Cooperative Environmental Monitoring for PTZ Visual Sensor Networks: A Payoff-based Learning Approach

TL;DR

This paper introduces a distributed, payoff-based learning method for cooperative environmental monitoring using PTZ visual sensor networks, enabling sensors to optimize coverage without prior environmental knowledge.

Contribution

It presents a novel game-theoretic framework and a payoff-based learning algorithm for distributed, cooperative environmental monitoring in PTZ sensor networks.

Findings

Effective in maximizing group monitoring objectives

Operates without prior environmental information

Validated through experimental results

Abstract

This paper investigates cooperative environmental monitoring for Pan-Tilt-Zoom (PTZ) visual sensor networks. We first present a novel formulation of the optimal environmental monitoring problem, whose objective function is intertwined with the uncertain state of the environment. In addition, due to the large volume of vision data, it is desired for each sensor to execute processing through local computation and communication. To address the issues, we present a distributed solution to the problem based on game theoretic cooperative control and payoff-based learning. At the first stage, a utility function is designed so that the resulting game constitutes a potential game with potential function equal to the group objective function, where the designed utility is shown to be computable through local image processing and communication. Then, we present a payoff-based learning algorithm so that the sensors are led to the global objective function maximizers without using any prior information on the environmental state. Finally, we run experiments to demonstrate the effectiveness of the present approach.