Underwater Distributed Detection And Data Fusion Methods For Binary Hypothesis

Whether it comes to aerial radar or underwater sonar,target-detection is an indispensable tactical function,and its related research has always occupied a place in the military or civilian field.How to improve detection performance has become an unavoidable problem in detection theory research.In recent years,the field of unmanned cluster cooperative detection is at the cutting edge,and all countries in the world are striving for technological superiority and leading power in this field.The United States,the European Union and other countries have done a large amount of scientific research and obtained practical results in this field.Unlike traditional single-sensor detection or centralized detection,multi-sensor distributed detection greatly reduces the computational complexity of the fusion center,thereby improving the overall performance of the detection system.This paper studies the multi-sensor distributed detection and data fusion technology based on Bayesian theory,because it can reduce the complexity of the fusion center decision and transmit the data to the bit level,which helps to compress the necessary information under the water to improve accuracy rate of transmission,and aims to implement a low-complexity,high-performance detection system for underwater complex and time-variant acoustic channel.This paper focuses on the methods of binary hypothesis detection and fusion of multi-sensor networks for different topologies,including parallel distributed detection and data fusion based on Bayesian minimum cost criteria,serial distributed detection and data fusion and hybrid distributed detection and data fusion and parallel feedback distributed detection and data fusion,and distributed constant false alarm detection based on Neyman-Pearson criterion.The paper first analyzes the theory and characteristics of the five distributed detection and data fusion methods,and compares the detection performance under different topologies.And the simulations related verify that the distributed detection fusion method will improve the detection performance of the original single sensor,and the performance of the distributed detection system will be better as the number of sensors increases.Finally,the paper discusses the constraints or limitations that may be encountered in implementing projects based on distributed detection and data fusion.The simulations of this thesis show that the Bayesian distributed detection and data fusion is the basic technology to reduce the computational complexity and improve the overall detection performance of the cluster.It is suitable for underwater transmission-rate-low and time-variant acoustic channels and can provide theoretical support for experiments or applications in the field of cluster cooperative detection in the future.