Adaptive Edge Processing via ML Resource Allocation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Cloud computing faces challenges with latency and bandwidth constraints, making it less ideal for real-time applications due to increased strain and congestion from numerous connections, while edge computing struggles to integrate effectively with cloud networks.
Innovation Solution
A system and method that adaptively allocate processing operations using machine learning to direct whether to use edge-only, hybrid edge-cloud, or cloud-only resources, optimizing resource allocation based on parameters such as application type, usage, latency, power consumption, urgency, and security to minimize latency and maximize efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If cloud computing is used to store and process applications and data, then users can access applications from any location, but latency and bandwidth constraints increase, reducing real-time application interactivity
Solution Approach 1:
The system segments processing operations into three categories: edge-only processing for time-sensitive operations, hybrid edge-cloud processing for operations requiring both local responsiveness and remote resources, and cloud-only processing for non-time-critical operations. This segmentation allows the system to maintain cloud-based accessibility while reducing latency for real-time interactions by handling them at the edge.
Solution Approach 2:
The patent introduces a spatial dimension to computing by deploying edge computing nodes geographically closer to end users, in addition to the traditional cloud data centers. This creates a multi-layered computing architecture where processing can occur at different physical distances from users, reducing latency while preserving cloud accessibility.
2Adaptability or versatility
If more devices are connected to the cloud, then more users can access applications, but strain and congestion increase, reducing network performance
Solution Approach 1:
The system segments network traffic and processing loads by routing time-sensitive and bandwidth-intensive operations to edge computing nodes, while cloud data centers handle less time-critical operations. This segmentation prevents congestion in the core cloud network even as device connectivity increases.
Solution Approach 2:
Edge computing nodes act as intermediaries between end users and cloud data centers, handling local processing and filtering traffic before it reaches the cloud. This intermediary layer reduces the strain on cloud network infrastructure while supporting increased device connectivity.
3Loss of time
If edge computing is used to reduce latency, then real-time interaction improves, but integration with cloud networks becomes technically difficult
Solution Approach 1:
The patent designs edge computing nodes with multi-functional capabilities that allow them to operate independently for low-latency processing while simultaneously integrating with cloud networks for resource sharing and coordination. This universality simplifies integration by allowing edge nodes to perform multiple roles rather than requiring specialized configurations.
Solution Approach 2:
The system implements feedback mechanisms where edge computing nodes continuously report their status, resource availability, and performance metrics to cloud management systems. This feedback enables dynamic coordination and resource allocation, simplifying integration by allowing the cloud to adapt to edge node conditions in real-time.
Data Source
AI summary
Aspects of the present disclosure provide for adaptively allocating processing operations according to an edge-only processing mode, a hybrid edge-cloud processing mode, and/or cloud-only processing mode, but are not so limited. In one aspect, the disclosure describes adaptively managing processing of applications or application modules by using machine learning in part to direct how the processing operations are to be performed whether using edge-only resources, hybrid edge-cloud resources, and/or cloud-only resources. Other aspects are described in detail herein.