A lightweight multi-parameter fusion link quality estimation method

Abstract

The invention discloses a lightweight multi-parameter fusion link quality estimation method. The method performs data collection and offline training in advance, and establishes a logistic regression model for link quality estimation; Kalman filtering preprocesses the two physical layer parameters of SNR and link quality indication, which solves the problem of large fluctuations in physical layer parameters obtained by traditional processing methods, and makes the link quality estimation method not sensitive to instantaneous link fluctuations. sensitive. The method of weighted Euclidean distance is used to effectively fuse the two physical layer parameters with low overhead, which solves the problem that the traditional link quality estimation method uses a single physical layer parameter or uses a multi-parameter fusion algorithm that is too complex. On this basis, the fusion parameters are mapped to the packet reception rate through the logistic regression model established offline, and then the quantitative estimation of the link quality is realized.

Description

technical field [0001] The invention relates to the field of wireless communication and network, in particular to a lightweight multi-parameter fusion link quality estimation method in a low-power wireless network. Background technique [0002] Low-power wireless networks represented by wireless sensor networks have been successfully applied in many fields such as military investigation, environmental monitoring, industrial production and medical care. Such networks usually use low-power radio frequency transceivers, so that the wireless link between nodes is greatly affected by the environment, so the link stability is poor and there are more fluctuations. In order to improve the transmission efficiency of the network and minimize the packet retransmission overhead caused by using low-quality links, real-time and accurate estimation of the link quality is required to find the optimal end-to-end route. Therefore, the performance of link quality estimation methods is critica...

AI Technical Summary

Problems solved by technology

[0004] Existing link quality estimation methods usually use window averaging or standard Kalman filtering. When there are many fluctuations in the link itself, the effect is not ideal.

In addition, the existing methods either use only one physical layer parameter to estimate the link quality, which makes it difficult to accurately describe the real quality of the link, or the multi-parameter fusion algorithm used is too complex or the estimation effect is not ideal, which cannot give consideration to accuracy and agility. requirements such as sex and low overhead

Images