.png){kind=icon}
.png){kind=icon}
Return
Quantized Distributed Estimation With Adaptive Combiner
.png){kind=icon}
PRE
AI
DOI：10.1109/TSIPN.2022.3156425.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Distributed estimation over networks has attracted great attention for its wide applicability in source tracking, environmental monitoring, etc. In the problem of distributed estimation, a set of nodes collectively estimates some parameter from noisy measurements. Most distributed algorithms assume that the information shared among nodes has full-precision. However, this assumption is unrealistic as communication channels among nodes have limited bandwidth. Considering this, in this paper, we impose quantization constraints on distributed estimation problem. That is, nodes can only receive quantized information from neighbors. The quantization inevitably brings in errors. Each node has to decide carefully which information provided by the neighbors is more trustworthy since these information could suffer from various degrees of distortion caused by the errors. To make good use of the aggregated information, we propose an adaptive combination strategy for quantized distributed estimation algorithms. We derive sufficient conditions for convergence of the resulting algorithm through mean-square analysis. The advantages of the proposed adaptive combiner over fixed combiners are illustrated experimentally.
Keywords:
Distributed estimation
distributed algorithm
diffusion
quantization
adaptive combiner
Journal
IF:
4.9
Papers:
727
Citations:
1.9K