CNN-Based Blockage Detection and Beamforming Design in Dual-Band Communication Systems

Abstract

In this paper, we consider a dual-band communication system, where a base station (BS) can communicate with a user equipment (UE) over either the sub-6GHz band or the millimeter wave (mmWave) band. We consider the joint detection of the blockage status of the BS-UE link and the optimization of the downlink rate through frequency band scheduling and beamforming design. Typically, the blockage detection requires the transmission and detection of an echo signal, and the downlink beamforming design relies on the estimation of the channel state information (CSI) of the downlink sub-6GHz and mmWave bands, the acquisition of which may result in high complexity and large overhead. To address these problems, we investigate solving the problem using only the CSI of the uplink sub-6GHz band. We decompose the problem into a blockage detection sub-problem and a beamforming design sub-problem. For the former sub-problem, we propose an efficient algorithm based on the convolutional neural network (CNN) technique. For the latter sub-problem, we propose an algorithm based on the MobileNetV2 module to achieve a high learning efficiency. Simulation results show that the proposed blockage detection algorithm can achieve over 99% detection accuracy, and the proposed beamforming design algorithm can achieve over 10% downlink rate gain, as compared to several benchmark schemes.

Keywords

• Beamforming design
• blockage detection
• convolutional neural network
• dual-band communication