What is Edge Computing in IoT and why is it required?

Edge Computing : Making analysis of data nearest to the spot where it was generated to take quicker action.

In the following diagram, we see that the total time for an IoT solution developed using cloud architecture , the response includes:

t1 = device response time

t2 = time to send to Cloud

t3 = time for analysis, processing, storing in the Cloud

t4 = time to send request/message from Cloud

In the diagram above, we can also consider the delay (added latency) of each part of the network:

latency = t1 + t2 + t3 + t4

This is added to every sensor and actuator event in the system.

Compounding this, for all the sensors actuators system will account to more latency.

Non-ip devices needs to be connected to gateway.These Gateways can have the capability to do data processing, and respond to events, closer to the device. This processing and ‘smarts’ off the Cloud is at the heart of Edge and Fog computing. With a Gateway, t1 remains the same, but t2, t3 and t4 are replaced by much faster interactions for time-sensitive responses; then data is propagated through the system only when required.

Here are some characteristics of Fog nodes (and Gateways):

• Storage: To give transient storage;
• Computing facility: to process data before it is sent to the Cloud, and make quick decisions;
• Network connectivity: to connect with other IoT devices, other Fog nodes, and the Cloud.

The distributed architecture of the Fog is highly suited to the following situations:

• When very short response times are needed;
• Where there are a large number of devices;
• When devices are physically distributed across a large area;
• Where devices are deployed in harsh conditions.

Think of a practical scenario , where car braking needs to take real time quick decision to avoid collision. Since the decision needs to be taken quick - the architecture used for automobile braking must be FOG/EDGE and not CLOUD.

To learn more about FOG/EDGE computing visit https://www.youtube.com/watch?v=EPAfduDwJA0&t=301s