| Summary: | 碩士 === 國立成功大學 === 電腦與通信工程研究所 === 98 === In this thesis, expectation-maximization (EM) channel estimation algorithms for multiple-input multiple-output orthogonal frequency division multiplexing (MIMO-OFDM) systems are investigated. This work can be mainly divided into two parts. In the first part, the condition that the channel length used in algorithms is greater than the real channel length is assumed and thus channel impulse response (CIR) estimated by the original EM algorithms may contain extra noise components which cannot be completely eliminated. To mitigate the effects of these noise components, a noise reduction method is proposed to increase the performance. Moreover, since the performance can be increased when more useful pilot symbols are applied, a two-dimension EM channel estimation method is proposed in the second part of this thesis. Simulation results show that the EM channel estimation with the proposed noise reduction method is insensitive to the effects of inaccurate channel length. Additionally, the proposed two-dimension EM channel estimation method also has better performance under static or time-varying channel environments.
|
|---|
