Research On Performance Optimization Of Stc Over Mimo Fading Channels

Comparing with conventional single-input single-output (SISO) communication systems, Multi-input Multi-output (MIMO) communication systems have two attractive advantages. These are multiplexing gain (or spectral efficiency gain) and diversity gain. MIMO technology can remarkably improve channel capacity and spectral efficiency in the premise of the same bandwidth. MIMO has been adopted as a key technology of standard for 3G and B3G wireless communications. Space-time coding (STC) has been popular since its inception. Specially, orthogonal space-time block code (OSTBC) becomes one prominent STC for its full diversity performance and simple maximum likelihood (ML) decoding algorithms. When the number of transmit antennas is larger than 2, the code rate of OSTBC is less than 1. Therefore, some new codes are designed to balance diversiy gain and code rate in STC family, such as general layered space-time code (GLSTC) and quais-orthogonal space-time block code (QOSTBC). In real wireless environments, reflectors and scatters are not uniformly deployed and directions-of-arrival (DOA) at antennas are not omnidirectional and not uniform. These cause spatial correlations among MIMO antennas. Like fading and noise, spatial correlations deteriorate channel capacity and error performance in MIMO communication systems. In this dissertation, several performance optimization algorithms have been discussed for STC over MIMO fading channels.When ML decoding is used for OSTBC, its computational complexity exponentially increases with the modulation order increases. Even ML decoding is respectively used for real part and imaginary part of one symbol, the computational complexity increases a lot with the increase of modulation order. If modulation order is high, computational complexity is an important obstacle which affects the application of STBC. A simplified orthogonal STBC decoding algorithm is put forward in this dissertation. At first, the communication equation is represented by equivalent channels matrix. An estimated vector of transmited symbols can be obtained by using zero-forcing (ZF) decoding. Sphere decoding is deployed to real part and imaginary part of the estimated vector, respectively. The orthogonality character among columns of the equivallent channel matrix provides sphere decoding has the same performance with ML decoding. With increase of modulation order, sphere decoding algorithm can reduce more computation than ML decoding algorithm.Training sequence and interpolation are fundamental channel estimation techniques in wireless communication systems. In MIMO communication systems, using plenty of training sequence for channel estimation and channel tracking will decrease bandwidth efficiency. If reducing training sequence, the error of channels estimation increases. In this dissertation, a recoding and interpolation scheme is proposed for channel estimation and channel tracking in OSTBC systems. At receiver, except for training sequence at the begin of one frame, one slot's data are recoded as training sequence for channels estimation in several slots. In other slots, the channels parameters are calculated by interpolation. Compared with the method using training sequence and interpolation, the recoding and interpolation scheme can achieve 1.5 dB gains at 10-6 bit error rate in the same environment. The recoding and interpolation can be used as a quasi-blind channels estimation algorithm.GLSTC is a combined technology of Bell Labs layered space-time (BLAST) and STBC which can provide both antenna diversity gain and spectral efficiency gain. At transimtter, antennas are devided into several groups and STC is implemented in each group. At receiver, interference suppression and interference cancellation are used for decoding. The errors in one group will deteriorate succedent decoding. Because the distributation of errors is very complex, error propagation has been ignored or overestimated in the existing literature. The effect of unsuccessful decoding and error propagation has been derived by using theoretic analysis combining with simulation in this dissertation. A new power allocation strategy considering error propagation is proposed by using Lagrangian multiplier method. The new optimal power allocation has the best error performance among all existing power allocation strategies from low singal-to-noise ratio (SNR) to high SNR scenarios. Furthermore, a multiple access algorithm using interference suppression and interference cancellation is designed for wireless personal area network (WPAN). The multiple access algorithm adapts to WPAN with few communication nodes and very high communication rate.QOSTBC is an attractive transmission and coding technology because it can achieve high transmission rate compared to OSTBC by scarifying partial diversity. In this dissertation, we present a parallel interference cancellation (PIC) algorithm for QOSTBC concatenated error correction code structure. At the receiver, transmitted symbols are detected by pairwise decoding algorithm at first. After error correction, the data are regrouped. PIC and dual OSTBC decoding are deployed. The proposed system can achieve full diversity same as OSTBC, full code rate same as QOSTBC and error correction code gain simultaneously. The concatenated structure has better performance than both QOSTBC and OSTBC schemes except at very low SNR scenarios. The proposed PIC algorithm can improve the error performance further.Time-varying and spatial correlations are two characters of MIMO channels. Fading and spatial correlations affect performance of communication systems. Adaptive modulation and code (AMC) can improve system performance by using corresponding transmission schemes. One realtime AMC scheme is proposed for time-varying spatial correlated fading MIMO systems. The decision algorithm is simple linear decision according SNR and error rate. Only a few indicator informations are required to feed back to transmitter. Antenna subset selection and QOSTBC are two candidate transmission schemes because they are suffered with SNR and spatial correlations in different degree. Simulation results show that the proposed AMC scheme can achieve considerable gain in low SNR and week spatial correlations at the expense of simple decision calculation and tiny feedbacks. The proposed AMC algorithm improves robustness of MIMO communication systems over slowly time-varying spatial correlated channels.