System and network architects beware – putting that IoT network in place isn’t as straightforward as it looks.
The Internet of Things has been hugely talked up, from the astronomical figures of devices and connections, to the promise of increased efficiency and new revenue streams.
Yet, as many IT teams would attest, putting in the infrastructure to support an IoT project is not easy. Deploying an IoT network requires everything from a proper tendering process for acquiring these devices, and making sure they are secure, to also choosing over what connection and protocol (over Wi-Fi, radio or fibre) these devices connect with one another.
Then there comes the conversation of monitoring the gateway/network, and establishing where the data should be stored, managed and secured. There is too the question of how and where data should be analysed – and by whom.
All of this represents a huge challenge for CTOs and network architects, not to mention application developers, in managing the device, the network and the back-end services.
With that in mind, Internet of Business looks at the five top technical challenges to building and deploying an IoT network:
There are various layers of information security as far as IoT is concerned, from the device and gateway to the network and the cloud, where most of the data will eventually reside.
Security is typically undone by vulnerabilities at the end point, or in the software itself. End-to-end encryption is vital at all times, both when the data is at rest and in transit.
The threats are almost limitless: IoT devices are small, inexpensive devices with no or little physical security, while some of the computing platforms (which are often constrained in memory and computing power resources), sometimes don’t support complex security algorithms owing to weak encryption or low CPU cycles.
There’s also the physical danger of the devices being tampered with, or stolen, and of the device software not being updated – thus making it more susceptible to cyber-attack.
Some say that there must be a stronger focus on identity with the IoT, and then there needs to be a strengthening on network-centric methods like the Domain Name System (DNS) with DNSSEC and the DHCP to prevent attacks.
Experts say that simply monitoring and controlling the flow of data packets to and from the IoT device is not going to be enough to guarantee security, with network management and data logging key to preventing and stopping attacks.
IoT platforms, whether designed internally or externally, need to be scalable and robust. Enterprises want control over their data, and to protect their intellectual property (IP), whilst ensuring that the platform can handle huge amounts of data and connect with existing legacy systems.
It’s important to carefully consider what suppliers you want to work with, as well as the costs involved, and to work out how these platforms will connect with your existing IT architecture.
“The main challenge facing network architects when it comes to IoT is future-proofing the network, and having confidence that they’re making the right choice when it comes to sensor hardware, radio technologies or cloud platforms,” says Adam Leach, director of R&D at Nominet.
“Network architects want to avoid being locked into one vendor, and finding themselves tied to an outdated piece of hardware or software when something better comes to market.
Jan Maciejewski, industry analyst and contributing editor at the Internet of Business, explained that there is simply too much choice.
“Everyone claims to have a platform so the debate is now about a choosing the right one…and getting backwards compatibility. The scalability of platform and the integration with legacy systems is vital to choosing the right one.
“The issue is there is far too many out there. It’s about choosing the right platform for the problem you’re trying to revolve.”
Interoperability and standardisation
Networks can connect over a wide variety of communication protocols, and experts say that standardisation is going to be crucial here if IoT is to survive and thrive.
“The first challenge is planning the physical design of the network, and working out exactly where you’re going to place the sensors. After that, you have to plan a connectivity strategy for these sensors, and then implement it,” says Leach.
Yet, he admits interoperability is essential in the long term. “In the short term, it’s often more important to just get a system deployed. However, in the long-term interoperability is definitely a good thing. It’s the key to a sustainable IoT.”
Bas Geerdink, IT manager at ING, adds: “There is no clear understanding and wide-accepted set of standard protocols yet, although various attempts have been made through Bluetooth, ZigBee, NFC, Wi-Fi, LoRA.”
“Interoperability is really important. It is essential to get a good understanding of the protocols and other network standards to make the IoT efficient. If not, every new project will have to go through the same hassle again of setting up connections, programming interfaces, etc. We’ve seen before that things settle after a while even though there are a lot of vendors and standards, so I have confidence in that the market finds a way.”
Finally, network architects need to address the integration of all these new systems with existing, legacy platforms and analysis tools. This is not an insignificant challenge, so proper planning is vital to ensure new and old systems are properly interconnected, integrated and providing seamless operation.
Data storage and analytics
IoT devices will generate in huge amounts of data, and businesses subsequently have to decide on storage, analysis and deriving insight on that data. For example, how much data can be kept at the edge and how much in the cloud to be monitored by a PaaS or data analytics provider? And how can this data be transferred and stored while adhering to industry governance and international laws?
Analysis of this data can help firms to make proactive, data-driven decisions, while delivering an insight into their network operations. Meanwhile, predictive network analytics tools, offered alongside network management systems, can provide reporting utilities that offer detailed network performance indicators.
This can be as simple as automatically prioritising data traffic, determining whether a new service or application being rolling out will exceed current network capacity, or agreeing a certain time when extra bandwidth will be provided for data-sapping processes.
IoT sensors and devices
Wireless sensor networks are a collection of distributed sensors that monitor physical or environmental conditions, such as temperature, sound, and pressure. Data from each sensor passes through the network.
The network engineers will need to measure and maintain these, while from an IT standpoint there is also the issue of procurement. Data management, security and accessibility all need to be considered in that initial tendering process.
“Network architects should be aware of the grand upscale that the IoT brings,” adds Geerdink. “Where currently an architect has to deal with only a few connections points (servers, computers/laptops), the IoT brings an enormous increase in endpoints.”
IoT Build is the only summit exploring the technical challenges of building the next generation of enterprise IoT networks. Tailored for CTOs, CIOs and Directors of Enterprise Architecture, you will learn deployment lessons from early adopter IoT “builders”. Featuring case studies as diverse as Hive, Bristol is Open, ING and Stanley Black & Decker.