Research On Data Collection And User Behavior Detection Based On WISP

With the rapid development of the Internet of things industry and Radio Frequency Identification(RFID)technology,people have broken through the limitations of traditional passive tags and developed Computational Radio Frequency Identification(CRFID)tags with integrated microprocessors that can not only identify objects but also have complex functions.Which is representative of the Wireless Identification and Sensing Platform(WISP)this kind of label,in addition to the small size,the cost is appropriate,the traditional passive tags such as passive work advantages,also integrated microprocessor and all kinds of sensors,such as acceleration sensors,temperature sensors and so on,greatly enrich the passive tag application scenarios,bring new development direction for the innovation of the Internet of things.Sports and health testing is one of the hot scenes of innovation application of Internet of things,such as the popularity of wearable devices such as xiaomi bracelet and Apple Watch,indoor antifall monitoring for the elderly and children,indoor motion sensing and games,etc.However,these devices will bring portability,energy consumption and charging problems.In view of the above problems,this paper uses the feature of WISP tags such as passive,calculable,wireless transmission and so on in combination with its built-in micro acceleration sensor to carry out the research on user behavior detection.The research mainly focuses on the following sections:(1)In this paper,the composition principle of WISP are firstly deeply explored,the research focuses on EPC C1G2 wireless communication protocol and completes the embedded software development of WISP,so as to make it meet the work requirements of sensor data collection and wireless communication transmission.At the same time also introduces the hardware composition,computing power,working principle and so on.(2)After the WISP tag is developed,according to its working principle,the environment construction of the whole WISP system is completed.The WISP system mainly consists of four components: reader,reader antenna,WISP tag and background computer.Then the platform system is tested and verified,and the acceleration data of various user behaviors are collected and saved locally.(3)The use of collected data set,in order to visualize the data processing,data filtering noise reduction,finally using Convolution Neural Networks(Convolution Neural Networks,CNN)model for training and analysis,so as to effectively identify the indoor and the different behavior of users,to achieve the purpose of user behavior detection,the experimental results show that this method can achieve more than 93% of the highest classification precision.To sum up,based on WISP tag,this paper develops the acceleration data collection function,and completes the collection and processing of user behavior acceleration data,and realized the detection of indoor user behavior through deep learning model training,expanded the application scenarios of WISP tags in real life,and enriched the research results of Internet of things.