| Modern wireless communications must cope with critical performance-limiting challenges that include limited availability of radio frequency spectrum and a complex time-varying wireless environment (fading and multipath). Meeting the increasing demand for higher data rates, better quality of service, fewer dropped calls, higher network capacity, and user coverage calls for innovative techniques that improve spectral efficiency and link reliability. The use of multiple antennas at the receiver and transmitter in a wireless system, popularly known as MIMO (Multiple-Input Multiple-Output) wireless is an emerging cost-effective technology that promises significant improvement in these measures. A growing acknowledgment of the performance gains from MIMO technology has spurred the insertion of this technology into wireless standards, notably the mobile standards such as UMTS and CDMA 2000. MIMO techniques are also under study in IEEE 802.16 and 802.11 standards for fixed and WLAN applications respectively. Broadly, MIMO system can used spatial multiplexing to increase data rate, and transmit diversity to increase link reliability. To these two aspects, the main contributions and innovations in this dissertation are: 1. The first experiment model of spatial multiplexing is the vertical BLAST (Bell Laboratories Layered Space-Time) or V-BLAST. But the classical detection algorithm they used, the ZF-DFE (Zero Forcing-Decision Feedback Equalization) algorithm, is not the optimum. And the performance of ZF-DFE suffered from the selection of the detection order of the symbol components and error propagation. We suggest an optimum detection algorithm ——the SSML (Symbol-by-symbol Maximum A Posterior Probability)——in terms of minimum bit-error-rate. The simulation result shows that the performance of SSML is much better that ZF-DFE. Furthermore, we show that the SSML algorithm can be implemented the radial basis function neural network. Therefore, by used of the advantage of the parallel computation and data restoration of the neural network, the SSML algorithm can be employed in the real world. 2. Research shows that the performance of transmit diversity depends strongly on MIMO channel matrix characteristics. Most existing works have focused on quasi-static flat rayleigh fading channel. But in the real world, there exist many channel effects such as rician fading, spatial fading correlation, temporal fading correlation and fast fading. This dissertation gives a general MIMO channel model which can be used under fast fading channel. Furthermore, we give a new fading correlation structure capable of capturing any combination of rician fading, spatial fading correlation and temporal correlation. 3. Under the general MIMO channel model, we analyze the performance of the space-time trellis code and get the close-form solution of the Chernoff bound of the average pairwise error probability. Furthermore, we conclude that the channel fading correlation has nothing to do with diversity advantage, but the code advantage will be affected. We provide the correlation between the loss in coding advantage and the fading channel covariance matrix. Finally, the simulation is shown. 4. Considering the fading conditions may change so rapidly that channel estimation is difficult or requires too many training symbols, it is useful to develop modulation techniques that do not require channel estimates at the transmitter or receiver. Differential space-time unitary code is one of the well-known differential space-time code. We derive the optimal and suboptimal non-coherent receivers of differential space-time unitary code under fast fading and get the exact pairwise error probability of them.
|
PDF Full Text Request