不依赖水声传感器信噪比的分布式量化检测融合方法

Distributed quantized detection fusion method independent of underwater acoustic sensor′s signal-to-noise ratio

针对水下跨平台协同系统信号检测融合的应用需求,本文提出了一种不依赖于水声传感器信噪比先验信息的水下目标分布式量化检测融合方法,不仅提升了检测融合系统的性能,而且避免了以往检测融合需要知道各传感器信噪比的难题。在奈曼-皮尔逊准则下,推导了量化检测融合的全局最优算法,在量化检测融合系统满足虚警概率要求的条件下,通过联合设计检测融合规则和各传感器判决规则,实现了对多位量化判决结果的最优判决。给出了量化检测融合的次优算法,通过独立设计融合规则与各传感器判决规则,使融合中心判决门限和各传感器判决门限仅仅依赖于与噪声方差密切联系的虚警概率,而不再依赖于信噪比。进行了计算机仿真,结果表明:所提算法性能优于参加融合的各传感器检测性能以及全局最优硬判决融合的性能,量化检测融合次优算法性能接近全局最优量化检测融合的性能。

To meet the needs of signal detection and fusion in underwater cross-platform collaborative systems,a new distributed quantized detection fusion method is proposed.This method is independent of prior information on the acoustic sensor′s signal-to-noise ratio.It not only enhances the performance of the detection fusion system but also avoids the need to know the signal-to-noise ratio of each sensor.First,using the Neyman-Pearson criterion,we derive the globally optimal algorithm for quantized detection fusion.This algorithm ensures that the detection fusion system meets the false alarm probability requirements.By jointly designing the detection fusion rules and individual sensor decision rules,we achieve the optimal decision for multiple quantized judgments.Subsequently,we introduce a suboptimal algorithm for quantized detection fusion.In this algorithm,the fusion center decision threshold and sensor decision thresholds are designed to depend only on the false alarm probability closely related to noise variance rather than the signal-to-noise ratio.This simplifies the process by using independent fusion rules and sensor decision rules.Finally,computer simulations show that the proposed algorithm outperforms the detection capabilities of individual sensors participating in fusion and globally optimal hard-decision fusion.The performance of the suboptimal quantized detection fusion algorithm closely matches that of the globally optimal quantized detection fusion algorithm.