Synchronization and interference suppression for OFDM based wireless communications

Abstract

High data rate mobile communications have several challenges. High speed transmission of information typically experiences longer delay spread in harsh wireless environments. In addition, the channel-variations introduced by motion can make the link much less reliable. Orthogonal frequency division multiplexing (OFDM) is a promising transmission technique suitable for broadband wireless communications. However, the performance of OFDM systems is extremely sensitive to carrier frequency offset (CFO) and timing synchronization error (TSE). Timing error and frequency offset introduce inter-symbol-interference (ISI) and inter-carrier-interference (ICI). Furthermore, for mobile applications, time-variations in channel gains within one OFDM frame also introduce ICI, which further degrades the performance. Therefore, synchronization is extremely important for OFDM system. Unfortunately, the doubly selectivity in high mobility wireless applications has destructive impacts on existing synchronization methods as the power delay profile of the channel can change rapidly due to the sporadic birth and death of the paths. To have an acceptable reception quality for applications that experience high delay and Doppler spread, the design of a more robust synchronizer, channel estimator and ICI mitigator is essential. This thesis takes an overall look at synchronization issues in OFDM systems. The effects of CFO and TSE on the performance of OFDM systems are first investigated and analyzed. Next, a timing synchronization scheme based on training sequence is proposed. A local synchronizing sequence (LSS) is defined based on one single training frame. An LSS correlator at the receiver is designed to perform the timing synchronization. The new scheme is proved to be robust in doubly selective channels. Analysis and simulation have confirmed that the proposed scheme has much better timing synchronization performance, in terms of estimation bias and mean square error (MSE), than existing methods. To estimate the CFO, this thesis proposes a two-step scheme. In the first step, the fractional CFO is estimated combined with the channel coefficients via EM algorithms. It is shown that the joint estimation improves the CFO estimation accuracy in doubly selective channels. The integer CFO is estimated in the second step based on the maximum likelihood criterion. Through these two steps, CFO estimation accuracy is improved and the acquisition range is significantly enhanced. Finally, an ICI suppression scheme based on second order polynomial Nyquist window is proposed. The new windowing scheme takes the advantage of the received signal itself to allow the use of large roll-off factor. Therefore, the sensitivity of the receiver to CFO is greatly reduced. A window parameter optimization scheme is also derived so that the receiver can adaptively maximize the SINR performance according to channel conditions.