IoT meets Industry
No doubt you’ve heard of the IoT – it’s seems hard to miss these days. Now we also have Industrial IoT - in vogue for describing its applicability to the broader industrial sector.
The IoT will make new data available in ways we couldn’t imagine before; it will spur new markets and connect existing markets that previously had little overlap, further driving the need for applications that can make sense of the available data to extract meaningful results. Automating this kind of information discovery and exploitation will be huge, IDC forecasts a $7.1 trillion market by 2020; McKinsey suggest a more ‘reserved’ $6.2 trillion impact by 2025. When you see numbers like this you realise that something rather dramatic is afoot.
So Why Now?
For starters, a number of technical advances have been made that will enable and then drive the IoT movement. They are:
- Wireless communication, at least in the developed world, is plentiful and cheap. New developments in 5g and low power radio communication standards will only make this even clearer. Mobile operators, tired of giving away 1000s of free minutes and text messages, are seeking new business models to boost their balance sheets and leverage their infrastructure investment – IoT is the answer.
- Cheap, distributed and scalable processing means that even a teen enthusiast can command a vast array of processing power (more than many governments would have had access to ten years ago) for little more than the cost of a can of soda. Not just in the Cloud, but practical processing is finding its way to the edge device emerging as a Fog Computing architecture.
- Low cost, ‘good enough’ sensors and devices such as the Raspberry Pi are beginning to litter the market driving both an excited hobbyist community – rekindling memories of the PC era of the late 70s and early 80s – and serious high-tech businesses. Electronic devices are coming off production lines with internet-connected sensors installed as standard. Whether we ask for it or not, things are getting sensored up. And if you need sensors they are readily available for less than a few dollars and a quick search turns up great open source code examples making integration painless.
M2M is not IoT
It is true, some sectors have been exploiting this sort of capability for years – aerospace, mining and Oil & Gas spring to mind. M2M (Machine to machine) has been used to describe this early work but let’s make it clear – that’s not IoT. Despite many industry stalwarts promoting this earlier work as "IoT", they’re fundamentally misunderstanding the real drivers of IoT. It's not simply the technology and interconnections.
The difference with the IoT is in the sheer scale and ambition. IoT is not just about connecting the devices, it’s about autonomy of menial activities and opening up and integrating datasets in new ways which will bring about much bigger changes. Silos will be broken down – or watch as your competitor’s leave you in their dust. Business models will change and news ones will emerge – think availability based contracting (or ‘power by the hour’ in the case of airliner engines).
From smart cities to intelligent manufacturing, it’s about organisational and personal change as much as it is about technology. That also means its impact will be felt far and wide. For instance, Smart City designers are wanting to use IoT to exploit city data to better inform citizens and provide a more transparent model of governance. “Yeah right”, I hear you say, “I’ll just be given more adverts for things I don’t need” but this is serious stuff, designed to make your everyday life easier.
The potential of IoT in manufacturing is clear. Unplanned downtime and reactive maintenance are undesirable cost drivers. I read that in car manufacturing, every hour of downtime can equate to $2m in losses. Fortune calculated that IoT could deliver maintenance savings of 25%, reduce unplanned downtime by 50% and extend the life of machinery by years. Sounds good?
With the increasing automation and computerisation of factories throughout the 90s and 00s, much of the core infrastructure is already in place. Many firms are now ‘upgrading’ this infrastructure to be IoT-ready – just look at the customers of ThingWorx and Xively to see who’s leading the charge. The benefits come from the analysis of the data and more fundamentally the prognosis of future machine degradation.
Big, asset intensive, industries have had good reason to implement predictive maintenance regimes for years and they’ve relied on complex and bespoke systems to meet that task, not to mention a small fortune on sensor and data acquisition technology. They don’t even think twice about using teams of diagnostic systems engineers to interpret the data, as they are highly technical organisations by nature. However, these high barriers have meant that smaller business or those in the adjacent sectors have been simply locked out. Moreover, this legacy technology fundamentally won’t scale as the Industrial IoT sweeps through the manufacturing sector and millions of small and medium sized companies come on-line.
This is also though where IoT adoption is really exciting and will play a big role in lowering the barriers as it forces down hardware prices and democratises data access. So the question is more likely to be when and not if prognostic capability will be within reach. That leaves the data science discipline…and I’m tired of reading about the ‘sexiest job of the 21st century’. These guys are on $250k..ok, yes, I’m super jealous – but that’s not the point.
I’m not disputing that this discipline isn’t core (at least at the moment) but as the IoT evolves how can it possibly meet demand based on this dependency for human analysis? I believe that working out clever ways of automating many of the data science tasks is absolutely fundamental, even critical, to a successful IoT and especially data science heavy solutions such as the prognostics required to make predictive maintenance a reality. Without that, some really cool technology will remain only available to the ‘big boys’.
You won’t be surprised to hear that at Senseye, were developing a product to make a more accessible predictive maintenance solution that can scale with the Industrial IoT. We want to deliver the benefits to organisations that have previously considered machine prognostics as unaffordable or too difficult to implement and work with.
It’s certainly a challenge – not just the application of cutting edge technologies – but the barriers that need to be overcome as the industry evolves and embraces IoT. Security is a key one although while the technology largely exists to deal with it, the commercial and political obstacles remain. So what do you do about that? Cultural challenges are also evident as maintenance teams, familiar with their reactive approaches to dealing with breakdowns, resist improvements which would ultimately lead to a streamlining in maintenance staff numbers.
However, there are three further areas that we are closely tracking that have more far reaching consequences beyond our own product. Firstly, the need for edge (or fog, as Cisco prefers to call it) devices to which some of the data pre-processing tasks can be offloaded to. It’s simply not feasible to send potentially vast quantities of raw data to the Cloud every second – especially in remote environments that rely on mobile connectivity. We’re seeing an increasing level of activity in this space – especially from innovative start-ups.
Secondly, sensor and node costs must continue to drop dramatically. In the industrial space where high quality accelerometers are needed, these are still in the thousands of dollars (per device!) which is a significant barrier to any firm, not just smaller ones. If the IoT is to become as pervasive as the hype has led us to believe, these need to come in at well below $100.
And thirdly, the almost complete lack of interoperability standards and protocols is once again creating proprietary silos and fiefdoms – the very thing that the IoT is supposed to be breaking down. This lack of compatibility stretches across the technology stack, from communications to application metadata. Some initiatives, like Hypercat, while potentially promising aren’t yet being adopted at a scale or rate to achieve critical mass.
It’s certainly a fascinating and dynamic time to be involved inside what can only be described as the IoT bubble. I am in no doubt that we as a company will need to bend and weave as it all evolves.
We’re currently running free trials of our solution for select facilities so give me a shout if you’d like to be considered.