5G • 5G-Advanced (Rel-18)
AI/ML Doppler Compensation
Neural Channel Doppler and Delay Spread Tracking
Neural network architecture designed to estimate and compensate severe Doppler shifts in High-Speed Train (HST) and Non-Terrestrial Network (NTN) channels.
Technical Explanation
Vehicular speeds exceeding 350 km/h or LEO satellites moving at 7.5 km/s induce massive time-varying Doppler shifts that distort OFDM subcarrier orthogonality. Rel-18 AI Doppler estimation uses convolutional and recurrent models to predict rapid phase rotations from DMRS symbols, performing active frequency pre-compensation before ICI occurs.
Key Functions
- Estimates continuous Doppler trajectory across fast-fading Rayleigh channels
- Mitigates Inter-Carrier Interference (ICI) in high-speed mobility environments
- Enables stable 5G communications at speeds up to 500 km/h and in LEO satellite orbits
- Reduces required DMRS pilot density, freeing up physical resource blocks for data
Specifications
3GPP TR 38.843, TS 38.211 (Rel-18)
Interfaces
Uu (DMRS)
Related 5G Concepts
Want to memorize 5G concepts like this one?
Study it with SuperMemo SM-2 spaced repetition flashcards.