Efficient low complexity partial crosstalk cancellation algorithm by using successive linear convex relaxation and branch and bound approaches in VDSL networks

Abstract

Partial crosstalk cancellation algorithm has been proposed to reduce the online complexity of crosstalk canceller in xDSL systems. However, it must have a very low pre-processing complexity to allow working efficiently in time-varying crosstalk environment. Also, much lower online complexity can be achieved by joint solution of the multi-user power control and partial crosstalk cancellation problems. Currently, this joint problem has been formulated as a constrained weighted sum rate maximization problem and solved by employing Lagrange dual decomposition method. However, it suffers from applying per-tone exhaustive search for its solution because of non-convexity of the problem. In this paper, the joint problem is reformulated as a mixed binary-convex problem by successive linear convex relaxation technique which can be solved by the efficient branch and bound method. The analytical and simulation results show that the proposed approach provides a solution with extremely low pre-processing complexity and close to the optimal solution.