Decision Fusion Using Three-Level Censoring Scheme in Sensor Networks under Rician Fading Channels

This work considers the problem of decision fusion using the three-level censoring scheme in wireless sensor networks. An optimal fusion rule is derived for the model of Rician fading channel. However, the optimal fusion rule requires instantaneous channel state information (CSI) which may be too costly for resource constrained sensor networks. Hence, a sub-optimal alternative with the knowledge of Rician fading statistics is proposed. To further simplify the fusion rules, two low signal-to-noise approximations are also derived. Performance evaluation confirms that by using the three-level censoring scheme, the goal of energy saving and performance improvement can be achieved. The sub-optimal fusion rule based on channel statistics exhibits only slight performance degradation compared with the optimal fusion rule. In addition, the impact of Rician K-factor on the performance of decision fusion has also been discussed.