TRFI-Net: Modified TRFI-Based Channel Estimation for Internet of Vehicles

In the evolving landscape of the Internet of Vehicles, ensuring dependable Vehicle-to-Everything communication is imperative for Intelligent Transportation Systems. The prevailing IEEE 802.11p protocol presents formidable obstacles to accurate channel estimation owing to its standardized pilot patterns. Conventional solutions, such as data pilot-aided methods, are encumbered by error propagation issues stemming from pilot limitations, particularly pronounced at higher speeds. To address these challenges, this paper introduces TRFI-Net, a novel residual network-based channel estimation method with timedomain and frequency-domain interpolation, demonstrating up to a 3 dB gain in high SNR conditions. By harnessing advanced deep learning methodologies and ResNet architecture, TRFI-Net enhances IoV communication resilience. Benchmarking against six channel models and assessing BER and NMSE in the “Roadside unit to vehicle Urban Canyon” scenario highlights TRFI-Net’s efficacy, marking a significant stride in dynamic vehicular channel estimation research.