Oruji F, Zayyani H, Kuhestani A. Communication Reduction in distributed estimation using correntropy as an information criterion. Journal of Iranian Association of Electrical and Electronics Engineers 2023; 20 (3) :97-106
URL:
http://jiaeee.com/article-1-1471-en.html
Qom University of Technology
Abstract: (693 Views)
Distributed estimation in communication networks is an issue that has received a lot of attention recently. One of the most challenging issues in these networks is the communication reduction of the network. Recently, researchers have proposed methods to reduce the communication load that while reducing the communication load in the network, have less impact on the final performance of the network. For signal processing in nonlinear space, correntropy has been regarded as an information criterion in recent years. Of course, in Gaussian space, the use of correntropy does not always work well, but if we have impulse noise, the cost function based on correntropy is better than the cost function based on square estimation error. In this research work, network noise is impulse noise. To have less destructive effect on the final performance of the network, a new solution is presented based on information criteria. Correntropy is a good criterion as a measure of similarity between two random variables that are in the same space and also because correntropy is not sensitive to sudden changes. Node deletion is performed intelligently and based on correntropy between the reference node and neighboring nodes. Also, a mathematical analysis shows that the condition of deletion of nodes is not dependent on the parameters of the correntropy kernel. The simulation results show that this method has less final MSD than other methods of reducing the network communication load.
Type of Article:
Research |
Subject:
Communication Received: 2022/05/22 | Accepted: 2022/12/26 | Published: 2023/05/24