Adaptive equalizer with coefficients determined using groups of symbols to compensate for nonlinear distortions in optical fiber communications
View Patent ↗An adaptive equalizer with coefficients determined by averaging an estimated filter coefficient over a number, N, of past and future symbols. Estimated filter coefficients may be optimized by optimization of the number N, an averaging window function and a scaling factor using a metric. The metric also allows estimation of the amount of noise that may be compensated by an adaptive equalizer consistent with the present disclosure.
1. An optical signal receiver comprising: a coherent receiver for receiving an optical signal and providing a receiver output representative of a plurality of symbols; and a digital signal processor (DSP) configured to receive a digital signal representative of the plurality of symbols, the digital signal processor comprising an adaptive equalizer comprising a plurality of delays for producing successively delayed versions of the digital signal and a plurality of multipliers, each of the multipliers multiplying the digital signal or an associated one of the delayed versions of the digital signal with an associated filter coefficient to produce an associated tap weight product, the tap weight products being summed to provide an output for the adaptive equalizer, each of the filter coefficients being determined by averaging an estimated value of the filter coefficients over a number of past symbols and future symbols.
2. An optical signal receiver according to claim 1 , wherein the averaging an estimated value of the filter coefficients over a number of past and future ones of the symbols is calculated as:
〈
G
t
〉
≡
1
N
+
1
∑
n
=
-
N
/
2
N
/
2
G
t
+
n
where G t is the estimated value, and N is the number of the past and future ones of the symbols.
3. An optical signal receiver according to claim 1 , wherein the estimated value of the filter coefficients is calculated as:
E
t
k
p
N
=
α
k
〈
Δ
t
·
y
t
+
k
p
exp
(
-
i
·
arg
(
y
t
+
k
)
)
W
t
k
〉
〈
y
t
+
k
p
+
1
〉
,
where k is a number of taps in the adaptive equalizer, t denotes a current update cycle, p is a variable parameter, indicates the averaging the estimated value of the filter coefficients over the number N of past and future ones of the symbols, α k is a scaling factor, W t k is an averaging window function, y t is an output of the equalizer, and Δ t is an error signal.
4. An optical signal receiver according to claim 1 , wherein the estimated value of the filter coefficients is calculated as:
E
t
k
1
N
=
α
k
〈
Δ
t
·
y
t
+
k
*
W
t
k
〉
〈
y
t
+
k
2
W
t
k
〉
,
where k is a number of taps in the adaptive equalizer, t denotes a current update cycle, indicates the averaging the estimated value of the filter coefficients over the number N of past and future ones of the symbols, α k is a scaling factor, W t k is an averaging window function, y t is an output of the equalizer, and Δ t is an error signal.
5. An optical signal receiver according to claim 1 , wherein the number of past and future ones of the symbols is optimized using a metric given as:
M
k
=
〈
Δ
t
-
δ
t
k
2
W
t
k
〉
〈
Δ
t
2
W
t
k
〉
where k is a number of taps in the adaptive equalizer, t denotes a current update cycle, Δ t is an error signal defined by the difference between an output the equalizer y t and a transmitted known reference signal x t , and δ t k is an estimate of Δ t for tap k, and W t k is an averaging window function.
6. An optical signal receiver according to claim 1 , wherein the adaptive equalizer is configured to iteratively calculate the filter coefficients using the output of the equalizer for each iteration.
7. A method of demodulating an optical signal, the method comprising: coupling a first digital signal representative of a plurality of symbols modulated on the optical signal to a digital signal processor (DSP) for processing by an adaptive equalizer; and determining filter coefficients for the adaptive equalizer by averaging an estimated value of the filter coefficients over a number of past symbols and future symbols.
8. A method according to claim 7 , wherein the averaging an estimated value of the filter coefficients over a number of past and future ones of the symbols is calculated as:
〈
G
t
〉
≡
1
N
+
1
∑
n
=
-
N
/
2
N
/
2
G
t
+
n
where G t is the estimated value, and N is the number of the past and future ones of the symbols.
9. A method according to claim 7 , wherein the estimated value of the filter coefficients is calculated as:
E
t
k
p
N
=
α
k
〈
Δ
t
·
y
t
+
k
p
exp
(
-
i
·
arg
(
y
t
+
k
)
)
W
t
k
〉
〈
y
t
+
k
p
+
1
〉
,
where k is a number of taps in the adaptive equalizer, t denotes a current update cycle, p is a variable parameter, indicates the averaging the estimated value of the filter coefficients over the number N of past and future ones of the symbols, α k is a scaling factor, W t k is an averaging window function, y t is an output of the equalizer, and Δ t is an error signal t .
10. A method according to claim 7 , wherein the estimated value of the filter coefficients is calculated as:
E
t
k
1
N
=
α
k
〈
Δ
t
·
y
t
+
k
*
W
t
k
〉
〈
y
t
+
k
2
W
t
k
〉
,
where k is a number of taps in the adaptive equalizer, t denotes a current update cycle, indicates the averaging the estimated value of the filter coefficients over the number N of past and future ones of the symbols, α k is a scaling factor, W t k is an averaging window function, y t is an output of the equalizer, and Δ t is an error signal.
11. A method according to claim 7 further comprising optimizing the number of past and future ones of the symbols using a metric given as:
M
k
=
〈
Δ
t
-
δ
t
k
2
W
t
k
〉
〈
Δ
t
2
W
t
k
〉
where k is a number of taps in the adaptive equalizer, t denotes a current update cycle, Δ t is an error signal defined by the difference between an output the equalizer y t and a transmitted known reference signal x t , and δ t k is an estimate of Δ t for tap k, and W t k is an averaging window function.
12. A method according to claim 7 further comprising iteratively calculating the filter coefficients using the output of the equalizer for each iteration.
13. An optical communication system comprising: a transmitting terminal for transmitting a plurality of optical signals; each at a different associated wavelength and having a plurality of symbols modulated thereon, on an optical information path; and an optical signal receiver coupled to the optical information path for receiving the at least one of the plurality of optical signals, the optical signal receiver comprising: a coherent receiver for receiving the at least one of the plurality of optical signals and providing an output in response to the at least one of the plurality of optical signals, and a digital signal processor (DSP) configured to receive a digital signal representative of the plurality of symbols, the digital signal processor comprising an adaptive equalizer comprising a plurality of delays for producing successively delayed versions of the digital signal and a plurality of multipliers, each of the multipliers multiplying the digital signal or an associated one of the delayed versions of the digital signal with an associated filter coefficient to produce an associated tap weight product, the tap weight products being summed to provide an output for the adaptive equalizer, each of the filter coefficients being determined by averaging an estimated value of the filter coefficients over a number of past symbols and future symbols.
14. An optical communication system according to claim 13 , wherein the averaging an estimated value of the filter coefficients over a number of past and future ones of the symbols is calculated as:
〈
G
t
〉
≡
1
N
+
1
∑
n
=
-
N
/
2
N
/
2
G
t
+
n
where G t is the estimated value, and N is the number of the past and future ones of the symbols.
15. An optical communication system according to claim 13 , wherein the estimated value of the filter coefficients is calculated as:
E
t
k
p
N
=
α
k
〈
Δ
t
·
y
t
+
k
p
exp
(
-
i
·
arg
(
y
t
+
k
)
)
W
t
k
〉
〈
y
t
+
k
p
+
1
〉
,
where k is a number of taps in the adaptive equalizer, t denotes a current update cycle, p is a variable parameter, indicates the averaging the estimated value of the filter coefficients over the number N of past and future ones of the symbols, α k is a scaling factor, W t k is an averaging window function, y t is an output of the equalizer, and Δ t is an error signal.
16. An optical communication system according to claim 13 , wherein the estimated value of the filter coefficients is calculated as:
E
t
k
1
N
=
α
k
〈
Δ
t
·
y
t
+
k
*
W
t
k
〉
〈
y
t
+
k
2
W
t
k
〉
,
where k is a number of taps in the adaptive equalizer, t denotes a current update cycle, indicates the averaging the estimated value of the filter coefficients over the number N of past and future ones of the symbols, α k is a scaling factor, W t k is an averaging window function, y t is an output of the equalizer, and Δ t is an error signal.
17. An optical communication system according to claim 13 , wherein the number of past and future ones of the symbols is optimized using a metric given as:
M
k
=
〈
Δ
t
-
δ
t
k
2
W
t
k
〉
〈
Δ
t
2
W
t
k
〉
where k is a number of taps in the adaptive equalizer, t denotes a current update cycle, Δ t is an error signal defined by the difference between an output the equalizer y t and a transmitted known reference signal x t , and δ t k is an estimate of Δ t for tap k, and W t k is an averaging window function.
18. An optical communication system according to claim 13 , wherein the adaptive equalizer is configured to iteratively calculate the filter coefficients using the output of the equalizer for each iteration.