Neat hacks avoid the need for lots of "stuff", by bootstrapping existing data and frameworks into rich new IoT datasets.
A depressing aspect of developing IoT systems - at least, in the past - was just how much "stuff" you had to put together just to deliver a complete service capable of doing anything useful: embedded software, electronics hardware, communications, cloud services, your application... and even when you've got all those pieces assembled, there's so much that could go wrong that the end result often isn't very reliable. So... what if there was a shortcut?
A few companies have spotted some cunning shortcuts to bypass all this hassle by simply repurposing existing infrastructure.
Turning analogue data into digital events
Audio Analytics can take that microphones (e.g. in a smart assistant in your home) and make it recognise not only your voice, but also many other ambient sounds, emitting a stream of events such as "smoke alarm sounding" or "glass breakage".
Vivacity can take a standard CCTV camera used for crime prevention and make it able to measure how crowded a street or railway platform is, measure the flow of pedestrians or vehicles, or recognise undesirable situations, and again emit a stream of events which can be used by other machines as part of an automation process.
Building systems to perform these tasks from scratch would require a lot of effort, but these companies product software which can use reuse the existing investment in hardware and communications.
These are examples of the exciting new area of "edge computing", of particular relevance to IoT where processing is done inside the edge devices rather than in the cloud.
An extreme version of this "reuse something expensive" theme is conducted by Orbital Insight and Descartes Labs, using cloud-based image-recognition on high resolution satellite images to automatically count parking lot occupancy, wind turbine installation, etc. with no new infrastructure.
Giving passive things an identity
In the world of product identity, there has long been tension between low-cost passive approaches like barcodes and RFID (where the "thing" is dead except when near a scanner) and high-cost active approaches such as Bluetooth and WiFi, LoRAWAN, SigFox, etc. (where the presence of a battery in the device allows it to spontaneously communicate, and at much larger distances).
Evrythng has found a way to bridge the divide, by combining their digital printing with smartphone apps. Now your whisky bottle or t-shirt can tell customers their provenance (e.g. proving that they're not counterfeit, and not made with slave labour, etc.) and also provide other avenues for vendors to track how and where their customers engage with their product... all without needing any pesky batteries on the device.
Borrowing someone else's network
Finally, the cleverest idea we've recently heard is from Micha Beneloiel of Nodle. He has a track record in peer-to-peer networks, such as FireChat - a P2P messaging application that works even when they're no internet available, like at the Burning Man festival (or, more controversially, if your local government has shut down the comms network to suppress dissent).
The Bluetooth 4.0 specification enables devices such as heart monitors or weighing scales to communicate small, infrequent data packets over a few metres to smartphones using extremely low power. Conceptually, it's similar to LoRAWAN or SigFox but without the range.
There are many IoT applications which only need to take occasional readings - like the product temperature in a fridge or the air quality of a building - or even just validating that the sensor is still within range of the phone, for security purposes. Often, the individual readings aren't particularly mission-critical or sensitive (if a few go missing or get delayed, it's not the end of the world), so one can use a "drive-by" paradigm to collect the data: as the smartphone travels around the world with its user, whenever it happens to be in range of a sensor, it grabs the data.
Nodle's cunning plan combines these two ideas. if you're a developer creating, say, a luggage tag application, then instead of using your app using Bluetooth directly to talk to the tags, it talks via Nodle's software, which in turn talks to the tag using Bluetooth. Initially so what?
The genius comes when lots of apps on lots of phones start running the Nodle software (and only a few months after launch there are already over a million!). As well as serving their own devices, each user is also unwittingly driving-by lots of other people's devices too. The Nodle software acts as a sort of global distributed lost property office - if your phone sees something it doesn't recognise, it posts it to Nodle to see if anyone else cares. And if someone else's phone sees one of your tags, Nodle will tell your app.
Peer-to-peer applications like this have the wonderful quality that as the number of users and devices increases they actually work better, unlike centralised resources such as 4G and the cloud which come under increasing strain.
Does any of this deserve to be called IoT?
Or perhaps one of the precursor names such as Ubiquitous Computing? We think so, because these are all examples of enabling machines to understand and automatically interact with the real word, to deliver new services at low cost. What marks out these innovators is that the cost has already been paid by someone else.