Research On Hybrid Beamforming Algorithm In 5G Massive MIMO System

With the rapid development of 5G wireless mobile communication,people have put forward higher requirements for mobile communication system capacity due to the emergence of various wireless data services and multimedia applications.Massive MIMO,as one of the most promising key technologies in 5G wireless communication systems,can significantly improve system performance and capacity.However,in the massive MIMO system,due to the large number of transmitting and receiving antennas,only using digital beamforming technology in the system will lead to high system implementation cost and energy consumption,which will hinder the application of the massive MIMO beamforming technology.In order to overcome this problem,scholars have proposed hybrid beamforming schemes,which is a combination of low-dimensional digital beamforming and high-dimensional analog beamforming.According to different antenna array structures,it can be divided into a fully-connected structure and a partially-connected connection structure.In this paper,we mainly study the massive MIMO hybrid beamforming algorithms based on these two structures.For massive multiuser MIMO systems in the fully-connected structure,this paper studies a low-complexity hybrid block diagonalization algorithm based on equal-gain transmission.The algorithm separates the digital domain beamforming from the analog domain beamforming.In the analog domain,using equal gain transmission obtains the antenna array gain which is only changing the phase of each element in the analog precoder and combiner.In the digital domain equivalent baseband channel,a traditional digital block diagonalization algorithm is introduced to eliminate inter-user interference and obtain the maximum spectrum efficiency.In addition,according to the computational complexity of the block diagonal algorithm,a low-complexity algorithm based on matrix pseudo-inverse,QR decomposition andLDL~H decomposition operations is proposed.The algorithm is examined in both the Rayleigh fading channels and millimeter wave channels.The results show that the proposed algorithm can achieve a spectral efficiency performance that is close to that of digital beamforming system,and it also reduces computational complexity of block diagonalization algorithm.For a massive single-user MIMO system under the partially-connected structure,this paper studies an alternating-optimization hybrid beamforming algorithm based on matrix decomposition,by decomposing high-dimensional unconstrained digital precoders and combiners into hybrid precoders and combiners.Due to the constant modulus constraints of the RF precoder and combiner,the matrix decomposition problem nonconve is caused.Based on an alternate optimization method,the nonconve matrix decomposition problem is decoupled into a series of convex subproblems.The algorithm is examined in both the Rayleigh fading channels and millimeter wave channels.Through simulation,the convergence of the algorithm is examined.At the same time,the performance of the algorithm can also approach the performance of the high-dimension digital beamforming system,and is much higher than that of the analog beamforming system.