Abstract
Abstract In this study, we propose two methods to estimate the time and frequency offset separately in an OFDM communication system, and then the two methods are integrated to solve the synchronization problem when both integer time offset and frequency offset exist. First, the OFDM system is modeled when only the time offset exists. The degradation of performance results from the inter-carrier interference (ICI) and inter-symbol interference (ISI) whose statistical knowledge are unavailable. Hence, an H_infty filter is employed to look for the time offset by sending a training sequence. The method proposed can estimate not only integer time offset but also any rational number. Second, we consider the case when only the frequency offset exists. We use the correlation of the samples at the output of the discrete Fourier transform (DFT) to estimate the frequency offset. To increase the tracking ability, an extended Kalman filter (EKF) is applied as a recursive estimation algorithm. Finally, we consider the synchronization problem with an integer time offset and a frequency offset. We propose an iterative algorithm to find time offset by an H_infty filter and an EKF first. Then the frequency offset can be estimated and corrected after the correction of the time offset.