An optimal spectrum‐balancing algorithm for digital subscriber lines based on particle swarm optimization

Abstract

SUMMARYThis paper presents a new algorithm for optimal spectrum balancing in modern digital subscriber line (DSL) systems using particle swarm optimization (PSO). In DSL, crosstalk is one of the major performance bottlenecks, therefore various dynamic spectrum management algorithms have been proposed to reduce excess crosstalks among users by dynamically optimizing transmission power spectra. In fact, the objective function in the spectrum optimization problem is always nonconcave. PSO is a new evolution algorithm based on the movement and intelligence of swarms looking for the most fertile feeding location, which can solve discontinuous, nonconvex and nonlinear problems efficiently. The proposed algorithm optimizes the weighted rate sum. These weights allow the system operator to place differing qualities of service or importance levels on each user, which makes it possible for the system to avoid the selfish-optimum. We can show that the proposed algorithm converges to the global optimal solutions. Simulation results demonstrate that our algorithm can guarantee fast convergence within a few iterations and solve the nonconvex optimization problems efficiently.

Enhanced self-adaptive evolutionary algorithm for numerical optimization

“…In the PSO algorithms [5], there is an important principle that all particles are guided by the global best and the local best particles. Thus, we design the PSO-GR-BL strategy based on the Baldwinian learning and PSO technique as follows:…”

“…Differential evolution (DE) [1,2], artificial immune system (AIS) [3,4] and particle swarm optimization (PSO) [5,6] have been the active fields of the evolutionary computation (EC) in the past decade. However, the existing algorithms usually perform very well on some special optimal problems whereas perform very poor on the other problems.…”