| 摘要 |
The forward filter of a decision feedback equalizer has a tapped delay line fox receiving an input signal to produce successively delayed tap signals, which are multiplied by respective tap-gain coefficients, and summed together. The outputs of the forward and backward filters of the equalizer are combined and a threshold decision is made on the combined signal to produce a decision output, from which a decision error is detected. The backward filter has a tapped delay line for receiving the decision output to produce successively delayed tap signals, which are multiplied by respective tap-gain coefficients, and summed together to product its output. Correlations between the tap signals of the forward and backward filters and the decision error are detected and the correlations are modified and supplied to the forward and backward filters as the tap-gain coefficients so that each of the tap-gain coefficients of the forward and backward filters converges toward zero. A spectral notch is formed in the frequency response of the forward filter to cancel continuous wave interference and intersymbol interference produced as a result of the spectral notch is canceled by the backward filter.
|
