Inteligência Swarm e Equilíbrio de Verhulst Aplicados à Alocação de Potência em Redes Ópticas CDMA Particionadas

The deployment of code division multiple access (CDMA) technique in optical networks was proposed taking into account the huge optical spectrum availability. In the optical CDMA (OCDMA) systems, multiple users simultaneously sharing the same optical asynchronous channel with neither the necessity of time coordination nor scheduling of resources. However, the attainable levels of multiple access interference (MAI) are a concern, due to the characteristic of the non-coherent power detection associate to the optical signal. The reduction in the MAI effects is crucial in order to achieve performance improvement in these systems. Hence, in this work we propose the use of resource allocation, particularly the power allocation within the lasers, that guarantee a minimum acceptable performance in terms of bit error rate (BER). In the centralized power control strategy, the central node coordinates the gain link of all active nodes in the system. On the other hand, in the partitioned power control optical networks (divided in access and broadcast network sides) the optical signals after the star coupler do not contribute to increase the MAI; as a consequence, the broadcast network side does not affect the power balance associated to each laser from access network side. In this work, the power control problem in partitioned optical CDMA (p-OCDMA) networks is analysed under two distinct approaches: a) based on the Verhulst mathematical model originally conceived for population growth, and discussed in [1]; b) heuristic approach named swarm intelligence (PSO – particle swarm optimization). Extensive simulations and numerical results have indicated that both the Verhulst as the PSO models are adequate in order to resolve the optimization power problem in p-OCDMA networks. The optimization was carried out under the single-objective optimization (SOO) framework, with suitable convergence results for both approaches and considering various p-OCDMA system operation scenarios. Finally, a computational complexity analysis is presented in terms of both iterations toward convergence and number of operations. As a consequence, a more realistic comparative framework is provided for both optimization methodologies, revealing clearly the superiority of the Verhulst analytical-iterative approach.