Beamforming and user scheduling for sub-array architecture in mmwave networks

Abstract

Wireless access networks are developing to provide high-quality voice calls and video calls as well as high-speed/high-capacity data transmission for accurate and rapid sharing of information in dense living spaces of the modern society and for providing services such as mobile video, Machine-to-Machine (M2M), Internet of Things (IoT), and etc. Millimeter Wave (mmWave) is one of the promising spectrum bands which can support huge data transmission with large bandwidth in 30-300 GHz frequency band. Thanks to the short wavelength of the mmWave, a large number of antenna elements can be employed in the limited dimensions of mmWave transceivers. In addition, data transmission in a narrow beam overcomes the large propagation loss in mmWave wireless networks. For efficient operation/beamforming exploiting many antennas, hence, hybrid beamforming that appropriately operates analog beamforming and digital beamforming is applied to the system.

Most of all, we propose a cooperative beamforming scheme to improve the data rate without increasing the hardware complexity of the sub-connected beamforming structure using many antenna elements. The proposed beamforming creates an antenna group of one sub-array and one or more antenna groups of a neighbouring sub-arrays as a group of antennas of a logical sub-array in order to combine individual beams. Beam combining via virtual beamforming attains higher beamforming gain by generating a narrower beam through beam combining and flexible operation without modification of the beamforming structure. In addition, a beam pattern according to a typical hybrid beamforming and a codebook according to a pattern are enhanced to provide additional beam patterns and codebooks according to these patterns.

Next, we propose efficient beam and user scheduling algorithms in time-varying mmWave wireless networks with combined beams and non-combined beams operating environments through virtual beamforming as well as with multi-beam transmission environments. First, in the virtual beamforming structure providing combined and non-combined beams, appropriate scheduling pairs of beam and user are selected to maximise the sum of data rates of all users at each scheduling time. In order to maximise the sum of data rates for all users, the optimal combined beams and non-combined beams are selected based on the beam combining expected transmission rate and the non-combined expected transmission rate. A precoding is constructed through beam combining to users expecting a high sum of data rates, by excluding for inappropriate beams where a small data rate is expected in order to avoid the high computational complexity due to the comparisons of all combined and non-combined beams. Second, we propose a user beam on/off scheduling and user scheduling algorithm considering inter-beam interference in a multi-beam transmission environment where serving beams are quantised in the beam-space spatially and adjacent beams overlap partially with each other. The user beam on/off scheduling avoids the interference between adjacent active beams. In other words, it is avoided that the simultaneous selection of dominant interference beams with values similar to the magnitude of the serving signal. The beam on/off scheduling algorithm excepting the dominant interference beams substantially reduces the computational complexity of the algorithm by constructing a special graph that considers inter-beam interference.

Finally, via simulations based on real millimeter wave environment, we verify that the proposed technique can achieve a high data rate without much loss of fairness. Simulation results also show that the proposed algorithm is close to the optimal performance with much higher computational complexity.