Beamwidth Design Tradeoffs in Radar-Aided Millimeter-Wave Cellular Networks: A Stochastic Geometry Approach

This paper develops a novel stochastic geometry model to investigate design tradeoffs in a large-scale radar-aided millimeter-wave cellular network to eliminate the beam training overhead. In the proposed system, each base station comprises two sub-systems: the sensing sub-system, where radar localizes users, and the communication sub-system, where directional antennas serve the detected users. Both sub-systems operate simultaneously to eliminate the beam training overhead and reduce misalignment errors. The system is modeled under realistic fading conditions with scatterers and interferers in the environment. System parameters, such as the density of mobile users and undesired clutter, radar cross-section fluctuations, radar search duration, antenna directivity, and bandwidth are considered in the analytical model. Two scenarios are studied; the first considers antenna misalignment for the communication sub-system, where error magnitude depends on the radar sub-system accuracy. The other scenario considers perfect alignment for benchmarking. It is demonstrated that the radar search duration, which controls the radar antenna beamwidth, is a crucial parameter for optimizing the system throughput. Moreover, the beamwidth of the communication antenna exhibits a tradeoff in which a narrow beam yields higher gain and reduces interference but increases the misalignment error impact. The results demonstrate that if the beamwidths of both sub-systems are carefully chosen, the proposed system can eradicate the beam training overhead without compromising the system's performance, in addition to significantly enhancing the average total system throughput.