Multi-criteria quadratic programming based low complexity nonlinear channel equalisation

We propose a low complexity multi-criteria quadratic programming (MCQP) based approach for nonlinear channel equalisation in wireless communications viewed as a classification problem. Compared to the standard SVM method, the proposed MCQP approach not only provides a better performance by introducing additional criteria to the objective function, but also requires a much lower complexity by solving a set of linear equations. Besides, the MCQP based equaliser achieves nearly the same performance as the optimal Bayesian detector. Furthermore, MCQP demonstrates its effectiveness in tracking the time variation of communication channels.