Abstract
The signal-to-noise (SNR) estimation is an essential technique in wireless communication systems, and many applications require the SNR as the prior information to optimize the system performance. This thesis presents a novel code-aided (CA) SNR estimation method for the trellis-coded systems. The proposed method estimates the SNR using the average symbol energy and the minimum path metric from the Viterbi decoding procedure. The simulation results indicate that the proposed method has better estimation performance than the other non-data-aided (NDA) SNR estimation methods in the literature, especially in low SNR conditions. Besides, the antenna diversity improves the estimation performance of the proposed method. In this thesis, we optimize the performance of two applications in wireless communication systems by using the proposed method. The first application is the power-aware space-time trellis code (STTC) multiple-input multiple-output (MIMO) detector. We presents a state-purging mechanism based on the T-algorithm to reduce the high complexity of branch metric calculations. The embedded SNR estimator using the proposed method provides the essential SNR information to determine the optimal threshold for the state-purging mechanism. Then, the mechanism dynamically decreases computational complexity and only degrades coding gains by less than 0.1dB. The power-aware STTC 4 times 4 MIMO detector is fabricated using 90nm 1P9M CMOS technology, and the power consumption with the clock rate of 83.33MHz is 7.94mW to 8.70mW for the QPSK modulation, 10.02mW to 10.95mW for the 8PSK modulation and 12.20mW to 13.10mW for the 16QAM modulation. The maximum power saving ranges from 13.45% to 17.62% when the SNR is 20dB. The other application is the high-accuracy RSS-based mobile positioning system. Each base station estimates the distance based on the RSS information, and the system collect all the estimated distances to locate the mobile station. The proposed method cancels the noise energy from the raw RSS and has better improvement on noise cancellation than the other NDA SNR estimation methods. The particle filter of each base station determines the distance based on the estimated RSS. We formulate the convex optimization problem based on the estimated distances and the positions of base stations. Finally, the system implements the sub-gradient method to solve the problem and locates the mobile station. Simulation results depict that the proposed method decreases the errors of the raw RSS estimation to 1% when the SNR is 0dB. The RSS estimator using the proposed method derives the RSS with noise cancellation and improves the localization performance of the positioning system. When the mobile station moves at the speed of 60km/hr, the localization errors of the proposed system are 30m in the SUI-1 channel and 40m in the SUI-3 and SUI-5 channels.