Readme
Purpose
Resource-Cell is a unit that divides the resources of a device, such as computing power, memory, and bandwidth in an edge computing environment.
Resource-Cell structure expresses a performance of the device as only one unit. So, it can standardize a performance of device.
In the edge computing environment, it is difficult to deploy complexed AI models due to resource constraints.
The purpose of Resource-Cell is to easily provide effective AI services
by comparing cells of devices and cells required by AI model in an edge computing environment.
Basic Concept
Devices and AI models can calculate resource requirements and resource supply through simple benchmarks.
This is expressed as the number of Resource-Cells. If demands and supplies are similar, efficient AI services can be deployed.
However, if demands are higher, the model is operated by gathering resources from other nearby devices.
Conversely, if there are sufficient supplies, multiple AI services are operated.
Goal
The goal of this research is to operate AI services by device network composed of trusted devices in the edge network.
A device network is a collection of device's Resource-Cells of in an edge network.
When service tasks are given, cells form computation tissues to provide the service. When the service ends, they are restored and waits for a new service.
Contact
- russell1113@kangwon.ac.kr