An Integrated New Deep Learning Framework for Reliable CSI Acquisition in Connected and Autonomous Vehicles
- 24 October 2023
- journal article
- research article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Network
- Vol. 37 (4) , 216-222
Abstract
According to society of automotive engineers (SAE)'s standard J3016, there are six levels of driving automation, ranging from no driving automation (level 0) to full driving automation (level 5). In order to make autonomous vehicles (AVs) fully automated when driving, they must be capable of comprehending the driving environment, which requires the assistance of 5G networks and multi-access edge computing (MEC). Through the use of advanced precoding techniques, 5G networks are able to provide low latency and high transmission rates. The real-time precoding is correlated with the quality of estimated instantaneous channel state information (CSI). However, two obstacles prevent acquiring accurate CSI. One is CSI feedback overhead, and the other is channel aging. The aforementioned challenge can be successfully solved through deep learning, but there are two aspects that have been overlooked. The conventional way to solve the challenge separately with two deep learning models lead to a sub-optimal results. Besides, deep learning models are static and unable to adapt to new data. In this article, we propose an integrated framework to address the performance degradation caused by acquiring CSI via two deep learning models. Further, we introduce continual learning to train the deep learning model to enhance its adaptability to new data. The simulation results demonstrate the integrated framework can address the challenge of CSI acquisition with a satisfactory performance. In addition, continual learning can help the deep learning model to adapt to changing data in a dynamic environment.Keywords
This publication has 10 references indexed in Scilit:
- Accurate Channel Prediction Based on Transformer: Making Mobility NegligibleIEEE Journal on Selected Areas in Communications, 2022
- Convolutional Autoencoder-Based Phase Shift Feedback Compression for Intelligent Reflecting Surface-Assisted Wireless SystemsIEEE Communications Letters, 2021
- Addressing the Curse of Mobility in Massive MIMO With Prony-Based Angular-Delay Domain Channel PredictionsIEEE Journal on Selected Areas in Communications, 2020
- An Adaptive and Parameter-Free Recurrent Neural Structure for Wireless Channel PredictionIEEE Transactions on Communications, 2019
- Deep Learning for Massive MIMO CSI FeedbackIEEE Wireless Communications Letters, 2018
- An overview of multi-task learningNational Science Review, 2017
- An Overview of Signal Processing Techniques for Millimeter Wave MIMO SystemsIEEE Journal of Selected Topics in Signal Processing, 2016
- Downlink Training Techniques for FDD Massive MIMO Systems: Open-Loop and Closed-Loop Training With MemoryIEEE Journal of Selected Topics in Signal Processing, 2014
- The evolution path of 4G networks: FDD or TDD?IEEE Communications Magazine, 2006
- Fundamentals of Wireless Communication2005