In this work, different aspects of the multi-band orthogonal frequency division multiplexing (MB-OFDM) based ultra wideband (UWB) high-speed communication systems are studied. A simplified MB-OFDM system is proposed, which has lower complexity at the both transmitter and receiver side by utilizing the up-sampling technique at the transmitter. The FFT processor at the receiver side, in the simplified system, can have higher hardware efficiency through time-multiplexing. The proposed system also kept the performance in the original MB-OFDM system. A Viterbi decoder architecture for the MB-OFDM system is proposed. By efficiently combining the sliding-block, look-ahead, pipelining and parallel techniques and carefully designed memory management algorithm, the proposed decoder can support all data rates required in the MB-OFDM system and may save power by disabling part of the circuit when operating in low data rates. Some modifications on the low-density parity check (LDPC) decoder architecture are proposed to make it suitable for the UWB system and support multiple quasi-cyclic LDPC codes with a single structure. To solve the simultaneously operating piconet (SOP) problem in the MB-OFDM system, the combination of sphere decoding algorithm (SDA) with the K-best algorithm is proposed. To further reduce the computational complexity, especially in the list sphere decoder (LSD) when combined with soft-input channel code decoder, some low complexity candidate list updating circuits and improved radius update algorithms are proposed. The proposed algorithms lead to acceptable performance loss and reduce the overall hardware complexity and latency significantly.Hence, compared with non-pulsed-OFDM, which uses a single 128-point FFT structure, the 4-parallel 32-point FFT ... POFDM system uses a rate-2/3 convolutional code compared to MB-OFDM system that uses a rate 1/3 convolutional code.
|Title||:||Architectures for OFDM-based Ultra Wideband Systems|
|Publisher||:||ProQuest - 2006|