High-speed digital signal processing in a coherent optical receiver
View Patent ↗In a coherent optical receiver, a frequency domain engine digitally processes at least two multi-bit sample streams of a received optical signal. The frequency domain engine includes a Fast Fourier Transform (FFT) filter for computing a complex vector representative of a frequency-domain spectrum of the received optical signal. A transpose and add block computes a vector sum of the complex vector and a transposed version of the complex vector, and an Inverse Fast Fourier Transform (IFFT) filter computes a complex output vector from the addition result. With this arrangement, parallel real filter operations are efficiently performed on each of the multi-bit sample streams, using a single back-to-back FFT-IFFT filter structure.
1. In a coherent optical receiver, a high speed digital signal processor (DSP) for digitally processing at least two multi-bit sample streams of a received optical signal, the digital signal processor comprising:
a Fast Fourier Transform (FFT) filter for processing the at least two multi-bit sample streams to compute an array {R A X } representative of a spectrum of the received optical signal;
a frequency domain processor for applying a selected frequency domain function to the FFT output array {R A X } to derive a modified array {V A X }, the frequency domain processor comprising:
a multiplier for multiplying the FFT output array {R A X } by a respective coefficient vector implementing at least a portion of the frequency domain function; and
a transpose-and-add block for emulating parallel real FFT-IFFT filters; and
an Inverse Fast Fourier Transform (IFFT) filter for deriving a time domain output vector based on the modified array {V A X };
wherein the frequency domain function is selected to support any two or more of: clock recovery, dispersion compensation, quadrature error correction and signal framing; and
wherein the transpose-and-add block comprises:
a vector addition block;
a direct path for supplying the FFT output array {R A X } to the vector addition block; and
a transpose path for supplying a transpose array { R A X } to the vector addition block;
wherein the multiplier comprises a respective vector multiplier disposed in each path, for applying a respective coefficient vector to each array.
2. A digital signal processor as claimed in claim 1 , wherein, in a signal acquisition mode of operation of the coherent optical receiver, the frequency domain function is selected to support any two or more of: clock recovery, dispersion compensation, quadrature error correction and signal framing.
3. A digital signal processor as claimed in claim 1 , wherein, in a steady state mode of operation of the coherent optical receiver, the frequency domain function is selected to support either one or both of clock recovery and dispersion compensation.
4. A digital signal processor as claimed in claim 1 , wherein the FFT filter, Frequency Domain Processor and IFFT filter are pipelined.
5. A digital signal processor as claimed in claim 1 , wherein the at least two multi-bit sample streams comprise respective In-Phase and Quadrature sample streams of a received polarization of the optical signal.
6. A digital signal processor as claimed in claim 5 , wherein each of the multi-bit sample streams has a resolution of at least 5 bits and a sample rate of at least 20 GHz.
7. A digital signal processor as claimed in claim 1 , further comprising an FDP port for outputting at least a portion of the modified array {V A X }.
8. A digital signal processor as claimed in claim 1 , further comprising a DSP port for outputting at least a portion of the complex output array independent of a main data path of the DSP.
9. In a coherent optical receiver, a method of processing at least two multi-bit sample streams of a received optical signal, the method comprising steps of:
processing the at least two multi-bit sample streams to compute an array {R A X } representative of a spectrum of the received optical signal;
applying a selected frequency domain function to the array {R A X } to derive a modified array {V A X }, wherein the step of applying the frequency domain function to the array {R A X } further comprises steps of:
multiplying the array {R A X } and a transpose array { R A X } by a respective coefficient vector implementing at least a portion of the frequency domain function; and
adding the multiplication results;
deriving a time-domain output vector based on the modified array {V A X }; and
wherein the frequency domain function is selected to support any two or more of: clock recovery, dispersion compensation, quadrature error correction and signal framing.
10. A method as claimed in claim 9 , wherein, in a signal acquisition mode of operation of the coherent optical receiver, the frequency domain function is selected to support any two or more of: clock recovery, dispersion compensation, quadrature error correction and signal framing.
11. A method as claimed in claim 9 , wherein, in a steady state mode of operation of the coherent optical receiver, the frequency domain function is selected to support either one or both of clock recovery and dispersion compensation.
12. A method as claimed in claim 9 , wherein the steps of computing the array {R A X }, applying the selected frequency domain function FFT filter, and deriving the time-domain output vector are pipelined.
13. A method as claimed in claim 9 , wherein the at least two multi-bit sample streams comprise respective In-Phase and Quadrature sample streams of a received polarization of the optical signal.
14. A method as claimed in claim 13 , wherein each of the multi-bit sample streams has a resolution of at least 5 bits and a sample rate of at least 20 GHz.
15. A method as claimed in claim 9 , wherein the step of computing an array {R A X } comprises a step of performing a Fast Fourier Transform (FFT).
16. A method as claimed in claim 9 , further comprising a step of outputting at least a portion of the modified array {V A X }.
17. A method as claimed in claim 9 , further comprising a step of outputting at least a portion of the complex output array independently of a main data path of the DSP.
18. In a coherent optical receiver, a high speed digital signal processor (DSP) for digitally processing at least two multi-bit sample streams of a received optical signal, the digital signal processor comprising:
a Fast Fourier Transform (FFT) filter for processing the at least two multi-bit sample streams to compute an array {R A X } representative of a spectrum of the received optical signal;
a frequency domain processor for applying a selected frequency domain function to the FFT output array {R A X } to derive a modified array {V A X };
an Inverse Fast Fourier Transform (IFFT) filter for deriving a time domain output vector based on the modified array {V A X };
a selector responsive to a select signal, for supplying a selected one of a set of two or more predetermined coefficient vectors to the frequency domain processor, wherein the set of two or more predetermined coefficient vectors comprises:
a first coefficient vector containing coefficients calculated to at least partially compensate dispersion of the received optical signal; and
a second coefficient vector containing coefficients calculated using a transform of a predetermined symbol sequence embedded in the optical signal, such that the complex output vector contains a cross-correlation of the sample streams and the predetermined sequence of symbols.
19. A digital signal processor as claimed in claim 18 , further comprising an FDP port for outputting at least a portion of the modified array {V A X }.
20. A digital signal processor as claimed in claim 18 , further comprising a DSP port for outputting at least a portion of the complex output array independent of a main data path of the DSP.
21. A digital signal processor as claimed in claim 18 , wherein at least one coefficient vector contains coefficients calculated to correct quadrature error of the multi-bit sample streams.
22. A digital signal processor as claimed in claim 18 , wherein the predetermined symbol sequence is a SYNC burst embedded in the optical signal.
23. A digital signal processor as claimed in claim 22 , further comprising a barrel-shifter for adjusting a position of the SYNC burst within the complex output vector, based on a detected timing of the SYNC burst.
24. A digital signal processor as claimed in claim 18 , wherein the predetermined symbol sequence is a pilot channel signature embedded in the optical signal.
25. In a coherent optical receiver, a method of processing at least two multi-bit sample streams of a received optical signal, the method comprising steps of:
processing the at least two multi-bit sample streams to compute an array {R A X } representative of a spectrum of the received optical signal;
applying a selected frequency domain function to the array {R A X } to derive a modified array {V A X }, wherein the step of applying the frequency domain function to the array {R A X } comprises steps of:
selecting one of a set of two or more predetermined coefficient vectors; and
multiplying the array {R A X } by the selected coefficient vector;
deriving a time-domain output vector based on the modified array {V A X };
wherein the set of two or more predetermined coefficient vectors comprises:
a first coefficient vector containing coefficients calculated to at least partially compensate dispersion of the received optical signal; and
a second coefficient vector containing coefficients calculated using a transform of a predetermined symbol sequence embedded in the optical signal, such that the complex output vector contains a cross-correlation of the sample streams and the predetermined sequence of symbols.
26. A method as claimed in claim 25 , further comprising a step of outputting at least a portion of the modified array {V A X }.
27. A method as claimed in claim 25 , further comprising a step of outputting at least a portion of the complex output array independently of a main data path of the DSP.
28. A method as claimed in claim 25 , wherein at least one coefficient vector contains coefficients calculated to correct quadrature error of the multi-bit sample streams.
29. A method as claimed in claim 25 , wherein the predetermined symbol sequence is a SYNC burst embedded in the optical signal.
30. A method as claimed in claim 29 , further comprising a step of adjusting a position of the SYNC burst within the complex output vector, based on a detected timing of the SYNC burst.
31. A method as claimed in claim 25 , wherein the predetermined symbol sequence is a pilot channel signature embedded in the optical signal.