Automotive Network Gateway Architecture for Low-Latency Data Routing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing software-based network gatewaying solutions in automotive communication networks fail to guarantee real-time performance, particularly in terms of latency, jitter, bandwidth, and throughput, especially for next-generation autonomous-driving vehicles with heterogeneous in-vehicle network technologies.
Innovation Solution
A hardware-driven network device with a central processing unit and multiple co-processors (frame normalization, ingress queuing, filtering and policing, intermediate queuing, gatewaying, and traffic shaping) that processes data in parallel and pipelined modes, ensuring efficient data switching, routing, and gatewaying across different subnetworks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a host device is assigned to each terminal device for processing data packets, then data transmission efficiency is improved, but network complexity increases significantly
Solution Approach 1:
The patent merges the functions of multiple host devices into a single shared host device. Instead of each terminal device having its own dedicated host device, the shared host device receives data packets from multiple terminal devices, processes them using a pool of worker threads, and sends them to the network. This consolidation maintains processing capability while significantly reducing network complexity.
Solution Approach 2:
The shared host device is designed to handle data packets from multiple different terminal devices simultaneously. The worker threads in the pool can be dynamically assigned to process packets from any terminal device, making the host device universal rather than dedicated to a single terminal. This multi-functionality allows one host device to serve the role previously requiring multiple dedicated hosts.
2Reliability
If data packets are buffered in the host device before transmission, then transmission reliability is improved, but memory usage increases
Solution Approach 1:
The patent implements dynamic buffering where the buffer size and allocation are adjusted based on current network conditions and traffic patterns. The host device buffers data packets temporarily when needed for reliability (such as when network conditions are poor or acknowledgments are pending), but releases buffer space promptly once data is transmitted successfully. This dynamic approach ensures transmission reliability while minimizing unnecessary memory consumption during periods of good network conditions.
3Speed
If multiple worker threads are used to process data packets in parallel, then processing speed is improved, but coordination overhead increases
Solution Approach 1:
The patent introduces a packet queue as an intermediary structure between the worker threads and the network interface. Data packets are placed in the queue by receiving threads, and worker threads independently retrieve and process packets from the same queue without direct communication or coordination with each other. This intermediary queue eliminates the need for complex inter-thread coordination mechanisms while maintaining parallel processing efficiency, as threads only need to access the shared queue structure with simple lock-free or minimal-lock operations.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The invention relates to a network device (101) for switching, routing and/or gateway ing data among different subnetworks of a communication network (100), wherein the network device (101) comprises: a central processing unit (103); one or more data ingress ports (105a-c) and one or more data egress ports (107a-d) configured to exchange data with a further network device (131) of the communication network (100); and a plurality of co-processors (109), wherein the plurality of co-processors (109) comprises one or more frame normalization co-processors (111a-c), one or more ingress queuing co-processors (113a-c), one or more filtering and policing co-processors (115a-c), one or more intermediate queuing co-processors (117a-c), at least one gateway! ng co-processor (119), one or more egress queuing co-processors (121a-d) and at least one traffic shaping co-processor (123). In an embodiment, the central processing unit (103) is adapted to configure and control the one or more data ingress ports (105a-c), the one or more data egress ports (107a-d) and the plurality of co-processors (109) to implement one or more data processing paths in parallel and/or in a pipeline between the one or more ingress ports (105a-c) and the one or more egress ports (107a-d). The invention also relates to a method (700) of providing the network device (101).