Research On Intelligent Gait Recognition And Wireless Transmission Application System

Biometric features such as faces and fingerprints have been widely used in various Internet of things devices.but they require user cooperation when unlocking and are vulnerable to imitation and theft.However,gait recognition can be completed when the user walks,without user interaction and the collection process is safer,which greatly improves the safety of wearable devices.At present,most electronic devices have built-in acceleration sensors,and using the acceleration sensors to achieve gait recognition has high practical feasibility.In addition,the transmission of gait information is often wireless.Efficient and reliable wireless transmission is of great significance to gait recognition.Gait recognition technology is still in the research stage,and there is no general mature equipment.Therefore,this thesis has carried out research on intelligent gait recognition and transmission application systems.The main work and innovations are as follows:1.To solve the problems of complex data acquisition devices and high sampling frequency in current gait recognition,a gait acquisition hardware platform with low cost,small size and easy assembly is designed.First,design a system solution according to system characteristics and requirements;then carry out its modular design,including software and hardware development of the data acquisition module Mpu6050,microprocessor module Arduino Nano,and Bluetooth transmission module HC-05,and finally complete the system development in combination with the designed circuit principle.Among them,the sensor sampling frequency is set to 40Hz to ensure low power consumption of the system.Moreover,the device is used to collect gait information and establish a 10-person gait database to provide technical support for subsequent gait recognition.2.Aiming at the problems of low gait recognition rate and single algorithm,an intelligent gait recognition software platform combining dynamic time warping and artificial neural network is proposed.First,using dynamic time warping,the different numbers of feature points are regularized to a fixed length,the waveform features are extracted,and the cumulative distance is calculated,thereby setting the threshold value of the cost function,and the positive and negative inputs of the neural network are obtained by comparison with it.Second,the Levenberg-Marquardt?LM?algorithm is used to improve the standard error Back Propagation?BP?neural network,introduce the characteristics of step length,step frequency,step speed and standard deviation,train them with the previous features,and finally complete the gait recognition.The simulation shows that the proposed improved gait recognition method maintains the average gait recognition rate and equal error rate at 91.5%and 9.1%,while effectively reducing the time delay,and improving the accuracy of gait recognition.3.In the transmission of gait information in the Internet of things,in view of the current Low-Density Parity-Check?LDPC?code decoding failure,which leads to the problem of signal error propagation,a coded collaborative soft information hybrid forwarding scheme was proposed.First,the decoding results are discriminated at the relay node.If the decoding is successful,the information is transmitted using LDPC encoding,otherwise the information is transmitted by calculating and forwarding soft information.Second,at the destination node,the joint iterative BP decoding algorithm is used to effectively improve the decoding performance and finally complete the information transmission.The simulation shows that the proposed improved information transmission scheme has low complexity and effectively reduces the system bit error rate.When the bit error rate is10^-2,compared with the single LDPC coded and uncoded Hybrid Decode-Amplify-and-Forward?HDAF?scheme,the proposed scheme obtains signal-to-noise ratio gains of about 2d B and 7d B,respectively.This thesis mainly carried out the optimization research of gait recognition software and hardware platform and code transmission,which improved the performance of the system.Moreover,the proposed solution takes into account both the sampling rate and the recognition rate,which effectively reduces the cost and has a lower complexity,meets the requirements of future Internet of Things device safety certification,and has application value in information security protection.