This dissertation focuses on the Fiber-Wireless (FiWi) access networks, more specifically on the problem of assigning frequencies to maintain connectivity and acceptable standards of service quality in face of changes in the pattern of traffic flows in the network. Mainly realized on radio and fiber technologies, these networks form an hybrid architecture comprising an optical section and a wireless section that provides a feasible paradigm for high bandwidth and ubiquity at new access network areas. In these FiWi scenarios, in particular when multi-radio and multi-channel configurations are used, an effective frequency assignment should be done to radios so that higher throughput and low delay can be obtained and the best of such architectures is achieved. However, traffic conditions may change over time, meaning that radio channel configurations may be outdated and new reconfigurations can be done to improve network performance. To cope with the increasing demand for bandwidth, fiber to the home/premises/building (FTTX) technologies were massively deployed at the back-end. These technologies are characterized by the huge bandwidth capacity and the absence of active devices on the network plant, which is an advantage for power saving. On the other hand, at the front-end, wireless mesh networks (WMN) are expected to provide mobility and converge different wireless technologies to provide high-speed and huge bandwidth connectivity to the end user. In this dissertation, the frequency reassignment problem in the context of FiWi access networks is discussed and a state-of-art on the subject is proposed. Also, two methodologies for frequency reconfiguration planning are proposed along with their mathematical formalization, and are evaluated by simulation. In one of the strategies, NBR, the algorithm prioritizes channel assignment according to the relative position of nodes and their gateways, while in the other, RBR, nodes are processed as their routes toward the gateways are traversed. A discrete event simulation model to evaluate the performance of the proposed frequency reassignment algorithms was developed using OMNeT++ framework. Simulation results showing that RBR is the algorithm that better exploits channel reconfigurations are presented and discussed.