Mean drop an effectual buffer management policy for delay tolerant network

Abstract

Traditional routing protocols e.g. TCP/IP, AODV cannot be configured in environments where end-to-end path is not stable due to network partitioned, dynamic topology changes and node mobility. Delay tolerant Network (DTN) has emerged as a technology by which the communication architecture is formed through intermittently connected mobile nodes. These networks are categorized as opportunistic since further availability of connections cannot be predicted. Thus, the routing protocols transmit the multiple copies of each message around the connected nodes. Each node then carries this message copy in its buffer until the transmission opportunity arises. This hop-by-hop mobility of message(s) makes the quick progress towards destination. However, such redundant diffusion of messages produces high congestion in the network and a situation occurs where new message arrives at a node and its buffer is full. In this case, the node must drop the buffered message(s) to sustain this new arrival. The function of buffer management policy is to minimize this drop. In this paper we propose a buffer management policy called as MeanDrop. This policy computes the mean of buffered messages at the congested node and drops only those message(s) which have the size greater than or equal to this mean value. The simulation results have shown that the proposed policy MeanDrop (MD) out performs well as compared to existing Evict Shortest Time First (ESTF) and Evict Most Forwarded Fist (MOFO) in terms of message relay, message drop, delivery probability and overhead.