Apparatus and method for processing a digital signal in a frequency domain linear equalizer

An equalizing method for processing a digital signal, the method comprising:

• transforming an input signal into the frequency domain by applying an FFT processing to obtain a transformed input signal;
• discarding a part of the transformed input signal;
• separating positive and negative frequency components of a remaining transformed input signal;
• separately processing respective ones of the separated positive and negative frequency components by respective digital filtering to obtain filtered frequency components, wherein the digital filtering comprise at least one of a chromatic dispersion (CD) filtering and a multiple-input multiple output (MIMO) filtering;
• updating coefficients of the MIMO filter based on the result of multiplication of decimated conjugate output data of the CD filter by decimated error samples;
• combining the filtered frequency components in the frequency domain using a down-sampling operation for down-sampling the filtered frequency components from an input number of samples per symbol to a different output number of samples per symbol;
• and
• converting the combined output components into the time domain by applying an IFFT processing,
• wherein the positive and negative frequency components are determined by the following positive and negative frequency components vectors;
• XFFTA(l)=XFFT(l)
• XFFTB(l+1)=XFFT(N−1−l)
• wherein l=0, . . . , L, and wherein L is determined as follows:
• L=(1+r)·M/2,
• wherein r is the maximal roll-off factor of a Nyquist filter and M is the data size of the IFFT.

A computer program product for processing a digital signal, the computer program product comprising a non-transitory computer readable storage medium having program code embodied therewith, the program code executable by a processor to:

• transform an input signal into the frequency domain by applying an FFT processing to obtain a transformed input signal;
• discard a part of the transformed input signal;
• separate positive and negative frequency components of a remaining transformed input signal;
• separately process respective ones of the separated positive and negative frequency components by respective digital filtering to obtain filtered frequency components, wherein the digital filtering comprise at least one of a chromatic dispersion (CD) filtering and a multiple-input multiple output (MIMO) filtering;
• update coefficients of the MIMO filter based on the result of multiplication of decimated conjugate output data of the CD filter by decimated error samples;
• combine the filtered frequency components in the frequency domain using a down-sampling operation for down-sampling said filtered frequency components from an input number of samples per symbol to a different output number of samples per symbol;
• and
• convert the combined output components into the time domain by applying an IFFT processing,
• wherein the positive and negative frequency components are determined by the following positive and negative frequency components vectors;
• XFFTA(l)=XFFT(l)
• XFFTB(l+1)=XFFT(N−1−l)
• wherein l=0, . . . , L, and wherein L is determined as follows:
• L=(1+r)·M/2,
• wherein r is the maximal roll-off factor of a Nyquist filter and M is the data size of the IFFT.