on
Volume 23, Issue 3, Page(s): 561 - 572, March
2005.
[10]Chenxi Zhu, M. Scott Corson, "QoS routing for
mobile ad hoc networks", Infocom 2002, Available
athttp://www.ieeeinfocom.org/2002/papers/121.pd
f, Last accessed on: 29 July, 1009.
[11]Lin, C.R. Gerla, M, "Asynchronous Multimedia
Multihop Wireless Networks", INFOCOM '97. Six-
teenth annual Joint Conference of the IEEE Com-
puter and Communications Societies. Proceedings
IEEE Vol: 1, page(s): 118-125, Apr1997.
[12]M. Wang and G.-S. Kuo, "An application-aware
QoS routing scheme with improved stability for
multimedia applications in mobile ad hoc net-
works," in Proc. IEEE Vehicular Technology
Con£,pp.1901-1905,Sep.2005.
[13]B. Zhang and H. T. Mouftah, .QoS routing for
wireless ad hoc networks: problems, algorithms and
protocols, IEEE Communication Magazine, vol. 43,
pp. 110.117, Oct. 2005.
[14]M. Sheng, J. Li, and Y. Shi, "Routing protocol with
QoS guarantees for ad-hoc network", Electronics
Letters, volume 39, pp. 143.145, Jan. 2003.
[15]L. Barolli, A. Koyama, and N. Shiratori, "A QoS
routing method for ad-hoc networks based on ge-
netic algorithm", In Proceedings of 14th Interna-
tional Workshop. Database and Expert Systems
Applications, pp. 175. 179, Sep. 2003.
[16]A. Misra and S. Banerjee, "MRPC: Maximizing
network lifetime for reliable routing in wireless en-
vironments", In Proceedings of IEEE Wireless
Communications and Networking Conference.,
March 2002.
[17]WLAN Quality of Service, Available at:
http://www.cisco.com/en/US/docs/solutions/Enterp
riselMobility/vowlan/41dg/vowlan_ch2.html, Last
accessed on 29th July, 2009.

557

You might also like