Development of a Reliable Multicast Protocol in Mobile Ad Hoc Networks

Mobile ad hoc network is a collection of mobile nodes forming dynamic and temporary network. The mobile nodes work in collaborative nature to carry out a given task. It can receive and transmit data packets without the use of any existing network infrastructure or centralized administration. Multicasting is among the pertinent issues of communication in such networks. The reliable delivery of multicast data packets needs feedback from all multicast receivers to indicate whether a retransmission is necessary. The Feedback Implosion Problem (FIP) states that reliable multicast in ad hoc networks suffers from redundant feedback packets, loss, duplication, and out-of-order delivery of data packets. To carry out this task, several reliable multicast protocols have been proposed to reduce the number of feedback packets from the receiver nodes. This is achieved by placing the responsibility to detect packet loss and initiating loss recovery timer on the receiver nodes which is complemented by feedback suppression. The initiating loss recovery timer depends on the number of hops between the nodes. As the dynamic nature of the number of hops between the nodes in ad hoc networks is unstable the loss recovery timer become inaccurate. Thus, the inaccuracy of the loss recovery timer, in return, causes extra overhead and more delays. The main objectives of this research are to enhance the FIP and decrease the recovery delays in reliable multicast protocol for mobile ad hoc networks using suggested approaches. First, the Source Tree Reliable Multicast (STRM) protocol adopting a novel technique to select a subset of one-hop neighbors from the sender node as its Forward Servers (FS). The key idea behind selecting this subset one-hop neighbors is to forward the retransmitted lost data packets and to receive the feedback packets from the receiver nodes. Second, proposed two algorithms to improve the performance of the STRM protocol. The first algorithm is developed to avoid the buffer overflow in the FS nodes. This is achieved by managing the buffer of the FS nodes; by selecting the FS nodes depending on the empty buffer size it has and reducing the amount of feedback sent from the receiver nodes to their FS node. The second algorithm is developed to decrease the number of duplicated packets in the multicast members in the local group. This is achieved by sending the repair packets only to the member that has requested it. The FS in the local group should create a dynamic and temporary sub group whose members are only the members that requested the retransmission of the repair packet. The approaches were tested using detailed discrete-event simulation model which was developed encompassing messaging system that includes error, delay and mobility models to characterize the performance benefits of the proposed algorithms in comparison to ReMHoc protocol. Our approaches achieve up to 2.19% improvement on average packet delivery ratio, 3.3% on requested packets, and 46% on recovery latency time without incurring any additional communication or intense computation.

Preview PDF FSKTM_2008_16a.pdf Download (654kB)

Actions (login required)

View Item