Effect of channel estimation error on MIMO systems and CDMA systems with multiuser detection

Wireless communication with mobile terminals is intrinsically more difficult than communication with fixed terminals through wirelines due to time-varying channels. Reliable channel estimation is critical for efficient transmission over wireless links. The effect of channel estimation error on Multiple Input Multiple Output (MIMO) systems and Code Division Multiple Access (CDMA) systems with multiuser detection is discussed in this dissertation. MIMO systems implement multiple antennas at the transmitter and the receiver. The capacity of MIMO systems increases linearly with the number of antennas in rich scattering environments. Orthogonal training sequences are transmitted periodically to estimate MIMO channels. We provide a complete analysis of the channel estimation error for continuous flat Rayleigh fading channels, including the effect of noise as well as the temporal variation of the channel. Training interval and training length can be optimized to maximize the throughput for certain target Bit Error Rate (BER). Numerical results are obtained through computer simulations as functions of Signal-to-Noise Ratio (SNR), Doppler frequency and the number of antennas. Channel estimation both with and without tracking by linear interpolation is considered. A sub-optimal detection algorithm is proposed and evaluated to reduce the computational complexity for channel estimation with tracking without significant loss in the throughput.; Analysis of channel estimation error shows that the error increases with the normalized Doppler frequency. This imposes a Doppler dependent lower bound on the symbol rate. On the other hand, system performance degrades with Inter-Symbol-Interference (ISI) related to the normalized delay spread. This imposes a delay spread dependent upper bound on the symbol rate. These fundamental limits on the symbol rate are verified through computer simulations for frequency selective channels and their use for system design is illustrated.; Multiuser detection algorithms have been proposed for CDMA systems to improve capacity. We invented a multi-stage pipelined parallel interference canceler and implemented it in a detailed physical layer computer simulation. The performance of ideal channel estimation, partial channel estimation, and complete channel estimation is compared for both uplink and downlink in frequency selective fading channels. A single user bound is derived as a performance benchmark. We show that channel estimation error significantly degrades the performance. Capacity and near-far resistance are reduced as a result.