WiFi identity recognition method fused with deep learning model

Abstract

The invention discloses a WiFi identity recognition method based on a fusion deep learning model. The WiFi identity recognition method comprises the following steps: S1, collecting WiFi channel data of 30 persons; S2, extracting an Ntx*Nrx*30 three-dimensional CSI matrix from the WiFi channel data; S3, performing data preprocessing on the extracted CSI matrix data; and S4, establishing a fusion deep learning model, and carrying out classification training on the CSI matrix data preprocessed in the step S3 to realize personnel identity recognition. According to the method, a user does not need to wear or depend on any sensor, only ubiquitous WiFi is needed, channel state information in the WiFi is processed, then a pixel transposition convolutional network and a fusion deep learning model are used for carrying out biological feature extraction on the channel state information, and identity recognition of multiple users is achieved.

Description

technical field [0001] The invention relates to the technical field of artificial intelligence identification, in particular to a WiFi identification method fused with a deep learning model. Background technique [0002] In recent years, human identification technology has been extensively studied because human identification plays an important role in human-computer interaction and can support many emerging applications, such as smart home, augmented reality, healthcare, etc. Many human identification systems have been proposed with different technologies, such as wearable sensor-based methods, computer vision-based methods, environmental device-based methods, etc. These methods require that the device is always on the body or use a camera to capture images of people. Experimental equipment is often expensive and may be affected by occlusions. Camera-based identification methods also have potential privacy issues. There are also studies using WIFI channel state information...

AI Technical Summary

Problems solved by technology

However, machine learning algorithms use shallow classifiers to achieve feature classification. They usually artificially select the time domain and frequency domain features in CSI samples in a heuristic and suboptimal manner. Therefore, shallow classifiers cannot delicately represent RF signals. The representative mode of biological characteristics hidden in the human body, the existing machine learning research can only realize the identification of a small range of people (2-10 people)

Images