Sunday, 19 June 2016

Where IOT is Going Wrong

Where is IOT Going Wrong ?
 We’ve all seen the predictions. Apparently IOT will be bigger than sliced bread, Ben Hur and everything in between. My earlier blog referred you to the hype curve and explained some realities but let me expand on why I think many companies will go broke or simply fizzle out. It boils down to this:

No-one is thinking clearly about the problems IOT should really be solving and how.
It comes back to the definition of IOT: what is it really? Once you strip out web sites, remote applications, phones, tablets, email and server to server functionality you are left with the “things” that IOT is really referring to - remote monitoring of small or single purpose devices. And this is where the concentration of IOT development is taking place. Let me explain where I believe it has all gone badly astray. There are four key problems I would like to explore with you:
·         An over emphasis on the value of LoRa
·         An inadequate understanding or visibility of true cost
·         Little genuine perception of IOT framework complexity
·         A complete lack of genuine applications making sense of it all
1.            The Inadequacy of LoRa
I almost wish no-one had thought of this: low powered WiFi devices that send tiny bits of data on an infrequent basis. Almost useless. 12 byte messages and 144 per day limits just won’t cut it for 90% of the remote monitoring requirements out there. There is no alert immediacy possible, no genuine data trending possible and no aggregation of multiple sensors possible. We can’t send commands back to the device or sensor and we can’t put intelligence at the point of collection. Even in agricultural scenarios, there are very few practical uses for this. You might think soil moisture or silo levels might be a viable sensor usage for LoRa but this only makes sense if you have software to aggregate and manage the readings off line. But it is useless for weather stations, water monitoring, hot house monitoring, stock monitoring and almost everything else. It is no good for building event monitoring, patient and medical alerts and just about any machine monitoring. LoRa really isn’t a useful data transmission methodology except in all but a very few specialist areas. Keep clear of it unless you really know how you are going to deal with the lack of data flow. Please make sure you know what LoRa solutions will NOT do for you.
2.            Nobody Told Me It Would Cost this Much!
The cost of solutions I see being constructed and pedalled over the web is going to horrify people and scare off any genuine take-up. Part of this is because software vendors are wrapping up solutions independently to hardware vendors. This always means two people need a cut and you will have two sides of the equation to deal with. But, did they tell you what the monthly costs will be? There are going to be Telco bandwidth charges and then Azure (or Amazon or Google or…) framework charges. Just dealing with data is fraught. 3G/4G data costs are still expensive the world over and if your solution doesn’t act nicely over the link, you will be up for data costs you didn’t expect. If you start from the software side (someone selling you a data solution), make sure you understand the cost of sensors and probes. Most websites will not detail all this up front because they know you will panic when you see the numbers.
3.            The Complexity is not for the Feint Hearted
If you are doing this yourself then be ready for the complexity and cost of collecting data for display. Making 24/7 auto scalable solutions is not easy on any of the Big Three platforms, despite the spin and hype they throw at you. For what it’s worth, only the Microsoft platform comes close to offering an API approach but this aspect of the industry is very immature. The platforms are not stable and their APIs have significant bugs. Your client is not going to listen to you telling him that Microsoft broke your connection last night and that’s why you didn’t get 12 hours of important data. The other complexity is having the software cooperate with the hardware in order to yield meaningful stats and alerts. Unfortunately, it requires a symbiotic relationship between the managing and reporting software and the hardware offering up the data. You simply will not be able to pick and choose from different vendors. This will dramatically complicate the design and installation of IOT systems.
4.            Where are the Apps!
This is my big question. Right now, they just aren’t out there. Basically we’ve put the cart before the horse. We need to know what we want and why before we go around spending millions on solutions to nothing. Perhaps if you want refrigeration alarm panels you can find something but what about all those supposed agricultural solutions? Where are they? What about those alarm monitors? There’s basically nothing out there yet so it is no point getting excited about a data collection capability until you have something that can deal with that data. Sounds simple, doesn’t it? Forget about the glitzy examples. Apparently there is an app out there that analyses driving behaviour and rates you against your friends. Duh! Well that’s going to save industry millions …not! So then the problem is that no-one is building practical apps that are solving actual problems and so saving people money or risk. This is a bit of a problem.
In summary, the IOT industry is busy tearing itself apart to be the first out there with really poor solutions. The customers are waiting for someone to build meaningful apps and no-one wants to tell the truth about cost less they scare them all away. We’ve got a long way to go…

Geoff Schaller

No comments:

Post a comment