New fusion rule for censored decisions in wireless sensor networks
Xiangyang Liu, Yingning Peng · 2006
A new mixed fusion rule (MFR) was developed to overcome the shortcomings of previous developed Likelihood ratio test based on channel statistics (LRT-CS) for the problem of fusing censored binary decisions transmitted over Nakagami fading channels in a wireless sensor network (WSN). LRT-CS relies on local sensor's detection probability, while the detection probability is a priori unknown for uncooperative targets. Meanwhile, for the Nakagami fading channel, the LRT-CS is involved in calculating an infinite series, which is cumbersome for real-time application. In contrast, the MFR is a combination of some simple fusion rules and do not require the detection probabilities of local sensors. Furthermore, it has small performance degradation in comparison with the LRT-CS. Considering the complexity of implementing the LRT-CS, the mismatch between nominal detection probability and the practical one of local sensor and the simplicity of our method, it deserves to adopt it in practice