As IoT gains pace, collecting and then analysing the vast quantities of data produced by the billions of devices that the IoT ecosystem will consist of, is vital. IoT data collection and analysis are now critical skills to have available within your enterprise.

Your business will be able to adopt an IoT strategy, but ensuring these investments deliver the expected commercial gains, will be dependent on your company’s ability to understand the information flowing into your enterprise from these devices.

Industries such as manufacturing where IoT will have the most immediate and tangible input must ensure high levels of data analysis are present. As IBM points out: “Cognitive manufacturing is powerful because it combines sensor-based information with machine learning and other artificial intelligence capabilities to find patterns in structured and unstructured data from plant, enterprise and industry systems. It pulls relevant information together in real-time and applies analytics to yield unprecedented levels of understanding and insights about the manufacturing process.”

The insights detailed data analysis delivers to any company using IoT can’t be overestimated. Indeed, according to IDC, around half of the Global 2000 companies will earn more from selling IoT sourced information, than through the sale of physical devices by 2020. Data has for some time, become the currency businesses can profit from. However, in an IoT dominated space, analysing data becomes even more critical.

Speaking to Silicon, Nelson Petracek, CTO, TIBCO explained: “Efficient data collection and analytics are key to the success of an IoT implementation. IoT is all about data generation and the subsequent collection of this data from the edge. Presumably, sensors are being placed to increase one’s visibility to particular data and behaviours, and to act upon this data at various stages in an IoT architecture.”

Petracek continued: “You may want to push the desired processing to the edge itself. Perform processing at the gateway level or, handle the IoT data in the data centre or cloud. If this collection is not timely, it introduces too much ‘decision latency’, delivers too much ‘noise’, or otherwise does not deliver the right data, then any related responses or analytics will be incorrect. Also, one is presumably collecting this data for the purposes of doing something ‘intelligent’ with it (not just for storing it). Whether that is to respond proactively to one or more conditions, have devices act together autonomously, or to make business decisions with increased visibility, the use of analytics is key to improving each of these elements.”

Analysing IoT

As many industries are on the cusp of what could be seismic changes to the environment in which they operate, analysing the data that is now becoming available is a commercial imperative. Supply chains are a clear area where these massive changes will take place. Sensor technology will transform how companies manage this vital component of their businesses.

Across the services sector, huge change is also taking place. The rich sensor environment IoT is creating will give companies unprecedented access to highly personalised information about the behaviour of consumers. Being able to analyse this data along with other data points, will be crucial to the long-term profitability of many companies in this new age of data analysis.

The type and quantity of the data being collected and analysed will also impact how IoT devices are supported by the networks they connect to. Speaking to Silicon, Andy Rowland, Head of Customer Innovation Energy, Resources, Manufacturing at BT’s Global Unit, explained: “The nature of connectivity required for individual deployments very much depends on the type of IoT data you want to collect and analyse.

Rowland continued: “Where you have small amounts of data, for example, status updates on water and silt levels in gullies which you need to monitor remotely to prevent flooding, a simple LoRaWAN network is sufficient as this involves very low amounts of data stored in the cloud. However, for something like Industrial IoT where you may have oil pumps generating 20Gb per day, you need to filter and analyse data at the edge and, be much more selective about what you send to the cloud.”

Finding patterns in the data that IoT devices are delivering to your company will enable your business to innovate. With so much information available, understanding where the value resides becomes a key business driver. Information and insight that was previously hidden from view, suddenly become the basis for what can be highly lucrative decisions.

Understanding data

Using the data that IoT devices will generate is a core driver for all businesses in this space. As the Vodafone Business, IoT Barometer 2019 acknowledges, many businesses are already readying themselves: “What’s striking is that a similar proportion (58%) are already seeking to draw meaningful insights from their data using analytics platforms — and that figure goes up among more sophisticated organisations. Within band A (early adopters of IoT), 80% are already using analytics platforms with IoT data to support decision-making.”

IoT, though, does deliver a number of challenges when interpreting the data being produced. Rich Pugh, Mango Solutions’ Chief Data Scientist explained to Silicon: “Over the last 20 years, data and the application of advanced analytics (including AI, Data Science and Machine Learning) has evolved into a strategic initiative, with companies investing heavily in the people and platforms to enable a data-driven business.

“However, the ‘rhythm’ of IoT with its streaming approach means it can be siloed, both in terms of technical solutions and the range of challenges it can solve. For the nature of the challenges IoT is currently being used to solve (spot change and react) this is acceptable, but, CTOs and CIOs are starting to understand that IoT offers a way for us to see data streams as a way to constantly learn and update our knowledge. It opens up many more possibilities, but to enable this we need to be able to connect IoT and streaming architectures in a way that allows us to consistently update our knowledge, which is held in more static forms.”

With Marc Canel, VP of Strategy Security, Imagination throwing AI and Machine Learning into the data analysis mix: “The challenge behind data collection is the fragmentation of systems. The machine learning algorithms will get better results as they iterate on data that gets generated by similar systems. For example, a machine learning system that studies the behaviours of hospital equipment will get better and better as it is fed more data coming from hospitals. It then becomes specialized in hospitals and cannot be used in another environment. We see the fragmentation of machine learning algorithms along with the environments from where the data comes from. Analytics becomes a fragmented market along the lines of the vertical markets because of the mechanisms used to process data.”

Ultimately, businesses will be able to see the value in the data they are collecting. IoT shifts not just the quantity of information being collected, but also, the quality. As more personalisation of goods and services becomes a key brand differentiator, this intimate data becomes vital to analyse.

TIBCO’s Nelson Petracek concluded: “I see a progression of approaches when it comes to IoT data collection and analytics. For example, people typically start by simply collecting the generated data (with minimal transformation, and often in micro-batches), dumping it into a data lake of some sort, and then executing historical reports or simple analytics against the data.

“People then tend to next look at pushing processing logic closer to the edge, applying more logic in real-time, or applying analytics against the inbound stream of data using more advanced techniques such as streaming analytics, event processing, or time series analysis. Some metadata associated with the devices may be maintained (and used to augment the inbound data), and device data may be enriched with other enterprise data in order to improve the context behind the analytics. Various combinations of these approaches tend to be utilised with varying degrees of maturity, and different technologies may be employed during this progression.”

Data continues to be currency many businesses are now based upon. Analysing that data is how enterprises move their goods and services forward. IoT will transform every business it touches. If your business can interpret the information flowing into it from the IoT ecosystem, this ensures your company is ready to take full advantage of this new data source.

Silicon in Focus

Founder, Owner and MD of PCG & Paratus People, having studied Embedded Engineering at University, White is the 500th member of the IoT Council and a member of the IoT Security Foundation. White has over 12 years’ experience working in broadcast technology, automotive, consumer electronics, embedded devices and IoT industries.

How important is efficient data collection and analytics within the IoT environment?

The number of products and services that are connected to IoT both directly and indirectly has increased at an exponential rate in the last decade. In fact from 2016 to 2020 devices are expected to increase by a CAGR of 21%.

Due to the rapid proliferation of these technologies, fast and efficient data collection and analytics are crucial to improving functionality of IoT devices, particularly as IoT is venturing into important markets such as healthcare. Efficiency is incredibly important within this sector, in order to quickly diagnose and function IoT powered machines and processes.

Collection of this kind of data reveals great insight into human behavioural patterns and trends as well as associations – this is how IoT processes will continue to develop. In addition, in order to stay at the forefront of this technological wave analytics can be used to help identify the next step or the next big thing. It’s a circular form of progression.

What are the current and future challenges when IoT data collection and analytics is considered?

There are many hurdles in the way of data collection at the moment, as is the case with any new technology, as we trial and error our way to full adoption. First and foremost, we don’t necessarily have the processing power to analyse the type of data that has evolved significantly versus traditional text-based data.

This means we need to build systems that better understand the types of data we are faced with now. If we look at mobile data as an example, this has progressed rapidly, encompassing formats such as video, image and voice search formats. The implementation of 5G will be the first step at tackling processing issues as it is built to handle various forms of data.

Another challenge is the sheer volume of data. As IoT is being developed in all sectors, from transportation to agriculture, education to retail – the amount of data being transferred is almost insurmountable. Therefore storage and essential data growth will be a challenge for the foreseeable future. IDC estimates that the amount of data is growing at an annual rate of 61% and this continues to rise. The velocity of IoT data must happen in real-time for functions and results to be carried out. IoT data is generated at such a velocity that this becomes a challenge, also.

Finally, the employment market is also affected. The rate of growth at which the IoT is expanding and the high level of proficiency needed to work within the sector, demand for skilled labour is through the roof. It’s clear that over the next few years that businesses will need to turn to specialist hiring services such as Paratus, in order to plug this skill gap.

Is the 5G network an essential component of efficient IoT data collection?

5G, rather like AI, is the support system that will be implemented to improve the functionality of the IoT. Fifth Generation technology will improve the ability to collect high velocity and high-volume data while responding much faster to inputs. It will also enable us to speed up all aspects of the data collection process from the devices and sensors collecting data, all the way to the storage and translation of the data. Processing large varieties of data will also be achievable post-5G.

However, there is a risk of relying too heavily on 5G for the effectiveness of the IoT. UK businesses alone are already losing three days a year to internet failures, setting them back around £7 billion in total. While 5G and transferring information via cellular data will aid processing, it can fail. This said, businesses are uncovering alternative ways to transfer data – for example, through the use of soundwaves. 5G will be crucial but will not be the only method of creating an efficient IoT.

How is data security being approached within the IoT space when data is collected and analysed?

If IoT security could have been invented before the IoT, we would certainly be in a better place. While we can’t go back in time, we can take the approach in which security is considered at the very beginning of initial development stages going forward. We are playing catch up, but we need to make sure that isn’t always the case.

There’s a lot that can be learned from the early days of personal computers. Similarly, security was an optional afterthought for consumers, rather than the product developers – remember having to buy Norton or McAfee to avoid viruses? PCs didn’t have the capacity to deal with security risks and, the IoT can learn from this by considering the risks pre-development.

There are four layers of IoT that need to have protection from attack. These are the device; the gateways that transmit data; servers; and the engineered infrastructure that allows the IoT to thrive. Therefore, there are a lot of stakeholders involved, rather than one straightforward format of security.

As well as a widespread security consideration, there is currently a real lack of compliance regulation, particularly in government use of IoT devices – this is a cause for concern. Education of the public will be the turning point for IoT security.

Right now, IoT can be seen as a mysterious innovation, that many people struggle to get their heads around – as was the case with cryptocurrency and blockchain. Offering simple explanation and education around these immensely important tech developments will allow consumers to use their devices in a safer way and, will improve the need for enhanced security through the design process – rather than responsive implementation.