eduzhai > Applied Sciences > Engineering >

Deep Learning for Distributed Channel Feedback and Multiuser Precoding in FDD Massive MIMO

  • king
  • (0) Download
  • 20210506
  • Save

... pages left unread,continue reading

Document pages: 15 pages

Abstract: This paper shows that deep neural network (DNN) can be used for efficient anddistributed channel estimation, quantization, feedback, and downlink multiuserprecoding for a frequency-division duplex massive multiple-inputmultiple-output system in which a base station (BS) serves multiple mobileusers, but with rate-limited feedback from the users to the BS. A keyobservation is that the multiuser channel estimation and feedback problem canbe thought of as a distributed source coding problem. In contrast to thetraditional approach where the channel state information (CSI) is estimated andquantized at each user independently, this paper shows that a joint design ofpilots and a new DNN architecture, which maps the received pilots directly intofeedback bits at the user side then maps the feedback bits from all the usersdirectly into the precoding matrix at the BS, can significantly improve theoverall performance. This paper further proposes robust design strategies withrespect to channel parameters and also a generalizable DNN architecture forvarying number of users and number of feedback bits. Numerical results showthat the DNN-based approach with short pilot sequences and very limitedfeedback overhead can already approach the performance of conventional linearprecoding schemes with full CSI.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×