RDMA Sensor Streaming Over Ethernet for Direct Memory Transfer

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing technologies face challenges in efficiently processing high-throughput, low-latency real-time data streams from remote sources, particularly due to limitations in scalability and resource requirements of Peripheral Component Interconnect Express (PCIe) cards and User Datagram Protocol (UDP) over networks, and the need for both ends to support Remote Direct Memory Access (RDMA) protocol capability.

Innovation Solution

A host device initiates a direct-to-memory RDMA transmission with a remote device using existing Ethernet or InfiniBand connections, transitioning into a Ready-to-Receive state and sending instructions to the remote device to enter a Ready-to-Send state, allowing streaming data to be sent directly to allocated memory, bypassing the operating system and leveraging RDMA stack capabilities on the host device.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If PCIe cards with FPGAs are used for real-time data stream processing, then processing capability is improved, but scalability is limited

Engineering Contradiction:
Improvereal-time data processing capabilityVSAvoidscalability
Core Design Contradiction:
ProductivityVSAdaptability or versatility

Solution Approach 1:

The patent introduces a host device as an intermediary that initiates RDMA transmissions to remote devices. The host device manages multiple data stream connections and coordinates RDMA operations, allowing scalable processing without requiring each remote device to have complex processing capabilities. This mediator approach enables centralized control while maintaining distributed data collection.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Adaptability or versatility

If UDP over network is used for data transmission, then scalability is improved, but throughput and latency performance deteriorates

Engineering Contradiction:
ImprovescalabilityVSAvoidthroughput and latency performance
Core Design Contradiction:
Adaptability or versatilityVSProductivity

Solution Approach 1:

The patent changes the transmission parameter from standard UDP to RDMA protocol, which fundamentally alters how data is transmitted at the network level. RDMA operates at a lower protocol level with direct memory access capabilities, changing the performance characteristics to achieve both high throughput and low latency while maintaining scalability through network infrastructure.

Inventive Principle:
Principle #35Parameter changes

3Productivity

If RDMA protocol is implemented on both ends of connection, then high-throughput and low-latency performance is improved, but device complexity and resource requirements increase

Engineering Contradiction:
Improvethroughput and latency performanceVSAvoidRDMA stack logic resources
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent segments the RDMA implementation responsibilities, with the host device performing full RDMA stack operations and the remote devices performing simplified data transmission. This segmentation allows remote devices to avoid implementing complex RDMA stacks while still benefiting from RDMA performance, dividing the complexity burden between central and distributed components.

Inventive Principle:
Principle #1Segmentation

4Ease of operation

If operating system processing is used for data streams, then ease of operation is improved, but processing speed and efficiency deteriorate

Engineering Contradiction:
Improveoperating system managementVSAvoidprocessing speed
Core Design Contradiction:
Ease of operationVSProductivity

Solution Approach 1:

The patent extracts the data transmission path from the operating system by implementing direct-to-memory RDMA operations. Data streams bypass the OS kernel and application layers, writing directly to allocated memory buffers. This extraction eliminates OS processing overhead while maintaining ease of operation through host-initiated connection management and memory allocation.

Inventive Principle:
Principle #2Taking out (Extraction)

Data Source

PatentUS20260037476A1Network transmission for data streams
Publication Date: 2026.02.05 NVIDIA CORP
  • US20260037476A1 patent drawing
  • US20260037476A1 patent drawing
  • US20260037476A1 patent drawing

AI summary

Approaches presented herein provide for the real-time backhauling of data streams direct to memory. In at least one embodiment, a host device can initiate a Remote Direct Memory Access (RDMA) transmission with a remote device for sensor data streams. Initialization data for the RDMA transmission can be sent from the host to the remote device, such as over an existing physical Ethernet or InfiniBand (IB) connection. The host device can enter a Ready-to-Receive (RTR) state and send instructions to the remote device to enter a Ready-to-Send (RTS) state. The RDMA transmission can be initialized between the host device and the remote device to transfer streaming data. Using a remote data manager, the streaming data can be sent over the existing Ethernet connection as a RDMA transmission and written directly to allocated memory. The system may be scaled to include higher throughput sensors, additional sensors, and multiple remote transmission devices.