Path prediction based on second-order Markov chain for the opportunistic networks

In the opportunistic networks, nodes carry and store the data and forward it until they encounter each other. How to choose an appropriate opportunity to forward data is pivotal for nodes' routing in this type of networks. Since nodes currently will keep a regular movement state in the scene of this paper discussed, forecasting a node's moving track in the near future would be very helpful. Through this way, the node can modify the forwarding strategy in time and increase the success rate of data routing. We proposed a path prediction model based on second-order Markov chain in this paper to analyze and predict node's next path choice at an intersection depending on its statistic of history tracks. Then we use the real data from Dartmouth College to verify the model and the result shows that it is more precise than normal one step Markov models and is more reasonable than some other models such as Manhattan model.