A Hybrid Meta-Heuristic Algorithm for Dynamic Spectrum Management in Multiuser Systems: Combining Simulated Annealing and Non-Linear Simplex Nelder-Mead

One of the major sources of performance degradation of current Digital Subscriber Line systems is the electromagnetic coupling among different twisted pairs within the same cable bundle (crosstalk). Several algorithms for Dynamic Spectrum management have been proposed to counteract the crosstalk effect but their complexity remains a challenge in practice. Optimal Spectrum Balancing (OSB) is a centralized algorithm that optimally allocates the available transmit power over the tones making use of a Dual decomposition approach where Lagrange multipliers are used to enforce the constraints and decouple the problem over the tones. However, the overall complexity of this algorithm remains a challenge for practical DSL environments. The authors propose a low-complex algorithm based on a combination of simulated annealing and non-linear simplex to find local (almost global) optimum spectra for multiuser DSL systems, whilst significantly reducing the prohibitive complexity of traditional OSB. The algorithm assumes a Spectrum Management Center (at the cabinet side) but it neither relies on own end-user modem calculations nor on messaging-passing for achieving its performance objective. The approach allows furthering reducing the number of function evaluations achieving further reduction on the convergence time (up to ~27% gain) at reasonable payoff (weighted data rate sum).