There are two main classes of decision fusion methods, namely hard decision fusion (HD) and soft decision fusion (SD), in which the number of bits transmitted by each local sensor to the fusion centre (FC) is always same, namely one bit in HD and n ( n ≥ 2) bits in SD. However, considering that there is always a limit of bandwidth in a distributed detection system, the number of bits sent by each local sensor to the FC does not need to be the same and should be allocated reasonably and suitably. Therefore, this study proposes an optimal bit allocation scheme based on the memetic algorithm, in which the number of bits transmitted by each local sensor could be different. This scheme aims to maximise the detection probability under the limit of bandwidth for a detection system with imperfect channels. The overall detection probability objective function about the number of allocated bits is derived. To optimise this objective function, an improved memetic algorithm with two local adjustment strategies, namely non-elite learning local adjustment optimisation strategy and elite greedy local adjustment optimisation strategy, is proposed to allocate the optimal number of bits. Simulation results show the efficiency and effectiveness of the proposed scheme.

中文翻译：

---

具有不完善信道的分布式检测系统的最优比特分配方案

决策融合方法主要有两类，即硬决策融合（HD）和软决策融合（SD），其中每个本地传感器传输到融合中心（FC）的位数始终相同，即一位在高清和 ñ （ ñSD中的≥2）位。但是，考虑到分布式检测系统中总是存在带宽限制，每个本地传感器发送给FC的位数不必相同，应该合理，适当地分配。因此，本研究提出了一种基于模因算法的最优比特分配方案，其中每个本地传感器发送的比特数可以不同。该方案旨在在信道不完善的检测系统的带宽限制下最大化检测概率。得出关于分配比特数的总体检测概率目标函数。为了优化此目标函数，采用了两种局部调整策略的改进型模因算法，提出了非精英学习局部调整优化策略和精英贪婪局部调整优化策略，以分配最优比特数。仿真结果表明了该方案的有效性。