The Study of Nonlinear Equalizer with Neural Network
碩士 === 國立彰化師範大學 === 電機工程學系 === 95 === A comparison study between the multilayer perceptron (MLP) based equalizer and the linear equalizer is conducted in this thesis. In addition to channel distortion, the effect of frequency offset between transmitting clock and receiving clock on the performance...
| Main Authors: | , |
|---|---|
| Other Authors: | |
| Format: | Others |
| Language: | zh-TW |
| Published: |
2007
|
| Online Access: | http://ndltd.ncl.edu.tw/handle/40628175453922924389 |
| Summary: | 碩士 === 國立彰化師範大學 === 電機工程學系 === 95 === A comparison study between the multilayer perceptron (MLP) based equalizer and the linear equalizer is conducted in this thesis. In addition to channel distortion, the effect of frequency offset between transmitting clock and receiving clock on the performance of the equalizer is observed. Timing error resulting from the frequency offset between the two clocks affects the bit error rate (BER) performance of the equalizer. In contrast to the linear structure, MLP is nonlinear in nature. Hence, the MLP based equalizer is capable of reducing the BER as a result of its nonlinearity in forming decision boundaries in the signal space. This thesis further investigates the effect of frequency offset on the BER performance. Computer simulations show that the MLP based equalizer outperforms the one using linear structure in different extent of channel distortion and frequency offset.
|
|---|