Bidding and Voting Strategy for Energy Efficient Collaborative Target Classification in Wireless Sensor Networks

Recent advances in wireless communications and micro-electro-mechanical systems have fostered the development of wireless sensor networks (WSNs) which consist of tiny, low-cost and low-power sensor nodes. Individual sensor nodes only have local views of the environment, but global views can be obtained by collaborative processing between densely deployed sensor nodes. Energy efficiency is critical for battery powered WSNs to prolong network lifetime. It is very challenging to concurrently achieve energy efficiency and collaborative processing in WSNs. Inspired by some commercial practices, we proposed the bidding and voting strategy (BVS), and apply it to vehicular target classification in WSNs. In BVS, bidding efficiently selects sensor nodes for collaborative classification while voting improves classification decision accuracy by combining local decisions. Vehicular targets are classified by means of support vector machine (SVM). Simulation experiments with real world data are conducted and the results show the proposed strategy can guarantee target classification accuracy while achieve energy efficiency.