Edge Resource Demand Estimation Using M-PCM-OFFD
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Edge systems face challenges in managing machine-to-machine workloads with limited resources and random workload demands, leading to inefficiencies and resource consumption spikes, which current methods like regression analysis and overprovisioning fail to address accurately.
Innovation Solution
Implementing a Multivariate Probabilistic Collocation Method-Orthogonal Fractional Factorial Design (M-PCM-OFFD) framework for resource uncertainty estimation (RUE) to provide accurate and scalable demand load predictions, enabling efficient resource allocation and scheduling in edge environments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional resource allocation methods are used in edge systems, then device complexity is reduced, but resource allocation accuracy deteriorates due to limited resources and random workload demands
Solution Approach 1:
The patent transforms resource demand estimation from direct measurement to statistical inference by changing the parameters from raw resource usage data to probabilistic distributions. The M-PCM-OFFD framework estimates resource demand parameters (mean, variance, skewness, kurtosis) rather than tracking every resource transaction, achieving accurate estimation with reduced data processing complexity
Solution Approach 2:
The patent introduces statistical moments (mean, variance, skewness, kurtosis) as intermediary representations between raw resource usage data and resource allocation decisions. These statistical parameters serve as compressed summaries that capture essential workload characteristics without requiring complete raw data, reducing system complexity while maintaining estimation accuracy
2Reliability
If real-time resource monitoring is implemented, then resource allocation reliability is improved, but loss of time increases due to data collection and processing delays
Solution Approach 1:
The patent performs preliminary statistical analysis by continuously maintaining estimates of resource demand parameters (mean, variance, skewness, kurtosis) as workloads arrive. This preliminary characterization allows the system to make rapid allocation decisions without real-time data collection delays, as the statistical framework is already prepared to evaluate new workload requests against historical patterns
Solution Approach 2:
The patent implements feedback through continuous refinement of resource demand parameter estimates. As new workload data arrives, the statistical moments are updated to reflect current conditions, allowing the system to adapt to changing workload patterns while maintaining rapid decision-making capability. The feedback loop operates at the parameter level rather than raw data level, reducing processing time
3Measurement precision
If comprehensive resource data collection is performed, then measurement precision is improved, but loss of information decreases due to limited access to real-time data in distributed environments
Solution Approach 1:
The patent extracts only the essential statistical characteristics (first four moments) from complete resource usage data. Rather than collecting and processing all raw resource data, the system extracts sufficient statistical information to characterize workload patterns, achieving accurate resource demand estimation with minimal data collection requirements. This extraction approach works effectively in distributed environments where complete data access is limited
Data Source
AI summary
Managing the resource demand load for edge systems is significantly more complex than for other systems, such as cloud environments. A time period in which an application or task is operating based on initial demand resource load values that are provided by a customer may be inaccurate, which may expose sub-standard execution. Embodiments herein seek to significantly mitigate the potential of sub-standard execution. Embodiments collect a repository of resource demand load usage data over a time period that can be used to accurately determine the statistical moments of uncertain resource demand load. In one or more embodiments, a repository of hypervector and/or hyperspace representations may be generated and used to help with resource demand load estimation.