These three innovations have made IOT really smart, useful
By Gregg Sanders, head of digital transformation at NEC XON
IOT devices on their own are dumb. Cameras, sensors, vehicle trackers, smart light bulbs – they’re useful to a degree, but on their own, they’re not breaking ground in how we live our lives and run our businesses.
IOT becomes exponentially more useful the more devices you integrate, the more you connect the sensors together, the more you analyse the information and apply some structure and cognition to it.
For example, a connected lightbulb is a novelty and being able to whip out our cellphone and switch the lounge light on and off is great for impressing your friends and perhaps for annoying your wife, from anywhere in the world, but how useful is it really? But, combine that ability to control the light with information from another completely separate system, like your vehicle tracker, and now you have a bulb that can sense your approach and activate as you arrive or leave.
First you have to connect the bulb and the car. You have to integrate the information from both and correlate the data via an intelligent platform. You also have to create, digitise and automate a process to switch on the light when your vehicle approaches. That’s digital transformation. It changes the way we interact with devices and sensors to create true benefit.
You can apply the same logic to virtually any industry, from telecoms to mining, agriculture, transport and logistics, even financial services and retail. The benefits of connecting previously isolated and disparate systems quickly become apparent. But that is just the starting point. To see really revolutionary change, you need a transformational mindset.
The same technology can be used to power a cloud-based facial and behavioural analytics system for security companies in South Africa that identifies known criminals in real-time. It transforms security from reactive to proactive.
We used this same technology to change the way we respond to faults in the telecoms industry. We traditionally monitor all of the systems on a cell tower through separate platforms that alert operators staring at screens who must make judgment calls on what caused problems and how they should respond. It leaves a lot of room for human error and relies on the experience of the operators.
Instead, we now collate the monitored data into one database, add human resources information such as skills, training, expertise, shift information, responsibilities, vehicle tracking and more, and integrate that with the financial and asset management systems.
The results are crazy. The ability to correlate events at the tower show where faults first occurred for root cause analysis. It shows us what components and systems are affected downstream. We can instantly see which service provider is responsible for affected systems, and which of their people are best qualified and geographically located to respond. Then we are able to automate the process.
Now, when a fault occurs, we alert operations teams that automatically deploy the best qualified engineer who’s closest. It cut the time to repair faults by 300% and human error is massively down. As a bonus, a by-product of monitoring all of the equipment on a site is that we now maintain a living asset register for those sites.
We now also use cameras to improve security and safety at towers. We use the same cameras that recognise faces and vehicle registration numbers to check that employees wear the correct protective gear. We remotely manage physical gate locks so unrecognised people or people wearing incorrect equipment cannot access the sites. We can even alert workers or managers when they get too close to dangerous equipment.
Mines use facial recognition technology to quickly and accurately identify thousands of workers during shift changes so they can pick out illegal miners trying to blend into the throng. It also provides mine operators and managers with vital information of who is on the mine during emergencies.
The applications are nearly limitless. Farms have been made exponentially more efficient, pests are being controlled in countries from Europe to the Americas and across Africa, cattle herds are being better managed in South Africa, macadamia farmers can benefit from models predicting optimum times to prepare crops, school kids can get personalised lessons and homework, mines are curbing theft, managing operations and providing real-time executive insights, and heaps more.
One of the greatest enablers of these capabilities are the low power, long range communication technologies such as LoRa (which stands for long range), which are enabling sensors to send data great distances while using negligible amounts of power. GSM operates over long distances, but it’s expensive and consumes a lot of power. LoRa devices typically have a battery life of between three and seven, whereas a GSM device using GPS would only last a few days at best. It’s therefore not feasible to fit GSM devices to every head of cattle in a herd to monitor movement and grazing patterns, and you certainly wouldn’t add the functionality of ingested temperature sensors that indicate when a cow may be pregnant – as well as a raft of other information that helps farmers be more profitable and better manage their herds.
But LoRa technology is being used for that and more because it’s cheap as chips.
Farmers are moving away from a gut feel, experience-based intuitive management process that involved a lot of phone calls and radio messages to fact- and predictive modelling-based methods with real-time communications. Low-cost vehicle trackers fitted to the farm vehicles, employees carrying LoRa sensors, and LoRa-based soil moisture, nitrogen, phosphorous and potassium (NPK) sensors across the farm all connecting to a single, Internet connected platform providing real-time information on soil conditions and staff location and upcoming weather conditions that help determine when irrigation or soil additives are required. And farmers know where resources are at any given moment so they can apply the right resources to any given job.
It’s the intelligence of the platform that really makes the difference. But for various reasons, we’ve come to think of artificial intelligence (AI) and machine learning (ML) as a tool you drop in and it just magically learns everything it needs to know, then faultlessly gets on with the job. That’s most definitely not the case. We’ve learned that you need experts in the specific sector to pre-bake the right hypotheses, run the tests, verify results, tweak and tune and continue that iterative process until it’s ready to roll out. That’s how you make it truly effective.
What’s fascinating is that it turns out you can take that exact process and apply it to virtually any industry or sector and come away with great results. In education, for example, kids at private institutions almost never attend class without a tool like a laptop or iPad. Using the same intelligent IOT platform, that device becomes a conduit of useful information. When a teacher asks a question, it’s no longer up to that one kid who gets picked on to provide the answer, giving the teacher anecdotal evidence of the level of understanding of only one kid. Now every kid answers via the device, the platform gets all the results, and together with the teacher’s skills and experience, children can get personalised homework or lessons that fill in the gaps.
IOT has become really powerful because the devices are capable, they’re more connectable using cheap technologies such as LoRa that use little power and operate over long ranges. We can integrate all the devices now through a common platform, then add layers of intelligence to automate a lot of what used to be manual stuff. And that’s why IOT isn’t just dumb technology anymore.