Environment adaptive activity detection method and system based on WiFi CSI

Abstract

The invention relates to the technical field of environment activity detection, in particular to a WiFi CSI-based environment adaptive activity detection method and system, the environment activity detection system comprises a CSI collector, a wireless router and a computing platform, the CSI collector is a WiFi terminal device with a CSI collection capability, the wireless router is a universal WiFi network device, and the computing platform is connected with the WiFi terminal device. A wireless network connection function is provided for the CSI collector; according to the method, the environment is perceived based on the WiFi CSI signal, the method is a non-intrusive detection scheme, the method has the advantages of low deployment cost, wide coverage range, no privacy leakage risk and the like, the unsupervised deep learning algorithm is applied to analysis of the WiFi CSI signal, the execution process of the VAE model training set does not need personnel intervention, and the detection efficiency is improved. By means of the method, a new model can be rapidly and automatically trained when the environment changes, the environment self-adaptive environment activity detection capacity is achieved, good generalization can be obtained under the condition that a training set is small, and the requirement for system computing power can be effectively lowered.

Description

technical field [0001] The invention relates to the technical field of environmental activity detection, in particular to a WiFi CSI-based environmental adaptive activity detection method and system. Background technique [0002] Environmental activity detection technology is widely used in various fields, for example, it can be used to detect intruders in the field of security; real-time monitoring of the density of people in public places to prevent stampedes or the spread of infectious diseases; in the field of intelligent transportation, By monitoring the density of vehicles on the road, dynamically adjust the duration of traffic lights, optimize road traffic capacity, etc. [0003] Environmental activity detection technology can be classified into the following schemes: wearable device schemes, video schemes, sensor schemes, and wireless schemes. Among them, the wearable device solution requires the detected object to wear special equipment to achieve accurate identifi...

AI Technical Summary

Problems solved by technology

This method requires data collection and model training for different human activities in different environments in advance, so the implementation is complex and difficult to apply in real environments

[0008] Although WiFi CSI-based applications are receiving more and more attention, limited by the complexity of algorithm implementation, there is no technical solution or system that can be actually deployed.

Images