¡@¡@In recent year, orthogonal frequency division multiplexing ¡]OFDM¡^ technology has been widely used in high-speed communication systems. One primary reason for the popularity of OFDM is its ability to provide good performance in multi-path channels than the other systems through the use of Guard interval(GI). By using the Guard interval, it can convert these inter-symbol interference (ISI) channels into ISI-free channels. But Guard interval without any information will caused inter-channel interference. In order to solve this problem, a guard interval using cyclic prefix (CP) is inserted to avoid inter-symbol interference from the adjacent symbols and inter-channel interference from other sub-channels. However, using long cyclic prefix will decrease the transmission rate, reduce the spectral efficiency, and increase the signal-to-noise power ratio¡]SNR¡^loss. If we choose a shorter one, the channel length may be longer than the cyclic prefix. The interference caused by insufficient cyclic prefix can seriously degrade the performance of OFDM systems. In order to solve this problem, a time domain equalizer¡]TEQ¡^is usually used in the receiver to shorten the channel length of OFDM transmission system, and therefore minimize the ISI and ICI. However, because of its high complexity, the optimum design of TEQ is hard to realize. ¡@¡@So we use an iterative channel estimation technique between time domain and frequency domain to mitigate the ISI and ICI which is caused by insufficient cyclic prefix. The iterative technique can remove ISI and hold the circular convolution property. By utilizing the iterative technique we can improve the channel estimation performance as the OFDM symbol used sufficient cyclic prefix. ¡@¡@In the computer simulations, we consider the worst case that the OFDM symbol is transmitted without guard interval. The results show that our proposed method can effectively suppress residual ISI. The comparison between our proposed method including both proposed channel estimation¡]PCE¡^and proposed data demodulation¡]PDD¡^, decision feedback channel estimation¡]DFCE¡^, avoid ISI preamble channel estimation¡]AISIP¡^, and residual ISI cancellation¡]RISIC¡^algorithm are made in this thesis. Finally, the performance improvement of the proposed algorithm under several channel conditions is considered and compared with other algorithms.