Fault-Tolerant Ethernet Network Hardware Switching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing Ethernet networks in industrial control systems lack fault tolerance, leading to potential network failures when even a single component fails, which is unacceptable for real-time industrial control systems, and existing solutions for increasing fault tolerance either fail to handle multiple faults or require complex software and reconfiguration that slows down fault correction.
Innovation Solution
Each end node in the network is equipped with two network connections having the same address, with hardware switching between them upon detection of a fault, utilizing standard IEEE 802.3 mechanisms and beacon packets for fault detection, and a redundant network topology with multiple paths between nodes to ensure rapid and seamless failure correction without the need for address changes or complex software reconfiguration.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If ring topology with ring manager device is used to provide fault tolerance, then network reliability is improved, but reconfiguration time is 100 microseconds to 500 milliseconds which is too slow for real-time control
Solution Approach 1:
The patent pre-configures alternate paths through multiple network infrastructures before failures occur. Each node maintains pre-established routing tables and path information for both primary and alternate networks, allowing immediate switching upon failure detection without requiring real-time path calculation or complex reconfiguration protocols.
Solution Approach 2:
The patent divides the network into separate redundant infrastructures (Network A and Network B) with independent paths. By segmenting the network into distinct redundant channels, a failure in one infrastructure does not affect the other, and switching between segments can occur at the network interface level rather than requiring entire network reconfiguration.
2Difficulty of detecting and measuring
If middleware software is used to control network connections and detect failures, then fault detection capability is improved, but reconfiguration speed is limited to less than one second due to software processing overhead
Solution Approach 1:
The patent replaces software-based failure detection and response mechanisms with hardware-based solutions. Network interface cards include dedicated hardware circuits that continuously monitor network health, detect failures, and trigger switching operations in hardware, eliminating the delays associated with software processing, context switching, and operating system interrupts.
Solution Approach 2:
The network interface card autonomously performs failure detection and switching operations without requiring software intervention. The hardware monitoring circuits continuously check network status and automatically switch to alternate paths when failures are detected, enabling sub-millisecond response times without software overhead.
3Reliability
If multiple redundant network infrastructures with middleware are used, then ability to tolerate multiple faults is improved, but device complexity increases due to multiple ports and middleware configuration
Solution Approach 1:
The patent combines multiple network interface cards into a single logical network interface from the perspective of the host system. The host sees one unified interface while the hardware automatically manages multiple physical paths and performs failover, eliminating the need for complex software configuration and management of multiple separate interfaces.
Solution Approach 2:
The network interface card acts as an intermediary between the host system and multiple network infrastructures. It abstracts the complexity of multiple redundant paths from the host, providing a simple single-interface view while handling all fault detection, path selection, and switching operations internally in hardware.
Data Source
AI summary
Fault-tolerant Ethernet is provided through the use of special interfaces providing duplicate ports that may be alternatively enabled with the same network address. A switching between the ports, corrects for single faults in a two-way redundant system without time-consuming reconfiguration of other end devices or the need for complex middleware in the end devices.