Research On Adaptive Algorithms Using MBER Criterion In Wireless Communication Systems

Nowadays, adaptive processing technology has been a research hotspot in the information age. The adaptive filtering technique is widely applied in wireless communication systems, includ-ing beamforming, multi-user detection, MIMO (Multiple-Input and Multiple-Output) receiver and channel equalization. The design of adaptive filtering is usually based on the traditional Wiener filter, or known as the minimum mean square error (MMSE) criterion. MMSE can be through least mean square (LMS) algorithm with low complexity and real-time computing characteristics to design and implement a modern high speed communication system. However, in the practical wireless communication systems, we are more concerned about the bit error rate (BER) perfor-mance, rather than the size of the mean square error. Thus, the algorithms based on minimizing BER criterion are worth studying.The research area of adaptive update algorithms are based on minimum BER (MBER) criteri-on in different communication system. The thesis mainly focuses on multi-user detection problem. The mentioned system models include DS-CDMA system, wireless sensor network and the large-scale multi user MIMO system. These three system models play a very important role in the 3G mobile communication system and development of the 5G mobile communication system, respec-tively. In addition, we study the joint power allocation algorithms and the joint reduced rank (RR) adaptive algorithms. We use the generalized Gauss kernel density estimate method with the high order modulation.The thesis is organized as follows:Firstly, the current research background and research status based on MBER criterion algo-rithm are introduced. Those four research key points in this thesis are introduced, including the principle of gradient method, the minimum error bit rate criterion, the rank reduction technology and the Kernel density estimation. In addition, we introduced the research background and the present situation about three different system models.Then, the joint power allocation and interference of an iterative elimination algorithm for cooperative DS-CDMA system model using amplify-and-forward (AF) scheme is studied in this thesis. The algorithm is based on minimizing the bit error rate criterion and has low complexity and low power performance. In addition, using the stochastic gradient method, the algorithm derive MBER cost function to joint update weight vector and power allocation vector.Next, two multi user detection algorithms are proposed in a two-hop wireless sensor network with multiple relay nodes where AF scheme is employed. The two algorithms jointly consider interference suppression and power allocation based on the minimization of the symbol error rate (SER) criterion. This criterion is a transform format of MBER criterion. Indeed, we change an study angle to further verify the proposed algorithm. Firstly, we develop stochastic gradient (SG) algorithms on the basis of the minimum-SER (MSER) criterion. Secondly, we develop conjugate gradient (CG) algorithms on the basis of the SER criterion. These two algorithm jointly update the parameter vector that allocates the power levels among the relay sensors subject to a total power constraint and the linear receiver. The power allocation codebook is designed at the fusion center. Destination nodes transmit the quantized information of the PA vector to the relay nodes by a limited-feedback channel. The complexity and convergence analysis of the proposed algorithms is carried out in the thesis.Finally, four proposed RR algorithms are based on joint iterative optimization of filters accord-ing to the MBER criterion in large-scale multiuser MIMO system. These algorithms employ the generalized Gaussian kernel density estimation. The generalized Gaussian kernel density estima-tion method can assist us to in measuring the distribution with heavier or lighter tails as compared to the method of normal Gaussian kernel density estimation. We calculate the optimal window width of the generalized Gaussian kernel. The proposed optimization technique adjusts the weights of a subspace projection matrix and a RR filter jointly. According to the difference of the SG and CG algorithm, the proposed algorithms will be divided into two categories. Then according to the modulation modes including BPSK and 16-QAM modulations, the proposed algorithms can be divided into four algorithms. The proposed adaptive RR algorithms are evaluated based on sim-ulations using Rayleigh block fading channels. The simulation results for large multiuser MIMO systems show that the proposed adaptive algorithms significantly outperform the existing schemes.