The Design Of Turbo Channel Estimation For Mullimeter Wave Communication System With Massive Antennas

With the abundant frequency spectrum available at millimeter wave(MMW)band,MMW communications can easily achieve multi-gigabit per second wireless transmission.In order to compensate for the severe propagation attenuation at such high frequencies,massive antenna arrays are often employed at both the transmitter and receiver to exploit array gains via beamforming technique.On the other hand,constrained by the cost,size and power consumption of MMW radio frequency(RF)devices,much fewer RF chains than antennas are widely deployed in the MMW systems.As one of enabling beamforming technologies,the channel estimation technology with high resolution and low latency is of great importance for MMW communications with RF chain constraints,which restricts the application of classic super resolution subspace methods such as MUSIC/ESPRIT-based ones.In this study,the efficient channel estimation algorithms are developed for MMW systems with massive antenna arrays and RF chain constrains.The main works of this paper include:ⅰ)by fully exploiting the sparse structure of MMW channel,an efficient Turbo compressive channel estimation method,is developed to achieve both significant reduction of the time slot consumption and great enhancement of the estimation reliability in the low signal-to-noise ratio(SNR)region;ⅱ)an enhanced channel estimation algorithm is developed for the off-grid problem which cause power leakage and then degrades the angle estimation performance severely.Firstly,a new channel model is formulated to catch the off-grid effect in channel quantization in the spatial domain.The new model divides the continuously-distributed angle into a quantization grid part,termed integral on-grid part,and an offset part,termed fractional off-grid part.Accordingly,an iterative algorithm is proposed to renew and upgrade the channel estimation between the integral and fractional part under the turbo principle,which enhances the angle estimation resolution greatly.