This work presents a method for finding diversity set operating points, which are Pareto optimal and diverse, to digital subscriber lines (DSL). Several works presented in the literature have proposed algorithms for optimizing data transmission in DSL lines, which results in a unique operating point for the modems. These works use spectrum balancing algorithms to solve the power allocation
problem, which differs from the approach presented in this work. The proposed method, called diverseSB, uses a hybrid solution that consists of the non-dominated sorting genetic algorithm-II(NSGA-II), based on a multi-objective optimization, and a spectrum balancing algorithm. The simulation
results showed that, for a given diversity, the computational cost for find the operating points with diversity using the diverseSB proposed algorithm is much smaller than “brute-force” search methods. In the proposed method, NSGA-II perform calls to the spectrum balancing algorithm adopted, so many tests involving the same number of calls to the algorithm were performed with the diverseSB proposed and with the brute-force search method, and the results of diverseSB proposed were better than brute-force search method. For example, to obtain a diverse set operating points the brute-force method performs 1,600 calls to the spectrum balancing algorithm and the diverseSB proposed performed 535 calls.
|
