噪声相关的带宽约束传感器网络融合算法

Bandwidth-Constrained Fusion Method for Sensor Networks with Correlated Noises

研究了一类相关量测噪声和带宽限制背景下的传感器网络融合问题.利用Cholesky分解方法将量测方程转化为量测噪声互不相关的等价模型.由于带宽的限制,各局部传感器节点的原始信息需量化成消息才能上传到融合中心.文中采用自适应的量化策略获得了局部测量新息的量化消息,并利用顺序滤波和强跟踪滤波技术设计融合方法.简要分析了基于量化新息融合算法的性能特点.通过一个计算机仿真实验验证了新算法的有效性.

Consider the fusion estimation problem of a bandwidth-constrained sensor network with correlated noises. First, the measurement equations with correlated measurement noises are transformed to the equivalent models with uncorrelated measurement noises by using the Cholesky factorization method. Due to limited bandwidth, only quantized messages of the original information from local sensor can be transmitted to fusion centre. In this paper, an adaptive quantization strategy is adopted for obtaining quantized messages of the local measurement innovations, whereas sequential filter and strong tracking filter are used to design fusion algorithms, Subsequently, a brief analysis of the performance characteristics of the fusion algorithm based on quantized innovations is introduced. Finally, a computer simulation shows the effectiveness of the proposed method.