In a recent meeting I attended a very well respected business leader asked a question about IoT. He mentioned that when he was an engineer, decades ago, he recalled designing a product based on a Texas Instruments device that when it detected a problem would use the PSTN and dial back to a waiting server to “cry” for help. He asked, so what is so different about IoT?
A few weeks later, talking with a friend who ran an company many years ago enabling smart advertisting signage using plasma screens, mentioned that the system would use the GSM data connection to report detected faults such that rather than time based servicing the service system could be forward looking and hence predictive. The example given of the temperature rising by an amount higher than usual above ambient would likely mean the air filters or cooling system needed attention. He mentioned at the time this did not have a name, but asked how does it differ from IoT?
In both cases I provided a different perspective, a non-engineering perspective. In fact IoT is not about engineering and devices and connectivity, it is not even about predictive maintenance, although that is a subset. It is really about establishing disruptive business models that enable new whole offers that are compelling to a broad customer base.
The example I like to use is that of Cascade3D, a provider of actional analytics. Historically they collected data from leisure centres and used their data scientists to analyse the static data to determine actions that can attract new members and retain current members.
With Hadoop, Spark and Storm running on arrays of servers to develop machine learning and intelligence for the IT arena and always on, connected operational technology the folks began development of a home monitoring care system.
If an M2M perspective was taken, that would likely be panic buttons, perhaps temperature sensors and possibly cameras. The control centre would be monitoring for an alert to come from the source and send in a nurse.
Taking an IoT perspective, the home is instrumented. The hot water pipe monitored, the cold water pipe, the fridge interior light, the front door, each room has a movement sensor, the chairs have pressure pads, the bed has pressure pads, the food shelves sense changes, the microwave, oven and hob, heating are monitored. The data is transmitted periodically to an actionable analytic system in the cloud.
Machine learning can determine “usual” behaviour, the home help arriving, the toilet flushing, the fridge opening, the kettle being used, movement between rooms, sitting watching TV or listening to the radio. No camera’s needed, no privacy breached, yet the disabled, aged or infirm person being cared for can be looked after.
The analytics and machine learning can then raise an alarm if unexpected behaiour is reported, or perhpas a new trend is noticed such as the toilet being used multiple times in the night.
The point of the example is to highlight that IoT is about bringing IT analytics to OT systems such that new, disruptive business models can be provided that bring greater value.
With web-services API’s an analytics engine can enquire of other services to aggregate something of additional value, bring in live data and innovations abound. Perhaps your home heating system could learn the thermal coefficient of your home such that it only heats when occupied and from the family members location can determine the optimal heating, or cooling pattern to ensure it meets the family needs.
A greater example is that of Telensa. On first look it would seem to be connected street lighting, reporting that a lamp has failed would be the M2M historic view. However, add lux meters, people sensors, traffic sensors,parking bay monitors, temperature, wind, humidity, COx, NOx and other pollution factors into every street light and a city can leverage analytics and machine learning to determine huge insight, with automated actionable analytics. Lux meters to set the output to provide uniform lux, so helping reduce the electricity useage, people sensors such that the lighting is only active when needed, parking bay monitors to direct a driver to a bay, pollution levels so redirecting traffic away from schools at key times.
Recently I sat in a room with motion sensors in the lights, having not moved, as I was reading email, the lights extinguished. Frantically waving my arms I triggered the light to return. However, if all the lights in the room were connected the system would know where I was last “seen” and would keep the lights on until I moved away, or exited through a door. Likewise, street lighting would know through their collective intelligence when they are no longer needed, and as LED light is instant, would illuminate as soon as needed.
In summary, M2M was just that, machine to machine. The immense power of servers, virtualisation, hadoop, spark, etc. and machine learning and analytics applied to live telemetry such that new business models can appear is the world of IoT.
My personal favourite is BrewLogix, the connected system that ensures a place of beer has beer. It instruments the barrel through a pad and monitors the draft beer levels, providing insight into trends, making historic recommendations connected to festivals, weather, promotional marketing activities and their effectiveness and reduces waste as a barrel change is when its really required and no longer subjective. It is more M2M than IoT, in that its not leveraging multiple sources of information, but for some reason it’s one I really like!