A flexible and real-time constrained controller for sparse linear zero-forcing based DSL vectoring

Abstract

Vectoring is recognized as a very efficient technique to tackle crosstalk interference in multi-user DSL systems. In this paper we focus on sparse linear zero-forcing based vectoring schemes with a shared resource pool, which are known to obtain a good trade-off between online computational complexity and near-optimal performance. We propose a new system setup with a controller that dynamically (in time) optimizes and allocates computational complexity resources taking into account variable data rate demands with real-time constraints. The controller is designed to minimize the online computational complexity over time and allows for flexible fixed and variable data rate allocations. Simulation results with variable real-time constrained video streams for a realistic downstream VDSL scenario demonstrate that this novel controller-based approach allows to significantly reduce the average as well as peak online computational complexity.