This is an extract from a speech presented by Dr Manjunath Iyer, principal consultant, Wipro, at iotshow.in 2019. It describes the emerging trends in the design of cognitive Internet of Things (IoT). Rather than just capturing data, these cognitive IoT devices can think, understand and accordingly provide data to the user.
Things were there even before the term Internet of Things (IoT) was coined, but these were not connected to the Internet. Today, with the introduction of the Internet, everyday objects (things) have become smart and capable enough to take decisions on their own.
The next step is transformation of these IoT devices into cognitive IoT devices. So, rather than just capturing the data, these devices should also have the ability to think, understand, and accordingly provide data to the user. IoT devices can be turned into cognitive IoT devices by incorporating cognitive computing technologies.
Living IoT devices
Yes, even an animal like buffalo or cow can be transformed into an IoT object. How? Just connect a location sensor to the ear of the animal. Based on the animal’s roaming information, the user can know grazing pattern of the animal on a particular day, how much milk it will provide, and thus calculate the amount of profit it will generate.
A living being as small as a honeybee can also be transformed into an IoT object. A small nanosensor can be placed on the body of the insect and, based on its flight path, one can deduce what kind of flowers that honeybee is interested in. This can lead to planting more of such flowers and production of greater amount of honey, resulting in more profit generation.
IoT is not free from issues, however. You cannot put anything and everything, anywhere and everywhere; there are a lot of connectivity issues that you need to deal with.
Let us first understand what all an IoT system consists of. It mainly has an object or thing to be connected, connectivity through wireless standards, and the Internet.
The object (things, organisms, etc) to be connected is the main component of the entire IoT system. Connectivity comes in various forms such as Bluetooth, Wi-Fi, Zigbee, NFC, RFID, and many other wireless standards. The data is put on the Internet for storage, processing, and analytics.
When 5G becomes operational, it will be easier to connect many more devices and put massive amount of data into a network.
IoT is a culmination of such technologies as augmented reality, signal processing, big data, mobile computing, object and voice detection, and many more. It can be used in various sectors like banking, energy, industrial, vending, security and surveillance, medical, consumer, telematics, etc. Due to such huge diversity, the chances of it going wrong increases. Therefore the right combination of components should be selected keeping in mind the end result and its applications.
Example of IoT based solution
Consider a mobile phone that remotely monitors when and how the home lights get operated.
This can happen by first registering the objects to be monitored (home lights in this case) on an app installed on the phone. Next, by object identification. Each and every object (light) can be subscribed to the installed application running on the Internet. Through this, one can easily control all the lights.
Sensors are at the lowest level in an IoT system. These can be temperature sensors, position sensors, velocity sensors, etc. They are connected to the nearest point to track data.
Once the data is collected from the objects through the sensors, it is processed and used in various applications. Application areas include surveillance, clinical infrastructure monitoring, health monitoring, smart transportation, etc.
IoT protocol layers
Now let us look at IoT’s working in a bit more detail. An IoT system comprises an application layer, information processing, a network, and sensing and identification. All these four components together make an IoT solution.
The hardware for an IoT system has the following:
Sensors, actuators, LEDs, relays, motors, linear actuators, lasers, solenoids, speakers, LCD, plasma displays, robots, etc.
Modem, gateway, router, satellite, tower, etc.
Grid, cloud, embedded processor, quantum computing, etc.
Cloud memory, flash memory and quantum memory.
The software for an IoT system requires a wireless sensor network (WSN) that has following characteristics:
- Smart and autonomous
- Can be configured automatically
- Self-monitoring and self-healing
- Capable of anomaly detection and tracking
The event notification service (ENS) of an IoT system acts as a common collector and distributor of events. Events come from multiple heterogeneous sources. Devices can subscribe to specific subsets of events.
In publish-subscribe software architectural style of ENS, whenever there is a change in an event, a notification is published stating the same. Then, accordingly, messages are generated to be sent to the user. For example, for a voltage change or current change, the event will be interpreted as switching on or off. This will be then conveyed to the subscriber (user) to act upon.
In object as a (Web) service ENS, this is an emerging paradigm. Nowadays, services such as food delivery happen over the Internet. Moving forward, thanks to AI, robots will be aligning themselves with Web technology to provide this service in a faster and efficient manner.
UPnP+protocol for IoT
This protocol is extensively used in present-day high-end IoT solutions. Going forward, it will be required in cognitive IoT because the objects adhering to this protocol can interact better.
UPnP+protocol involves following steps:
Device advertises itself within a network and enables discovery.
After the discovery, the control point hits the URLs to get the description.
Then the address is auto-generated for ease of use.
The focus is on human-machine interfaces.
This step is closely coupled with control actions.
Here the device is required to perform some action over other devices.
However, some issues are faced by the presence of a multitude of IoT devices. In a large distributed system this can sometimes create confusion. Besides, sensors that are placed on a moving object are not made robust enough and at times provide low output.
Issues can arise because of:
- Multiple services, multiple users
- Heterogeneous services
- Bursty traffic requests, that is, people sending large amounts of data to the cloud at the same time. This causes an overload resulting in system/server crash.
How to handle these issues?
Do not put all applications to use at once. Put some applications to use first so that the moment the device captures data, it also processes it. And divide the data into parts before transferring it to the cloud.
Dr Manjunath Iyer has over two decades of experience and has filed about 82 patents, authored 180 research papers and a book