Replication probability-based routing scheme for opportunistic networks

Opportunistic networks offer a delay-tolerant end-to-end message delivery in networks with intermittently connected nodes and communication islands. Routing based on knowledge in those networks is classified into two main types which are utility-based and flooding-based approaches. In general, flooding-based approaches result in very high resource consumption and buffer congestion. In this paper, we propose a probabilistic forwarding scheme for opportunistic networks, called Replication Probability-based Routing Scheme (RPRS), based on controlled replication of messages aiming to keep the high delivery ratio while drastically reducing the message overhead. It uses from each single message two local parameters, replication count and hop count, to calculate the desired replication probability, which is used to prioritize the message for replication. Also, the proposed RPRS has its own drop policy whose utility function is calculated as a function of replication count, hop count, and the buffer time of a message which is considered being an estimate of the end-to-end delay. Through this, RPRS allows us to decide when it is desirable to further spread a message and thus reach a high delivery ratio without congesting the network with unnecessary message copies. In simulation results, we analyze the performance of RPRS and compare it with well-known routing protocols such as Epidemic routing and Spray & Wait, with different scheduling and drop policies based on the same variables. Conducted scenarios show that our scheme has better performance regarding delivery ratio, delay, buffer delay and overhead.