FPGA TCP Reassembly Middleware for High-Throughput Low-Latency Messages

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing FPGA systems face challenges in efficiently processing high-frequency, low-latency TCP segments from multiple clients, particularly in high-frequency trading applications, due to bottlenecks in memory utilization and latency, which can result in significant financial losses.

Innovation Solution

Implementing a re-assembly middleware within the FPGA architecture that operates as a modular, plug-and-play component between the TCP Offload Engine and the application layer, utilizing memory optimization techniques such as horizontal and vertical slicing, and a message writer and reader to process TCP segments into application layer messages at pre-defined frequencies, thereby improving performance.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If traditional FPGA systems process TCP segments directly, then the system structure is simpler, but memory utilization and latency performance deteriorate

Engineering Contradiction:
Improvememory utilizationVSAvoidsystem structure
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent introduces a re-assembly middleware as an intermediary component between the TCP Offload Engine and the application layer. This middleware layer handles the complex tasks of TCP segment re-assembly, memory management, and data buffering, thereby improving memory utilization and reducing latency without requiring the entire system to be overly complex. The middleware acts as a mediator that optimizes resource usage while maintaining manageable system architecture.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Productivity

If processing speed is increased to handle millions of orders per second, then throughput improves, but latency increases

Engineering Contradiction:
ImprovethroughputVSAvoidlatency
Core Design Contradiction:
ProductivityVSLoss of time

Solution Approach 1:

The re-assembly middleware performs preliminary actions by pre-assembling TCP segments into complete application layer messages before they reach the application layer. It maintains buffers and pre-processed data structures that prepare data in advance, allowing the application layer to receive ready-to-process messages. This preliminary re-assembly operation reduces the processing time and latency when high-volume orders arrive, enabling the system to maintain both high throughput and low latency performance.

Inventive Principle:
Principle #10Preliminary action

3Ease of operation

If a modular re-assembly middleware is implemented, then ease of operation and maintenance improves, but device complexity increases

Engineering Contradiction:
Improveease of maintenanceVSAvoidmiddleware structure
Core Design Contradiction:
Ease of operationVSDevice complexity

Solution Approach 1:

The patent implements the re-assembly middleware as a modular, segmented structure with independent components for TCP segment handling, message re-assembly, buffering, and data transmission. Each module can be independently configured, maintained, and optimized. This segmentation approach improves ease of operation and maintenance by allowing targeted updates without affecting the entire system, while the modular design itself becomes manageable through clear interface definitions and encapsulation.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentEP4033715B1Re-assembly middleware in FPGA for processing TCP segments into application layer messages
Publication Date: 2025.07.16 TATA CONSULTANCY SERVICES LTD
  • EP4033715B1 patent drawingFigure 1~2
  • EP4033715B1 patent drawingFigure 3
  • EP4033715B1 patent drawingFigure 4

AI summary

A method and system of a re-assembly middleware in FPGA for processing TCP segments into application layer messages is disclosed. In recent years, the communication speed in digital systems has increased drastically and thus has brought in a growing need to ensure a good/high performance from the FPGA services. The disclosure proposes a re-assembly middleware in the FPGA for processing TCP segments into application layer messages at a pre-defined frequency for a good/high performance. The pre-defined frequency is a high frequency performance feature of the re-assembly middleware, wherein the FPGA's implementation frequency is at atleast 300 MHz based on a memory optimization technique. The memory optimization technique includes several strategies such registering an output and slicing memories.