Edge Node Segmentation for Low Latency and Resource Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In edge computing, deploying applications on single MEC nodes leads to resource constraints at base stations, causing high response delays and resource competition with radio access network events, while deployment at data centers results in longer delays due to distance from users, and application migration disrupts services with long interruption times.
Innovation Solution
Deploying an application's input nodes near base stations and computational nodes near area data centers, allowing for separate resource management and minimizing service interruptions during migration by only migrating input nodes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If applications are deployed on single MEC nodes co-located with base stations, then response delay is reduced, but computational resource requirements cannot be met and resource competition with radio access network events increases
Solution Approach 1:
The application is segmented into two types of nodes: input nodes deployed at MEC nodes co-located with base stations for low-latency data exchange, and computational nodes deployed at area data centers for heavy processing. This segmentation allows each node type to be optimized for its specific function, resolving the contradiction between low response delay and sufficient computational resources.
Solution Approach 2:
The deployment architecture transitions from a single-location deployment to a multi-dimensional distributed architecture spanning both edge (base station) and cloud (area data center) dimensions. This dimensional expansion enables the system to simultaneously achieve low latency for data exchange and sufficient computational power for processing.
2Power
If applications are deployed at data centers, then computational resource requirements are met, but response delay increases due to distance from users
Solution Approach 1:
The application is divided into input nodes placed at distributed MEC nodes near users for fast data exchange, and computational nodes placed at area data centers for powerful processing. This segmentation allows the system to achieve both low response delay and sufficient computational resources by separating these functions spatially.
Solution Approach 2:
MEC nodes co-located with base stations serve as intermediary nodes between users and area data centers. These intermediaries handle data exchange and preliminary processing locally, reducing the distance-related latency for users while maintaining connection to the powerful computational resources at area data centers.
3Adaptability or versatility
If applications are migrated to different locations, then resource optimization is achieved, but service interruption time increases
Solution Approach 1:
The application is segmented into input nodes and computational nodes that can be independently migrated. During migration, only the input nodes need to be moved to new MEC nodes, while computational nodes remain at area data centers. This segmentation significantly reduces service interruption time compared to migrating the entire application.
Solution Approach 2:
The migration process extracts and migrates only the necessary input nodes to new locations while leaving the computational nodes at their original area data center locations. This selective extraction minimizes the scope of migration and reduces service disruption.
4Loss of time
If applications are deployed at base stations, then user proximity is maximized, but resource competition with radio access network events increases
Solution Approach 1:
The application workload is segmented and divided between input nodes at base station MEC nodes and computational nodes at area data centers. This segmentation allows the system to maximize user proximity for data exchange while offloading heavy computational tasks to area data centers, reducing resource competition at the base station level.
Solution Approach 2:
The architecture adds a dimensional separation between data exchange functions (at base station edge) and computational functions (at area data center). This dimensional division allows both functions to coexist without direct resource competition, as they operate in different spatial and functional dimensions.
Data Source
AI summary
Methods, systems, and computer program products for deploying an application. One method includes acquiring a computational graph corresponding to an application. The computational graph includes a plurality of nodes corresponding to operations performed by the application, the nodes including at least one input node for exchanging data with at least one terminal device and at least one computational node for processing data. The at least one input node is deployed to a first group of edge nodes co-located with a first base station serving the at least one terminal device, and the at least one computational node is deployed to a second group of edge nodes co-located with an area data center. Illustrative embodiments of the present disclosure enable an input node to be as close to a user as possible, thereby achieving low response delays while meeting the requirements of a computational node for storage and computational resources.