In this paper, we propose a novel frequency scheduling strategy for an OFDM switched-beam system. We deal with dynamic frequency allocation based on the distance of the user from the serving base station (BS) or on the power of the channel response received by the user. The key idea is to reduce the interference among the neighboring cells. To do so, we consider a time-space-frequency allocation scheme where users are assigned with the appropriate set of subcarriers according to their serving beam at any given time. We evaluate the performance of the proposed frequency allocation scheme when Kalman filtering is used for joint channel estimation and beam selection. A system-level simulator is developed which computes the signal to interference ratio (SIR) for a reference mobile user with different resource allocation strategies. Previously obtained link-level throughput vs. signal to noise ratio (SNR) results are then translated into a system-level cumulative distribution function (cdf) of the user throughput. Link-level frame error rate (FER) results suggest that the proposed Kalman-based OFDM switched-beam system offers high performance over slow fading channels. System-level SIR results show that the proposed frequency scheduling scheme reduces significantly the interference. Our scheme enhances the system throughput compared to other allocation schemes.