Soft Buffer Partitioning for Limited Memory Bandwidth

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

LTE systems face challenges in handling soft buffer memory access bandwidth limitations, particularly when supporting multiple carriers and high data rates, which restricts the efficiency of HARQ operations and system performance.

Innovation Solution

A method is introduced to prioritize memory access for soft buffer handling in User Equipment (UE) by determining and optimizing memory access bandwidth, allowing for selective reading and writing of bits based on redundancy versions, thereby maximizing overlapping bits between previous and retransmitted data, even under limited memory access conditions.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If soft buffer memory access bandwidth is increased to handle retransmissions efficiently, then system performance improves, but device complexity and cost increase

Engineering Contradiction:
Improveretransmission handling efficiencyVSAvoidmemory access bandwidth requirement
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent segments the soft buffer into multiple partitions, each dedicated to storing soft bits from different redundancy versions. This segmentation allows the receiver to selectively access only the relevant partition during retransmission combining, rather than accessing the entire soft buffer. Consequently, the memory access bandwidth requirement is reduced while maintaining efficient retransmission handling, as the system only needs to access the specific partition containing the initial transmission's soft bits combined with the retransmission data.

Inventive Principle:
Principle #1Segmentation

2Reliability

If all soft bits are read and written to soft buffer during retransmission, then combining accuracy improves, but memory access bandwidth consumption increases beyond limits

Engineering Contradiction:
Improvecombining accuracyVSAvoidmemory access bandwidth
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent extracts and processes only the essential soft bits needed for combining during retransmission. Instead of reading and writing all soft bits to the soft buffer, the system identifies and processes only the relevant soft bits from the initial transmission that need to be combined with the retransmission data. This extraction approach maintains combining accuracy by focusing on the critical data elements while significantly reducing the memory access bandwidth consumption to within acceptable limits.

Inventive Principle:
Principle #2Taking out (Extraction)

3Ease of manufacture

If soft buffer is implemented in off-chip memory to reduce complexity, then ease of manufacture improves, but memory access bandwidth is limited

Engineering Contradiction:
Improvesoft buffer implementationVSAvoidmemory access bandwidth
Core Design Contradiction:
Ease of manufactureVSDevice complexity

Solution Approach 1:

The patent segments the soft buffer into multiple partitions stored in off-chip memory, with each partition dedicated to a specific redundancy version. This segmentation strategy enables the receiver to access only the relevant partition during retransmission combining operations, thereby reducing the effective memory access bandwidth requirement. The off-chip implementation maintains ease of manufacture while the selective partition access approach ensures that the limited bandwidth of off-chip memory is sufficient for efficient HARQ operations.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentEP3522424B1Soft buffer handling with limited memory access bandwidth
Publication Date: 2022.07.06 TELEFONAKTIEBOLAGET LM ERICSSON (PUBL)
  • EP3522424B1 patent drawingFigure 1~2
  • EP3522424B1 patent drawingFigure 3
  • EP3522424B1 patent drawingFigure 4~5

AI summary

Systems and methods for soft buffer handling with limited memory access bandwidth are provided herein. In some embodiments, a method of operation of a User Equipment (UE) of a cellular communications network includes receiving a retransmission that would require a total number of memory accesses to process that is greater than a memory access bandwidth of a soft buffer of the UE. The method also includes prioritizing which bits should be read from or written to the soft buffer of the UE when processing the retransmission based on the memory access bandwidth, any previously received redundancy versions, and/or a currently received redundancy version. According to some embodiments, this achieves higher system performance than Chase Combining alone while keeping the receiver implementation limitation as low as alternative solutions.