What is the Difference Between Edge and Fog Computing?
First, let make an assumption
That a sensor produces data.
The data shall be processed to get a piece of information.
This piece of information shall be presented to an End-user.
Next, let’s focus on processing the data. Which device shall process the data?
To answer this simple question we have to check where is the best place to do it.
Even though our devices have broadband connection to a cloud services we have to do the Left Shift Processing
Long story short, a rule – process the data in a place of origin.
So data shall be processed on an Edge device. That’s why we need the Edge Computing
What if we have to process data from several devices in a place of origin?
In this case, devices have to discover each other and make computing in a mesh network.
This self-organizing sub-net is considered a Fog.
Fog Computing can leverage mesh computing capabilities and can process “heavy” data.
Why do I see circles on this picture?
Let’s imaging that you have thousands of geo-located Edge devices which can form Fog computing networks.
And these Fog networks can cooperate with another Fog network closer to the Cloud.
So the Cloud can get aggregated data pre-processed by several levels of Fog Networks.
This concept is the core concept of IoT – processing data goes from Cloud to Edge via Fog levels/layers.
This leads to offline/disconnected processing where the Cloud is used for hosting web applications for End-User.
Cloud could manage, via “intends” unconfirmed downstream broadcast commands, Edge devices to update e.g. device configuration.
In such cases where a device doesn’t have a permanent connection to the cloud or the bandwidth is low, we usually see Fog and Edge computing. Next, if we don’t have to have a permanent connection between devices and/or cloud then we can use different radio modulation to send date over long distance. The Long Range (LoRa and Thread) networks are the future of IoT