An adaptive censored summation fusion scheme for distributed detection

Abstract

An adaptive censored summation fusion (ACSF) is proposed to improve the detection performance of a distributed detection system based on local test statistic (LTS) when the SNR at each local sensor is unknown and even scan-to-scan variant. The proposed ACSF first estimates the SNR of each sensor and then discards the LTSs with the smallest SNRs. Finally, the proposed ACSF compares the sum of the SNRs in the remaining LTSs with a detection threshold to make a global decision. Numerical results show that, when there is a great deal of variation among the SNRs at local sensors, the ACSF performs much better than the sum fusion, the optimum scheme in the sense of the generalized Neyman–Pearson criterion. Conversely, when the SNRs of all the sensors are identical, the ACSF has almost the same detection performance as the sum fusion. Therefore, the proposed ACSF is robust with high performance.