I was talking to the leadership of an IoT company in Vancouver, who is producing sensors and what are called data loggers for the water industry. They attach sensors to a pipe, or a pump, to connect them to what's called the narrowband IoT network, which is constantly sending data back to the mothership. What happens now, or before today's IoT technology, the mothership would only notify a human when something broke and they'd have a 'big' problem to fix. The mothership would monitor sensors saying, "you're working, you're working, you're not working. Emergency."
Now, the mothership can say, "you're working, but you've slowed down, I wonder why you've slowed down", or "you're working, but you're working intermittently," similar to the way your car works. "You're working, but you're very lumpy," or "you're working, but you're very smelly." I need to know what's going on.
Your Car's IoT
When the engine light on your car's dashboard comes on, that is one machine telling you there is a problem. If you're fortunate enough to own a Tesla, a more advanced Ford, a Mercedes Benz, or a Daimler, your car is always talking to the dealership and the factory. The motherships are constantly saying, "These Ford F-150s are having this problem right now. We need to let our service departments know that these Ford-150 drivers need to bring their trucks in to fix the problem, before it breaks." It can also let the factory know there is a problem in a certain section of the assembly process.
Vancouver and IoT
Back to the water sensor company in Vancouver. The technology enables the mothership to know what happened over the last month, and the mothership will be able to tell you what's likely to happen in the next seven days. If you're a water pipe, and the water pipe is running parallel to a sewage pipes, what happens when the water pipe gets a huge amount of rain? It overflows into the sewage. What happens to the sewage when it gets too much water? It either backs up into people's basements or, God forbid, goes out into the rivers and ocean.
We've got problems in North Vancouver, Vancouver, Seattle, and Toronto, where sewage discharges into the waterways. This company uses the Internet of Things to connect pipes to to a computer using sensors, which says, "this pipe is not working very well right now and we can see that there's going to be more rain coming down in the next seven days. Let's send our maintenance crews to fix this pipe before it rains and causes a big mess."
Sewage is harmful, stressful, and difficult to clean up, but what if a tailings pond fails? The pond might overflow, tumbling toxic liquids into the ecosystem, food system and water system, now, you've got a bigger problem. These sensors will mitigate possible disasters by identifying areas that are not working at optimum levels.
There's a place for IoT and it's beginning to have a massive impact. We shouldn't dismiss it as marketing, or a trend, it is going to benefit every single thing we do. We've already had machine-to-machine connection for 10 years, maybe more, IoT and AI is only the next step.
Predictive Analytics and AI
Predictive analytics is what is going to happen in the future. Before, a machine was not able to say, "I'm going to get more of this work tomorrow" because it doesn't know what is going to happen, it's an inanimate object. However with artificial intelligence, which is looking backwards, the machine knows that for the last 10 days it constantly received work. It also knows that on Fridays, it can't cope because it receives too much work. Now it doesn't need somebody in a corduroy suit and brown pair of shoes to manage the workload. The machine can now say, "every day for the last month, my volumes have got bigger, and bizarrely, there's always a peek on a Thursday that I can't handle and I fall over on Friday." It can then look forward at what's happening in the field to discover, that the sales team sold another four contracts. It knows that the four contracts are going to the one machine, the workload has now gone up 15 times in volume, which the machine won't be able to handle. You don't need a person to figure that out. It is one ledger connected to salesforce.com, which triggers an event when a potential sale gets to an 80% gateway with a high likelihood of closing in the next month. Then the machine can tell the accounts department, the accounts department can tell the operations department, and the operations department can go out and buy another machine before they get slammed and yell, "it would have been really nice if you told us we needed to buy a new machine six months ago."
This is a wonderful illustration of predictive analytics at work. That's all it is.
Too Much Data on the Highway
The only problem is, data produces erroneous information. As Spam is to email, anomalies is too big data. For every one piece of useful data, you're getting tens of thousands of pieces of rubbish. The big trick at the moment is to subtract, remove, or scrap the anomaly, to clean the data, so you can analyze real data. This scraping should be done as near to the device as possible, so that you're not using the internet to transport rubbish.
Remember dial up, and those funny noises when it tried to connect to the internet. I was living in Europe at the time and we used to say, "oh, gosh, the Americans are online," because at about two o'clock in the afternoon, our internet would just slow down until it was like surfing in porridge. It is purely volume of traffic, nothing more than that. It's like a motorway with one car on it, or a motorway with 1000 cars. It's the same motorway, the same thickness, the same number of lanes, but the motorway going across Canada has one car, and the motorway going across the US has 1000 cars. The connectivity in the US is as good as Canada's, theirs is slowed down by what's called contention, which means the internet is slowing down purely because of volume. Now, if conservatively 60% of those cars do not need to be on the road, that is congestion. Take them off the road, and what needs to be on the road gets from A to B quicker. It's the same for Internet traffic, take the anomalies out, take the spam out, and you will get your email quicker. There's nothing worse than getting an invitation to a party, the day after it happened. You want to get your invitation two days before, so you can do something about it.
That's predictive analytics.
Let us know what you think about IoT and the future of technology in the comments section. Be a part of the conversation.
Have questions about IoT, Smart City, AI, or Big Data? Ask in the forum or in the comment section.
Nicholas Jeffery, a Vancouver based Smart City Expert, contributes regularly to IoT Economist. Nicholas pivoted in the new millennia to bring his business acumen and strategic thinking to bear in the technology, media and telecommunications markets, helping companies flesh out their growth strategies, business development and sales operations.
ความคิดเห็น