Probabilistic Replica Allocation in Multihop Networks

In this paper, we propose a probabilistic replica allocation scheme for reducing the total number of messages in multi hop networks. When data is distributed uniformly into the network such that every node has its own data, there are per-node and a network-wide problems caused by increasing the network traffic. To overcome the problems, although the traditional method accepts all incoming and relaying data for replication, the proposed method randomly accepts the data for replication so as to minimize the probabilistic distances between the data holders and the data requesters, and is supposed to reduce approximately 10% of the total number of messages in the network. We believe that our method requires less computing and communication resources, because our method is a sort of randomized algorithm, easy to implement, and has no communication overhead. The simulation results on wireless ad hoc networks using the JiST/SWANS network simulator show the effectiveness of the proposed method.