PROJECT TITLE :

Distributed Coverage Control of Mobile Sensor Networks in Unknown Environment Using Game Theory: Algorithms and Experiments - 2018

ABSTRACT:

This Project studies the coverage drawback in an unknown surroundings by a Mobile Sensor Network (MSN). Every agent in the MSN has Communication, sensing, moving, and computation capabilities to complete sensing tasks. These agents would have some limitations on time and energy to accomplish their tasks that need to be thought of by the designers. Here, the agents need to relocate themselves, from their initial random locations, to their optimal configuration. An algorithm based on game theory is proposed, where a collection of distributed agents communicate with local neighbors and use their local data build decisions. A state-based mostly potential game is outlined in which every agent's utility perform is designed to contemplate the trade-off between the worth of the lined space and the energy consumption. The agents employ binary log-linear learning to update their actions in each iteration in order to converge to the Nash equilibrium. Because the agents do not have the information of the sensing area, a Gaussian Mixture Model (GMM) is used to model the distributions of the value in the sensing space. To estimate the unknown parameters of the GMM, a Most Likelihood (ML) estimation scheme is utilized, where an expectation-maximization algorithm is used as a tool to solve the ML recursively. Then, so as to feed the estimation algorithm with additional informative information, a mutual data term is added to the agents' utility functions. The mutual info is used to determine that observation can improve the agent's information of the unobserved space a lot of. Each simulation results and experimental results on a multi-robot platform are provided to validate the performance of the proposed algorithm.