Optimal fusion scheme in wireless sensor networks under sum-rate capacity constraint

In this paper, we investigate optimal fusion schemes for decentralized detection problems in which a network of wireless sensors provides relevant information about the state of nature to a fusion center. Upon independent reception of the signal, each sensor performs distributed binary detection and sends the decision as well as the reliability of the decision to the fusion center. The capacity region of the corresponding multi-access channel (MAC) from wireless sensors to the fusion center is identified and the sum-rate capacity is characterized. Further-more, it can be shown that when the number of wireless sensors is big enough, the sum-rate capacity of the corresponding MAC channel converges to a constant due to ¿channel-hardening¿. Based on this result, we analyze the performance of quantizing the reliability of the decentralized decision under the sum-rate capacity constraint. The corresponding optimal fusion scheme is derived and the analysis is conducted under the low signal-to-noise (SINR) regime to better reflect the reality.