Edge and Fog Computing in IoMT

This flashcard set covers the fundamental definitions, architectural differences, and performance characteristics of Edge and Fog computing based on the IoMT Part $$01$$ lecture.

Cloud Computing

A model for storing data online, effectively making a user's computer accessible in the "cloud".

Edge Computing

A distributed information technology (IT) architecture where client data is processed at the periphery of the network, as close to the originating source as possible.

Edge Device

A device, such as the $\text{ESP32}$, that performs computing "on-the-spot" at the point of data collection.

Fog Computing

An extension of cloud computing that serves as a unique layer between the edge and the cloud layer.

Fog Nodes

Intermediate components that act as filters, controllers, or intermediates to manage data flow to the cloud or store it locally.

Edge Computing Scalability

Scales by adding more edge devices, although each individual device has limited compute and storage capabilities.

Fog Computing Scalability

Scales as an aggregation layer by adding fog nodes to serve multiple edge devices and balance loads.

Edge Endpoints

Typically consists of billions of devices including sensors, cameras, and wearables.

Fog Endpoints

Consists of fewer nodes than the edge layer, which can be geographically distributed.

Edge Proximity to Cloud

The layer farthest from the cloud, running directly on or very near the device.

Fog Proximity to Cloud

Closer to the cloud or data center than the edge, often located at gateways, PoPs, or MEC sites.

Backhaul Bandwidth (Edge)

Lowest requirement because most data is processed locally, with only summaries or alerts sent upstream.

Backhaul Bandwidth (Fog)

Medium requirement as it filters and aggregates data before forwarding selected data to the cloud.

Edge Operational Cost

Can be high due to the need for device-by-device management of many distributed nodes.

Fog Operational Cost

Potentially lower per device due to centralized aggregation and management at the fog nodes.

Edge Privacy

Highest level of privacy as data can remain local on the device, resulting in a smaller attack surface per node.

Fog Privacy

Medium level of privacy where data is aggregated at fog nodes to reduce exposure compared to cloud-only processing.

Edge Typical Location

Located on the IoT device, gateway-on-device, or an on-prem controller.

Fog Typical Location

Located at local gateways, ISP PoPs, base stations, or micro-data centers.

Edge Power/Compute

Lowest per node with constrained CPU and RAM, optimized for quick decisions.

Fog Power/Compute

Higher per node than edge, utilizing stronger CPUs or GPUs to process data from many devices.

Edge Primary Benefit

Provides ultra-low latency decisions made right where the data is created.

Fog Primary Benefit

Provides filtering, aggregation, and regional analytics to reduce latency and cloud load.