Decentralized Computing Reliability via Instance Validation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing decentralized computing systems face challenges in ensuring the reliability and integrity of processing units, which can lead to incorrect responses to function requests, especially in time-critical applications like vehicle control systems.
Innovation Solution
A method is implemented where computing instances are validated and executed only in processing units with a predetermined minimum level of reliability, and results are compared and weighted based on confidence values to determine a reliable response, with redundant algorithms and test call-ups used to identify and mitigate faulty units.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If computing instances are executed in decentralized processing units, then computing resources are made available and flexibility is improved, but reliability and integrity of results cannot be ensured
Solution Approach 1:
The system implements a feedback mechanism where results from multiple computing instances are collected, compared, and evaluated. A provisioning unit receives results, determines whether they meet predefined criteria, and provides feedback by selecting valid results or requesting re-execution. This closed-loop feedback ensures reliability in decentralized computing by validating outputs against expected standards.
Solution Approach 2:
The system performs preliminary actions by establishing trust relationships and validity criteria before executing computing instances. The provisioning unit pre-configures validation rules, trust levels, and result criteria that computing instances must meet. This preliminary setup ensures that only reliable results are accepted, preventing unreliable decentralized processing from compromising system integrity.
2Reliability
If multiple computing instances are used to determine a response, then redundancy is improved and reliability is enhanced, but device complexity increases
Solution Approach 1:
The system segments the computing task into multiple independent computing instances that can be executed in parallel across different processing units. Each instance handles a portion of the computation independently, and results are collected and evaluated by a provisioning unit. This segmentation enables redundancy without requiring a single complex system, distributing complexity across simpler, independent components.
3Reliability
If computing instances are validated and executed only in processing units with minimum reliability level, then integrity of results is ensured, but the number of available computing resources is reduced
Solution Approach 1:
The system dynamically changes the reliability parameter threshold for executing computing instances based on the specific task and context. Instead of using a fixed minimum reliability level, the provisioning unit can adjust acceptance criteria, trust levels, and validation requirements according to the criticality of the computation. This parameter flexibility allows more processing units to be utilized while maintaining appropriate integrity standards for each specific computing task.
Data Source
AI summary
A method for providing a function via a network of processing units using multiple computing instances. Each computing instance implements the function and is configured, when it is executed, to determine at least one result in response to a call-up of the function. The method includes validating the computing instances, each being checked as to whether the computing instance corresponds to a respective predetermined state; determining a respective level of reliability for each of the processing units; starting execution of successfully validated computing instances in the processing units, which have a level of reliability that is equal/greater than a predetermined minimum level of reliability; receiving a function request from a user; calling up the function in at least a portion of the executed computing instances corresponding to the function request, in order to determine multiple results; determining a response based on the results; and sending the response to the user.