EV Compute Scheduling for Unreliable Node Availability
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Modern vehicles with advanced computing resources are underutilized as node agents in distributed computing systems due to their unpredictable availability and unreliable internet connectivity, limiting their effectiveness in performing computing tasks.
Innovation Solution
A centralized management system that utilizes machine-learning models to predict the compute mode and network availability of electric vehicles, enabling them to perform computing tasks during periods of low usage, such as charging, and ensuring secure data processing by maintaining user-specific data onboard, thereby increasing resource utilization and reducing energy consumption.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If electric vehicles are utilized as node agents in distributed computing systems, then computing resources are underutilized due to unpredictable availability and unreliable internet connectivity
Solution Approach 1:
The system performs preliminary actions by predicting vehicle availability and network connectivity before assigning computing tasks. The availability predictor determines future time windows when vehicles will be available, and the work package generator creates tasks in advance that can be executed during these predicted windows, ensuring tasks are assigned only when vehicles are reliably available.
Solution Approach 2:
The system dynamically adapts to changing vehicle availability conditions by continuously updating predictions of compute mode and network connectivity. The centralized management system adjusts task assignments in real-time based on predicted availability windows, transforming the static, unreliable vehicle resources into dynamic, reliably utilized computing nodes.
2Use of energy by moving object
If computing tasks are assigned to electric vehicles during charging periods, then energy consumption is reduced, but task completion reliability depends on accurate availability prediction
Solution Approach 1:
The system implements feedback mechanisms where the availability predictor continuously monitors vehicle state and updates predictions based on actual charging patterns and usage behavior. This feedback loop improves prediction accuracy over time, enabling reliable task assignment during charging periods while optimizing energy utilization.
Solution Approach 2:
The system performs preliminary prediction of vehicle availability during charging periods before assigning tasks. By predicting future availability windows in advance, the system can confidently assign computing tasks that will execute during low-energy-consumption charging periods without compromising task completion reliability.
3Loss of information
If user-specific data is maintained onboard vehicles, then user privacy is protected, but data processing capabilities are limited by vehicle computing resources
Solution Approach 1:
The system segments data processing functions by maintaining user-specific data onboard vehicles for privacy protection while offloading complex processing tasks to the centralized management system during predicted availability windows. This segmentation allows privacy-preserving local data retention combined with powerful remote processing capabilities.
Data Source
AI summary
Methods, computing systems, and technology for scheduling computing tasks to be performed by an unreliable device are presented. For example, a task management system may receive, from a first electric vehicle, vehicle data comprising computing capabilities of the first electric vehicle and an availability calendar of the first electric vehicle. Additionally, the system can generate, based on the computing capabilities of the first electric vehicle, a work package for the first electric vehicle. The work package can include a computing task for the first electric vehicle to complete. Moreover, the system can determine, based on the availability calendar, one or more transmission parameters for communicating the work package to the first electric vehicle. Subsequently, based on the transmission parameters, the system can transmit a request for the first electric vehicle to perform the computing task of the work package.