Erlang Based Buffer Management and Routing in Opportunistic Networks

In opportunistic networks the network's topology is not connected due to the intermittent links. Thus, to deliver messages there is not a route between the source and destination node. To solve this problem in opportunistic networks, the routing and forwarding are carried out concurrently. In this approach, which is called store, carry-forward, the selection of the next node for message forwarding is based on the appropriateness of the next node to deliver the message. One of the most appropriate message routing and forwarding algorithms for opportunistic networks is spray and wait algorithm. The selection of the node encountered as the next relay and the number of tokens sent to the next node depend on factors such as the probability of message delivery, buffer status (the probability of message deletion), and the delivery time of the message. In most studies carried out, one of these factors has been effective in selecting the next node. In some others the number of tokens sent to relay node has followed a fixed function. This has resulted in the reduction of the ratio of delivery and it has increased delays. In this paper the proposed algorithm has considered the probability of message delivery, buffer status, and message delivery time concurrently in selecting the relay node and in allocation the tokens sent to the node. The results of simulation have shown that the proposed algorithm has improved the delivery ratio and the delay in delivering messages with a trivial overhead.