Pooling Computing Resources via User Presence Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing computing systems lack efficient methods to optimize resource usage across multiple devices, leading to suboptimal performance and energy consumption, as they do not effectively predict user presence and device proximity to offload computing tasks.
Innovation Solution
A method that identifies inference information about user presence and device capabilities to optimize resource usage by pooling computing resources across devices, allowing tasks to be delayed or offloaded to more capable devices when they are in proximity, thereby saving energy and improving performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Power
If computing tasks are performed on mobile devices with limited resources, then device portability is maintained, but processing performance and available computing power are insufficient
Solution Approach 1:
The patent merges computing resources across multiple devices by establishing a resource pool that includes mobile devices, fixed computing devices, and cloud resources. The system combines CPU cycles, memory, storage, and other computing resources from different devices to provide unified computing power that exceeds the capabilities of any single device, thereby resolving the contradiction between maintaining device portability and achieving high computing power.
Solution Approach 2:
The patent implements a universal resource management system that can allocate computing resources from any device in the network to any task. The resource pool serves multiple functions: it provides computing power to mobile devices when needed, offloads tasks to fixed devices or cloud when more power is available, and dynamically adapts to changing device availability and task requirements, thereby achieving multi-functionality that resolves the computing power limitation.
2Productivity
If computing tasks are continuously performed on mobile devices, then task completion is maintained, but battery energy consumption increases
Solution Approach 1:
The patent applies preliminary action by predicting user presence and device availability before tasks need to be executed. The system uses machine learning models to forecast when users will be present at different locations and which devices will be available, allowing tasks to be scheduled in advance for optimal execution. This enables the system to perform calculations before battery power becomes critical or before more powerful devices become available, thereby maintaining productivity while reducing overall energy consumption.
Solution Approach 2:
The patent introduces a cloud-based resource management system as an intermediary between mobile devices and computing tasks. This intermediary coordinates task offloading by determining when and where tasks should be executed based on user presence predictions and device availability. The intermediary manages the complex decisions of task scheduling, resource allocation, and device selection, enabling mobile devices to conserve battery power while ensuring task completion through coordinated resource sharing.
3Productivity
If device resources are allocated independently without coordination, then device autonomy is maintained, but overall system efficiency and resource utilization are suboptimal
Solution Approach 1:
The patent implements feedback mechanisms where devices continuously report their resource availability, usage status, and operational state to the centralized management system. The system uses this feedback to dynamically update resource pool information, predict user presence patterns, and optimize task scheduling decisions. This feedback loop enables the system to learn from historical data and improve resource allocation efficiency over time, resolving the contradiction between maintaining device autonomy and achieving optimal resource utilization.
4Measurement precision
If user presence is predicted using machine learning models, then task scheduling accuracy is improved, but computational overhead and model training requirements increase
Solution Approach 1:
The patent extracts the machine learning model training and complex computational tasks from individual mobile devices and relocates them to centralized cloud servers. The mobile devices only need to collect and transmit basic data (location, usage patterns, device state), while the cloud-based system performs the heavy-duty machine learning analysis to predict user presence and optimize resource allocation. This extraction significantly reduces the computational overhead on mobile devices while maintaining high prediction accuracy, thereby resolving the contradiction between measurement precision and computational complexity.
Data Source
AI summary
Pooling computing resources based on inferences about a plurality of hardware devices. The method includes identifying inference information about the plurality of devices. The method further includes based on the inference information optimizing resource usage of the plurality of hardware devices.