Probabilistic Workload Allocation for Edge Carbon Footprint
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In Multi-Access Edge Computing environments, existing technologies face challenges in efficiently allocating workloads to minimize carbon footprint due to uncertainty in observability data and heterogeneous server conditions, leading to increased carbon emissions.
Innovation Solution
The method involves using Probabilistic Timed Automata to model workload allocation policies, quantify risk associated with uncertainty metrics, and determine the probability of adherence to carbon-footprint metrics, with a Probabilistic Model Checker to recommend policies that maximize carbon footprint reduction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If workload allocation policies are based on traditional deterministic methods, then the allocation process is simple and fast, but the carbon footprint requirements cannot be satisfied due to uncertainty in observability data
Solution Approach 1:
The patent transforms the workload allocation problem by changing the mathematical framework from deterministic to probabilistic. It uses Probabilistic Timed Automata (PTA) to model allocation policies, incorporating probability distributions for carbon footprint metrics and observability data uncertainty. This parameter change enables the system to satisfy carbon footprint requirements by explicitly modeling and reasoning about uncertainties in the allocation process.
Solution Approach 2:
The patent introduces a Probabilistic Model Checker as an intermediary tool that verifies whether allocation policies satisfy carbon footprint requirements. This model checker acts as a mediator between the PTA-based policy models and the actual workload allocation execution, providing formal verification of policy correctness under uncertain conditions without requiring complex real-time calculations.
2Reliability
If probabilistic models are used to account for uncertainty in observability data, then carbon footprint requirements can be satisfied, but the computational complexity increases
Solution Approach 1:
The patent applies preliminary action by pre-defining Probabilistic Timed Automata models for workload allocation policies and using Probabilistic Model Checkers to verify their correctness before actual deployment. The uncertainty metrics and probability distributions are characterized in advance, allowing the system to make reliable allocation decisions without performing complex probabilistic calculations in real-time, thus reducing policy determination time.
3Object-generated harmful factors
If workload allocation considers heterogeneous server conditions and data uncertainty, then carbon footprint is reduced, but the allocation accuracy decreases due to uncertainty in observability data
Solution Approach 1:
The patent implements beforehand cushioning by incorporating uncertainty margins into the workload allocation decision-making process. The Probabilistic Timed Automata models include probability distributions that account for potential variations in observability data. This cushioning approach allows the system to make robust allocation decisions that satisfy carbon footprint requirements even when observability data has uncertainties, effectively preparing for potential data inaccuracies in advance.
Data Source
AI summary
Provided are a method, system, and computer program product in which a plurality of edge computing nodes are provided in a multi-access edge computing environment. Workload allocation policies are recommended in the multi-access edge computing environment by determining which policy to use to allocate workloads to edge sites to maximize the probability of carbon footprint requirements being satisfied given the uncertainty with observability data.