Deep-Learning-Based Radio Channel Prediction for Vehicle-to-Vehicle Communications

Details

• Creators:
  Molisch, Andreas F ; Chu, Lei
• Corporate Creators:
  University of Southern California ; METRANS Transportation Center (Calif.) ; Pacific Southwest Region 9 UTC, University of Southern California
• Corporate Contributors:
  United States. Department of Transportation. University Transportation Centers (UTC) Program ; United States. Department of Transportation. Office of the Assistant Secretary for Research and Technology ; California Department of Transportation
• Subject/TRT Terms:
  • Vehicle to vehicle communications
  • Machine learning
  • Spatial analysis
  • Radio transmitters
  • Networks
  • Neural networks
  • channel modeling
  • spatial-temporal prediction
  • meta learning
• Publication/ Report Number:
  22-10 TO 065
• DOI:
  https://doi.org/10.25554/8ed4-fw73
• Resource Type:
  Tech Report
• Geographical Coverage:
  California ; United States
• Edition:
  Final report (8/17/2023-6/15/2024)
• Corporate Publisher:
  Pacific Southwest Region 9 UTC, University of Southern California
• Abstract:
  Reliable and efficient V2V communications are essential for driver-assistance systems to reduce accidents and improve energy efficiency through efficient convoying. A key challenge is that resource allocation for communication must rely on the current propagation channel state, but vehicles only have past measurements. Therefore, effective channel prediction methods are crucial. Traditional methods using simplified models and classical tracking/extrapolation perform poorly in real-world environments. This motivates the use of Machine Learning (ML), which handles complex data well but faces challenges like limited channel measurement data and mismatched neural network structures for V2V channels. Our project addresses these issues by leveraging extensive past measurements available in our lab and developing new neural network structures tailored for multi-dimensional channel prediction in various vehicular scenarios, such as campus and city roads. Finally, we validate the effectiveness of these predictions in actual resource allocation.
• Format:
  PDF
• Funding:
  69A3551747109 ; Caltrans 65A0674TO065
• Download URL:
  https://rosap.ntl.bts.gov/view/dot/77399/dot_77399_DS1.pdf
• File Type:
  [PDF - 5.86 MB]
• Collection(s):
  Artificial Intelligence Deployment Center of Excellence (AI COE) ; University Transportation Centers
• Main Document Checksum:
  urn:sha-512:107966632c30043cc39a21ae6cbd627dbd7d25b0e6bb956e3c55265ad55d533fa9007042239cce88323cb8b1cc79110b6331b8ef4b6ed0779078a290e5bdc3fe