This post is the third in a series of seven chapters intended to help individuals and organisations on the journey to deploying a successful Internet of Things (IoT) transformation. The first chapter covered preparation, the second covered how to choose the right hardware. This particular chapter covers the Information challenges. If it’s helpful to go back to the beginning and read in sequence you can start with an introduction to this series here, along with a list of the other topic areas you will need to assess on your journey.
A key objective of the digitisation process is to collect data using devices/sensors that you then put through analytics to deliver business benefits. Depending on your particular IoT project, you may be collecting a small or large amount of Data. Either way, you need to consider the 5 “V”s of Big Data: Volume, Velocity, Variety, Variability, Veracity. I will also touch on two more “V”s: Visualisation further below in this blog and Value in an upcoming blog.
The five key focus areas for data/information are:
- Volume: For example, while an ultrasonic sensor reading could result in a small data packet (e.g. <50KB), an image based sensor could deliver large picture or video files of the size 1000KB and above. Estimating the volume of data being generated is an important task
- Velocity or the “duty cycle”: Data will need to be uploaded at certain intervals of time. This could range from real time (always on) to once every 24 hours or more and anything in between. Ask your vendor for the number of connections per battery lifetime and check how often you need to collect this data from the field.
- Variety if you are using different sources of data from different devices vendors, or different systems then you need to assess how to derive value from each source, individually and collectively. Most of the time your new IoT solution is part of your overall picture…make sure it fits and you can derive business value by combining with other data as you need to extract full value from your investment.
- Variability: Is the data reliable when it comes to availability and/or interval of reporting? Does it describe the event reported in a correct way? or in the event that data is variable, assess whether it is just noise and should be disregarded or if you must interpret each set of data differently.
- Veracity or the accuracy of the data. If automated decisions are made based on this data, you need to be sure that the data is always reliable, not just sometimes.
The five Vs could have various impacts on your IoT project such as storage, connectivity, infrastructure, cost, business continuity, choice of technology, choice of supplier to name but a few.
Therefore, when you choose your solution, make sure you understand: what data is being collected? Who owns the data? With whom the data is being shared? How long is it retained? Who is it retained by? Who has access to the data and for what purpose? How secure is the data in question? What measures are taken to ensure its security? And how accessible is your data to you and your organisation and in what format?
Trust is a fundamental issue in selecting your solution (or ideal solution provider) and these questions will get you closer to the right answers. Ultimately this is your data, you own it and you must decide how and who you share it with, what risks you can tolerate and what you cannot. Trust extends beyond the solution into the supplier and the people you are dealing with and this is something I will cover further in a future blog.
What is the business value of the data generated by your IoT solution? The answer will be found in the analytics and visualisation. Many businesses are creating new business models thanks to the power of data. When they are able to use real-time or near real-time data to anticipate and predict behaviours of devices or even customers, this can transform commercial models.
While your IoT solution will generate large quantities of data, analytics will allow you to discover the hidden value and visualisation is the art and science of making such value easily understood.
You need to assess how much your preferred solution allows you to manipulate the data to create visual intelligent reports, match your operations and identify the value to your business. It’s also essential that different kinds of users get the outputs they need. Does this solution simply produce raw data that requires you to integrate into other tools and do it yourself? Be clear on what you are buying into. Are you paying to collect data and then pay someone else to derive value from this data using a different solution? Or are you paying once to get value aligned with your use case?
What insights and knowledge do the analytics capability of this solution offer? Does it go as far as predictive analytics, allowing you to proactively take decisions and actions, or does it provide descriptive analysis, exploring and confirming analysis? The latter allows you to process large amounts of data quickly and the other changes the dimension of your operation into a proactive stance as opposed to a reactive one.
If no significant analytics are offered you need to separately assess the layering of another Business Intelligence (BI) or analytics engine and understand the ease and cost of this additional layer.
Don’t be fooled by an attractive dashboard presenting basic information, but look much deeper into the value you can derive from the data.
As an example, predictive asset management applications are delivering huge value today. The ability to predict breakdowns and deploy assets accordingly versus reacting after a breakdown takes place is a major change in the way business is being conducted. This not only impacts on your operational efficiency but enables the streamlining of your entire business including the supply chain, order fulfilment, maintenance workforce distribution and sizing, customer charging model, service level agreement, etc.
Data analytics and visualisation often result in new opportunities that will require cultural shifts to reap the benefit otherwise you risk breaking your traditional operations, processes and methods.
If you are reading the news or you follow this topic, you will know that many IoT devices are not fully secure and can be hacked. The reason for this is a mixed bag of rushing to push products into a hyped market, bad judgement, cost control, or early release of technology with security planned for later product updates.
As you investigate your security requirements, consider the points of failures or vulnerabilities. If you have 1000 sensors then you may have 1000+ points of vulnerability. Trust, reliability and integrity of the device and the data generated are paramount. Additionally the data transport, collection and storage must be safeguarded from any alteration or misuse, whilst also not breaking any privacy guidelines.
If I have a security system that is transmitting video or pictures from different points in my house, I want to make sure that no hacker can access these cameras physically or electronically, that the data transmitted is protected and will not be interrupted and copied on the way, that it cannot be tampered with or altered, and most importantly that is only used for the sole purpose that I approved and visible only to me. This applies in principle to any device installed anywhere! Wearables, sensors measuring tank fill levels, measuring pollution, etc… after all I want to be confident that any action I take, is based on real data and not fake data.
So, the transfer of data and the source of the data (the sensor) today is most likely vulnerable to hacking and other security attacks. Privacy (if the device holds information people do not want to reveal) is also a major concern.
While I agree all security threats are not to be tolerated, some are more urgent than others. Some are not a genuine privacy issue, while others are more serious and need resolving. Some things to consider include anonymous data versus well-identified ownership of data, critical device function versus non-critical function. Authorization, encryption, safety and well-being are all critical… but if you have a sensor measuring how damp the soil is in order to switch on your sprinklers and water your garden, security is less critical (although you don’t want anyone hacking them and flooding your garden!)
However, this does not mean that you shouldn’t make sensors reliable and secure but that you must decide if you can live with the associated risks. To make such a decision you need to be clear on the risks you face by adopting a non-secure solution as well as the delays incurred until it can be made secure. Can you afford to wait? What are the risks of doing nothing compared to those of an IoT solution that is not fully secure… the probable vs the possible!
Privacy, Data protection and security are 3 critical requirements. Be clear on what you can and cannot tolerate and accordingly investigate and decide if it meets your criteria.
Always play what-if scenarios to assess the outcomes and impact on your business and customers and ensure your solution is designed to give you peace of mind and/or in the case of breach, is offering you an acceptable way to isolate, manage and permanently eradicate the problem.