Bi-partition based analog beamforming for millimeter wave backhaul systems

Analog beamforming (ABF) provides a low-cost, high power efficiency solution for millimeter wave (mmWave) backhaul. Typical ABF algorithms rely on BF training to exploit the array gains using the Channel State Information (CSI). However, most existing ABF methods either produce a large dynamic range or low Signal-to-Noise Ratio (SNR) during the training, which deteriorate the BF gains in practical systems. A bi-partition based beamforming (BP-BF) method is proposed here to address this problem, in which the full array BF is decomposed into a series of dual-element-array (DEA) BFs with specifically designed weights to improve the training signals. Then a practical antenna aiming scheme with two joint transmit-receive (JTR) BP-BF methods are also proposed for different system setups in this paper. The simulations show that the BP-BF method achieves higher BF gains in realistic system setups. The algorithm is also implemented and validated in our prototype equipment.