IOT Architecture & Reference Models (Vol-1): The importance of an IOT Architecture

The concept of IOT even if it seems simple but in reality, it’s more like an Iceberg where the components we talk about are only small pieces of a much larger ecosystem. There are actually a lot of moving instruments that need to work together in order for the Internet of Things to function as intended. It's vital for all of these different cogs to work together if the IoT implementation is to function as desired.

One thing that can bring all these instruments together & link the physical components, the functional organization and configuration of network elements together with the operational procedures, use cases and data items is an IOT architecture.

BACKGROUND

When we talk about IOT architecture the process of defining one single reference model or one architecture of the IoT ecosystem is not at all simple. This domain touches upon a wide area of technologies and the components travers through all layers of an IT network. That’s the reason there is no single consensus on architecture for IoT, which is agreed universally. Different researchers, companies, technology groups and forums have proposed different reference models and different solution for us to choose.

CHALLANGE's IN DEFINING AN ARCHITECTURE

While the IOT is accelerating digital transformation and smart systems, showing us an exciting future but it does not come without significant challenges. In this article we will talk about such challenges which prevents us from using the existing IT architectures and models.

1. INTEROPERABILITY: - As we understand that IOT solution is an amalgamation of different devices, technologies, protocols, services from different vendors & the end solution needs a full-bodied integration of all to deliver value. For example, different companies develop Android based phones and applications but they all follow a common structure of development. A study by the International Standards Organization (ISO) revealed that there is more than 400+ such standards existing related to IoT. Even different product companies build products not following a defined architecture and framework. An end-to-end solution needs a common language, framework & standardization of various assets made by different companies. Hence defining a common architecture and framework is difficult at this stage.

2. LIMITATIONS of EXISTING IT NETWORKs (difference between IT & OT Networks): - The existing IT network which supports business applications, mails, web, databases, CRM etc are networks where data is initiated / generated primarily by a computer. Whereas IOT is all about the Data generated by sensors and how that data is used to take actions. The factors of differentiation between IT & IOT networks are

• Scale: - The scale of a typical IT network is in order of several thousand of devices at the max – typically, computers, printers, mobile or wireless devices. Traditional networks with Lan’s, WAN’s or Wi-Fi’s structures can still manage the requirements. Whereas an average size IoT network can scale from a few thousands to millions+ end points. Structuring a network with so many routable endpoints need a different architecture and IPv6 should be the natural foundation for the IOT network layer with suitable protocols.

• Constrained Devices: - The IOT end points are designed for specific purposes, small, non-expensive, with less than 10 KB of memory and 100 KB of storage space, battery powered, with extremely low computation power in comparison to the computers with multi gigabyte of storage and unlimited commutation capacity. This is a completely different situation on top of that these devices are usually installed in external environments with lossy network connections.

• Security: - The IT security systems which is designed for protecting Servers, corporate networks, Data intrusion with firewalls, strong encryption methodologies can’t be applied for IOT devices as these devices are not in a secured physical environment, it works on a wireless, low bandwidth, Lossy network and also doesn’t have the required computation power for complex encryption and authentication methods. IOT networks needs a different architecture and modified processes for authentication, encryption and intrusion prevention.

• Volume of Data: - IOT devices generates 1 KB to terabytes of data in hours based on use cases. This networks not only processes structured data but a almost all the amount of Data in an IOT network is unstructured. As more and more devices are connected to the network data generated by this system becomes overwhelming. It not only consumes network bandwidth but also chokes server resources. Hence the transport, management, processing and analysis of that needs special care and architectural changes.

• Support of Legacy Systems: - in our IT world computers and software’s are constantly being upgraded, replaced, security patches and updates gets uploaded on a daily basis interrupting our work. But in the world of IOT there are machines which are decades or older which can’t be replaced, neither be interrupted for months, neither be upgraded. IOT works with such devices and hence supporting, managing operations, connectivity, data collections with all such devices needs a different technology & architecture.

3. DIVERSE PROTOCOLS & TECHNOLOGY STACK

The picture here represents a ready reckoner on the leading protocols & technologies available at various layers of the OSI 7-layer architecture which are primarily applicable for IOT. The normal protocols like HTTP etc are not suitable for the constrained devices and hence new protocols are developed across the years and even now modified versions of the same are being developed continuously. It’s difficult to manage such a wide range of technologies unless and until a proper architecture is built which will pick the right technology and will deliver the end result.

4. REAL TIME DATA PROCESSING

In the world of IOT Cloud Computing has been used for processing Internet of Things (IoT) data primarily. This works fine for the analytical and batch processing jobs but most of the IoT applications demand real-time response which cannot be achieved through Cloud Computing mainly because of inherent latency. Real-time utilization of massive IoT data streams suggests a paradigm shift to new horizontal and distributed architecture. Edge computing is a solution that facilitate data processing at or near the source of data generation in the context of IoT. Another concept of Fog Computing also solves this problem by offering cloud-like services at the edge of the network. The computationally powerful edge devices have enabled realising this idea. Witnessing the exponential rise of IoT applications, the IOT architecture should take required measures for providing solutions to this aspect.

4. BIG DATA & ANALYTICS

IOT solutions are ultimately an intelligent system which functions automatically based on events and triggers but all of them are an output of the analysis of DATA received. While the system should be capable of managing structured data sent from IOT sensors in the form of values such as pressure, temperature, humidity, acceleration but it should also be capable of handling data like, text, speech, images, Video which doesn’t fit a predefined model. The complexities of the IOT architecture perspectives takes care of the following:

• Machine Learning - Once you have the data what do you do with it?

• Big Data Analytics tools & Technologies - technologies and protocols

• Edge streaming analytics -Processing closure to the source

• Network Analytics - Analysis of Network flow

CONCLUSION

The solution to the above-mentioned challenges and limitations are a structure & framework which is flexible, modularised, scalable, independent and still can talk to different systems extended across Vendors, Technologies, protocols data types and use cases.

Hence, we need to establish IOT architectural reference modes for the interoperability of Internet-of-Things systems, outlining principles and guidelines for the technical design of its protocols, interfaces, and algorithms. It also defines an efficient way of Integrating different components into different service layers, allows scalability of IOT resources.

Industry bodies and forums has developed few such Architectures. The most common of them are oneM2M, IoTWF, ETSI, IIC. I will talk about all of them in future articles.

In the next volume we will talk about different Layered structure proposed as the baseline for different IOT architecture and reference models by different industry bodies, technology forums and organizations.

---