eduzhai > Applied Sciences > Engineering >

A Deep Learning Framework for Hybrid Beamforming Without Instantaneous CSI Feedback

  • Save

... pages left unread,continue reading

Document pages: 13 pages

Abstract: Hybrid beamformer design plays very crucial role in the next generationmillimeter-wave (mm-Wave) massive MIMO (multiple-input multiple-output)systems. Previous works assume the perfect channel state information (CSI)which results heavy feedback overhead. To lower complexity, channel statisticscan be utilized such that only infrequent update of the channel information isneeded. To reduce the complexity and provide robustness, in this work, wepropose a deep learning (DL) framework to deal with both hybrid beamforming andchannel estimation. For this purpose, we introduce three deep convolutionalneural network (CNN) architectures. We assume that the base station (BS) hasthe channel statistics only and feeds the channel covariance matrix into a CNNto obtain the hybrid precoders. At the receiver, two CNNs are employed. Thefirst one is used for channel estimation purposes and the another is employedto design the hybrid combiners. The proposed DL framework does not require theinstantaneous feedback of the CSI at the BS. We have shown that the proposedapproach has higher spectral efficiency with comparison to the conventionaltechniques. The trained CNN structures do not need to be re-trained due to thechanges in the propagation environment such as the deviations in the number ofreceived paths and the fluctuations in the received path angles up to 4degrees. Also, the proposed DL framework exhibits at least 10 times lowercomputational complexity as compared to the conventional optimization-basedapproaches.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×