IoT – You Ain’t Seen Nothing Yet

May 22, 2017, anysilicon

This title, borrowed from Bachman Turner Overdrive’s classic 1974 hit is very relevant today if we apply it to the “Internet of Things”. I’ve been reading and using the term “IoT” for a long time, but realized I needed a clear definition. The Internet is full of similar choices but I selected this from TechTarget: The Internet of Things (IoT) is a system of interrelated computing devices, mechanical and digital machines, objects, animals or people that are provided with unique identifiers and the ability to transfer data over a network without requiring human-to-human or human-to-computer interaction.


Does the concept ‘without human interaction’ bother anyone other than me?


Technological advances are exploding and suddenly everyone is trying to jump on the IoT band wagon. Junko Yoshida’s recent article, “Spiraling Data Costs Imperil IoT” (EETimes, April 14, 2016) points out a critical issue; information overload and its associated cost. I was amazed, for example, to learn that a smart meter produces 400mB of data per day. Does my utility really need sixty messages per minute from my smart meter? Furthermore, much of the actual “information” being communicated by IoT devices is often quite small… a few tens to a couple of hundred bytes. But the address and security headers can often be much larger, making the entire payload unmanageable and as Junko rightly points out, prohibitively costly. As applications expand, the information content will only become larger. Today we already have hundreds of IoT applications in service: iris scanning for immigration, shot spotting technology for high crime neighborhoods, traffic cameras, NFC payment systems, RFID item location, remote automotive diagnostics, intrusion detection, fall detection for the elderly….and much more including climate controlled clothing. Really?


Yet today’s applications are miniscule compared to the challenges that lie ahead. Compounding an uncertain IoT future is a proliferation of protocol spinoffs. XMPP, CoAP, MQTT and others in the alphabet soup seem to divide users into several distinct camps. Are we heading toward another VCR/Betamax showdown with a dozen or more players?


Fast forward to the future to see how complicated it may get. What will happen to IoT when Artificial Intelligence becomes more main-stream? Perhaps the most obvious AI example is the future automobile. We are getting close to having fully automated automobiles. Here in Mountain View, we see Google’s efforts daily. Apple may soon follow. Replacing human decision making is a huge undertaking. Yet these cars, sophisticated robots; Roombas on steroids…lots of steroids, will need to communicate massive payloads of data (more than my smart meter) in real time in order to sense, act and react to both predictable and unpredictable road events. Look at the ads on television. We have cars that sense collisions and automatically apply the brakes. Good for safety reasons, but what if a hacker gets in? Do I want someone to apply my brakes while I’m merging into the fast lane? Recall the early flaws within Tesla’s Model S that allowed hackers from Lookout Security in San Francisco to remotely open and close the front and rear trunks, lock and unlock the doors and start and stop the car by intercepting the entertainment system. Entertaining to some perhaps, but worrisome to me. Yet we blindly move forward.


Advances in Artificial Intelligence, in parallel with the billions of sensors and actuators being dispersed into IoT endpoints, feeds an insatiable set of hungry global databases from which better (more accurate) information can be retrieved at higher and higher speeds. Today we joke and say all knowledge is available on the Internet. Not even close…yet. But as we gather more data about all things physical, AI systems will make better, faster decisions than we ever dreamed of because they can easily access this repository of knowledge.


Passenger planes can take off, fly and land themselves. How soon before the Internet of Aviation (my term) will be allowed to dispense with that unnecessary weight in the pilot/co-pilot seats? Surely, they are already redundant. And I’m not talking remote control like the “Eye in The Sky” drone strike piloted by soldiers in Nevada.


On a more peaceful note, we are already compiling medical information databases at an alarming speed that AI mining tools can sort through to winnow down possible illnesses and cures to a manageable few. How do we protect patient privacy while communicating all these symptoms and possibilities digitally over the internet? How soon will AI patient diagnostics replace doctors with a biointelligence beyond our comprehension? (If you think about it, a doctor is nothing more than an encyclopedia of medical information.) We already have designer drugs to fight cancers. How long before we inject nanobots and direct them (via the IoT?) to remove a blood clot or destroy, repair or rebuild cells in specific organs? How do we keep all this from being hacked or misdirected if everything is connected digitally? Think about it for a moment. A sophisticated Tesla gets hacked thru its entertainment system and you want me to connect to the same Internet (of Things) a cheap Chinese toaster that only knows how to burn bread?


Junko’s article got me concerned about future of IoT, cross purpose protocols and the confluence of information in my own home. When my father-in-law visits, will his pacemaker be able to communicate with my refrigerator to tell my LED light to send a message to my electric car that must then tell my thermostat to notify the water heater to call 911 when he has an issue he may not even be aware of? I can see it now. “911. What’s your emergency?”…. “WE….ARE…OUT…OF…BUTTER.”



This is a guest post by Bob Frostholm – JVD Analog ASIC Semiconductors