Limited Buffer Rate Matching

Description

Limited Buffer Rate Matching (LBRM) is a sophisticated physical layer process defined in 3GPP New Radio (NR) specifications for the downlink shared channel (PDSCH) and uplink shared channel (PUSCH). It is an integral part of the channel coding chain, sitting between the Low-Density Parity-Check (LDPC) encoder and the modulator. The core function of LBRM is to perform rate matching—the process of creating a codeword of the exact length required for the allocated time-frequency resources—under the specific constraint that the receiving User Equipment (UE) has a finite soft channel bit buffer for storing log-likelihood ratios (LLRs) during Hybrid Automatic Repeat Request (HARQ) combining.

How LBRM works is fundamentally tied to HARQ operation. When a transport block is transmitted, the LDPC encoder produces a mother codeword. A conventional circular buffer rate matching scheme would select bits from this mother codeword. However, if the UE's soft buffer size is smaller than the full mother codeword size (which is common for lower-complexity UEs), the gNB must limit the set of bits it can potentially transmit across all HARQ retransmissions to those that the UE can store. LBRM defines this limited set. The gNB's rate matching process selects bits only from this predefined, buffer-size-limited window within the circular buffer. This ensures that any transmitted bit, whether in the initial transmission or a retransmission, can be stored and combined by the UE.

Key components involve the configuration of the limited buffer size (Nref), which is derived from the UE category and its reported capability. The gNB uses this knowledge to determine the LBRM rate matching output. This technique plays a critical role in the network by enabling efficient support for UEs with vastly different processing capabilities and cost points, from high-end smartphones to simple IoT devices, all within the same NR carrier. It ensures that the advanced channel coding gains of LDPC are maintained even for UEs with limited memory, preventing performance degradation that would occur if the gNB transmitted bits the UE could not process.

Purpose & Motivation

LBRM was introduced in 3GPP Release 14 for the feasibility study of NR and solidified in later releases to solve a specific problem arising from the new 5G NR channel coding scheme. NR adopted LDPC codes for data channels, which are excellent for performance but generate very large mother codewords, especially for the large transport block sizes supported by 5G. The soft buffer memory required to store the LLRs for the full mother codeword across multiple HARQ processes would be prohibitively large and costly, particularly for low-complexity, low-power IoT devices.

The purpose of LBRM is to decouple the UE implementation cost (dictated by soft buffer size) from the peak data rate performance of the system. Without LBRM, the network would have to assume all UEs can buffer the full codeword, or else severely restrict the coding and modulation schemes for capable UEs. LBRM allows the gNB to know exactly which subset of coded bits a specific UE can store. This enables the network to schedule high spectral efficiency transmissions to advanced UEs while still supporting low-cost devices on the same network, all without compromising the HARQ combining gain for any device. It addresses the economic and practical necessity of a diverse device ecosystem in 5G.

Key Features

• Enables efficient HARQ operation for UEs with limited soft buffer memory
• Integrates with NR LDPC channel coding and circular buffer rate matching
• UE capability dependent, with the limited buffer size Nref signaled by the UE
• Ensures the gNB only transmits bits that can be stored by the target UE's buffer
• Critical for supporting a wide range of UE complexity and cost points
• Maintains link performance and coding gain despite memory constraints

Evolution Across Releases

Defining Specifications