ARM Based Embedded Smart Home System

With the development of the Internet of Things(IoTs)technology,the connection between things and things,people and things becomes possible,and it has broaden application prospects in the fields of smart retail,smart city,intelligent security and so on.As an application of IoTs,smart homes have developed rapidly in recent years due to their relative low costs.A large number of products have emerged and the level of automation has been greatly improved.However,because of the low level of intelligence and passively accepting people's instructions,further promotion of smart homes has been affected.In order to make the smart home system mine the user's behavior patterns and usage habits based to their usage history to predict user's intention and automatically make corresponding actions,which will further facilitate people's lives.Based on some researches,this dissertation builds a smart home system according to user behavior.The system mainly consists of two parts:(1)Smart home control system.To realize device control and data acquisition,CC2530 chip of Texas Instrument is used as a microprocessor for various smart home devices.Besides,the Zigbee wireless communication protocol is used to form a mesh local wireless local area network to realize remote control of the device.Moreover,the stability of the wireless network is improved by instruction retransmission mechanism and network status self-test mechanism.(2)User behavior mining system.Firstly,the IMTP4412 module based on ARM architecture is used as the central processor to establish a TCP Server open port for making communication with local clients such as mobile APP,to realize various interactions with users.Then,connecting the Zigbee network coordinator through serial port to realize collection of control and status information of Zigbee devices.Besides,by connecting ITOP4412 module to a remote server,device status can be sent to sever and devices can accept remote command.Finally,collaborative filtering algorithms can mine user's usage patterns and behavior habits on the basis of the usage history of collected users.(3)This dissertation builds a real smart home device network,1)stability test experiments show that the system can run stably for a long time,and can restore the previous state after special conditions such as network disconnection and power failure.2)The limitation of embedded device memory is well solved by using improved collaborative filtering algorithm and estimating the overall historical data by sliding average and iterative methods.Simulation experiments show that the improved collaborative filtering algorithm is effective on embedded devices and recognition accuracy of user mode reaches 90%.