Uncertain Networks Optimization Based on Random ADD Algorithm

Abstract

In order to solve the traditional network optimization problem, the dual gradient descent algorithm is adopted. Although the algorithm can be realized by a distributed method, its convergence rate is slow. Acceleration dual decline algorithm mainly uses the distributed calculation of the approximative Newton step to improve its convergence rate. Due to the uncertainty characteristic of communication networks, its convergence is difficult to be guaranteed in the presence of restriction uncertainty. An ADD algorithm based on random modality is proposed to resolve the network optimization problem. It is proved theoretically that ADD algorithm can converge to an error neighborhood of optimum value with higher probability when the uncertainty mean square error is bounded; the ADD algorithm can also converge to the optimum value with higher probability when there is the more stringent restriction condition. Experiment results demonstrate the convergence rate of the proposed algorithm is more than 100 times faster than the random gradient descent algorithm