Joint Optimization Design For Two-Way MIMO Multi-relay AF Systems

In recent years, wireless sensor networks are extensively concerned. Sensor nodes transmit andexchange information through wireless communication, however limited processing capacity ofsensor nodes restrict the application of wireless sensor networks, it needs cooperative relay anddistributed information processing technology to realize the long distance communication. Mostrelays work in the half-duplex mode, two-way relay protocol complete one round of informationexchange in two time-slots, which avoid the spectral efficiency loss, and with combination oftwo-way relay and multiple input multiple output (MIMO) technology, then a better transmissionperformance can be obtained. Therefore, wireless sensor networks can employ cooperative two-wayrelay to improve system performance.The major efforts of this thesis are focused on the joint optimization design of two-way MIMOmulti-relay amplify-and-forward system. Firstly, briefly introduce some elementary knowledgeincluding the characteristic and application of wireless sensor networks, the system model andcapacity of MIMO, the cooperative protocol, forward mode of relay technology, and the two-wayrelay protocol, furthermore, several optimization methods adopted in this thesis are introduced.Secondly, propose three optimization algorithms for two-way MIMO multi-relay system under thetransmitted power constraint condition, which aims to minimize the total mean square error;Through calculating the gradient of sum-MSE with respect to source precoding matrix and relayprecoding matrix and applying the existing gradient descent algorithm, a locally optimal solutioncan be obtained; Then, consider the structure of optimization variable with respect to block diagonalrelay precoding matrix, and convert the matrix optimization into scalar form, according to theoptimization theory, prove that the optimization is convex; On this basis, propose two optimizationalgorithms, one is a low complexity suboptimal scheme based on geometric algebra inequality,which can obtain a suboptimal solution; another is an optimal scheme based on the globalconvergent interior penalty function, the simulation of two methods all prove their advantageousperformance on MSE with comparison of the existing gradient descent algorithm.Research shows that a great improvement of system performance can be obtained throughoptimization design of cooperative communication in wireless sensor networks.