A shift from the cloud to the edge might signal a real autonomous revolution in IoT connectivity. While previously, we witnessed how cloud computing allowed for centralization and collaboration — edge devices are all about abilities to work offline, autonomously, without sending data to the cloud for processing and storage. Here is the future of IoT devices and what it means for connectivity.
Does edge connectivity mean we’ll outgrow Cloud-based connectivity and that we are heading towards the era where edge computing takes central place? Good question.
When we say IoT, what do we mean?
When the Internet of Things term was first introduced around 20 years ago, it alluded to the Internet, which was a big thing back then.
The concept of miniature sensors sending and receiving the data from the cloud over WiFi was huge and breathtaking. When talking about the Internet of Things today, we mean a remotely controllable ecosystem of devices connected to the cloud and to each other with some kind of connectivity.
Most importantly, these devices must be able to perform some actions.
In terms of smart homes, we talk about smart speakers/voice assistants like Alexa or Google Echo, that can issue commands to switch on the lights, tune the conditioner or order a pizza at the nearest Domino or Pizza Hut.
The connected-concept can be rigged up to smart systems controlling commercial real estate across a variety of scenarios. When talking about the Industry 5.0 factories and other industrial installations like wind farms, the IoT means an ecosystem of devices capable of communicating with each other and able to perform some actions based on the commands received.
However, as the technology evolves, the meaning of the terms like IoT and connectivity broadens, and we must take into account this updated image of what connectivity is today — and what it will become in the future.
Why IoT is not enough anymore
The concept of IoT as an independent development entity centered at the gathering, sending, and receiving data has overstayed its welcome. In short, the IoT, in its original meaning, is long dead.
Such systems must provide much more business value to be feasible nowadays. They must enable the users to analyze the data gathered and perform meaningful actions based on the results of this analysis.
The focus of the IoT and connectivity has shifted from the brilliance of myriads of sensors to the value of data they gather. The data, not the sensors, is king. There are surely more sophisticated sensors to come, but their main value is the data they can gather — and the actions we can perform based on this data.
Of course, we only need a smart kettle to be simply switched on when we are close to home so that we can get a cup of tea or coffee faster.
But an autonomous car must be able to react to the changes in the road situation around it, and a smart factory must be able to adjust complex working scenarios should something go awry.
Therefore, the IoT alone as a concept of Digitally Connected Assets, or DCAs, is not viable. It cannot exist in a vacuum, as such systems must be able to process the data quickly and make use of it either through analytics or through issuing some commands.
Performing the task in the cloud means too large latency — so we need something faster. “Faster” is where the edge computing concept comes into play.
Edge computing — the next stage of the IoT evolution
The edge computing term refers to the concept of local computational nodes that form the hearts of the sensor networks in some locations. These sensor networks can be a server node on a factory or in an agricultural complex, an aforementioned Google or Amazon smart home system.
The system can also be the smart utility control system for commercial real estate like malls or office buildings.
In short, edge computing provides a Local Area Network connection for sensors, enabling lightning-fast data transmission. It is also connected to the cloud to enable centralized data gathering and analysis, storage of historical data, and training of AI/ML models on this data.
But most importantly, edge computing nodes provide sufficient computing capacity to host Artificial Intelligence / Machine Learning algorithms locally, which allows these models to issue the needed commands based on the data received from the sensors.
Let’s imagine the fully-automated Industry 5.0 factory equipped by various sensors (movement, temperature, humidity, etc.), a fleet of robots, and multiple actuators.
The robots perform the production operations while the sensors monitor the situation — and one sensor signals the drastic overheating in one of the conveyor belt engines.
The local edge computing node receives the signal, and the AI/ML algorithm running it enacts one of the response scenarios. The scenario can shut down the engine, apply the coolant if possible, disconnect the engine from the conveyor belt (if there are backup engines – start them).
To minimize the production disruption — or reroute the flow of production to other conveyors. All of the functions are done within milliseconds, preventing fire and saving the manufacturer millions in potential damage.
To make operations possible, the edge computing nodes must have three key abilities:
- To control the processes in the physical world. Edge computing nodes must be able to gather the data, process it, and enact some response actions.
- To work offline. Deep underground mines or sea installations far from the shore can have issues communicating to the cloud, so their systems must be able to operate autonomously.
- Zero-second response time. With automated production or utility operations, a delay in several seconds can results in huge financial losses, so the response scenarios must be enacted and executed immediately.
The future of IoT: cyber-physical, contextual and autonomous objects
As we can see, the meaning and the value of the IoT have shifted from the ecosystem of interconnected devices for gathering data to the ecosystem of devices able to gather the data, process it, and act based on this data. Therefore, we can define three main categories of existing and future IoT devices:
The sensors that collect physical signals and transform them into digital data. Think of smart wearables that track our vitals, digital printers, many machine-to-machine and telematic equipment, various smart home systems like thermostats, etc.
All the consumer devices that can perform only a single function like switching the light on/off or rolling the blinds up/down also belong to this group.
Simple cyber-physical DCAs just provide the data or execute single commands, but more complex systems allow understanding the context in which these sensors and actuators operate and make better decisions.
As an example, let’s imagine an agricultural complex, where DCAs control the irrigation systems or the location and operations of a fleet of automated machines.
By supplementing this with an edge computing node, the farmer can consolidate this data to a single dashboard and augment it with weather forecasts and other crucial information, which will help get much more value of the data and control all the systems effortlessly.
- Autonomous objects: the highest level of the “gather-process-react” chain, these systems combine the sensor networks, edge computing nodes, and the AI/ML algorithms to form autonomous objects that take the responsibility from humans to machines. An example is the factory incident we mentioned earlier.
Summing up: call it as you wish — connectivity will not die
We must operate in the real world and use the tools available to us. Basic gateway devices provide ample capacities for data gathering, storing, and processing within an edge computing node.
These nodes enable the ML model in it to take action. Nevertheless, they cannot provide sufficient computing resources for training a model like this, as it requires processing mounds of historical data over hundreds of computational cycles, which can be done only in cloud data centers.
Connectivity is still crucial for connecting edge computing nodes to the cloud, gathering statistical data, training new AI algorithms, and updating the existing ones. It is an integrated ecosystem, where every component plays its role.
What are we going to call this new and exciting ecosystem?
IoT 2.0? Cyber-physical edge computing-enabled objects? The terms itself matters little, while we understand what stands behind it. These objects will have the ability to connect the physical and digital worlds, gather the data with sensors, process it in context with other input, and take actions based on this analysis.
While this ecosystem works and is feasible, it matters little what we call it.
Most importantly, connectivity is still crucial for connecting edge computing nodes to the cloud, so connectivity will never die.
What do you think of the future of IoT and the importance of connectivity? Please let us know in the comments below.