Estimation of Random Mobility Models using the Expectation-Maximization Method

Random mobility models (RMMs) capture the statistical movement characteristics of mobile agents, and have been widely used for the evaluation and design of mobile wireless networks. In many RMMs, the movement characteristics are captured as stochastic processes constructed using two types of independent random variables. The first type describes the movement characteristics for each maneuver, and the second type describes how often the maneuvers are switched. In this paper, we develop a generic method to estimate RMMs that are composed of these two types of random variables. In particular, we formulate the dynamics of movement characteristics generated by the two types of random variables as a special Jump Markov system, and develop an estimation method based on the Expectation-Maximization principle.