New Energy Aware MPR Selection for Securing OLSR Routing Scheme Under Black Hole Attack: A Machine Learning Approach

Mobile Ad Hoc network (MANET) is a collection of high speed mobile nodes and exchanges the data packets wirelessly using electromagnetic waves or radio transmission links. However, the data transmission through wireless channel is vulnerable to the intrusions in the network and may lead to disabling of seamless network services. Therefore, security is another essential parameter for successful dissemination of the data besides design of optimized Routing in MANET. Towards this, Machine Learning frame work known as Deep Belief Network Based Secure Optimized Link State Routing (DBN-OLSR) Protocol is presented in MANETs. The proposed DBN-OLSR approach includes series of steps for networking of data packets: firstly, Multipoint relay nodes are selected using real time metrics of the nodes like, hop-count and residual energy to reduce the unnecessary transmission of beacon messages in route selection. Secondly, secure data transmission is achieved by preventing single black hole attack using Machine learning based Deep belief network (DBN) and Naive bayes classifier. The performance of the proposed DBN-OLSR protocol is compared with traditional OLSR and on demand routing (AODV) schemes. The simulation results were tested for different nodes and illustrates that the proposed DBN-OLSR outperforms in terms of intrusion detection accuracy, routing overhead, packets lost, throughput, execution time and average end-to-end delay over normal OLSR and AODV routing schemes.