Constrained maximum-likelihood detection for high-speed fiber-optic systems

The authors present design techniques for constrained maximum-likelihood detection (MLD), a receiver electrical signal processing technique that provides near-optimum interference compensation, yet can be implemented at gigabit-per-second data rates. The technique determines the most likely transmitted bit given N₁ previously detected bits and N₂ received signal samples, using decision boundaries formed by (N₂-1)-dimensional planes which are implemented with comparators. Two examples are used to illustrate the design procedure and the tradeoffs, showing the application to coherent and direct detection systems with polarization and chromatic dispersion, as well as to systems with nonlinearities. The results show that the performance of these techniques approaches that of MLD