Abstract
Orthogonal frequency division multiplexing (OFDM) is a wideband transmission technique capable of high-rate multimedia services and has been widely adopted in wireless communication systems. A major problem in OFDM systems is that system performances are highly sensitive to frequency offsets. The mismatch of local oscillators at the transmitter and the receiver causes single frequency offset in the received signal, while a time-varying channel results in a spread of frequency shifts. Each type of frequency offset destroys the orthogonality among subcarriers and induces intercarrier interference (ICI), which could limit overall system performance. To deal with the ICI problems induced from the frequency offset in OFDM systems, two-path parallel cancellation schemes known as conjugate cancellation (CC) and general phase rotated conjugate cancellation (PRCC) were proposed recently. However, since the frequency offset considered in these schemes is modeled as time-invariant, they cannot provide satisfactory performance under time-varying channels, especially those with high frequency offsets. In this thesis, we have proposed an adaptive receiver design based on the conjugate transmission for ICI self-cancellation in OFDM systems. The proposed scheme utilizes an adaptive normalized BLMS algorithm to adjust the artificial phase rotation factor with frequency offset variations which usually occur under time-varying channels. Unlike PRCC, the proposed receiver is able to employ the frequency offset estimate directly, rather than feed it back to the transmitter and result in an overhead of signaling. Besides, with the adaptive phase rotation design, the tolerable range of frequency offsets can be maximized. Computer simulation results show that the proposed scheme outperforms the CC scheme and the PRCC scheme under time-varying channels in terms of both BER and CIR performances. The robustness of the proposed scheme is also shown to be superior than that of previous works when the frequency offset estimation error is considered.