Network Energy Consumption Assessment via Computational Work Correlation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Accurately assessing energy consumption in large computer networks is challenging due to the high cost of utility-grade electrical metering equipment, making it expensive to track energy usage for each computer, which is crucial for pricing computational capacity effectively.
Innovation Solution
A system that correlates energy usage with logged information about computational work, allowing for detailed inference of energy consumption while minimizing the number of power meters required, by using a network structure with a logging computer, power control units, and an energy meter to measure aggregate energy consumption, and solving linear equations to estimate baseline and computational increment power.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If utility-grade electrical metering equipment is used for each computer, then measurement precision of energy consumption is improved, but device complexity and cost increase significantly
Solution Approach 1:
The patent merges multiple measurement functions into a single energy meter that measures aggregate power consumption for the entire network. Instead of having separate meters for each computer, the system combines all measurements at the network level and uses computational work logging to allocate energy consumption to individual computers, thereby reducing device complexity while maintaining measurement precision.
Solution Approach 2:
The patent introduces an intermediary approach by using logged computational work information as a mediator to infer individual computer energy consumption. Rather than directly measuring each computer's power consumption with separate meters, the system uses the relationship between computational work performed and energy consumed as an intermediary method to calculate individual consumption from aggregate measurements.
2Measurement precision
If individual meters are installed for each computer, then measurement precision is improved, but cost increases considerably
Solution Approach 1:
The patent merges the function of multiple individual meters into a single energy meter that measures total network power consumption. By combining all measurement functions at the network level and using computational work data to allocate consumption, the system reduces the quantity of meters from N (one per computer) to just one, thereby significantly reducing cost while maintaining the ability to measure precision energy consumption.
Solution Approach 2:
The patent creates a virtual copy of the measurement function through computational accounting. Instead of physically installing multiple meters, the system uses software-based tracking of computational work and applies this information to calculate what each computer's energy consumption would be, effectively copying the measurement function through computation rather than hardware.
3Device complexity
If aggregate energy measurement is used, then device complexity is reduced, but measurement precision for individual computers deteriorates
Solution Approach 1:
The patent uses logged computational work information as an intermediary to bridge the gap between aggregate measurement and individual precision. The system measures total energy consumption with a single meter, then uses the intermediary relationship between computational work performed and energy consumed to accurately allocate and determine individual computer energy consumption, thereby maintaining precision without increasing complexity.
Solution Approach 2:
The patent changes the measurement parameters by tracking computational work output (hashes performed, computational tasks completed) in addition to energy input. By introducing this additional parameter and establishing the relationship between work done and energy consumed, the system can derive precise individual computer energy consumption from aggregate measurements, resolving the precision loss that would normally occur with simplified metering.
Data Source
AI summary
An article of manufacture, a machine, process for using the articles and machines, processes for making the articles and machines, and products produced by the process of making, along with necessary intermediates, directed to assessing the energy consumption of networks, typically computer networks, and/or applications of assessments made thereby. Industrial applicability is representatively directed to energy consumption/conservation and efficiency, such as in networks, along with control and implementation therefrom, as well as in networking, control systems communications and related systems, receiver systems, and components used in assessing and carrying out the same.