Blind channel estimation method for an MLSE receiver in high speed optical communication channels
A method for performing blind channel estimation for an MLSE receiver in a communication channel, according to which Initial Metrics Determination Procedure (IMDP) is performed using joint channel and data estimation in a decision directed mode. This is done by generating a bank of initial metrics that assures convergence, based on initial coarse histograms estimation, representing the channel and selecting a first metrics set M from the predefined bank. Then an iterative decoding procedure is activated during which, a plurality of decision-directed adaptation learning loops are carried out to perform an iterative histograms estimation procedure for finely tuning the channel estimation. Data is decoded during each iteration, based on a previous estimation of the channel during the previous iteration. If convergence is achieved, ISI optimization that maximizes the amount of ISI that is compensated by the MLSE is performed.
1. A method for performing blind channel estimation for an MLSE receiver in high speed optical communication channel, comprising:
a) performing Initial Metrics Determination Procedure (IMDP) using joint channel and data estimation in a decision directed mode, by:
a.1) generating a bank of initial metrics that with at least one metric having convergence tendency, based on an initial coarse histograms generated by a set of FIR filters representing parameters of said channel;
a.2) selecting a first metric from said bank of initial metrics;
a.3) activating an iterative decoding procedure during which, a plurality of decision-directed adaptation learning loops are carried out for a selected metric, to perform an iterative histograms estimation for finely tuning the channel estimation, while during each iteration, decoding samples of the signal received via said channel by an MLSE decoder, based on a previous estimation of said channel during the previous iteration;
a.4) checking whether the resulting metrics are converged using sampled standard deviation of the central moments, and if convergence is not achieved, selecting the next metrics set from said bank, otherwise;
a.5) performing ISI optimization by said MLSE receiver using metrics for which convergence has been achieved;
b) if the initial metrics bank is run out of metrics sets, repeating said IMDP over again; and
c) using said decision-directed adaptation loops for tracking variations of said channel during steady state operation.
2. A method according to claim 1 , wherein the checking whether the resulting metrics are converged is performed using a Z-test.
3. A method according to claim 1 , wherein the convergence tendency of the histogram set is monitored by using the sampled standard deviation of the central moments after a predetermined number of iterations.
4. A method according to claim 1 , wherein the convergence tendency of the histogram set is monitored, based on a training sequence.
5. A method according to claim 1 , wherein the ISI optimization is performed by:
a) collecting several channel estimates, while each time, setting a different Match Point (MP)-shift between the stream of ADC samples and the stream of the corresponding decision bits; and
b) selecting the MP-shift that yields the minimal variances-average of the histograms.