Tag Archives: IoT

Why doing IoT right is so difficult

If you’ve been following CES 2014 to any degree, you’ve already heard that the Internet of Things (IoT) and wearable devices are the next big thing. Experts cite that by a certain year it will be worth so many billions. I love IoT, I believe in it and I work on it, but I can tell you that the IoT revolution needs a lot more work. For years people have been working on IoT using Bluetooth, Wi-Fi, Zigbee, 6LowPan and anything else you can imagine. One of the considerations of IPv6 was because of the sheer number of IP addresses that would be needed, and that was a long time ago. The IoT revolution has been next year’s for many many years.

The reason I say this is that IoT assumes that smart devices make our life easier, when in fact at the moment they’re adding more complexity. Removing complexity is, ironically, very hard. IoT devices are technologically rich and by extension very complex. This complexity increases failures, and failures are frustrating to users.

Looking into why IoT is tough to do I see at least four factors – Installation Cost, Use Complexity, Increased maintenance as well as failure cost. To see how these factors play , consider the light switch and bulb in your home and the equivalent IoT device where you can control a lightbulb from your phone.

When you move into the house, the installation “cost” of having the light bulb and switch has been taken by a builder who (usually) knows what he’s doing.  You never have to play
around with the wires to get your bulb to work. On the other hand, since IoT devices are so new they require the consumer to install it. This can involve multiple devices like the gateway that allows your home network to communicate with the smart device. The sheer details and unknowns can cause even the most technology literate consumer to stay away. Check this light, check that connection, provide network name and password (oops, I lost it and now I have to spend a few hours). This device is marketed as making our lives easier, yet we have to pay a pretty steep installation “cost”. Clearly we’re now off to a bad start.

As for failure cost, have you ever had a light switch fail?  In all my years I’ve never heard of this. A light bulb may fail every few years  but the switch itself is 99.999% reliable or more. Replacing the lightbulb is a no brainer and doesn’t require any external information. When an IoT device fails, its not always obvious. If it does fail, the replacement cost can be almost as bad or even worse as the installation cost I discussed above. You might have to reprogram the new device and sometimes this is worse than a new installation because some parameters need to be checked. Perhaps you named your old part something and need to figure out how to reuse that. This is made even worse when manufacturers fail to consider this and don’t provide instructions. A standard lightbulb may be a few dollars, but a smart one will cost much more. As for all the silicon based devices, your Bluetooth might only be tested for use at 25% duty cycle for 5-7 years. Many other devices are not designed to operate for extended periods of time, and are much more susceptible to heat, moisture and the elements.

Perhaps the most stark issue is the use complexity. Lets say you managed to install it, or it was installed for you. Using these devices  might not always be as easy as they’re made to be. In order to turn on or off the light I would have to get my phone, which might be inconveniently placed in my pocked. I then have to open the phone with the PIN, find the App from among dozens others, open it, and then control the device. The amount spent doing this is not trivial. Never mind that any number of issues can pop up: No network service or the device is too far for the Bluetooth to connect, interference, low battery on our phone, etc.

The security aspect is something that is hard to get right. The more technologically complex the device is, the hardware it is to protect it. New features can easily introduce backdoors, allowing a hacker to easily open or close your doors. The low power aspect of many of the IoT devices is directly opposed to security and encryption. Encryption requires significant resources, but battery powered devices cannot always afford to perform encryption and other security schemes that can protect the device. What’s more, Denial of Service (DoS) attacks against these devices can be very effective since you’ll kill the battery and the monitoring capabilities are limited.

My discussion shows we still have a lot of work before IoT gets to where it should be. For people to actually use IoT enabled devices, its value must be very high. You have to be able to provide something that the customer currently can’t do and would really make a difference in their lives. Turning off lights at home might be important, but as humans we tend to ignore and forget what’s not in front of us or isn’t critically important. We would like to think that we’re good enough to turn off the lights we left at home, but the cost of electricity is not high enough to motivate everyone. I feel this scenario might be better left to automation. Second, you have to have solved all the issues that the user must have. If Real Estate depends on location, location, location, then IoT depends on Testing Testing and Testing. Until IoT is reliable enough, normal approaches (light switch) can’t be discarded.