正交分頻多工系統在時變通道下同時域與頻域之等化

Abstract

Increasing demand for higher data rates under high mobility has catalyzed sev- eral new techniques and systems. One of the most popular techniques is orthogonal frequency-division multiplexing (OFDM). The OFDM system is widely adopted in many modern communication systems because of its spectral e±ciency and its ro- bustness over multipath channels. For many mobile applications, the end users might be moving and this will introduce the doppler eRect. Due to the doppler eRect, chan- nel is not time-invariant anymore and the orthogonality among the subcarriers of the OFDM systems will be destroyed. As a result intercarrier interference (ICI) is intro- duced. ICI will degrade the performance of the system signi‾cantly. Thus, in order to be a reliable communication system, some ICI mitigation schemes are important to the system. There are many ICI mitigation schemes. One technique is to design an equalizer. By having an equalizer at the receiver, the ICI could be eRective suppressed. But the computational complexity of the conventional zero-forcing (ZF) or minimum mean square error (MMSE) equalizer will be high when the channel is time-varying. Sev- eral low cost equalizers have been proposed in the literature. In particular, it was shown [16] that most ICI comes from neighbor subcarriers. From this conclusion, a reduced complexity equalizer, Q-tap equalizer, is proposed without sacri‾cing much performance. To reduce the complexity of equalizer, another method is to add a time domain window at the receiver [17]. The purpose of adding the time domain window is to prone the channel response into a desired form. However, this idea does not reduce additive channel noise, nor does it reduces interference from other subcarriers. The result of performance seems to be less than satisfactory. Presented in this thesis is a novel method of channel equalization, and ICI reduc- tion based on the time domain window and Q-tap equalizer. By jointly designing the window and Q-tap equalizer, the system can better reduce ICI and its performance is greatly improved without a large computation. Besides, we also extend the time domain window using banded matrix. Finally, Simulation results are provided to demonstrate the performance of the proposed methods.