I'm with you on the hype aspect, but it's hard to say how many "sensors" will exist. The vast majority will most likely NOT be directly connected to the Internet. They will instead be embedded in any number of products.
You need sensors when you build automated control systems, and especially so if you're after greater operational efficiency. So for sure, that's on the increase, in factories, planes, trains, cars, trucks, buses, amusement parks, movie theaters, elevators, HVAC systems, kitchen appliances, and on and on and on. In some countries, even bathrooms have their own set of sensing equipment.
In most cases, for Internet access to these systems, you wouldn't really need unique IP addresses for each sensor.
Jack, well done on an informative, fact based analysis. More sensors isn't necessarily a good thing. Go on to Weather underground and find how many weather stations are available for your area. Most are run by "non engineering" types and as such , the accuracy is questionable. Just the other day, I was seeing a report of rain and a very low temp when I switched to a nearby airport sensor (maintained by professionals) it was much more believeable. If you look at most temperature sensors for consumers, try to find an accuracy spec. So the internet of things will likely just flood us with more useless data.
Sure there's hype. Bu maybe there's some truth to those claims:
In embedded.com Jack continues to talk about $15T over the rest of the decade being 4% of GDP. I can see how new systems with sensors(coupled with AI) creating savings/value greater than 4% everywhere in the economy, be it robots, labs-on-a-chip or sensor networks or plain old smartphones(altough i';m not sure it would appen that fast)
And i can even see some of those getting to the poor , the way smartphones got there:for example, assuming the lab-on-chip vision is true, treating poor people in places with lack of medical expertise will be possible and probably cheaper using lab-on-chip devices.
And for many industries , the 4% savings/value number is pretty low. Putting sensors in garbage cans saves more than 30% of transport costs(onavo.com). Labs-on-chip could really transform healthhcare. Smart lighting probably saves much more than 4% in total lifetime costs of a bulb.
But sure , that's just savings - it speaks nothing about how much of the value would go towards the electronics.
I think out of this "noise" would come opportunities for the creation of innovative algorithms to extract useful information. In my view, one key to enabling this is standardization of information formats and sensor IDs and capabilities. So when I hear about billions or even a trillion connected sensors, I say bring-it-on!
True, today's car is packet with sensors, but I don't necessary want to know your engine's per-cylinder gas-air mixture distribution status over the Internet. Will that kind of information useful to anybody? (Car manufacture? Gas stand? Environmental agency?) It is hard to say what is useful and what is useless. Somebody may hit gold mine (or "Killer Appliation") from pile of data, or can be just useless meaningless digital garbage. I think "Trillion Sensors by 2020" is optimistic outlook that hoping somebody find gold mine, but nobody know what it is (or is it real) at this point.
I can see how, possibly, 200 billion sensors a year would come about, but ironically that would not lead to 1 trillion active sensors after 5 years;
Sensors in consumables.
For example, perishable food, sticking a cheap sensor on the package at the factory. Being able to track temperatures during transport could lead to later "consume before dates" and thus less waste. (In the EU alone, enough food to feed Spain is discarded everyday).
These sensors could communicate RFID like with the shop inventory system when sitting on the shelf, making them connected.
However, being throwaway, they would not accumulate to 1 trillion connected after 5 years...
So first of all, the trillion sensors is by 2050, not in the next 5 years. Only talking 50 billion devices connected to the internet by 2020. (3-6 billion will be people, add all the servers in and maybe you are left with half that goes to sensors. Just spitballing here: 25 billion sensors in the next 6 years.)
Second, my phone has 13 sensors in it already. It's one phone out of 6 Billion in the world ... do the math. (Maybe the 50 billion by 2020 is low?)
RFIDs on consumables shouldn't be included, as they aren't on the internet ... maybe the readers, but that would be orders of magnitude fewer.
Cars will definitely be on the internet, and they have 10's, maybe 100's of sensors. (Although, a good cyber-security requirement would be to leave them OFF the internet.) Again, over a Billion cars in the world, so that multiplies through rather quickly.
Smart Meters, weather stations, air quality sensors, etc., etc. are all coming down in cost and going up radically in deployments. 50 years ago the US Weather Service struggled to get a weather station in every city. Now days, every neighborhood school has a station on the internet.
Finally, don't forget where we are on the Hype Cycle:
|...........____ <--- Plateau of Productivity
|........./ <--- Slope of enlightenment
|...^.../ <---- Peak of Inflated Expectations
|/.....v <--- Tough of Disillusionment
/<-------------- Trigger Technology (YOU ARE HERE)