To make the most efficient use of scarce bandwidth, channel assignment methods for wireless mesh networks (WMNs) should try to minimize the number of frequency channels used while achieving maximum network throughput. Beamforming is a well-known technique that improves spatial reuse in wireless networks. However, there are no channel assignment methods for WMNs that use beamforming to reduce the number of frequency channels. We develop the first channel assignment method for dynamic WMNs that incorporates beamforming in the conflict graph and matrix. This reduces co-channel interference significantly, thereby reducing the number of frequency channels required (NCR) to ensure interference-free communication among the mesh nodes while achieving maximum network throughput. Our novel Linear Array Beamforming-based Channel Assignment (LAB-CA) method significantly increases the spectrum utilization efficiency of WMNs at the expense of increased hardware complexity. It outperforms classical omni-directional antenna pattern-based channel assignment (OAP-CA) in terms of NCR. In a heterogeneous WMN where mesh nodes have differing numbers of radio interfaces, LAB-CA also outperforms OAP-CA in terms of NCR in both sparse and dense scenarios. A further significant reduction in NCR is achieved when the number of antennas in the linear antenna arrays of mesh nodes is increased.
