4 minute read 8 Apr 2019
Young woman using laptop paying with credit card

How data could undermine trust in IoT implementations

4 minute read 8 Apr 2019

Show resources

Trust is a core component in platforms of all kinds, from the internet to blockchain. But IoT data calls trust into question.

We might not know it, but we face a major problem with IoT every day: trust.

For the internet, with explosion of users and the increasing numbers of malevolent agents – hackers, spammers, malware bots – make cybersecurity the number one internet-related concern for organizations.

The internet was built for human communication, via a connection protocol across the world wide web. But in IoT, machines are the users. And they are generating terabytes and petabytes of data every day. 

While cybersecurity, confidentiality and privacy are prime concerns for organizations and internet users, for IoT the most important challenge is the integrity of the data it generates. 

Who says IoT data is correct?

What does integrity mean in this context? Simply having assurance that data is correct and unadulterated. 

As a parallel, think about a bank transaction: if you transfer €1,000 to someone, not only will the recipient confirm they have received it, but the bank uses sophisticated IT tools and logical instruments to make and record the payment. The bank is the trusted third party.

But in IoT, data is the trust of decision-making systems – and there is no human agent involved to confirm anything. In some circumstances, this could initiate false or unnecessary procedures. For example, if you put a smart thermometer under your arm and it came back with a reading of 40 degrees, you would understandably panic; however, you would feel very ill already. If you didn’t feel unwell, you would reason that the thermometer has made a mistake. But if a smart thermometer transmitted your temperature data to the cloud, it may trigger a call to the emergency services and, before you know it, an ambulance could be sent to your home.

To take another example: some cars are fitted with an SOS button in the event of a crash, and this signal can be triggered by a collision. It could also be accidentally triggered if, say, someone crashed a shopping trolley into your car at the supermarket. In this scenario, you could intervene to stop the SOS call. 

But in the very near future, millions of sensors and machines will take readings of all kinds of parameters, and the ecosystems and decision-making engines they feed will take the accuracy of the data for granted. So, in an IoT ecosystem there is a need for trust to achieve data integrity. 

A second element of integrity is whether to trust the algorithms in devices and ecosystems.

Testing is essential: but if there are a million devices in an IoT ecosystem, all the trillions of connections between them need to be tested. 

How is it possible to trust an IoT ecosystem based on a test of just one or two devices within it?

As artificial intelligence (AI) becomes a core component in decision-making, the integrity of the IoT ecosystem becomes even more acute. 

What are the consequences?

The gravity of this challenge varies according the IoT implementation. Quality control slippage in manufacturing from faulty temperature sensors could result in expensive losses and wasted time. But in a fully automated hospital, a connected car or critical infrastructure such as refineries, there could be waves of losses, including power blackouts and loss of life.

In predictive maintenance implementations, data about everything from a train’s brakes to aircraft engine turbines is detected by sensors and shared with a digital twin in the cloud. But what if a sensor suddenly malfunctions, that data is no longer accurate, resulting in a test not being performed correctly? 

The potential consequences are incredibly serious. Which is why we are working on a solution to confirm the integrity of data, algorithms and testing within IoT ecosystems. You can learn more about this in the next instalment. 


IoT sensors are generating terabytes of data daily, which can tell us about everything from our health to the quality of manufactured goods as they roll off the production line. Can those sensors, the data they generate and the decision-making ecosystems be trusted to be accurate all the time?

About this article