Intelligent Mobile Video Surveillance System as a Bayesian Coalition Game in Vehicular Sensor Networks: Learning Automata Approach

In a mobile video surveillance system (MVSS), an efficient approach is required, so that captured video can be transmitted to its final destination under tight constraints of delay and accuracy. This paper presents a new intelligent MVSS using the concepts of Bayesian coalition game and learning automata (LA). These LA are assumed to be the players in a game and are deployed on vehicles. Coalition among players is formed using the Bayesian Coalition Game Theory. To decrease the delay that occurred during transmission of captured video to the nearest access points, the best path is chosen based on a new metric called Path Score, which is computed by each player in the game. For each action performed by the automata, their actions may be rewarded or penalized by a value, which is defined as a sequence, with respect to the inputs provided from the stochastic environment. According to the reward or penalty received from the environment, the automata update their action probability vector. After 15 iterations, a Nash equilibrium is achieved in the game by defining a twice-differentiable function in Banach spaces, and convergence of sequence is proved using the Cauchy convergence theorem. The performance of the proposed scheme is found to be better in comparison to the other state-of-the-art schemes, with respect to various performance evaluation metrics.