Dynamic resource allocation strategy for frequency-based passive optical networks

Dynamic resource allocation presents a key issue in passive optical networks (PONs) for efficient and fair utilization of system resources in order to ensure leveled quality-of-service (QoS) requirements. This paper investigates the problem of power and bandwidth allocation for the maximization of the worst weighted capacity/rate in frequency-based PONs. By deriving a convex optimization problem, we propose an optimization algorithm which gives an upper bound of the weighted capacity and maintains the fairness related to users' capacity requirements. In addition, the theoretical analysis is applied to practical scenarios by taking the discrete nature of modulation techniques into account. An allocation strategy in terms of discrete optimization is presented to improve user weighted transmission rate. Numerical results validate the convergence and the performance of the proposed resource allocation strategy, which is consistent with the analytical results.