Dumb sensor vs smart sensor

Question

Everytime when i read about the "Internet of Things" it is suggested that devices are smart but when it comes to an implementation of an IoT ecosystem, it is not so clear to me anymore. So i need some help for explanation of the term "Internet of things" and "Smart devices"

For me, there are two cases

1. Devices are very smart. Example: The classical smart fridge which orders milk if its empty. The fridge is able to perform a order by itself and does not need any help from intermediate logic between the device (fridge) and the webshop.

2. Devices are dumb as much as possible. The fridge does not know if the milk is empty. It just know that sensor A ("milk detector") is status "empty" or "not empty". Or just have a sensor B which counts the pieces of milkbags in the fridge and push this to a queue or a REST Service. This Service is an intermediate layer between other systems like webshops or other devices.

Does "smart device" even mean that itself is able to evaluate its own data and therefore is able to perform actions without any external help?

I can't imagine that a simple light bulb should be able to go to my webshop and order itself automatically ... or is this exactly that for what "Internet of things" is standing for?

For 1. i can see the main drawback is that you have to reprogram the fridge if it should have a different behaviour (order new milk if count of milk < 2 instead of < 1). The upside is, that you don't need any additional layer between webshop and device.

For 2. the main drawback is the additional logic layer between webshop and device. But it is easy to modify the smart behaviour in a central way on a common platform where all "smart devices" are registered.

Answer 1

In the Internet of Things, you have Smart Objects and not-Smart Objects.

The not-Smart Objects are sensor and actuators. Sensors allow you to obtain different measures from the environment: the light fluctuation using a photoresistor, the temperature using a thermistor, to detect flames, sounds, movements, or any other fluctuation in the environment. The actuators can do an action. Examples of mechanic actuators could be motors, servomotors or hydraulic bombs, and examples of actions could be to send a message, control LEDs, turn on lights or control the movement of a robot or any other available robot’s actions.

On the other hand, you have the Smart Objects. These one are composed of not-Smart Objects, maybe only sensor, maybe only actuators, or a combination of both. However, Smart Objects also have the capacity to think, because they have a processor. For instance, a smartphone, a microcontroller like an Arduino, etc.

You can see this in one of the existing definitions: 'A Smart Object, also known as Intelligent Product, is a physical element that can be identified throughout its life and interact with the environment and other objects. Moreover, it can act in an intelligent way and independently under certain conditions. Furthermore, Smart Objects have an embedded operating system and they usually can have actuators, sensors, or both. This allows Smart Objects to communicate with other objects, process environment data, and do events'.

Besides, we can classify the Smart Objects according to the different level of intelligence that they can have. They have three levels of intelligence, three levels of the location of intelligence, and three levels of the aggregation of this one.

Here you can see a better explanation with more details and the classification in: https://www.researchgate.net/publication/307638707_A_review_about_Smart_Objects_Sensors_and_Actuators

Here you have an explanation that I did before about the Internet of Things: What's the difference between the Internet of Things and the traditional Internet?