CG-DFI

Configured Grant - Downlink Feedback Information

Radio Access Network
Introduced in Rel-16
CG-DFI is a feedback mechanism in 5G NR for uplink Configured Grant (CG) transmissions. It enables the gNB to provide acknowledgment (ACK/NACK) feedback to the UE for its uplink data sent without dynamic scheduling grants. This improves reliability and efficiency for low-latency, grant-free uplink traffic, crucial for URLLC and industrial IoT applications.

Description

Configured Grant - Downlink Feedback Information (CG-DFI) is a physical layer signaling mechanism defined in the 5G New Radio (NR) specifications. It operates within the context of uplink Configured Grant (CG) Type 1 and Type 2 transmissions, where a User Equipment (UE) is pre-configured with periodic resources for uplink data transmission without requiring a dynamic scheduling grant (DCI) for each transmission instance. While this grant-free approach minimizes latency and control signaling overhead, it traditionally lacked a direct, low-latency feedback channel for the gNB to acknowledge successful reception. CG-DFI addresses this by defining a dedicated Downlink Control Information (DCI) format, specifically DCI format 0_2, which the gNB can use to convey Hybrid Automatic Repeat Request (HARQ) feedback for multiple CG transmissions from one or more UEs.

The mechanism works by associating each configured grant resource with a specific HARQ process. When the UE transmits data on its pre-configured PUSCH resources, it monitors a designated search space for DCI format 0_2, scrambled with a specific Radio Network Temporary Identifier (RNTI), such as the Configured Scheduling RNTI (CS-RNTI). The DCI payload contains a feedback bitmap. Each bit in this bitmap corresponds to a specific CG transmission opportunity or a combination of CG configuration and HARQ process. A set bit (e.g., '1') indicates a positive acknowledgment (ACK) for that transmission, while a cleared bit (e.g., '0') indicates a negative acknowledgment (NACK), prompting the UE to initiate a retransmission. The mapping between bitmap positions and the UE's CG transmissions is configured via higher-layer RRC signaling, ensuring the UE can correctly interpret the feedback.

Architecturally, CG-DFI is a function of the gNB's Medium Access Control (MAC) and Physical (PHY) layers. The gNB's MAC layer, upon decoding a PUSCH transmission received on a CG resource, determines the ACK/NACK status. It then instructs the PHY layer to generate and transmit the corresponding DCI format 0_2 in a downlink control channel (PDCCH). On the UE side, the PHY layer decodes the DCI, and the MAC layer processes the HARQ feedback, updating the state of the relevant HARQ process. This closed-loop feedback is critical for enhancing the reliability of grant-free uplink, which is inherently less reliable than scheduled access due to potential collisions and lack of link adaptation per transmission. By providing timely HARQ feedback, CG-DFI enables quick retransmissions, improving packet delivery success rates for latency-sensitive services without resorting to higher-layer retransmission protocols that introduce greater delay.

Purpose & Motivation

CG-DFI was introduced to solve a fundamental limitation of the initial 5G NR Configured Grant uplink scheme. Configured Grant was designed to support Ultra-Reliable Low-Latency Communications (URLLC) and massive Machine-Type Communications (mMTC) by allowing UEs to transmit immediately on pre-configured resources, eliminating the scheduling request and grant exchange latency. However, in its original form, the gNB had no efficient, low-latency method to inform the UE whether a transmission was successfully received. The UE would only discover a failure through higher-layer (e.g., RLC) retransmission timers, which are too slow for URLLC's stringent latency budgets (often 1 ms or less). This gap made true high-reliability communication challenging for grant-free uplink.

The creation of CG-DFI was motivated by the need to close this feedback loop at the physical/MAC layer with minimal delay. It addresses the problem by providing a dedicated downlink control channel that carries HARQ acknowledgments specifically for CG transmissions. This allows for rapid retransmissions at the HARQ level, significantly improving the reliability of the link while maintaining the low-latency advantage of grant-free access. It solves the problem of silent failures where a UE would repeatedly transmit new data while previous packets were lost, wasting resources and degrading reliability. By integrating HARQ feedback, CG-DFI makes Configured Grant a viable solution for mission-critical applications like industrial automation, wireless control, and vehicular communications, where both ultra-low latency and high reliability are non-negotiable requirements.

Key Features

  • Provides HARQ ACK/NACK feedback for uplink Configured Grant transmissions via DCI format 0_2
  • Supports feedback for multiple CG transmission occasions and/or multiple UEs within a single DCI message
  • Uses a configurable bitmap for feedback mapping, defined through RRC signaling
  • Enables low-latency retransmissions at the MAC layer, crucial for URLLC service guarantees
  • Scrambled with identifiers like CS-RNTI for targeted feedback to specific UEs or groups
  • Works with both Configured Grant Type 1 (fully RRC configured) and Type 2 (activated via DCI)

Evolution Across Releases

Rel-16 Initial

Introduced CG-DFI as a new capability for 5G NR. Defined the fundamental mechanism using DCI format 0_2 to carry HARQ-ACK feedback for configured grant PUSCH transmissions. Established the framework for bitmap-based feedback mapping via RRC configuration to support reliable, low-latency uplink for URLLC services.

Defining Specifications

SpecificationTitle
TS 38.212 3GPP TR 38.212