Vehicle Computing Parameter Checks for Random Error Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current computing devices in vehicles face challenges in handling random errors during algorithm execution, particularly with high-performance processors, where existing solutions like Hardware Lockstep are expensive and not commercially viable for complex algorithms.
Innovation Solution
A device comprising a first calculation unit with high integrity and a second calculation unit with greater computing power but lower integrity, which executes calculation functions redundantly and compares results to detect errors, ensuring operational safety without duplicating hardware resources.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If Hardware Lockstep is used to detect random errors, then reliability is improved, but device complexity and cost increase due to full hardware duplication
Solution Approach 1:
The system divides error detection into two segments: a first computing unit that executes instructions and a second computing unit that verifies results through redundant execution. This segmentation allows error detection without duplicating the entire hardware system, reducing complexity while maintaining reliability.
Solution Approach 2:
Instead of duplicating hardware resources, the patent uses software-based copying by executing the same algorithm multiple times on the same hardware with different initial states or parameters. This virtual copying achieves error detection without physical duplication.
2Productivity
If high-performance processors are used to increase computing power, then productivity is improved, but susceptibility to random errors increases
Solution Approach 1:
The system implements feedback mechanisms where the second computing unit monitors and verifies the outputs of the first computing unit. When errors are detected through comparison of multiple executions, the system can request re-execution or correct the error, maintaining reliability while using high-performance processors.
Solution Approach 2:
The patent performs preliminary error detection by executing algorithms multiple times before final output is generated. This preliminary action of redundant execution with different initial states prevents error propagation and ensures correctness before results are used.
3Reliability
If redundant execution is performed to detect errors, then reliability is improved, but execution time increases
Solution Approach 1:
The system performs partial redundancy by executing algorithms multiple times only when necessary for verification, rather than always performing full redundant execution. The second computing unit selectively verifies results based on error probability assessments, reducing time loss while maintaining reliability.
Solution Approach 2:
The patent changes execution parameters such as initial states, input data variations, or computational paths across multiple executions. This parameter diversification allows error detection through comparison while potentially reducing the number of repetitions needed, thereby minimizing time loss.
Data Source
Figure 1~2
Figure 3~4
Figure 5
AI summary
Embodiments of the invention provide a computing device (10). The device may advantageously comprise a first unit (100) and a second unit (200). In response to receipt of a request corresponding to a computing function, the first unit (100) may be configured to: - determine an execution context; - trigger a first execution of the function on the second unit (200), this delivering a first comparison parameter, a first temporal execution parameter being associated with the first comparison parameter; - trigger a second execution of the function on the second unit (200), this delivering a second comparison parameter, a second temporal execution parameter being associated with the second comparison parameter; - compare the first and second comparison parameters, a temporal comparison parameter being associated with the result of the comparison; - determine a computing status.