Sensor Deployment for Wireless Sensor Networks: A Conjugate Learning Automata-Based Energy-Efficient Approach

Abstract

Wireless sensor networks (WSNs) boost the development of the Internet of Things by their ability to monitor environments and flexibility. It is desirable to study an efficient configuration of a WSN that balances its energy consumption and its functionality. In this article, we propose to formulate the sensor deployment task as a combinatorial optimization problem and introduce an effective sensor deployment paradigm in which both the randomness and the dynamics of the environment are captured. Following the activity scheduling mechanism, we adopt a powerful non-associative reinforcement learning method, conjugate learning automata (CLA), to learn the optimal sensor deployment strategy. Compared to conventional methods, the proposed CLAbased sensor deployment method yields good performance by activating only a subset of all sensors and does not lean on prior expertise about environments. Meanwhile, the learning process is efficient, and thus energy is saved in multiple aspects. Comprehensive experiments under different settings demonstrate the effectiveness of the proposed method.