Precise, Robust and Least Complexity CD estimation

Efficient and low-complexity chromatic dispersion estimation in coherent optical systems

“…In all six investigated cases the estimation error was 0. Unlike methods using autocorrelation [3][4][5], the new method performs well in case of ROF=0 and in FTN systems.…”

“…However, when fast start-up is required their complex hardware realizations may prohibit the use of these methods. On the other hand, blind methods based on the received signal (or signal modifications) autocorrelation functions [3][4][5] (directly related to the clock tone quality) completely fail in Nyquist systems with a roll-off factor (ROF) equal to 0. Additionally, none of these algorithms can work in faster than Nyquist systems (FTN) [6] that has become very attractive in strongly bandlimited systems applications.…”

Maximum-Likelihood-Based Blind Dispersion Estimation for Coherent Optical Communication

“…As described in Section IV-C, this selection of the set S leads to a small performance penalty for any amount of DGD. However, the penalty can, if necessary, be made arbitrarily small by increasing the number of matrices in the set S. The increase of the computational complexity is marginal as a larger number of M matrices is included, which is seen by writing (31) in terms of´(Y a (f )) * Y b (f + 1/T )e −j2πηT f df for a, b ∈ {x, y} and the matrix elements of M. Then, for a given value of η, the four integrals can be calculated first, and the results are used to form a linear combination with the elements of the matrix M. We notice that the final result (31) bears resemblance to the estimator suggested in [6], which was not derived directly from the ML criterion. However, some differences exist, e.g., (i) [6,Eq.…”

“…However, the penalty can, if necessary, be made arbitrarily small by increasing the number of matrices in the set S. The increase of the computational complexity is marginal as a larger number of M matrices is included, which is seen by writing (31) in terms of´(Y a (f )) * Y b (f + 1/T )e −j2πηT f df for a, b ∈ {x, y} and the matrix elements of M. Then, for a given value of η, the four integrals can be calculated first, and the results are used to form a linear combination with the elements of the matrix M. We notice that the final result (31) bears resemblance to the estimator suggested in [6], which was not derived directly from the ML criterion. However, some differences exist, e.g., (i) [6,Eq. (1)] is for a scalar field and no explicit algorithm is given for a polarization-multiplexed signal; (ii) while DGD and PMD are mentioned in [6], no detailed method for handling DGD is given; (iii) The spectral shift is ±1/T in [6], while it is 1/T in (31); (iv) two different cost functions are suggested in [6], while the ML approach results in one single algorithm.…”

Anatomy of a Digital Coherent Receiver

“…Diverse variants of this procedure, each using a different metric, have been shown so far. A criterion derived from constant modulus algorithm (CMA) is used in [10], [11], delay-tap sampling estimator [12], [13], autocorrelation of signal power waveform [14], [15], clock tone search [16]- [18], Gardner time error detector variance [19]. Recently, another technique has been demonstrated, where the sweep over CD values is performed automatically when applying FFT on the autocorrelation of discrete spectrum [20].…”