Page 113 International Journal for Research in Science & Advanced Technologies Bullibabu * et al. Issue-1, Volume-3, 113-117 ISSN 2319-2690 Nov-Dec 2012 Edition TRAFFC CONGESTION CONTROL IN MOBILE AD-HOC NET WORKS R. Bullibabu 1 and J.V.N. Ramesh 2 1 Associate Professor, Department of CSE, St. Anns College of Engineering and Technology, Chirala 2 Associate Professor, Department of CSE, Narasaraopeta Institute of Technology, Narasaraopet. Abstract —This paper presents a multi-rate multicast congestion control scheme for Mobile Ad-hoe Networks (MANETs). Not only does the proposed scheme overcome the disadvantages of existing multicast congestion control protocols which prevent them from being used in MANETs, but it also achieves good performance in other aspects such as fairness with TCP, robustness against misbehaving receivers, and traffic stability. Besides achieving the above advantages, the proposed scheme does not impose any significant changes on the queuing, scheduling or forwarding policies of existing networks. I. INTRODUCTION Existing multicast congestion control schemes generally fall into two categories: single-rate and multi-rate. Multi-rate schemes (e.g., [ 1] [2] [3] [4]) usually offer much more freedom to receivers in choosing appropriate receiving rate than single-rate schemes (e.g., [5] [6] [7]). Because the links of a multicast tree are usually heterogeneous, receivers in a multicast session may have diverse amounts of available bandwidth. So multi-rate schemes have a great advantage over single-rate schemes in catering to every receiver in a multicast session. This paper presents a new multi-rate multicast congestion control scheme suitable for Mobile Ad- hoc Networks (MANETs). For transport protocols not specifically designed for MANETs, the main sources of problems in MANETs are high link error rates, limited bandwidth, link access delays, and hand-offs. Almost all existing multicast congestion control schemes will suffer from the same problems as TCP suffers in MANETs (e.g., unnecessarily reducing the transmission rate in response to link errors). This is because they use losses as the indication of congestion but cannot distinguish between link-error losses and congestion losses. Another specific problem for multi-rate schemes is the link access delay in MANETs caused by access competition. Because of the inherent design of the IGMP protocol, the layer-drop latency is already a significant problem in wireline networks for multi-rate schemes [3] [4]. The link access delay in MANETs caused by competition will exacerbate the layer-drop latency problem, because pruning information can reach a upstream router only after the upstream link has been successfully accessed, and in congested situations, there is a significant delay before the upstream link becomes available. Although some schemes such as [3] [4] have made a significant progress in combating this problem, they usually introduce considerable control traffic overhead, which is a serious disadvantage in MANETs (e.g., valuable bandwidth and power are wasted.). Besides the disadvantages specific to MANETs, most existing schemes still have problems in sharing bandwidth fairly with TCP [8] [9] [3] [4] and dealing with misbehaving receivers. To deal with the above disadvantages of existing schemes, instead of depending on individual receivers to detect congestion and adjust their receiving rates, the scheme proposed in this paper adjust multicast traffic rate right at each bottleneck of a multicast tree. Specifically, when congestion occurs or is about to occur at a branch, some layers of the multicast sessions traversing the branch are "blocked" from entering the branch; when the branch is lightly utilized, some blocked layers are "released" to traverse the branch. The proposed scheme overcomes most of the disadvantages of existing schemes. First, link errors cannot cause the proposed scheme to wrongly block a layer, because the queue state at a bottleneck, instead of the loss information at receivers, is used as the metric to adjust the multicast traffic rate at the bottleneck. Second, the link access delay caused by competition in MANETs cannot hinder the rate adjustment of the proposed scheme, because, instead of depending on receivers to request pruning to drop layers, the scheme blocks multicast layers right at each bottleneck of a multicast tree. Third, the proposed scheme only introduces very limited control traffic overhead because of the on-the-spot information collection and rate control. Besides the above features that enable it to work effectively and efficiently in MANETs, the proposed scheme also has good performance in fair bandwidth sharing with TCP, robustness against misbehaving receivers, and traffic stability. Moreover, the proposed scheme does not impose any significant changes on the queuing, scheduling or forwarding policies of existing networks. The rest of the paper is organized as follows. Section II introduces the proposed scheme, and Section III analyzes the proposed scheme for fairness, effectiveness, and cost. Simulation results are presented in Section IV The summary appears in Section V. II. THE MULTI-RATE CONGESTION CONTROL SCHEME The proposed scheme operates in the following way. When multicast sessions traverse a link, the scheme agent starts to observe the output queue of the link and the traffic passing the link. When the number of packets in the queue, NQuPkt, exceeds a threshold, QuThreSh2, some layers of multicast sessions are blocked from entering the link.