Early-Exiting DNN MIMO Detector Design with Fast Signal Estimation Process
- Authors
- Hua, D.-T.; Lee, D.; Tran, A.T.; Lakew, D.S.; Do, Q.T.; Cho, S.
- Issue Date
- 2023
- Publisher
- Institute of Electrical and Electronics Engineers Inc.
- Keywords
- deep convolutional neural network; deep neural network; early-exiting model; Multiple input multiple output; signal detector
- Citation
- 5th International Conference on Artificial Intelligence in Information and Communication, ICAIIC 2023, pp 505 - 507
- Pages
- 3
- Journal Title
- 5th International Conference on Artificial Intelligence in Information and Communication, ICAIIC 2023
- Start Page
- 505
- End Page
- 507
- URI
- https://scholarworks.bwise.kr/cau/handle/2019.sw.cau/67583
- DOI
- 10.1109/ICAIIC57133.2023.10067114
- ISSN
- 2831-6991
- Abstract
- Multi-input multi-output (MIMO) receiver design has been continuously investigated in order to satisfy the massive requirements of six-generation (6G) wireless networks regarding the accuracy-centric signal estimation under the dynamic channel state information. In addtion, the architectures MIMO detector is a pioneering work for applying artificial intelligence techniques. Eventhough the DL-based MIMO detector can be able to provide accurate signal estimation, huge execution time and energy consumption of immensely large deep neural network (DNN) models in size has not been studied. To this end, we introduce a concept of early-exiting DNN MIMO detector and fast signal estimation, which outputting the estimated transmitted signal without a full forward and back propagation computation of the conventional DNN model. The early-exiting model and the fast signal estimation concept is believed to be suitable for scenarios of practical time sensitive and resouce limited applications. © 2023 IEEE.
- Files in This Item
- There are no files associated with this item.
- Appears in
Collections - College of Software > School of Computer Science and Engineering > 1. Journal Articles
![qrcode](https://api.qrserver.com/v1/create-qr-code/?size=55x55&data=https://scholarworks.bwise.kr/cau/handle/2019.sw.cau/67583)
Items in ScholarWorks are protected by copyright, with all rights reserved, unless otherwise indicated.