Key Technical Points of 5G Radio Group RAN1 in R18

3GPP Release 18 is the first 5G-Advanced release, focusing on AI/ML integration, ultimate performance in XR/Industrial IoT, mobile IAB, enhanced positioning, and spectrum efficiency up to 71GHz. RAN1 further promotes AI/ML enhancements in RAN optimization and artificial intelligence (PHY/AI) through physical layer evolution.

I. Key Features of RAN1 (Physical Layer and AI/Machine Learning Innovations)

1.1 MIMO Evolution: Multi-panel uplink (Level 8), MU-MIMO with up to 24 DMRS ports, multi-TRP TCI framework.

• Operating Principle: Extends Type I/II CSI reporting through a unified TCI framework across multiple TRP panels. The gNB schedules up to 24 DMRS ports for MU-MIMO (12 in Rel-17), enabling each UE to use Level 8 UL links; DCI indicates joint TCI status; UE applies phase/precoding across panels.
• Progress: The lack of unified signaling in Rel-17 multi-TRP resulted in a 20-30% loss of spectral efficiency in dense deployments; level restrictions limited the UL throughput of each UE to layers 4-6, thereby achieving a 40% increase in uplink (UL) capacity for stadiums/music festivals.

1.2 AI/ML Applications to CSI Feedback Compression, Beam Management, and Positioning.

• Working Principle: The neural network uses an offline-trained codebook to compress Type II CSI (32 ports → 8 coefficients). The gNB deploys the model via RRC; the UE reports the compressed feedback. Beam prediction uses the L1-RSRP mode to pre-position beams before handover.
• Project Progress: CSI overhead consumed 15-20% of DL resources; in high-mobility scenarios (e.g., highways), beam management failure rates reached as high as 25%.
• Improvement Results: Channel State Information (CSI) overhead reduced by 50%, handover success rate improved by 30%.

1.3 Enhanced Coverage (Uplink full-power transmission, low-power wake-up signal).

• Operating Principle: The gNB sends a signal to the UE, enabling it to apply full power output across all uplink layers (without tiered power backoff). An independent low-power wake-up receiver (duty cycle controlled, sensitivity -110dBm) receives the wake-up signal (WUS) before the main receive cycle. The WUS carries 1 bit of indication information (monitoring PDCCH or sleep).
• Project Progress: Rel-17 uplink coverage is limited by tiered power backoff (4th order MIMO loss of 3dB); the main receiver consumes 50% of the UE's power during DRX monitoring.
• Improvements: Uplink coverage extended by 3dB; IoT/video streaming applications saved 40% of power.

1.4 ITS Band Sidelink Carrier Aggregation (CA) and Dynamic Spectrum Sharing (DSS) with LTE CRS.

• Operating Principle: Sidelink supports CA across the n47 (5.9GHz ITS) + FR1 bands; supports autonomous resource selection for Type 2c coordination among UEs. Due to a round-trip time (RTT) greater than 500 milliseconds, NTN IoT disables HARQ (only supports open-loop repetition); pre-compensation is implemented for the Doppler effect in DMRS.
• Project Progress: Rel-17 Sidelink only supports single-carrier (50% throughput loss); NTN IoT HARQ timeouts result in 30% packet loss.
• Improvements: V2X formation sidelink throughput is increased by 2x, and NTN IoT reliability reaches 95%.

1.5 Extended Reality (XR)/Multi-sensor Communication (High Reliability, Low Latency Support).

• Operating Principle: New QoS procedure, latency budget less than 1 millisecond, supports multi-sensor packet tagging (video + haptic + audio stream). gNB prioritizes data through a preemption mechanism. UE reports attitude/motion data for predictive scheduling.
• Project Progress: Rel-17 XR support only supports unicast; haptic feedback latency exceeds 20 milliseconds (unusable for remote operation).
• Improvements: End-to-end latency of AR/VR + haptic in industrial remote control is less than 5 milliseconds.

1.6 NTN Functionality Enhancement (Smartphone Uplink Coverage, Disabling HARQ for IoT Devices).

• How it Works: Rel-18 improves the uplink coverage of smartphones in non-terrestrial networks (NTNs) by optimizing physical layer transmission, allowing for higher transmit power and better link budget management to accommodate satellite channels. For IoT devices on NTNs, traditional HARQ feedback is inefficient due to long satellite round-trip times (RTTs), therefore HARQ feedback is disabled, and an open-loop repetition scheme is adopted instead.
• Project Progress: Previously, due to insufficient power control and link margin, the uplink coverage of smartphones on NTNs was limited, resulting in poor connectivity. HARQ feedback caused throughput reduction and latency issues for IoT devices due to satellite latency. Disabling HARQ eliminates feedback latency and improves the reliability of constrained IoT devices. This enables robust global connectivity for IoT and smartphones beyond terrestrial networks.

II. RAN1 Project Applications

• Dense Urban XR (Multi-TRP MIMO technology reduces AR/VR latency to below 1 millisecond);
• Industrial Automation (AI/ML beam prediction reduces handover failure rate by 30%);
• V2X/High Mobility (Sidelink CA improves reliability).

III. RAN1 Project Implementation

• gNB PHY (Base Station Physical Layer): Integrates an AI model for CSI compression (e.g., neural networks predict Type II CSI based on Type I CSI, reducing overhead by 50%). Deploys Multi-TRP TCI via RRC/DCI and uses 2 TAs for uplink timing.
• Terminal Equipment (UE): Supports low-power wake-up receivers (independent of the main RF link) for DRX alignment signaling.