I know I will draw calls of “dinosaur” or worse, but I see many parallels to previous technologies here, rebranded under this generic new name, the ‘Internet of Things’. My own experiences with this sort of technology actually pre-dates the Internet by a couple of years, and was in the UK Utilities Industry. It was known as Telemetry or SCADA (System Control and Data Acquisition). Others will have had similar experiences, maybe through alternative avenues such as Industrial Automation or Process Engineering.

Nowadays we take high bandwidth networks and powerful computers for granted, but not that long back this was not the case. Modems were used to squeeze data down networks that were designed for voice traffic and so were relatively low bandwidth. Computers to do the back-end processing just weren’t as powerful as today and were much larger. Sensors were often connected to computers by VERY slow connections. For Telemetry systems, 9,600 baud was considered fast and connections were certainly not reliable, as many were radio link based. Have a removal van park in front of your directional antenna for a few hours and you could kiss goodbye to your flowmeter readings. 

For this reason, ‘store and forward’ messaging was used – a bit like SMS, where if the recipient cannot be reached, then data is stored in some network node until it can be successfully transmitted. This made the ‘removal van’ scenario OK, as long as it didn’t stay too long, or else the data buffers would over write, and data would be lost.

For connecting up the sensors, ‘Outstations’ as they were then (and maybe still are) called were developed. These were based on popular microprocessors and microcontrollers, such as Motorola’s 6800 and 68000 series. Open the outstation up, and you could pretty easily deduce the architecture – there’s the processor with its crystal, here is RAM and EPROM, some ADCs and DACs just there, a bank of opto-isolators and output relays next to them, and usually a dirty great mobile vehicle radio hooked up with a modem, to act as a radio data connection. Programming these outstations was via RS232 link with either via Psion Organisers, or a laptop.

A typical outstation – did the job, but not pretty

The sensors themselves were large, ruggedized and pretty expensive things; a pressure transducer for example could set you back near on three hundred pounds, a helix flowmeter over five hundred pounds. Specialist sensors such as chlorine residual indicator transmitters (“CRIT”) well over a thousand pounds.

Industrial Pressure Transducer

These things also needed periodic recalibration, as they were based on mechanical or chemical action; a strain gauge in a pressure transducer, spinning magnets in helix flowmeters and the current inducing effect of chlorinated water upon the copper and platinum electrodes in a CRIT. Nearly always, they were single sensors with single purposes. 

These sensors pretty much all produced different outputs; some gave a pulse for every 100 litres of water that had passed them, some gave analogue signals, some gave non-linear signals. Almost universally there would be some sort of box of tricks attached to each sensor that would take its output signal and convert it into a linear ‘live zero’ 4-20mA signal, that the outstation could accept and pass on.

And battery power? Solar Power, Wind Power? Forget it – Sensors and their control boxes all consumed a fair bit of power (just for the 4-20mA signal alone) so pretty much everything had to be mains powered, as even a car battery wouldn't power them and the outstation for long and solar cells were simply too expensive.

Security was virtually non-existent, both on physical and logical planes. Anybody with a commercial VHF receiver tuned to around 80MHz could receive the squawks and warbles of the outstation signal and would be able to decode it into 0’s and 1’s if they really wanted.

Oddly enough, the thought that some crank would actually want to listen in on digitized raw data coming from a flowmeter was never really brought into question. 

And what to do with this mass of raw data that was coming in? Storing the data at anything but summary level required huge amounts of storage (tapes, and hard disks and RAM were expensive, and flash drives something out of the movies). The tables and columns in relational database management systems ruled the day. Clever things like Map-Reduce and Hadoop weren’t even thought of, and even if they had have been, most companies didn't have a supercomputer upon which to run such Big Data and Analytics.

So, if this technology upon which it is based is not radically new, what has changed with the advent of this thing called the ‘Internet of Things’, and why is it causing so much stir?

A number of things spring to mind: 

• The Internet itself, and the standardisation it brings;
• Advances in radio spectrum usage (including mobile radio);
• Bandwidth increases;
• Cost reductions;
• Improvements in sensor design;
• Back-end computing capability, and;
• Cyber security.

It goes without saying that the Internet itself has provided the single most important advance.

The availability of an accessible, high-resilience, low cost and standardized data communications medium has meant that the DIY approach with vehicle radios, modems, and leased lines is now the exception rather than the norm.

The Standardisation that the Internet brings with it means that modern sensors and their mini ‘outstations’ are all able to use a relatively small number of protocols such as Zigbee and Wimax, to communicate (see here). Largely driven by a previous industry move to TCP/IP, IoT has followed suit, and companies are rolling out dedicated networks that a whole range of sensors from many different manufacturers can connect to and start communicating.

Advances in the way the available radio spectrum is managed, its coverage and technology means that the ‘last hop’ can be via low power radio, rather than via expensive cables. Many countries have overhauled their radio frequency spectrum allocations to utilise them more effectively, and this has opened up whole new radio communications possibilities. By more closely controlling who can use which frequencies, and how they do it, then far more efficient use of the available radio spectrum has resulted.

Through careful power emission control, frequency hopping and advanced modulation techniques (again made possible by advances in electronics), more efficient use of the available radio spectrum has made new communications channels possible. These can be used for IoT, and include including packet radio / GPRS, WiFi, 4G and specialist IoT networks such as Sigfox, LoRa, FlexNet, NWave, Telensa, M-Bus, IEEE802.15.4 and other Narrow Band IoT (NB-IOT) based systems.

RFID technology has been around for a long time now, and increasingly this is being used in IoT infrastructures too.

In many countries, regulators have also allowed the use of ultra low power radio frequency on an un-licenced basis. This has made possible some very convenient short-range data exchange for applications such as 'drive by' or 'walk by' meter reading (Automatic Meter Reading, AMR); the business benefits here are huge, as traditional meter reading (which usually required a site visit) can be conducted far more quickly and conveniently. If the link is bi-directional, then meters can also be programmed, for instance with special tariff rates or similar. Many European countries have adopted protocols which operate on 433MHz and 868MHz (an old article, but see here)

And also here - I particularly liked the bit about how the city Council of Corpus Cristi in the US implemented AMR after one of its meter readers was attacked by a dog :-0

See here for a view on how Singapore's IDA 'Radio Spectrum Master Plan' is managing its national radio spectrum, and here for a pictorial version, as shown below.

Of course light is also a part of the electromagnetic spectrum, and it too has been roped into IoT service under the generic moniker 'LiFi', since it is ideal for line of sight communications and offers tremendous bandwidth. The availability of ultra cheap laser diodes (as used in those cheap laser pointers) has meant that line of sight laser links are now cheap and ubiquitous. Even LEDs are being used for very short range ‘to the lamp post’ communications.

Massive increases in the amount of Bandwidth has meant that this shot through the roof of late. When 4G first came out in Singapore, I got ‘faster’ speeds on my mobile phone than I did on my work network connection! 

Source: McKinsey

This increase in bandwidth has meant that huge amounts of data can now be shifted about conveniently (it does of course mean that people design with this in mind, so the amount of traffic also rises correspondingly).

The Cost of pretty much everything mentioned so far has fallen dramatically, from the sensors themselves through the costs of communication to the platforms which crunch the numbers and the software used to do so. It’s all got much, much cheaper, so more and more people are designing systems very economically.

Sensor Design – with dedicated chipsets (see here) now able to handle all of the functions that the early outstations used to do (and many more), and with advances in electronics manufacturing (surface mount, etc) then the physical size of sensors and their associated box of tricks has plummeted.

Coupled with this are significant advances in low power design and higher power density batteries mean that it is now viable to make a sensor with a 5-year battery life that can just be replaced with a new one very economically (See here). 

Add to this power from now cheap solar cells and wind generators, and modern sensors can last a long, long time. Furthermore, manufacturers are capitalizing of space and many are introducing multi-sensors in a single package that can deliver several monitoring channels at once. Maybe flow, pressure, and turbidity for water companies, or temperature, humidity, pollution level, and visibility for Smart Cities.

Singapore's Public Utilities Board has implemented a system known as 'WaterWise' (see here), which is based on collecting and analysing data from a network of multi-sensors. This data is analysed and used (amongst other things) for leak detection and prediction.

Improvements in GPS technology (accuracy, cost, miniaturisation etc) has also meant that for mobile IoT devices, then adding real time positional data is a given.

Cyber Security of course has become a huge area of concern in recent years, and there has been a growth in activity and a sharpening of focus which has spawned an entire sub-industry.

Organisations are rightly paranoid about IT security, and numerous events ranging from state sponsored security attacks, organized groups, to ‘lone wolf’ hacks have been commonplace. A particularly interesting one from an IoT perspective was the Stuxnet attack on the Programmable Logic Controllers (PLCs) that ran 984 of Iran's enrichment centrifuges (see here).

So securing data at rest and ‘in flight’ has become important of IoT, and a variety of tools and techniques have emerged to assure this, often based around encryption and intrusion detection systems.

Finally, the availability to rent of powerful and affordable Cloud Computing and storage infrastructure, coupled with the development of Big Data type Analytics means that a powerful backend Analytics infrastructure can be run up very quickly and very cheaply. It can also be taken down equally easily. Indeed certain companies have offerings intended exactly for this purpose that include much of the specialized software and hardware configurations to be able to mine data, spot hidden trends and patterns and generally analyze masses of IoT data.

It seems to have been a gradual parallel development in a number of areas that have made the IoT a viable technical proposal, boosted along the way by a few landslide events, such as the Internet coming along. Fascinating stuff for us tech lovers!

There remains a key question though, that is proving somewhat slow to crack – how to make money out of IoT – how to ‘Monetize’ it? 

It’s all very well to connect up masses of sensors and analyze it, but what are the business benefits of doing so? Are they of sufficient size to warrant the investment? I did hear of an Australian Mining Company that successfully applied IoT and Big Data techniques to uncover significant new mineral deposits to mine, so in that case their investment seems to have paid off. 

But what of the other touted applications of IoT – of Smart and Connected Cities, Smart Grids, Wearables, Connected Car, Connected Health, Smart Retail, Smart Supply Chain, Smart Farming, the Industrial IoT, and the like – how to Monetize those?

It feels like we are on the brink of a huge area of development and hopefully Monetization of IoT. I for one really hope take up of this technology ramps up soon, because it's a fascinating area to work in and offers tremendous opportunity.

Citations:

Components, RS. "11 Internet of Things (IoT) Protocols You Need to Know About." 11 Internet of Things (IoT) Protocols You Need to Know About. N.p., 20 Apr. 2015. Web. 18 Mar. 2017. 

VIO, Isabelle. "Radio Communications Protocol for Meter Reading." Metering.com. N.p., 31 Dec. 2002. Web. 18 Mar. 2017.

"Automatic Meter Reading." Wikipedia. Wikimedia Foundation, 14 Mar. 2017. Web. 18 Mar. 2017. 

"RADIO SPECTRUM MASTER PLAN." www.ida.gov.sg. N.p., Aug. 2014. Web.

"IoT Chips and Modules." 2017 Guide to Options and Vendors. N.p., n.d. Web. 18 Mar. 2017.

"Internet of Things | Sensors." Internet of Things | Sensors. N.p., 21 Apr. 2015. Web. 18 Mar. 2017. 

Paganini, Pierluigi. Securityaffairs.co. N.p., 18 Jan. 2016. Web.