Direct Perceptive Routing Protocol for Opportunistic Networks

Abstract

Setting up a path from source node to destination node in an Opportunistic Network (OppNet) proves to be a very strenuous task because of two reasons, lack of infrastructure and constantly changing environment. In OppNet the message gets transferred in a store-carry-forward way. Security issues caused by malicious nodes such as Sybil Attack and Selective Forwarding Attack create an abrupt drop in packets tending to cause a lower delivery ratio and greater latency. Hence, we require a smart and secure store carry forward technique. In this paper a Deep Learning based routing protocol called Direct Perceptive Routing(DPR) has been proposed. It uses memory from individual nodes and gathers past experiences to make decisions. The protocol proposed gives an improved message delivery ratio, average hop count and overhead ratio when compared with other high performing protocols such as PRoPHET, Epidemic, HBPR and KNNR.