摘要
无线传感器网络环境下受限于能量和带宽,传感器的观测往往需要经过量化后才发送,针对大规模不一致传感网络环境且带宽受限下的分布式状态估计,探讨了量化比特分配和估计性能评估问题。首先给出一种二进制概率量化方案,并基于量化观测构造线性无偏量化估计器。然后,考虑一个总的传输比特率限制下,得到了传感器的最优量化比特率及比特分配方案,由其信噪比(SNR)和总带宽确定。同时,对提出的量化估计器的均方误差上界进行了分析,发现其与理论下界仅相差一个小常数因子。最后,仿真结果表明,相比于一般的均匀比特分配,文中所提出的最优比特分配方案估计性能更优。
Due to energy and bandwidth constraints in wireless sensor networks,sensor observations are usually quantized before transmission. We discuss the problems of quantization bit allocation and evaluation of estimation performance for distributed state estimation in large-scale inhomogeneous sensing environment and bandwidth-constrained networks. First,a binary probabilistic quantization scheme is given and the corresponding linear unbiased quantization estimator is established based on quantized observations. Subsequently,under a total bit rate constraint,we obtain the optimal quantization bit rate of each sensor and corresponding bit allocation scheme,which is determined by the signal-to-noise ratio(SNR)and total bandwidth. Furthermore,the mean-square error upper bound of proposed quantization estimator is analyzed,and it is shown that the proposed scheme achieves a MSE performance that is within a small constant factor of the theoretical lower bound. Finally,simulations show that the proposed optimal bit allocation scheme has better estimation performance when compared with traditional uniform approaches.
作者
陈军勇
邬依林
CHEN Junyong;WU Yilin(School of Information and Electronic Engineering,Zhejiang University of Science and Technology,Hangzhou 310023,China;Department of Computer Science,Guangdong University of Education,Guangzhou 510310,China)
出处
《传感技术学报》
CAS
CSCD
北大核心
2020年第2期286-292,共7页
Chinese Journal of Sensors and Actuators
基金
浙江科技学院科研启动基金项目(F701106J02)
广东省普通高校特色创新项目(自然科学类)(2018KTSCX163)
国家自然科学基金项目(61273214)
广东省广东第二师范学院计算机实验教学示范中心(2018sfzx01)
广东省科技计划项目(2016A010106007,2016B090927010)
广东第二师范学院网络工程重点学科项目(ZD2017004)。
关键词
无线传感器网络
分布式状态估计
最优量化比特率
信噪比
均方误差
wireless sensor networks
distributes state estimation
optimal quantization bit rate
signal-to-noise ratio
mean-square error
