Beam Alignment And Channel Estimation For Millimeter Wave Communication Systems

Accompanying with the diversified demands of mobile uses as well as the drastically amplified structural complexity of future wireless communication systems,the 5th generation mobile communication system has introduced millimeter wave(mm Wave)technology,which provides an ultra-wide frequency band,to empower the ever growing data transmission needs.Meanwhile,large-scale antenna techniques,which are mainly based on the multi-input multi-output structure,is capable of forming a narrow beam with high gain such that the intrinsic huge path loss and attenuation caused by millimeter wave transmissions can be compensated.However,a narrow beam pattern naturally calls for a precise beam alignment(BA)method to obtain the angle information of the LOS path of the channel so as to establish a strong communication link with a large signal-to-noise ratio(SNR).Besides,the channel information is also useful when designing the hybrid precoder/combiner weight which serves as a critical step to maximize the array gain.All in all,channel estimation(CE)is indispensible.However,in the mm Wave background,the high dimensionality of the channel coefficients preclude the direct application of many traditional methods.As such,it is urgent to study low-overhead and high-performance BA and CE methods.Regarding the beam alignment problem,a computationally efficient BA scheme that is based solely on the magnitude(power)information is proposed.The proposed method is free from the influence of the phase noise.Firstly,by a properly adaptation of the subcarrier domain,a set of multi-peak beam codebooks satisfying the prescribed time overhead is designed.Afterwards,the distribution of the power of the received signal is derived according to the central limit law.At last,using the received signal power,a maximum likelihood method is deduced to estimate the direction of the best channel path.To make a solid and comprehensive comparison with existing methods,we have also cast the BA into a traditional phase retrieval problem,which can be solved by some existing phase retrieval algorithms.In the simulation part,we have compared the proposed algorithm with traditional BA methods as well as phase retrieval method.It turns out that the proposed algorithm achieves a higher BA accuracy and a larger beamforming gain than the existing methods.Beyond the above efforts,we have additionally proposed a high resolution BA scheme to realized gridless channel estimation.Simulation results show that this scheme can overcome the energy leakage problem caused by grid mismatch.With regard to the channel estimation problem,we first conduct a thorough discussion on several existing modelings of carrier frequency offset(CFO).Our aim is to study the joint estimation of the CFO as well as the time-domain channel coefficients.The core idea is to decompose such a complex problem into two relatively independent subproblems,i.e.,CFO estimation and time-domain channel estimation.For the CFO estimation,a maximum likelihood method is adopted to enable a fast and accurate estimation of the value of the CFO.Once the CFO is known,the channel estimation can be formulated as a block sparse signal recovery problem,which is solved by the pattern coupled sparse bayesian learning(PC-SBL)algorithm.Simulation results show that the proposed method is capable of accurately estimating the CFO as well as the channel coefficients.Regarding the computational overhead,the proposed methods is advantageous over comparing methods such as the G-SBL algorithm.