On Interference Alignment for Crosstalk Mitigation in VDSL2

One of the main impairments for advanced digital subscriber line (DSL) systems is crosstalk. Crosstalk can be effectively cancelled using vectoring. However, some challenges such as implementation and computational complexities asso- ciated with full users’ coordination when the number of DSL lines is large or when they are not co-located at any end make full vectoring impractical in many scenarios. In this paper, we consider interference alignment (IA) as an alternative technique when full vectoring is not practical. We apply this technique to both non-coordinated and partially coordinated very fast digital subscriber line 2 (VDSL2) systems. Th e computational complexity of IA can be prohibitive, however, we reduce it by applying IA to subsets of tones resulting in a computational complexity much smaller than that of vectoring. We also use iterative IA algorithms for performance improvement. Simulation results show that IA increases the achievable rates of VDSL2 loops considerably without a need for signal coordination among users.Moreover, when IA and partial vectoring are applied together, the users can achieve bit rates moderately close to that of a crosstalk-free network