Cluster-based fusion detection of soft and hard decisions for underwater non-cooperative targets

Abstract

Underwater target detection performs the essential contribution in marine exploration. Unfortunately, the little energy available in the underwater environment makes detection difficult. To address the aforementioned obstacles, this paper proposed a detection scheme based on soft and hard decision fusion. The detection scheme consists of two stages: network topology control and the fusion of local and global detection. The first stage, an improved low energy adaptive clustering hierarchy (LEACH) algorithm is introduced into the underwater sensor network to choose the most suitable cluster heads and to introduce relay nodes to establish multi-hop routes between clusters, effectively balancing node energy consumption and extending network lifetime. With the network topology, the generalized likelihood ratio rule (GLRT) is employed by each sensor to detect the underwater non-cooperative target and make the corresponding decision during the second phase. Specifically, the member nodes make the multi-bit local soft decision, and then the cluster head node fuses the decision information from each member nodes to make the one-bit hard decision and transmits it to the fusion center, and finally the fusion center makes the global decision, such that the tradeoff between detection accuracy and energy consumption may be balanced by generating the soft and hard decision fusion technique. Simulations show that the strategy maintains detection performance extremely comparable to that of centralized and conventional multi-bit soft decisions while eliminating communication overhead.