Introduction
The Internet of Things is gradually being integrated into the everyday life of everyone, more and more devices are communicating with each other, collecting data, and then processing it to solve certain problems. However, users can see only the "tip of the iceberg," IoT systems are much larger than we can imagine. These include collecting, converting, processing data, transporting and storing information, highlighting certain features, using neural networks and big data algorithms, and at last reproducing the end result in a user-friendly way. Modern Internet movements are so infrastructurally complex that it is almost impossible to develop without using one or another IoT platform.
1. What is a cloud platform? How does it work and why do we need it?
In one of the previous articles, we looked at IoT platforms in general and said that they come in different types depending on what part of the infrastructure they provide. This time we will look at cloud-based IoT platforms, which provide, among other things, data processing and storage and are most often used in business projects and startups. The main market players in such types of IoT platforms are, of course, the big three cloud service providers, namely Microsoft, Amazon, and Google.
IoT platforms are often referred to as so-called intermediate products that allow the digital world to interact with the real world. With their help, developers can create programs specifically for IoT purposes, as well as digital shadows of real physical devices for further modeling and predicting the state of the system in the future. Cloud platforms also give developers the ability to quickly connect and deploy numerous devices and complex infrastructures, test, and scale IoT-specific software systems. The use of IoT platforms is very wide. They are used by suppliers and manufacturers of smart devices, are used to automate the production and control of traffic flow, widely used platforms in medicine, and for monitoring of ecology state.
Simply put, the IoT platform is a set of services that offers you the tools you should move physical objects to the Internet or Cloud. It must support millions of simultaneous device connections and easily configure devices for inter-machine communication.
2. Trends and descriptions of the main players on the market
The development of the Internet of Things in general and the actual platforms is not dictated by the development of technology, but by market requirements. A business sets tasks that need to be completed, and engineers and developers try solving the problem, thus developing technology. This approach leads to the fact that some technological solutions, even from the list of very promising, are simply washed out of the market, if there is a more common analog. The IoT industry does not yet have uniform international standards but is working to create them. And while the market culls the vendors, protocols and standards are most widely used.
However, new challenges require new approaches, which, however, are often partially or fully compatible with previous ones. A good example of such evolution is the Narrowband-IoT, which we talked about in one of the articles. This standard allows transmitting information from numerous devices on already established networks with minimal changes in the physical infrastructure, and, at the same time, devices consume less power and have a larger coverage area. This example illustrates well the prevailing trends in IoT in terms of hardware: reduced power consumption; increase number of devices that operate in one network; increase geography of coverage. It is worth noting here that, although new current protocols and standards are displacing fewer effective ones, there are still many devices and ready-made infrastructures (especially state ones) that use outdated equipment. It can be concluded that an effective IoT platform should support both state-of-the-art and outdated data transfer protocols but may be limited to the most common ones. In terms of data processing, the main methods now are and in the near future will be neural networks (including deep and ultra-deep NN), artificial intelligence and current Big Data algorithms. It is clear that IoT platforms should provide tools for the effective use of these technologies, including hardware acceleration for neural networks.
Important parameters of IoT platforms are data preservation and protection because data are the basis of modern business, as well as the scalability of solutions depending on the development and objectives of the business. That's why all majors build platforms based on their cloud services. In this context, the issue of the security of IoT systems should also be mentioned. The platform must ensure the security and integrity of data at all levels of the architecture. Now this problem is one of the keys because security solutions are often costly, and the customer does not always agree to so high additional cost. Of course, one of the main factors is also the monetization of the platform and payment form.
Let’s look at the main players in the market and what they offer now for businesses and developers in all aspects mentioned above.
3. Amazon Web Services
The most common cloud platform for the Internet of Things today is the solution from the global public cloud market leader, Amazon, whose services are included in the AWS list. Amazon was the first to assess the possible market growth of cloud services for IoT and introduced specialized services. There are now 13 IoT-focused cloud services on the AWS list and the company continues to add new ones. This approach allows the developer to choose services with optimal functionality for current tasks. Also, the specialization of services should simplify the setup process and increase efficiency compared to general-purpose cloud services. It is the ease of learning, thanks to helpful tutorials, and working with IoT services that AWS users highlight when talking about the pros of the platform.
Application management services are presented, among others, the new AWS IoT TwinMaker and AWS IoT Roborunner, the latter specializing, as the name implies, in robotics, and we know the use of robotics to optimize warehousing logistics is a strength of Amazon itself. AWS IoT TwinMaker allows you to create a digital twin of both individual devices and complex systems, like factories, plants and so on. This technology incorporates both the process of obtaining data from specific devices and simulations, which allows assessing the state of the real physical system and at intervals when information from sensors is not received.
AWS includes service for device management with the same name, and a whole set of tools for data processing tasks: AWS IoT SiteWise, AWS IoT Analytics, AWS IoT Greenpass, and others. Among the services offered by Amazon are those that provide protection of devices from unauthorized access and support to work with the real-time operation.
Probably the main disadvantage of the AWS IoT platform is many different services with repeated and overlapped functionality. It will be difficult for the user or not experienced developer to determine at the stage of choosing the necessary tools, and the reprocessing then involves financial losses, so already at the stage of planning of using the platform, it is better to involve relevant experts.
We should keep in mind that based on management rhetoric, the IoT industry is a priority for cloud computing of Amazon, and based on trends, they are making the right bet. Therefore, we can expect improved IoT of AWS services and further expansion of the range.
4. Microsoft Azure
4. Microsoft Azure The Microsoft Azure platform is practically not far behind the market part and the set of opportunities. Priority of the software giant is the corporate IoT market and the organization of smart industries, smart buildings, and smart cities. The advantage in this segment is achieved due to the widespread use of other cloud services, such as Office 365 and Microsoft Dynamics. It is easier and more profitable for customers to add new cloud services on top of already existing tools, instead of transferring the system to another vendor. The popularity of Microsoft services is primarily due to two factors: an aggressive market policy to attract new customers in the corporate segment and the integration of services with industry-specific services such as Microsoft Cloud for Retail or Microsoft Cloud for Manufacturing and other business intelligence tools.
In a list of Microsoft Azure services are actually nine for specific IoT tasks. There are two services for application management: Azure IoT Central and Azure Digital Twins. The first is designed to simplify the interaction between devices using IoT Plug and Play, and the second is an analog of AWS IoT TwinMaker to create digital copies of real devices and monitor their status.
Azure IoT Hub is the only service for device management. This is a key service in any IoT system based on solutions from Microsoft. This includes data from device-related services and manages the deployment of services on the Azure Kubernetes Service cluster. Services such as Azure IoT Edge and Azure Percept are used for data management and processing. Other specific services can also be used, such as Power BI for business intelligence.
The disadvantage of Azure IoT is a non-prompt detection and elimination of bugs in existing software and the slow implementation of new developments and features.
5. Google Cloud Platfrom
5. Google Cloud PlatfromThe third-largest market share is the Google Cloud platform. Representatives of the search giant have taken a fundamentally different path, so, unlike others, Google has only one service for all tasks of the Internet of Things – IoT Core. This is possible by using a container structure and a powerful orchestrator that manages containers and is responsible for the integrity of the structure as a whole. Containers can accommodate blocks that perform tasks: work with devices and applications, data processing using various algorithms and artificial intelligence, and so on. The suchlike approach is the most universal, as it allows developers not only to use ready-made solutions from the vendor but also to create new services for their own specific tasks. On the other hand, the increasing complexity of the system (and we are moving to billions of devices in one system in the near future) increases the complexity of the orchestrator and debugging the system in all, which can lead to stability problems.
The advantages of IoT Core include the overall strengths of Goggle itself. It has extensive experience in systems and algorithms for artificial intelligence, machine learning, analytics, and Big Data. Of course, as always, searching and ranking algorithms are strong. And the open-source container orchestration platform Kubernetes used in the service is already a cloud industry standard.
At the same time, Cloud IoT is probably not a priority for Google, as it does not develop IoT-specific tools and algorithms that rivals already have: there is no implementation of Digital Twins, and the promising real-time operating system Android Things is shutting down this year.
Conclusion
Conclusion Choosing the right cloud platform provider can be a daunting task. All of these platforms have common features - the complexity of the structure, numerous components, and settings. So, the development of solutions requires some experienced engineers who are familiar with a particular platform. If the IoT solution is planned as an extension of an existing system, it is advisable to use the platform of the same vendor whose cloud services are already in use. For new projects, we can recommend AWS IoT as the fastest growing platform and the first to introduce new functionality.
