Link Prediction for Opportunistic Network Based on the Attraction between Nodes

Abstract

Opportunistic network is a type of self-organizing network that uses node movements to bring about encounter opportunities for communication, and characterized by sparse node connections and frequent network topology changes. The prediction of node link probability is the key to study the topological changes of opportunistic network. We propose a node-attraction-based link prediction method of opportunistic network (OLPA), which analyzes node properties and the connection relationship of node pairs to obtain the attraction between nodes, and introduces a time decay factor to weight the evolution sequence of attraction to obtain the probability of future connections between nodes. By comparing with LSTM, E-LSTM-D, GCN-GAN link prediction models on multiple real opportunistic network data sets, the propose link prediction method we mentioned has good accuracy.