In frequency-division duplexing systems, the downlink channel stateinformation (CSI) acquisition scheme leads to high training and feedbackoverheads . In this paper, we propose an uplink-aided channelacquisition framework using deep learning to reduce these overheads . Unlike existing works that focus only on channel estimation or feedback modules, this is the first study that considers the entiredownlink CSI acquisition process, including downlink pilot design, channelestimation, and feedback . Our results reveal that, with the help of uplink, directly feedingback the pilot signals can save approximately 20% of feedback bits, which provides a guideline for future research . The former framework estimates andfeeds back the channel at the user equipment subsequently. The latter one directly feeds back the received pilot signals to the basestation. The user equipment directly feeds Back the received Pilot signals to The basestation . The results reveal the pilot signal can save about 20%

Author(s) : Jiajia Guo, Chao-Kai Wen, Shi Jin

Links : PDF - Abstract

Code :

Keywords : pilot - channel - signals - downlink - acquisition -

Leave a Reply

Your email address will not be published. Required fields are marked *