Вы находитесь на странице: 1из 7

2010 Second International conference on Computing, Communication and Networking Technologies

Agent Based Adaptive Multicast Routing with QoS guarantees in MANETs

G. Santhi1 and Dr.Alamelu Nachiappan2


1 2
Assistant Professor, Department of IT, Associate Professor, Department of EEE,
Pondicherry Engineering College. Pondicherry Engineering College.
shanthikarthikeyan@yahoo.co.in nalam63@yahoo.com

Abstract network, a routing protocol is used to discover routes


between nodes.
The emergence of multimedia applications demands 978-1-4244-6589-7/10/$26.00 ©2010 IEEE
multicasting in MANETs. For such applications QoS
provisioning is a complex and challenging issue. This The primary goal of such an ad hoc network routing
paper proposes a QoS aware multicast routing scheme protocol is to provide an efficient route establishment
uses a set of static and mobile agents and ensures QoS between a pair of nodes so that messages may be
guarantees in terms of bandwidth reservation, delay delivered in a timely manner. Route construction
constraint, delay-jitter constraint and packet loss to should be done with a minimum of overhead and
multicast session. Here the mobile agents move bandwidth consumption. Quality of service (QoS) is
around the network and collect the routing an important consideration in networking, but it is
information. A hybrid method to support an adaptive also a significant challenge. QoS is more difficult to
multipath routing scheme under multiple QoS guarantee in MANETs than in other type of networks,
constraints is introduced. The scheme operates in the because the wireless bandwidth is shared among
following steps. i) Clustering of nodes and then select adjacent nodes and the network topology changes as
the QoS aware cluster heads. 2) Identify and join the nodes move. This requires extensive collaboration
multicast group members to the backbone. 3) Identify between the nodes, both to establish the routes and to
the intermediate nodes and discover multiple paths to secure the resources necessary to provide the QoS.
satisfy the multiple constraints of a call. 4) Set up a With the extensive applications of MANETs in many
QoS aware path for the required multicast route. 5) domains, the appropriate QoS metrics should be used,
Maintain backbone and alternate paths in case of such as bandwidth, delay, packet loss rate and cost for
node mobility. Extensive simulations have been multicast routing. Therefore, QoS multicasting routing
conducted to evaluate the performance of the protocols face the challenge of delivering data to
proposed scheme in terms of packet delivery ratio, destinations through multihop routes in the presence
control overheads and node mobility. The simulation of node movements and topology changes.
results show that the new scheme performs better than
MAODV and ODMRP in terms of reduces the network In recent developments, agent technology is making
overhead and improves packet delivery ratio. its way as a new paradigm in the areas of artificial
intelligence and computing which facilitates
Keywords: MANETs, Multicasting, Clusters, End-to sophisticated software development with features like
End delay, Packet loss ratio. flexibility, scalability, adaptability and efficiency.
Agents are the autonomous programs activated on an
1. Introduction agent platform of a host. The agents use their own
knowledge base to achieve the specified goals without
An ad hoc mobile network is a collection of disturbing the activities of the host.
mobile nodes that are dynamically and arbitrarily
located in such a manner that the interconnections The mobile agents are simple packets, which move
between nodes are capable of changing on a continual around the network and collect useful information
basis. In order to facilitate communication within the such as node id, link latency, congestion level etc. as
they visit different nodes. The information carried by
the mobile agents helps multicast routing protocol to group sequence numbers for tracing the freshness of
find a route for a given destination, when no route the group situation. MAODV however is a hard state
exists in the multicast table to the destination. By this protocol and uses flooding of data packets for data
way, the protocol overcomes the additional delay discovery that affects the overall performance metrics.
which would have been required, in finding a new
route to the destination and also reduces the control ODMRP [7] is a mesh architecture protocol, i.e., it
traffic generated. has multiple paths from the sender to the receivers. A
node that has information to send but no route to the
The QoS multicast routing discussed in this paper is destination, broadcasts a Join Query message. The
a new scheme which utilizes mobile agents for finding next node that receives the Join Query message
multiple paths and to meet the requirement of a single updates its routing table with the appropriate node id
call under multiple QoS constraints. The rest of the (identification) from which the message was received
paper is organized as follows. Related work is for the reverse path back to the sender (backward
presented in the next section. In Section 3, the learning). The node checks the value of the TTL (time
proposed QoS aware multicast routing is explained in to live) and if the value is greater than zero, the node
detail. Section 4 presents the agent model. Some rebroadcasts the Join Query message. A multicast
simulation results are presented in section 5. The group member node broadcasts a Join Reply message
paper concludes with section 6. in response to Join Query message. A neighborhood
node that receives a Join Reply consults the join reply
2. Related works table to see if its node id is the same with any next hop
node id. If it is the same, then the node understands
Multicast routing protocols in ad hoc networks must that it is on the path to the source and sets the
deal with typical limitations of these networks, which forwarding group flag. However, to counter for the
include high power consumption, low bandwidth, and link failures, ODMRP maintains more than one path
high error rates. These protocols maybe divided into increasing the complexity of the routing process.
two main categories: Table and source initiated
(demand-driven). Table-driven routing protocols A Framework for QoS Multicast (FQM) to support
attempt to maintain consistent, up-to-date routing QoS multicast applications for MANETS that is able
information from each node to every other node in the to support QoS enabled applications is proposed.
network. It includes Destination-Sequenced Distance- Aggregated QoS multicast (AQoSM) [5] to provide
Vector Routing (DSDV) [2], Cluster head Gateway scalable and efficient QoS multicast in Diff-Serv
Switch Routing (CGSR) [8], and the Wireless Routing network is also proposed. The key idea of AQoSM is
Protocol (WRP) [15]. A different approach from to separate the concept of groups from the concept of
table-driven routing is source-initiated on-demand distribution tree by mapping many groups to one
routing. This type of routing creates routes only when distribution tree, so that multicast groups can be
desired by the source node. It includes Dynamic routed and rerouted very quickly by assigning different
Source Routing (DSR) [4], Ad hoc On demand labels to the packets.
Distance Vector protocols (AODV) [3].
ABMRS [11] introduces the agent technology for
MAODV [6] is an on-demand routing protocol that multicast route discovery. This hybrid protocol works
discovers the route only when a node has something to with static and mobile agents that are integrated with
send. It is a hard state protocol, i.e., if a member node existing on-demand multicast routing protocols such
of a multicast group desires to terminate its group as multicast ad hoc on-demand distance vector
membership, it must request for termination. When a (MAODV) routing protocol, on-demand multicast
mobile node wants to join a multicast group or send a routing protocol (ODMRP) routing protocol and
message but does not have a route to the group, a others. Here, the mobile agents move around the
Route Request (RREQ) is originated. All the nodes network and collect the routing information. This
that are members of a multicast group together with routing information assists the on-demand multicast
the nodes that are not members of the group but their routing protocols in discovering the route. But, the
position are very critical for forwarding the multicast system still suffers in case of random failures of links
information, compose the tree structure. Every and nodes, excessive control overheads and absence of
multicast group is identified by a unique address and a methodology for multi-group transmissions. Also,
flooding of agents in all direction, absence of cluster
management provides room for enhancement of the 3.1. Network environment
protocol.
A MANET comprising of several nodes that are
It is observed from the literature that still multicast distributed across a given geographical area is
routing with QoS needs much more attention so that considered. Multicast group members may be located
the routing scheme must be robust, maximize packet in any part of a given geographical area. Mobile nodes
delivery ratio and adapt dynamically to changes un may randomly move in any directions. All the nodes
MANET topology and environment. This paper maintain their QoS parameters. All the reliable nodes
proposed a QoS aware multicast routing scheme that support multicast operations whereas intermediate
ensures QoS guarantees such as bandwidth nodes just forward packets from one node to another.
reservation, delay constraint, packet loss and
minimum cost to multicast sessions The existing zone based clustering algorithm is
used for topology construction. Cluster heads are
3. Proposed work chosen based on Reliability Factor (RF). RF is
computed based on the availability of power,
The proposed scheme in MANETs employs a set of bandwidth, memory and mobility of a node. The
static and mobile agents. The scheme assumes reliability factor of each node is recorded and the node
availability of an agent platform at all the mobile which has more reliability factor is selected as cluster
nodes. However, in case of agent platform head. The mobile agents resides at every cluster head
unavailability, traditional message exchange collects information from all the other cluster heads
mechanisms can be used for agent communication. through intermediate nodes of the multicast group if it

Figure 1. Architecture of the system


Figure 2. Nodes organization in the cluster
This scheme operates in the following sequence. (i)
Hello messages are exchanged by all the nodes in the exists. This cluster head maintains the control
entire network. 2) Construct a backbone for information between clusters and for individual nodes.
multicasting using agents. 3) Use of agent technology The mobile agents also monitors for a change in the
to find the multicast tree with multiple QoS reliability factor to declare a new cluster head thus
constraints which may deal with the delay, delay-jitter, accounting for dynamic topological changes.
bandwidth and packet-loss metrics. 4) Agents are
used to discover multiple feasible paths during route 3.2. QoS aware multi constrained routing
discovery and select the path with minimum cost as model.
the primary route. 5) Due to mobility, when a link
failure occurs, the agents use the alternate routes for The proposed multicast routing is an on-demand
data transmission. Cluster management and link routing protocol that builds disjoint multiple routes
failures are handled by agents to significantly reduce using request/reply cycles. The network is represented
the overhead of constructing a multicast tree with as a graph G(V,E), where ‘V’ denotes the set of nodes
multiple QoS constraints. and ‘E’ denotes the set of communication links
connecting the nodes. For any link eєV, the allowed to have nodes in common, but the links still
characteristics of an edge is described by four tuples must be unique. To discover link-disjoint paths, each
(D, J, B, C), where D, J, B and C denote delay, delay- node forwards only one route request (RREQ) towards
jitter, bandwidth and cost respectively. Among the the destination during the route discovery process.
delay, delay-jitter and cost are additive metrics and The destination node sends a route reply to each of the
bandwidth is a concave metric. For any link, let unique previous hops from which it received a route
assume that D(i,j) is a delay metric, J(i,j) is a delay- request.
jitter metric, B(i,j) is a bandwidth metric and C(i,j) is
a cost metric. For a path ‘p’ = (s, i, j, k …d), let

D(p) = D(s,i) + D(i,j) + . . . + D(n,d) (1)

B(p) = min {B(s,i), B(i,j), . . ., B(n,d)} (2)

J(p) = J(s,i) + J(i,j) + . . .+ J(n,d) (3)

C(p) = C(s,i) + C(i,j) + . . . + C(n,d) (4)

Where D(p) represents the accumulated value of the Figure 3. Finding Link Disjoint paths
Delay, B(p) represents the minimum bandwidth on the
path, J(p) represents the accumulated value of the In the above Fig. 3, node ‘s’ is the multicast source.
delay-jitter, and C(p) represents the accumulated value D1 and D2 are the destinations. The network’s edge
of the cost. are described by four tuples (D,J,B,C) . In this
example, suppose delay constraint D=15, delay-jitter
The QoS based multicast route should satisfy the J=30, Bandwidth constraint B=45 and Cost C = 60.
following constraints. When D1 is the destination, it computes the paths
according to the multiple QoS constraints. The paths
D(p) ≤ Dt, , B(p) ≥ Bt , J(p) ≤ Jt , and S n1n2n3D1, Sn1n2n3n8d1
and path Sn6n7n8D1 does not satisfy the
C(p) ≤ Ct . Where Dt, is delay constrain of a route, Bt delay constraint. The paths sn1n6n7n8d1
is the bandwidth constraint, Jt is the delay-jitter and sn5n6n7n8d1 satisfies the delay
constraint and Ct is the cost constraint of a route. constraint, delay-jitter constraint, bandwidth
Meanwhile, the cost C (p) should be minimum. constraint and cost constraints. Furthermore, the path
sn1n6n7n8d1 has minimum cost among
3.3 Adaptive multi path routing mechanism. these paths. Therefore the primary path should be the
path sn1n6n7n8d1. When d2 is the
MANETs are typically characterized by high destination, it computes the path
mobility and frequent link failures that result in low sn1n6n7n8d2 which should satisfy delay,
throughput and high end-to-end delay. To reduce the delay-jitter, bandwidth constraints and also have
number of route discoveries due to such broken paths, minimum cost.
multi path routing can be utilized so that alternate
paths are available. In this approach multiple paths 3.4. Maintenance of alternate paths
are formed during the route discovery process. All the
paths are maintained by means of periodic update When a link failure occurs due to the node mobility,
packets along each path. At any point of time only the the corresponding node invalidates its routing table
path with minimum cost is chosen for data entry for that destination and sends and Route Error
transmission as primary path. When the primary path (RERR) message towards the source. Each node along
breaks due to node movement, one of the alternate the active path receives route error and invalidates its
paths can be chosen as the next primary path and data corresponding route table entry. Once the source node
transmission can continue without initiating another receives RERR, it switches its primary path to the next
route discovery. Here, we use the link-disjoint best alternate link-disjoint path. If no alternate path is
approach for route discovery. In which paths are
available, at the source it initiates a route discovery. generated at every node. The agents update their
knowledgebase with more recent values. The agent
based architectures provide flexible, adaptable and
asynchronous mechanisms for distributed network
management, and also facilitate software reuse and
maintenance

5. Simulation

The proposed scheme has been simulated in various


network scenarios using NS-2 simulator. A discrete
event simulation is done to test operation effectiveness
Figure 4. Finding alternate paths due to node of the scheme. In this section we describe the
mobility simulation model and the simulation procedure.

In this example (Fig. 4) when node ‘n1’ moves out


of the transmission range, the path 5.1. Simulation model
sn5n6n7n8d1 could be used as alternate
A mobile ad hoc network consisting of ‘n’ nodes is
route to reach d1.
generated by using a random placement of the nodes
and allowed for the free movement within the area of
4. Multicast routing scheme using agents ‘l x b’m2. Each node starts from a random location
and moves in all directions. A maximum number of
Agents are the autonomous programs activated on movements allowed per node every period ‘per’ is
an agent platform of a host. Agents use their own ‘move_max’. The communication range for each node
knowledgebase to achieve the specified goals without is selected as ‘C_ran’. All nodes are considered to be
disturbing the activities of the host. The primary goal non-malicious and are included in the clustering
of an agent is to deliver information of one node to scheme. All nodes must support an agent server,
others in the network. Here, the agents are used to find interpreter and transport mechanism. Every node has
the multicast routes and to create the backbone for enough memory to support the agent’s knowledge
reliable multicasting. An agent consists of three database. Every mobile agent is only allowed three
components. (a) agent identifier (b) agent program (c) hops from the parent node to avoid network
agent knowledgebase. The knowledgebase maintains congestion.
a set of network state variables such as status of the
node (cluster head, source, intermediate, others) RF 5.2. Simulation Procedure
value, available power, bandwidth, number of
movements made in recent interval, delay between To illustrate some results of the simulation, we have
adjacent nodes etc. Agents use the knowledgebase for taken n = 50, l = 1000 m and b = 1000 m, per = 100 s,
reading and updating. move_max = 2, C_ran = 225 m. D=16, J=30, B=45,
C=60 are given as user input for various scenarios.
To initiate multicasting, the mobile agent which The proposed routing scheme is evaluated in terms of
resides in the source node provides group ID and packet delivery ratio, group reliability, control
distributes multicast key to all the group members. overhead and packet delivery latency.
The source node examines all the QoS metric values
of next intermediate node using the knowledgebase of 5.3. Analysis of results
agents. If it satisfies all constraints data transmission
takes place among all multicast members. When an Packet delivery ratio of the proposed multi agent
intermediate node either moves out of the range or system is higher than MAODV and ODMRP (Figure.
fails, the static agent resides at the node that 5). As the nodes move with high speed, the packet
monitored such a situation will find out the new delivery ratio decreases since the reliable nodes may
alternate path with minimum cost between the nodes. change and cause some losses triggering
The alternate path and its connectivity to the network retransmissions.
are broadcast, so that the new forwarding table is
Figure 5. Packet delivery ratio vs. mobility
Figure 7. Packet delivery latency vs. mobility
Control overheads are reduced than MAODV and
ODMRP. (Figure 6). This is due to the fact that Packet delivery ratio decreases with increase in
flooding of agents in all directions is avoided as the group size (Figure. 8) for fixed mobility value (5 m/s).
cluster management is employed. Packet delivery ratio with varying group sizes is
higher than that of MAODV and ODMRP. This is
because of computing and caching multiple paths
during a single route discovery process.

Figure 6. Control overhead vs. Group size

Packet delivery latency is significantly less than


MAODV and ODMRP (Figure. 7) even at higher node
mobility for constant group size. This is due to the fact Figure 8. Packet delivery ratio vs. group size
that the regular maintenance of the paths leads to an
increased availability of valid alternate paths when the On the whole, the performance metrics in terms of
primary path breaks. QoS parameters such as packet delivery ratio, end to
end latency and control overheads are improved when
compared to the existing systems such as MAODV
and ODMRP.

6. Conclusions
This paper proposes a multi path multicast routing
algorithm with multiple constraints based on mobile
agents. It effectively routes data packets to group
members even in case of high mobility and frequent
link failures. It has higher packet delivery ratio as [8] C.C. Chiang, Routing in clustered multi hop mobile
compared to MAODV and ODMRP and reduces the wireless networks with fading channel, in: T.S. Chua, H.K.
network delay and the overhead of control messages Pung, T.L. Kunii (Eds.), Proceedings of the IEEE Singapore
International Conference on Networks, Springer, Singapore,
for routing. This work can further be extended to
1997, pp. 197-211.
include the mobile prediction scheme for multiple
multicast trees. [9] J.H. Cui, L.Lao, M. Faloutsos, M. Gerla, AQoSM:
Scalable QOS multicast provisioning in Diff-Serv networks,
7. References Computer Networks 50 (2005) 80-105.

[10] Luo Junhai, Xue Liu, Ye Danxia, Research on multicast


[1] Luo Junhai and Ye Danxia et al., “A survey of multicast routing protocols for mobile ad-hoc networks. Computer
routing protocols for Mobile ad-hoc networks” IEEE Networks 52(5): 988-997, 2008
Communication surveys & Tutorials, vol.11, no.1, pp.78-90,
First Quarter 2009. [11]. S. S. Manvi, M. S. Kakkasageri, Multicast routing in
mobile ad hoc networks by using a multiagent system. Inf.
[2] C.E.Perkins and P.Bhagwat, Highly dynamic Destination Sci. 178(6): 1611-1628, 2008
Sequenced Distance-Vector routing (DSDV) for mobile
computers, in: T. Imielinski, H.Korth (Eds.), Proceedings of
SIGCOMM’94 Conference on Communications [12] Mohammed Tarique and Kernal E. Tepe et al., Survey
Architectures, Protocols and Applications, ACM, London, of multipath routing protocols for mobile ad hoc networks,
UK, 1994, pp. 234-244. Journal of Network and Computer Applications 32(2009)
1125-1143.
[3] C. Perkins, Ad-hoc On-Demand Distance Vector
(AODV) routing, RFC3561[S], 2003, [13] Xiaoyan ZHU and Jin LIAN A QoS multicast routing
http://www.ietf.org/rfc/rfc3561.txt. protocol with mobile prediction based on MAODV in
MANETs, International Conference on Computer Science
[4] D.B. Johnson and D.A. Maltz et al. The dynamic source and Software Engineering, 355-357, 2008.
routing protocol for mobile ad hoc networks, Internet draft,
draft-ietf-manet-dsr-10.txt, 2004. [14] Huayi Wu and Xiaohua Jia QoS multicast routing by
using multiple paths/tress in wireless ad hoc networks, Ad
[5] J. H. Cui and L.Lao et al. AQoSM: Scalable QoS Hoc Networks 5 (2007), 600-612.
multicast provisioning in Diff-Serv networks, Computer
Networks 50 (2005) 80-105. [15] M. Shree, J.J. Garcia-Luna-Aceves, An efficient routing
protocol doe wireless networks, ACM Mobile Networks and
[6] E.M. Royer and C.E.Perkins, “Multicast operation of the Applilcation Journal 1 (1996) 183-197.
ad-hoc on-demand distance vector routing protocol”, in Proc.
ACM MOBICOM, Aug. 1999, pp. 207-218. [16] M. Saghir. T.C Wan, R. Budiarto, A new cross-layer
framework for QoS multicast applications in mobile ad hoc
[7] M. Gerla and S.J. Lee et al. “On-demand multicast networks, International Journal of Computer Science and
routing protocol (ODMRP) for ad-hoc networks” Internet Network Security 6 (2006) 142-151.
draft, draft-ietf-manet-odmrp-02.txt, 2000.

Вам также может понравиться