Comparison of strategies for traffic optimization in multiservice networks

Communications through networks, at present, involves the transfer of information in different amounts and times for each type of service. In a telephone conversation a small loss of information can be tolerated, since in this case, is more important the rate with which the data reaches the receiver. When it is required to transfer sensitive information such as banking data, longer times of arrival may be tolerable, but high reliability is required. Multiservice networks can transport the different flows, so as to have availability of information with the requirements of each case. Traffic planning on those networks implies solving a combinatorial optimization problem. The heuristic techniques, applied to solving such problems, have demonstrated good performance getting good solutions in acceptable time. In this paper, we propose the comparison of two strategies to address a traffic optimization problem in multiservice networks: the first based on a mechanism of estimation of probability distributions and the second based on the incremental learning populations.