Cluster-Based Routing Protocol To Improve QosIn Mobile Adhoc Networks
Prof. M.N. Doja Mohd. Amjad
Department of Computer Engineering Department of Computer EngineeringFaculty of Engineering & Technology Faculty of Engineering & TechnologyJamia Millia Islamia, New Delhi, India Jamia Millia Islamia, New Delhi, India
Abstract: AnAd Hoc network is a collection of wireless mobilehosts dynamically forming a temporary network without theaid of any existing established infrastructure. Quality of Service (QoS) is a set of service requirements that needs to bemet by the network while transporting a packet stream from asource to its destination. QoS support for Mobile AdhocNetworks (MANETs) is a challenging task due to the dynamictopology and limited resources. Characteristics of Mobile AdHoc Networks (MANETs) such as lack of central coordination,mobility of hosts, and limited availability of resources makeQuality of Service (QoS) provisioning very challenging.Limited resource availability such as battery power, averageenergy consumption of the network by all of the nodes andinsecure medium are some of the major QoS issues to be dealtwith. In this paper we have suggested a clustering of participating nodes with minimum energy consumption by theoverall network by hierarchical cluster-based routingAlgorithm. In this algorithm we have introduced a new metric,next hop availability, which is a combination of two metrics. Itmaximizes path availability and minimizes travel time of packets and therefore offers a good balance between selectionof fast paths and a better use of network resources withminimum energy consumption. In the conclusion it providessimulation result to evaluate the performance on a networksimulator.
Keywords : - Power saving protocol, clusters, Quality of service support, Ad hoc network
.
1 INTRODUCTION
In an ad hoc network the mobile nodes agree to serve asboth routers and hosts. The nodes can dynamically join andleave the network, frequently without warning, and possiblydisrupting communication amongst other nodes. Moreover,the limitations on power consumption imposed by portablewireless radios result in a node transmission range that istypically small relative to the span of the network. Thislimits the propagation range of a mobile node[7]. In such anenvironment, it may be necessary for one mobile host toenlist the aid of others in forwarding a packet to itsdestination. These networks can be formed on the fly,without requiring any fixed infrastructure. As these areinfrastructure less networks, each node should act also as arouter. These characteristics of MANETs such as lack of central coordination, mobility of hosts, limited availabilityof resources and insecure medium make Quality of Service(QoS) provisioning very challenging. QoS is usually definedas a set of service requirements that need to be met by thenetwork while transporting a packet stream from a source toits destination(s)[11]. This can be achieved by incorporatingquality of service (QoS) metrics such as energyconsumption by the network, battery life of the nodes andsecurity measures into the routing decisions as opposed tochoosing a shortest path. Efficient resource managementmechanisms are required for optimal utilization of thisscarce resource i.e. battery power. In ad hoc network thisoperation is called clustering, giving the network ahierarchical organization. A cluster is a connected graphincluding a cluster head responsible of the management of the cluster, and (possibly) some ordinary nodes. Each nodebelongs to only one cluster. Some MANETs, such as mobilemilitary networks or future commercial networks may berelatively large (e.g. hundreds or possibly thousands of nodes). A way to support the increasing number of nodes inMANET is to subdivide the whole network into groups, andthen create a virtual backbone between delegate nodes ineach group[16]. In ad-hoc network this operation is calledclustering, giving the network a hierarchical organization.
2 CLUSTERING IN MANETS
A way to support the increasing number of nodes inMANET is to subdivide the whole network into groups, andthen create a virtual backbone between delegate nodes ineach group. In ad-hoc network this operation is calledclustering, giving the network a hierarchical organization.Several cluster based adaptations has been proposed forexisted routing protocols and other protocol as ZRP (zoneroutingProtocol), CBRP (cluster based protocol) haveoriginally exploited this concept[6][10]. Clustering forsecurity can simplify the management of CertificateAuthority in a Public Key Infrastructure (PKI) by affectingthe full or a subset of Certificate Authority services tocluster-heads, ensuring in this way the availability of theCertificate Authority.The Hierarchal organization consists of:
Cluster Head
: A cluster head, as defined in the literature,serves as a local coordinator for its cluster, performing inter-cluster routing, data forwarding and so on. In our self-
(IJCSIS) International Journal of Computer Science and Information Security,Vol. 10, No. 1, January 201295http://sites.google.com/site/ijcsis/ISSN 1947-5500
.
organized clustering scheme the cluster head only serves thepurpose of providing a unique ID for the cluster, limiting thecluster boundaries [3].
Cluster Gateway
: A cluster gateway is a non cluster-headnode with inter-cluster links, so it can access neighboringclusters and forward information between clusters.
Cluster Member
: A cluster member is a node that is neithera cluster head nor a cluster gateway.
3 LIMITATIONS OF EXISTING LGORITHMS
None of the two well known cluster based routing protocolsi.e ZRP or CBRP leads to an optimal election of cluster-heads since each deals with only a subset of parameterswhich can possibly impose constraints on the system.However, a cluster-head may not be able handle a largenumber of nodes due to resource limitations even if thesenodes are its immediate neighbors and lie well within itstransmission range. In other words, simply covering the areawith the minimum number of cluster-heads will put moreburdens on the cluster-heads. On the other hand, a largenumber of cluster-heads will lead to a computationallyexpensive system [2][[8]. Although this may result in goodthroughput, the data packets have to go through multiplehops thus implying high latency.
4 OUR ALGORITHM
Assumptions1. The network is divided into cluster of nodes with a singleclusterhead per cluster.2. No two clusterheads can be one hop neighbors of eachother.3. Overlapping clusters are connected through Gatewaynodes.4. All the ordinary nodes are one-hop from their clusterheads.5. Each node that requests for an entry permit must keeptrack of the respectiveWeights broadcasted by the neighbor nodes.6.Battery power is reduced in proportion to the number of packets sent.
4.1DATA DICTIONARY4.2 THE DESIGN APPROACH
In this section, we will describe our proposed clusteringalgorithm called Cluster Base Algorithm (BCA). BCA is aweight based clustering algorithm which uses a weightcomputed from a set of parameters to elect cluster-heads[13].The main basic concepts used to derive the neededparameters are given below:
The Max Value
:
Represents the upper bound of the number of nodes that can simultaneously be supported by a cluster-head. Since mobile nodes have limited resources, thereforethey can’t handle a great number of Nodes. This value isdefined according to the remainder of resources of thecluster-head.
Min Value
: represents the lower bound of the number of nodes that belong to a given cluster before proceeding to theextension or merging mechanisms. This value is global andthe same for the entire network. The
Min Value
may avoidthe complexity due to the management of great number of clusters [18][19].
D hops Cluster
s:
As one hop clusters aretoo smallfor largead hoc networks, therefore BCA creates
D
hops clusterswhere
D
is defined by the underlying protocol or accordingto the cluster-head state (busy or not). By the way, thediameter of the cluster can be extended in some situations.
Identity (ID):
It
is a unique identifier for each node in thenetwork to avoid any spoofing attacks or perturbation in theelection procedure. We propose to use certificate as identity,therefore we suppose the existence of an online or offlinePublic Key Infrastructure managing the certificatedistribution.
Weight
:
Each node is elected cluster-head according to itsweight which is computed from a set of system parameters.
W
A
Combined weight of each node A.PC
WT
Minimum weight among all W
A
.PC Possible Clusterhead.X[] Neighbor cluster heads in the transmission range of PC.CHMsgCluster head selection message.Weight[]Weights of all neighbor cluster heads in thetransmission range of PC.th1 Threshold value 1 (associated with weights of newlyselected cluster head).th2 Threshold value 2 (associated with weights of newlyselected cluster head and existing neighbor cluster headsin the transmission range of PC).nTotal number of existing cluster heads in the wholenetwork.cThe total number of existing cluster heads in the wholenetwork whose weights are greater than a specifiedvalue.B Any neighbor node in the transmission range of node A.TABLE 1: DATA DICTIONARY
(IJCSIS) International Journal of Computer Science and Information Security,Vol. 10, No. 1, January 201296http://sites.google.com/site/ijcsis/ISSN 1947-5500
The node having the greatest weight is elected as cluster-head.
4.3 ELECTION CRITERIA
The following parameters define the criteria on which BCArely to elect the cluster-head.
Trust value
:
it measures how much any node in the network is trusted by its neighborhood. It’s defined as the average of trust values received from each neighboring node. In orderto compute the trust value, we suppose that every mobilenode has an intrusion detection mechanism to determine if anode is considered as trust or not by periodically collectinginformation about the behavior of each neighbor.
Degree
: is the number of neighbors of a given node, withina given radius. This parameter is used to choose as cluster-head the node having the maximum neighbors to serve themore number of nodes.
Battery power
:
This factor is the capability of a node toserve as long as possible. Since cluster-head has extraresponsibility and it must communicate as far/long aspossible, thus it must be the most powered node.
The Max Value
: as defined above, this parameter is used inthe election procedure to elect as cluster-head the nodewhich can handle the maximum of nodes.
Link availability:
This is the Number of nodes connected atone time from the CH.
Stability
(Mobility):
This is a useful parameter whenelecting the cluster-head.Stability is defined as the difference between two measuresof M
D
(mean distance)at
t
and
t-1
, it becomes large whenthe node goes far from its neighbors or whenever itsneighbors are going in other direction than the one taken bythe considered node. This value is compared with
D
and anode is considered as most stable if it has the less value of stability.
STA= M
Dt
– M
D(t-1)
In order to elect the most stable node as cluster-head,avoiding frequent roaming, we have computed the
stability
using the following metrics:
•
The distance
:
The
distance
between two nodes A,B(
D
A
,D
B
), is the number of hops between them, which can beobtained from the packets sent from one to other, or hellomessage used in routing protocols. The possibility of obtaining the number of hops between two nodes is evidentand simple within all existed routing protocols.
•
The mean distance
:
This is defined as the average of distances between node A and all its neighbors.
Weight Factors
:
Each of the previous parameters is calledpartial weight. Each parameter is affected a weight factordefining its degree of importance for the underlying protocolor the network. Since only a subset of these parameters canbe used according to the requirements of the network andthe underlying protocol, these factors provide moreflexibility and large scale of use to our algorithm. Forexample trust value may take the great value if theunderlying protocol is a key management protocol. Factorsare given values between 0 and 1, so that the sum of factorsis 1.
Global Weight
:
using all parameters cited above every nodein the network computes its global weight. Depending onthis weight a given node can be elected as cluster-head ornot.When cluster formation is to be performed by BCA, thenodes can change their position randomly (moves awayfrom each other) due to mobility. The communicationamong them may become difficult when they placethemselves outside the transmission range (tx) of the nodefrom which data has to be transferred. For this reason,transmission power of each node is required for weightcalculation. Mobility produces the randomly changedposition of each node. But the rates of data transfercapability (Tr) are not same for all the nodes in a clusterformation procedure. It shows the amount of data can bedelivered in a certain period of time by a node to all theother nodes in its transmission range. These two parametershave been considered for overall improvement inperformance. Using the following formula we calculate thecombined weight Wm for each node m, whereWe denote W1 , W2 , W3, W4 ,W5 and w6 the partialweights factors corresponding respectively to Trust value,Degree difference, Battery power , Max Value, Stability(Mobility) and Link Status in such a way that the sum of allthe factors are 1. The global weight is computed as follows:Wm=
)*6*5(
*4*3*2*1
txwtr W
BmwVmw Dmwmw
++++∆
Where
m
∆
: Degree differenceFor each node m ,
m
∆
is defined as
m
∆
=
δ−
N
for every node m, where,
= The total number of ideal neighbors of nodeN = Degree (Number of neighbors) of node m=
∑
<
}),({
range x ji
t mmdist
for mj
i j
mm M
≠∈
,
BCA finds the sum of the distances, Dm, for every node andits neighbors asDm : Mean distanceDm = Sum of distances=
∑
−−−
22
)()(
mjmimjmi
Y Y X X
for (m
i
,m
j
M
∈
andAverage relative distance
N D D
mm
/
=
Vm: Average relative speed
∑
=
=
N it mjmi
t mjVmi N V
1,
),,( / 1,
Relative velocity with its neighbor nodes = V
(mi, m j, t)
=V
(mi, t)
- V
(m j, t)
Running average of the velocity
(IJCSIS) International Journal of Computer Science and Information Security,Vol. 10, No. 1, January 201297http://sites.google.com/site/ijcsis/ISSN 1947-5500
Search History:
Result 00 of 00
00 results for result for
p.
17 Paper 31101132 IJCSIS Camera Ready Paper Pp. 95-100
34
Reads
0
Readcasts
0
Embed Views
Get Scribd Mobile
To get Scribd mobile enter your number and we'll send you a link to the Scribd app for iPhone & Android.We've sent a link to the Scribd app. If you didn't receive it, try again.
We'll never share your phone number.
More From This User
Notes
Load more
