5G • 5G-Advanced (Rel-18)
AI-Based RACH Detection
Neural Preamble Detection & Collision Resolution in RACH
Deep neural detector at the gNodeB that isolates overlapping random access preambles in massive IoT scenarios, resolving collisions that fail standard matched filters.
Technical Explanation
When thousands of IoT devices wake simultaneously, multiple UEs transmit identical or overlapping PRACH preambles, causing severe collisions that traditional matched filter correlation cannot untangle. In Rel-18, deep learning-based PRACH detectors learn non-linear interference signatures, successfully resolving collided preambles and saving massive signaling retries.
Key Functions
- Resolves multiple simultaneous overlapping PRACH preamble transmissions
- Expands effective random access capacity by up to 300% without adding spectrum
- Reduces connection establishment latency for massive IoT and smart grid bursts
- Operates robustly in low SNR environments where matched filters fail
Specifications
3GPP TR 38.843, TS 38.211 (Rel-18)
Interfaces
Uu (PRACH)
Related 5G Concepts
Want to memorize 5G concepts like this one?
Study it with SuperMemo SM-2 spaced repetition flashcards.