Publication

Abstract : In Opportunistic Mobile Network, routing remains as a challenging issue since participating nodes are strangers to each other and are not trustworthy. An efficient routing model entitled Socialized Proficient Routing (SPR) using Machine Learning (ML) technique is proposed in this paper. In SPR, the relay nodes are selected based on human-social characteristic of the nodes, in-order to attain high trustworthiness. SPR model embodies three phases. In feature selection phase, the significant features are extracted from the training dataset using Boruta wrapper algorithm. Naïve-Bayes, Decision-Tree, Neural-Networks, Support-Vector-Machine, and Random-Forest (RF) are the different ML classifiers used in the training phase. Testing phase accurately selects the trusty neighbour (friendship) nodes for routing. This model is investigated over MIT reality mining dataset and is evaluated using Opportunistic Network Environment simulator. Experimental results prove that SPR_RF performs the best among the classifiers with 0.93 Message-Delivery-Probability, 894.91s AverageDelivery-Delay, 3.08 Average-Hop-Count, Zero Dropped-Message and 45.15 Overhead-Ratio.