Sensor Selection With Constrained Beam-pattern Via Convex Optimization Based On WBAN

With the development and application of wireless communication technology, the Internet of Things is coming into various fields of human life. Wireless communication is developing towards miniaturization, short distance and personalization. As one of the most important branches of the Internet of Things, on the basic research of wireless sensor network, wireless body network has received increasing attention. With some chronic diseases from spreading, people pay more and more attention on health. WBAN(Wireless Body Area Network), which uses the sensors nodes placed in, on or around human body to monitor the physiological parameters for forecasting the disease possibility which is of great importance to human’s health. Hence, WBAN can be used in practice with many advantages.Considering the variety and the huge amount of data in WBAN, and also the requirements of low latency, real-time and other characteristics, this paper presents the concept of big data in WBAN.Wireless body area network uses sensor node to node,node to AP node to transmit data for exchanging information. However, considering the power of the sensor nodes is limited, especially the sensors placed in the body are not easy to charge or change the battery. Moreover, the energy consumption of WBAN mainly depends on the information transmission between nodes. Hence,control the number of the sensors in WBAN is an essential issue for the efficiency of the network. The traditional WBAN optimization problems usually revolve around the following parameters — the lifetime of sensor nodes, the reliability of data transmission, the efficiency of data transmission and so on. In this paper, from reducing node energy consumption to increase efficiency, formular the node arraymodel of human body and design optimization algorithm to complete the optimization of WBAN.The main work of this paper is as follows:First, we propose the overall WBAN big data processing framework, and apply Map Reduce and HBase to process, store and update the WBAN big data. Traditional data characteristics and processing methods are summarized, and combined with the characteristics of WBAN to improve it, this paper puts forward the WBAN data processing framework.Second, node array model of WBAN is put forward.According to the geometrical feathers of human body, on the basis of planar antenna node array and the sparse array,build the circular sensor array to simulate the head and planar rectangle node array to simulate the body. Furthermore, we propose the different formation of the node array for planar array. Finally, concentrates on simulation results using our proposed algorithm and comparison with some different formation.Third, the sensor node optimization algorithm via convex optimization is put forward. We propose the three-dimensional structure model of human body and the modified optimization algorithm. First, make a primary introduce of the basic advantages of convex optimization. Then combining practical matters, we converse non-convex problems to convex problems. In order to get the minimum number of sensor nodes and the more concentrated energy of the mainlobe, we set the weight vector parameters and PSL(Peak of Sidelobe Level) as object functions to improve the energy efficiency of WBAN. And whether the constrain function and object function are both convex is proved. Aiming at the uncompletely optimizating of number of sensor nodes and concentration aforementioned, we proposed modified iterative algorithm of weight vector. Every time the calculation of weight vector is based on the precious result of weight vector. In the end, according to the comparison of the simulation in planar array, we can get the conclusion that the method based on human body model via convex optimization to minimize the number of the sensors subject to the energy of the mainlobe and sidelobe is pretty reasonable and effective for optimizing the sparse array.