COMSP: Correlated contact and message scheduling policy in DTN

The routing algorithms of DTN (Delay- and Disruption-Tolerant Networks) have not been efficient enough until now, which is one of the hot spots for DTN research. In this paper, aiming at fully using the knowledge inside the DTN environments both locally and globally, we propose a forecast method based on time-varying graph using time series analysis theory. By processing the history contact data with ARMA(Auto Regression Moving Average) model of time series analysis theory, we can predict the start time and the duration of future contact between specified nodes, without needing to know the movement model of the nodes. Then we devise an algorithm based on correlated contact and message scheduling policy (COMSP) with the utility of messages and contacts taken into consideration. From the simulation result, the algorithm performs higher efficiency under the different transmit speed situations than most of the ones for comparing in both the cost and performance(e.g. delivery ratio, latency, hops, buffer time, overhead and so on).