Orthogonal Frequency Division Multiplexing (OFDM) is a bandwidth efficient modulation technique. Using multiple carriers and guard time between OFDM symbols, OFDM transmission can be easily demodulated even in the presence of frequency selective fading, often experienced in wireless wideband communication systems. However, the constructive and destructive superpositions of the multiple carriers lead to a large variation in the signal envelope of the resulted OFDM signal. The large peak-to-average power ratio (PAPR) of OFDM signal requires linear amplification and is often achieved by non-linear amplifier with a large back-off. Unfortunately, this configuration has very low power efficiency and as a result, many methods had been proposed to reduce the PAPR of OFDM transmission....[ Read more ]

Orthogonal Frequency Division Multiplexing (OFDM) is a bandwidth efficient modulation technique. Using multiple carriers and guard time between OFDM symbols, OFDM transmission can be easily demodulated even in the presence of frequency selective fading, often experienced in wireless wideband communication systems. However, the constructive and destructive superpositions of the multiple carriers lead to a large variation in the signal envelope of the resulted OFDM signal. The large peak-to-average power ratio (PAPR) of OFDM signal requires linear amplification and is often achieved by non-linear amplifier with a large back-off. Unfortunately, this configuration has very low power efficiency and as a result, many methods had been proposed to reduce the PAPR of OFDM transmission.

Existing techniques to reduce the PAPR include clipping, coding, and phase alignment approaches. Clipping is the simplest, but leads to performance degradation. Coding and phase alignment approaches do not lead to any performance degradation, but need additional overhead and complexity. Some of the phase alignment methods require multiple IFFT computations to search for the best transmit sequence to minimize the PAPR.

In this thesis, we propose a multi-layer phase alignment PAPR reduction approach that takes advantage of the Cooley Tukey FFT structure to reduce the complexity. We find that swapping the intermediate values within the IFFT computation can reduce the PAPR of an OFDM symbol. As only a few "butterfly" computations are needed to test a new phase rotation vector, the computation complexity is much lower. Moreover, unlike most phase alignment methods where all phase rotation vectors within the selected set are tried before picking the best vector, this method selectively tries the phase rotation vectors based on the input data pattern. This allows the search to find the best vector quickly. As the approach is an iterative process where the PAPR reduces step by step as the process continues, we can stop at any given iteration based on the complexity limitation. We also proposed a complexity-controlled version of this method, which allow flexible tradeoff between complexity and performance.