Ten things that need sorting out before the IoT takes off

Entrepreneurs and analysts met recently to discuss the barriers in the way of making the connected future work

The Internet of Things (IoT) faces a number of challenges, some quite formidable, if its potential as a force for progress is to be realised. A group of entrepreneurs and analysts met up at a roundtable event in London recently to thrash out the issues.

1) You can't solve a problem if you can't define it

The IoT is coming. Or perhaps it's already arrived. Or maybe it's been here for years. It really just depends on who you talk to.

The consumer IoT of smart thermostats and wearables may be picking up, with driverless cars waiting just around the corner, but the industrial IoT, using sensors for predictive maintenance and to manage production lines, has been with us for two decades or more.

Those present at the roundtable event, which was organised by database firm Basho, were largely in agreement that "real IoT" implies cross-matching data from many different sources rather than simply using sensor networks for a pre-defined project.

"I wouldn't call what we do IoT," said Declan Caulfield of his firm Intellicore's work with Formula E electric car racing.

"It's too specific and outside that context it wouldn't be that useful, so I think it's about having sensors that have the capability of having a network stack attached to them."

"We should be talking about IoT in terms of new opportunity," said Manu Marchal, EMEA managing director at Basho.

"So with fleet management all the cars need to be tracked,which means they have to have sensors, and those sensors can tell you the temperature and that data can be sold to weather companies to give a fine-grained view of temperature on the road. That's an example of a real-world IoT use case."

So let's settle on a definition of taking data from many disparate sources, processing it on a general-purpose platform and then extending the value of that data by using it to act on the physical world.

2) Silos of things

Following directly from this is the need for insight to be released from the narrow confines of sector-specific use cases where much of it resides right now. Most sensors are used for a defined purpose, for example monitoring a golfer's swing, measuring wear on a drill bit, tracking traffic congestion and pollution, but as the fleet car example shows, the real added value is to be had by combining these sources in novel ways.

"We need to liberate the entrepreneurs and allow innovation on top," said Dave Page, CEO of analytics-as-a-service firm Actual Experience.

"At the moment it's centred on connectivity management, which is quite vertically stacked. There's a need to bring in a horizontal management that's standards based, an IoT IP layer if you like, an open place where these businesses can build and genuinely start to leverage the scale of the data and the sensors and do so at marginal cost."

In other words, the IoT needs to become an open platform like the internet rather than the set of siloed operations that it is at present. This will democratise the IoT making it accessible to business people, in the same way that cloud has allowed marketing departments to be responsible for CRM systems - once very much the preserve of IT.

3) Too many standards

To allow for increased interoperability the industry must consolidate around a number of secure open standards, the way the internet did with the World Wide Web Consortium (W3C).

"The IoT has evolved with lots of different proprietary standards and there's a latency overhead because we have to parse and serialise the data as it comes in," said Intellicore's Caulfield.

A bigger problem with multiple competing connectivity standards is that it is hard for security vendors to know which ones to focus on, said Page.

"There's an issue with proprietary hardware and software and operating systems because security companies have to hedge their bets about who they integrate with."

Ten things that need sorting out before the IoT takes off

Entrepreneurs and analysts met recently to discuss the barriers in the way of making the connected future work

4) More open source

It is important that the IoT standards are open source, argued Ian Hughes of analysts 451 Research.

"There's a company like GE for example that has made the whole Predix [industrial IoT] platform open source because they felt that that way anybody can integrate with us and the more open source we use the better the security gets. An open source platform may develop a better trust."

"Open source is fundamental to building in horizontal layers," agreed Page, reiterating his point that the platform must be accessible to all.

"Business people don't care about technologies they care about a few key business interests."

5) Joined-up thinking

Outside of obvious life and death scenarios, current use cases tend to emphasise cost and convenience over safety and security. This needs to change, said Hughes, who urged manufacturers to learn from best practice in the nuclear industry.

"The reason nuclear works is that they are very careful about it. But vendors seem to be a few steps behind so you get [insecure] whimsy smart-home products, or car manufacturers that stick an API on the air conditioner just because they can. There is a clash of worlds where things can happen but the expertise is elsewhere."

"There are very few real life examples about what's going on and very little published best practice," said Tony Cripps of Current Analysis.

"There are at least tens of thousands of IoT deployments and billions of devices already but we're looking around and saying where are these things? It's genuinely happening but we know very little about what's happening and how it's being done.

"The problem is these things are being done but they're not being called IoT; they are being called something specific to the industry they're in," he added.

6) Humans don't scale

The only way that the IoT can scale up economically is to automate everything, leaving people out of the equation.

"We're going from the world where there are thousands of things to a place with billions of things and we're riding into a problem," said Page.

"Manageability is simply an enormous problem and unless we solve the operational complexity of keeping things doing what they're supposed to be doing it's going to be absolutely prohibitive. It has to be frictionless, zero touch, no human intervention."

Page continued: "If you have to do a truck roll to fix a home set-top box you just eroded your margin. Now multiply that by all the homes in a street. The avoidance of human intervention is the only way you can crack that problem."

Wireless over-the-air updates are needed but these have often been error prone so far, he said.

"The IoT is dead in the water if you can't manage it without human intervention. No one's even talking about that much."

Ten things that need sorting out before the IoT takes off

Entrepreneurs and analysts met recently to discuss the barriers in the way of making the connected future work

7) Intellectual property and privacy rights

For the IoT to connect disparate resources usefully, those resources must be willing to share some of their data. By and large this is not happening at present as that data is tied up with IP and share value. For example, it would be useful scientifically to test the health benefits claimed by wearables manufacturers by measuring how many users have actually abandoned their wearable and confined it to the back of a drawer, but vendors like FitBit are very reluctant to release such information.

It is not just commercially confidential data but personally identifiable information (PII) too. Huge benefits might be realised by opening up health data for analysis, but this must be done with regard to its sensitivity, which raises issues of trust about how the data has been anonymised and aggregated and to whom, ultimately, the benefits will accrue.

NHS 2020 is a plan to digitise the NHS with the aim of making it more responsive and personalised and ultimately to extend healthcare into people's homes.

"A huge element is using wearables, smart devices and sensors to help pre-empt disease or provide and extend healthcare to the home, but that will be not possible if you don't have a connectivity between all these devices and the internet and the backbone services," said Basho's Marchal.

"It requires a more horizontal way of working in this industry currently doesn't exist."

But how far should this connectivity go? How much do we want institutions monitoring us in our own homes?

"Perhaps you could have a one way API so the NHS could see if someone's been sitting in front of a smart TV for a week and can then check up on that person, but with the data still NHS owned," suggested Alexandra Deschamps-Sonsino, founder of Designswarm.

8) Cultural acceptance

The IoT means massively increased data sharing and this will only be acceptable if people get something back and if systems are reliable and secure and their operations transparent. Already social media has had an enormous impact on how we interact and the IoT will increase the network effect by orders of magnitude.

"Will it be like social media where it explodes and everyone gets on board, then everyone gets worried because someone judges you by what your stupid friend did, and then you might not be able to get insurance just because of who you know?" asked Hughes.

Then there is the issue of reliability and tenability. Things must "just work" in the way they're supposed to or they'll be rejected, argued Marchal.

"At the moment the smart home is only suitable for single people because if you are in a relationship you'll drive each other crazy. They don't really work well yet."

Think too of the disruption to planning and insurance that will be caused when it is easier and cheaper to order a driverless car to come to you than to own a vehicle. These changes will take time to work through.

Ten things that need sorting out before the IoT takes off

Entrepreneurs and analysts met recently to discuss the barriers in the way of making the connected future work

9) The politics of data ownership

A smart city, the classic IoT use case, is a complex collection of thousands of projects by hundreds of vendors and other organisations. As such it likely requires the services of a large integrator to glue all the bits together. But to whom should the masses of potentially sensitive data about citizens and their interactions accrue - to the integrator, the city authorities or the individual projects?

That sort of data would offer the owner a huge amount of political power, enough to profoundly alter the democratic process. Do we want that power in the hands of an unaccountable corporation? How will personal data be screened out? Will citizens be able to opt out of being observed?

The roundtable seemed unsure of the answer to this complex issue, with some vague talk about cryptography, data masking and signal jamming. Certainly opting out seems unlikely.

"To a certain point it's inescapable. It's passive so as you pass through it detects you. It does throw some security issues around passive versus active IoT," said Caulfield.

10) Security

Last but certainly not least is the vexed question of ensuring data security. This has proved very difficult on the internet and will be even harder on the IoT unless some serious thought is given to the problem.

A number of recent DDoS attacks have been carried out using botnets of hacked security cameras and routers. And the prospect of the sort of malware that can play havoc on the web breaking out into the physical world of connected cars and national infrastructure is truly frightening. Stories of hacked automobiles, coffee machines and baby monitors are already a staple of the tech press and the IoT has barely got off the ground yet.