Performance of Decentralized Detection in Large Sensor Networks: Impact of Different Binary Modulation Schemes and Fading in Sensor-to-Fusion Center Link

摘要

We study the performance of a decentralized detection system in which a large number of independent and identical sensor nodes transmit their binary decisions regarding the presence or the absence of a phenomenon of interest (POI) to a fusion center. Each link between a sensor and the fusion center is modeled as an independent and identically distributed slow Rayleigh-fading binary modulation signal received in AWGN. The fusion center counts the number of decisions favoring the presence of POI in order to arrive at a final decision. For three standard modulation techniques we study (ⅰ) the impact of the sensor-fusion center link on the quality of the decision received at the fusion center and (ⅱ) the minimum required sensor decision quality, given the availability of a minimum sensor-to-fusion link SNR, in order that the asymptotic (large number of sensors) error in the counting rule classification goes to zero. With a proper choice of threshold for noncoherent OOK detection, it is shown that an asymptotic performance comparable to that of FSK, while achieving some energy saving, is possible.