Redundant Processing Circuit Error Recovery via Lockstep Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Programmable integrated circuits (ICs) face errors due to single event upsets (SEUs) caused by atmospheric radiation, which can alter configuration memory cells and bi-stable circuits, leading to functional failures without efficient recovery methods that allow continued operation without repeating processing steps.
Innovation Solution
A redundant processing system with three or more processing circuits operating in lockstep, where a control circuit detects and recovers from non-fatal errors by disabling faulty circuits, storing and restoring states, and continuing execution from the error point, allowing seamless recovery without resetting the entire system.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If three or more redundant processing circuits operate in lockstep to detect and recover from non-fatal errors, then system reliability is improved, but device complexity increases due to the need for additional circuits and control mechanisms
Solution Approach 1:
The system divides processing circuits into redundant segments (three or more copies) that operate in lockstep. Each circuit segment can be independently disabled when an error is detected, allowing the system to maintain functionality through the remaining segments while isolating the faulty portion.
Solution Approach 2:
The error recovery mechanism applies local quality by selectively disabling only the specific processing circuit that experiences an error, rather than shutting down the entire system. The control circuit identifies and isolates the faulty local segment while maintaining operation of healthy segments.
2Reliability
If the entire system is reset when an error occurs to ensure correctness, then reliability is improved, but productivity decreases due to loss of processing state and need to repeat processing steps
Solution Approach 1:
The system performs preliminary actions by maintaining multiple redundant processing circuits in a ready state before errors occur. These circuits operate in lockstep ahead of time, so when an error is detected in one circuit, the other redundant circuits already contain the correct processing state and can immediately take over without requiring a full system reset.
Solution Approach 2:
The system discards the faulty processing circuit when an error is detected and recovers by continuing execution on the remaining functional redundant circuits. The processing state is preserved in the healthy circuits, allowing the system to recover from errors without repeating processing steps or losing computational progress.
3Reliability
If redundant processing circuits are used to mitigate SEUs, then reliability is improved, but loss of time increases due to the overhead of maintaining and managing multiple circuit states
Solution Approach 1:
The redundant processing circuits operate continuously in lockstep, maintaining identical processing states at all times. This continuous synchronized operation ensures that when an error is detected in one circuit, the correct state is already present in the redundant circuits, eliminating the need for time-consuming state recovery or reprocessing.
Solution Approach 2:
The control circuit continuously monitors the processing circuits for errors and provides immediate feedback when a non-fatal error is detected. This real-time feedback mechanism enables rapid error identification and triggers the recovery process by disabling the faulty circuit and continuing on healthy circuits, minimizing the time lost to error handling.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Enables recovery from errors without repeating previous processing steps, ensuring continuous operation and maintaining system integrity by isolating and resetting faulty components within the redundant processing system.
Implementation Method 1
an SEU may induce errors in configuration memory cells used to configure programmable logic circuits to perform specific functions
Implementation Method 2
When a single heavy ion strikes a silicon substrate, it loses energy through the creation of free electron hole pairs. This results in a dense ionized track in the local region, generating a current pulse that can upset the circuit.
Data Source
AI summary
Methods and circuits are disclosed for error recovery in redundant processing systems. Respective instances of a software program are executed in lockstep on redundant processing circuits. Using a control circuit, in response to detecting a non-fatal error, an interrupt is generated and non-functioning ones of the processing circuits are disabled. The interrupt is serviced using the functional processing circuits operating in lockstep. In servicing the interrupt, a processing state of the processing circuits is stored and a reset of the processing circuits is triggered. Following the reset, the processing circuits are configured to operate in lockstep. The state of the processing circuits is restored to the stored processing state and a return from the interrupt is signaled. In response to the signaled return from interrupt, execution of the software program is resumed on the processing circuits in lockstep at a point at which the non-fatal error was detected.