Signal processing method and device for a memory system interface circuit
A signal processing method for a memory system interface circuit is provided. The memory system interface circuit comprises at least one signal pin each being configured to receive a transmission signal via an individual signal link and to generate a received signal at a receiving node of the signal link. The signal processing method comprises: pre-processing the received signal to obtain an input signal; removing a weighted feedback signal from the input signal to obtain an output signal, wherein the weighted feedback signal is provided by a finite impulse response (FIR) filter in a feedback path; deciding on the output signal based on a predetermined base signal to generate a digital output signal; weighting, on a scale of a filter coefficient matrix of the FIR filter, the digital output signal to obtain the weighted feedback signal; comparing the output signal with a reference signal to generate an error signal; and determining a set of optimal filter coefficients of the filter coefficient matrix of the FIR filter and the reference signal according to correlation between the digital output signal and the error signal, to minimize inter-symbol interference (ISI) of the received signal introduced by transmission characteristics of the signal link.
1 . A signal processing method for a memory system interface circuit, wherein the memory system interface circuit comprises at least one signal pin each being configured to receive a transmission signal via an individual signal link and to generate a received signal at a receiving node of the signal link, and wherein the signal processing method comprises:
pre-processing the received signal to obtain an input signal;
removing a weighted feedback signal from the input signal to obtain an output signal, wherein the weighted feedback signal is provided by a finite impulse response (FIR) filter in a feedback path;
deciding on the output signal based on a predetermined base signal to generate a digital output signal;
weighting, on a scale of a filter coefficient matrix of the FIR filter, the digital output signal to obtain the weighted feedback signal;
comparing the output signal with a reference signal to generate an error signal; and
determining a set of optimal filter coefficients of the filter coefficient matrix of the FIR filter and the reference signal according to correlation between the digital output signal and the error signal, to minimize inter-symbol interference (ISI) of the received signal introduced by transmission characteristics of the signal link, wherein determining a set of optimal filter coefficients of the filter coefficient matrix of the FIR filter and the reference signal according to correlation between the digital output signal and the error signal further comprises:
determining whether to update the reference signal according to signs of the digital output signal and the error signal at a current time moment; and
updating the reference signal at a predetermined reference step according to an indication to update the reference signal.
2 . The signal processing method of claim 1 , wherein the pre-processing further comprises continuous time linear equalization, variable gain adjustment or the combination thereof.
3 . The method of claim 1 , wherein updating the reference signal at a predetermined reference step further comprises:
determining the reference signal using least mean square algorithm or sign-sign least mean square algorithm according to the digital output signal and the error signal.
4 . The method of claim 3 , wherein determining the reference signal using sign-sign least mean square algorithm further comprises updating the reference signal using the following equation:
dLev
n
+
1
=
dLev
n
+
u
dLev
*
sign
(
e
n
)
*
sign
(
d
n
)
wherein dLev n+1 is an updated reference signal at the current time moment, dLev n is the reference signal before updating at the current time moment, u dLev is the predetermined reference step, e n is a value of the error signal at the current time moment, d n is a value of the digital output signal at the current time moment, and sign( ) refers to a mathematical sign function.
5 . The method of claim 1 , wherein the FIR filter comprises m stages, and m is a positive integer, and wherein determining a set of optimal filter coefficients of the filter coefficient matrix of the FIR filter and the reference signal according to correlation between the digital output signal and the error signal further comprises:
determining signs of ISI at k time moments prior to the current time moment according to correlation between a sign of the error signal at a current time moment and signs of the digital output signal at k time moments prior to the current time moment, and determining whether or not to update a filter coefficient of a k th stage of the FIR filter according to the determination of the signs of ISI at k time moments prior to the current time moment, wherein k is a positive integer smaller than or equal to m; and
updating the filter coefficient of the k th stage at a predetermined filter step according to an indication to update the filter coefficient of the k th stage.
6 . The method of claim 5 , wherein updating the filter coefficient of the k th stage at a predetermined filter step further comprises:
determining the filter coefficient of the k th stage using least mean square algorithm or sign-sign least mean square algorithm according to the digital output signal and the error signal.
7 . The method of claim 6 , wherein determining the filter coefficient of the k th stage using sign-sign least mean square algorithm further comprises updating the filter coefficient using the following equation:
w
[
k
]
n
+
1
=
w
[
k
]
n
+
u
dfe
*
sign
(
e
n
)
*
sign
(
d
n
-
k
)
wherein w[k] n+1 is an updated filter coefficient of the k th stage at the current time moment, w[k] n is the filter coefficient of the k th stage before updating at the current time moment, u dfe is the predetermined filter step, e n is a value of the error signal at the current time moment, d n−k is a value of the digital output signal at k time moments prior to the current time moment, and sign( ) refers to a mathematical sign function.
8 . The method of claim 1 , wherein the transmission signal is a randomly generated signal or a pseudo random sequence.
9 . The method of claim 1 , wherein the method is repeated at a predetermined interval to re-determine the set of optimal filter coefficients of the filter coefficient matrix of the FIR filter and the reference signal.
10 . The method of claim 1 , wherein the memory system interface circuit is integrated in a memory interface of a memory controller.
11 . The method of claim 1 , wherein the memory system interface circuit is integrated in an interface of a memory module.
12 . A signal processing device for a memory system interface circuit, wherein the memory system interface circuit comprises at least one signal pin each being configured to receive a transmission signal via an individual signal link and to generate a received signal at a receiving node of the signal link, and wherein the signal processing device comprises:
a pre-processing module configured for pre-processing the received signal to obtain an input signal;
a decision feedback equalizer (DFE) comprising an output path coupled with an output sampler and a feedback path coupled with a finite impulse response (FIR) filter; the DFE being configured for removing a weighted feedback signal from the input signal to obtain an output signal, and the output sampler being configured for deciding on the output signal based on a predetermined base signal to generate a digital output signal; wherein the weighted feedback signal is obtained by weighting on a scale of a filter coefficient matrix of the FIR filter the digital output signal;
an error sampler configured for comparing the output signal with a reference signal to generate an error signal; and
an adaptive processing module configured for determining a set of optimal filter coefficients of the filter coefficient matrix of the FIR filter and the reference signal according to correlation between the digital output signal and the error signal, to minimize inter-symbol interference (ISI) of the received signal introduced by transmission characteristics of the signal link, wherein the adaptive processing module is further configured to determine a set of optimal filter coefficients of the filter coefficient matrix of the FIR filter and the reference signal according to correlation between the digital output signal and the error signal through:
determining whether to update the reference signal according to signs of the digital output signal and the error signal at a current time moment; and
updating the reference signal at a predetermined reference step according to an indication to update the reference signal.
13 . The signal processing device of claim 12 , wherein the adaptive processing module is further configured to update the reference signal using the following equation:
dLev
n
+
1
=
dLev
n
+
u
dLev
*
sign
(
e
n
)
*
sign
(
d
n
)
wherein dLev n+1 is an updated reference signal at the current time moment, dLev n is the reference signal before updating at the current time moment, u dLev is the predetermined reference step, e n is a value of the error signal at the current time moment, d n is a value of the digital output signal at the current time moment, and sign( ) refers to a mathematical sign function.
14 . The signal processing device of claim 12 , wherein the FIR filter comprises m stages, and m is a positive integer, and the adaptive processing module is further configured to determine the filter coefficient using the following equation:
w
[
k
]
n
+
1
=
w
[
k
]
n
+
u
dfe
*
sign
(
e
n
)
*
sign
(
d
n
-
k
)
wherein w[k] n+1 is an updated filter coefficient at the current time moment, w[k] n is the filter coefficient of the k th stage before updating at the current time moment, u dfe is the predetermined filter step, e n is a value of the error signal at the current time moment, d n−k is a value of the digital output signal at k time moments prior to the current time moment, and sign( ) refers to a mathematical sign function.
15 . The signal processing device of claim 12 , wherein the transmission signal is a randomly generated signal or a pseudo random sequence.
16 . The signal processing device of claim 15 , wherein the transmission signal is a pseudo random sequence generated by a circulating shift register.
17 . The signal processing device of claim 12 , wherein the adaptive processing module is configured to re-determine the filter coefficients of the FIR filter at a predetermined interval according to the transmission signal that is randomly generated.
18 . A memory controller comprising the signal processing device of claim 12 .
19 . The memory controller of claim 18 , wherein each signal pin of the memory controller comprises an individual signal processing device of claim 12 .
20 . The memory controller of claim 18 , wherein the memory controller is a register clock driver or a data buffer.
21 . A memory module comprising the signal processing device of claim 12 .
22 . The memory module of claim 21 , wherein each signal pin of the memory module comprises an individual signal processing device of claim 12 .