Indoor Signal Analysis And Feature Extraction For Water Leakage Detection Based On Home WiFi

The continuous development of wireless sensor devices and wireless network technology has made wireless sensing technology widely used in various fields.The use of wireless signals for human-computer interaction,behaviour recognition and noncontact sensory detection is receiving more and more attention in the field of wireless sensing,and the relevant theories and technologies are becoming hotspots of research nowadays.In households and public places,due to the deployment of a large number of Wi Fi access points,Wi Fi wireless networks have become a kind of universal wireless local area network,and Wi Fi wireless signals have been widely used in wireless sensing and indoor positioning.Occasional water leakage accidents occur in the domestic environment,and the process of detecting leakage status by commercial devices is complicated and expensive,while non-commercial devices detect inaccurate results.To address the difficulties or shortcomings in the current leak detection process in the home,this paper proposes a solution for real-time leak detection using Wi Fi signals,which are common in the home environment.By receiving and capturing the Wi Fi signals reflected from a leaking or non-leaking environment and analysing the changes in the signal channel characteristics,the aim of detecting water leakage in the home environment is to innovate the traditional home leakage detection solution.Compared to traditional Wi Fi signal-based target identification techniques,which require the use of a combined camera or infrared imager,this solution is more advanced,simple and reliable.The proposed solution consists of three main steps: data pre-processing,feature extraction and machine learning classification.Experimental results show that the proposed solution is accurate,simple and feasible for the detection of household water leakage.In this paper,research work is carried out in the following three areas: outlier detection,Kalman smoothing,empirical modal decomposition and wavelet filtering algorithms are investigated for the pre-processing of the received Wi Fi signal data to reduce environmental noise;regression quality analysis algorithms are investigated to filter the effective subcarriers in the irregular cyclic oscillation signal using the regression graphs generated by the algorithms;principal component analysis compression algorithms are used to compress the input signal;and the regression quality analysis algorithms are investigated to reduce the environmental noise.dimensions to accurately extract feature components;support vector machine machine learning algorithms and deep learning algorithms for convolutional neural networks were investigated for the detection and identification of water leakage states in indoor environments.