Optical Fiber Communications

Product Accumulate Codes for Optical Fiber Channels:

Bandwidth- and power-efficient forward error correction codes are desired for optical fiber communications. We investigate the performance of several advanced codes on optical fiber channels where amplitude spontaneous emission (ASE) noise dominates all other noise sources. Three channel models are considered: asymmetric Chi-square, asymmetric Gaussian and symmetric Gaussian.

  • Product accumulate (PA) codes are proposed as a low-complexity alternative to turbo or LDPC codes for AWGN channels. We investigate their potential for use in optical fiber communications. We show that they can achieve impressive coding gains on optical fiber channels also. (paper).

Maximum Likelihood Bounds:

Theoretical analysis is conducted to facilitate the evaluation of FEC for optical fiber communications. Pairwise error probability (PEP) is derived for three channel models and union bounding technique is exploited.

  • To evaluate the performance of PA codes at high SNR regions that are beyond simulation capabilities, we explore an ensemble average bound under the assumption of ML decoding (paper).

Capacity of Optical Fiber Channels:

We compute the Shannon limit for optical fiber channels with hard and soft decision. Both the ultimate channel capacity (optimal channel input) and the practical channel capacity (equal probability of channel input) are considered.

  • Asymmetric Chi-square channel is by far the most accurate channel model for optical fiber communications employing optical amplifier. We investigate the capacity of Chi-square channels with continuous output (soft-decision) and binary output (hard-decision) (paper).