IoT is a Different Animal

In the past fifteen years the growth in connected devices has come from mobile phones.  As we near 8 billion cellular subscribers that engine is slowing down.  The next engine of growth of course will be the Internet of Things.  In Ericsson's 2015 Mobility Report the forecast is for 10 billion M2M devices to come online by 2021.  Just stage two of the rocket, right?  Not exactly.  The fact of the matter is that IoT is a different animal from consumer cellular, and only the companies that adapt to the new model will reap the benefits.  Let's look at the difference between mobile phones and the typical IoT device.

Bandwidth:  mobile phones are huge bandwidth hogs.  I'm sporting a 12GB plan for two phones at the moment and often it's not enough.  My monthly bill hoovers around $200 (including phone payment plan).  Most IoT devices use very little data.  Aeris' Neo plan gives you 750 kB of data per month.  SIGFOX gives you much less than that and has millions of devices connected to its network.  

Product Lifespan: smartphones are replaced every two years, but IoT devices such as cars, gas meters, and soil sensors are in the field for decades.  This is a problem because networks change and shut down in that time frame.  The alarm industry spent over $3B upgrading old 2G alarm panels to 3G.  There are still GM car owners fuming because their 1997 Cadillac with analog AMPS connectivity no longer works.

Connectivity Expectations:  consumers understand that when you take your phone to the boonies there might not be coverage.  It's accepted.  If you take your Wi-Fi only device on the streets the expectation is that it is no longer connected.  However if a soil sensor on a remote farm cannot get connectivity, rendering the high-priced smart farming solution useless, there's a much bigger price to pay.

Unit volume:  Apple sells 200m phones a year, split among a handful of models.  Even Asus sells 5m tablets a year.  Many IoT devices will be lucky to see volumes in the tens of thousands.  Lower unit volumes have implications with regards to the type of product development utilized.

What are the implications of these differences?  Firstly, carrier business models need to change.  Spending huge dollars on advertising and brick-and-mortar stores makes sense when each subscription nets $100/month and the buying habits of the target customer (you and I) are similar.  However, how do carriers sell to the myriad of applications and buying profiles in IoT?  The sales pitch to General Motors for connected cars is completely opposite to Edison Electric for smart meters, and the revenue per device is a lot less.  Furthermore, in the US carriers require devices to go through certification before they are allowed to connect to the network.  This is manageable for a couple dozen phones, but for the thousands of IoT devices out there?  Luckily the carriers seem to get it.  Rather than trying to own the customer as they do with cellular retail they partner with re-sellers, aggregators, and implementation partners.  Carriers have a 50% margin on their consumer cellular plans.  In other words, half of what you pay to Verizon goes to those brick and mortar stores, etc.  Carriers are smart to use a wholesale model for IoT and let smaller companies spend the money on sales, marketing, and support.  In fact, if carriers discount their data by 40% to wholesalers their margins are actually higher than the smartphone plans.

The second implication is that the way industrial IoT products are designed needs to change.  Lower unit volume means integrating a higher-price module to save in the fixed development and testing costs rather than spending $$$ to put wireless chips directly on the PC board.  IoT devices also need to be extremely flexible and support a wide range of networks to guarantee connectivity in all locations.  What networks are available along the route of an oil pipeline?  Wouldn't it be nice to have a pipeline sensor that is quickly adaptable to the networks available in a particular remote location?  And while we're at it, wouldn't it be nice if you could upgrade the connectivity module in the field, just like a battery, when the network shuts down?  

When I worked in automotive it was always fascinating to hear all the advice from friends about what should be done to improve cars.  Some of the suggestions were actually pretty good, because cars are an everyday object to most.  Unfortunately the consumer experience with wireless connectivity has very little relevance to IoT for all the reasons I just wrote about.  The danger for aspiring device manufacturers is to make connectivity decisions based on their consumer experience (e.g.- let's just put in WiFi, and maybe LTE for some models).  Eventually lessons are learned, but the process is painful.

By Alfred Tom on May 2, 2016