• News
  • Columns
  • Interviews
  • BW Communities
  • BW TV
  • Subscribe to Print
BW Businessworld

IoT: Sensors Everywhere

And they’re busy networking. Welcome to Industrial IoT, which could add $37 billion to India’s GDP in the next 15 years, according to Accenture

Look around you: how many sensors can you see? Perhaps none: they’re hidden.

But wait, that door just slid open as you approached. That escalator started moving as you walked onto it. Walk further, and that darkened passage lights up as you cross more motion sensors.

Drive out, and dozens of sensors in your car track fuel and oil levels, speed, temperature, tyre pressure, location. That display showing 233 parking slots at the mall changes to 232 as you drive in.

Your life depends on sensors. Think of what would happen if sensors in an elevator failed, shutting the door on a person entering, or worse, allowing the lift motor to operate without the door having closed?

We’ve depended on sensors for over a century. But over the past decade, they’ve gotten omnipresent, digital, and networked, with lots of software and intelligence managing them.

The Internet of Things is projected to comprise 50 billion objects by 2020. A majority of these will be in the industrial sector, making up Industrial IoT—IIoT.

Accenture says IIoT could add $37 billion to India’s GDP in the next 15 years, with business as usual. “But with additional measures such as improving the telecom infrastructure, this GDP boost could grow to $47 billion.”

Accenture analyzed 20 countries “to identify their national absorptive capacity for IIoT technologies, products and services”, and came up with some staggering numbers: a positive impact of $10.6 trillion by 2030 for these 20 countries, with ‘business as usual’ policies, or $14.2 trillion with greater investment and policy tweaks.

So where and how’s IIoT changing our lives? Everywhere, including in the air.

On a wing and a prayer—with 10,000 sensors
On March 8, 2014, Malaysia Airlines flight MH370 took off from Kuala Lumpur and flew into history books as aviation’s biggest mystery. Two years later, we still don’t know where it went. But we do know that its aircraft monitoring system, ACARS, stopped stopped responding to ground data requests a few hours into the flight. The Boeing 777’s engine monitoring system reported back, twice, to Rolls Royce’s global engine health monitoring center in the UK.

And so it appeared that MH370 may have crashed soon after last radar contact at 02:22 local time somewhere in the Andaman Sea. But then we learnt one more thing. Satellites were pinging the Inmarsat receiver on the airliner, once an hour, and the aircraft answered 7 times. MH370 had continued to fly for 7 hours after the aircraft vanished. It could have flown to another part of the world.

MH370 triggered huge interest in connected systems on-board aircraft to stream data in real time. Such systems exist, but cost as much as $100,000 for one aircraft. They could save lives, preparing search and rescue even ahead of a crash. After a crash, they’d help avoid the frantic hunt for black boxes, or, indeed, a global hunt for a missing airliner, in an age where a free ‘find my phone’ app can locate a missing phone.

An airliner is a massive flying IoT ecosystem. There are over 5,000 sensors on a modern airliner: sensors for fuel quantity and quality, flight attitude and angle of attack, flow sensors, ice detection, liquid level, pitot probes, air temperature. The Airbus A350 has nearly 6,000 sensors, and generates 2.5 terabytes of data per day. The Airbus A380-1000, scheduled for deployment in 2020 with each aircraft carrying 1,000 passengers will have 10,000 sensors in each wing.
Big data analytics firms focusing on aviation, such as masFlight, help airlines analyse flight data to improve operations and safety. But if data from many of these sensors on board aircraft could be streamed to and analysed in real time in ground stations, aviation safety could be drastically enhanced, with most air incidents—barring terror attacks—anticipated well before they occur. MH370 will push airlines and civil aviation further toward that holy grail.

Autonomous Vehicles Everywhere
How far away is real driverless tech? Why, it’s been around for decades.

First, in the air, with autopilot, then autoland, with which aircraft using Cat III instruments can land automatically in zero visibility, keep the aircraft aligned dead center on the runway post landing, and slow down the aircraft with autobrake systems. (In practice, 50 meters visibility is mandated for Cat IIIB, mainly because taxiing to the gate in zero visibility is difficult.)

Autoland was developed in the 1950s in the UK, home to fog-afflicted Heathrow, and entered commercial airliners in the 1970s. It depends on high-precision sensors and instruments such as radar altimeters, given the low tolerance to error when a 747 is coming in to land at 240 kph, a few meters above a runway. For commercial airliners, take-offs are not yet automated, and pilots are required by law to be in the cabin for landings, but drones have been flying autonomously for over a decade, with remote pilots supervising route guidance and weapons engagement.

Trains were next in line. London’s Victorial Line was the first with automatic train operation (ATO) in 1967. This is automation at “grade 2”, where a driver remains in the cab, but does not drive. This also applies to the entire Delhi Metro system, and even the older Kolkata Metro Line 1, which uses ATO, with a driver to supervise. Dozens of subway and other train systems worldwide have moved to grade 4 automation, where no driver is required, and the Delhi Metro will start using such autonomous, driverless trains later this year.

Think of the sheer numbers of sensors in a subway train. Apart from the usual speed, location and other parameters, there’s multiple sensors in every door, networked to the train control system. The train won’t move unless all the sliding doors are shut.

Or think of what happens if a driver sees something on the tracks, and hits the brakes—and there’s another train a minute behind. A buffer segment of track behind the train ‘goes red’, and the brakes are applied in any trains close behind. The system is automated. Additional sensors on the train sense proximity to other trains on the track.

The toughest nut to crack is the driverless, autonomous car. But everyone’s on it: from Google to Audi, Mercedes, Tesla to Delphi testing them for years. The sheer number of variables especially on Indian roads would pose a serious challenge to autonomous cars, which depend heavily not just on sensors and technology but on rules and dependable cues such as lane markers, and accurate and detailed maps. The human driver is a tough element to replace in chaotic, mixed road traffic. A wide range of proximity and other sensors, cameras and radar, and software, will help bridge that gap.

‘Sensor Journalism’ and Fixing
Our Dirty Air — In our increasingly polluted cities, networks of sensors report air quality, and the reports are grim. On a network of sensors run by IndiaSpend, a data journalism startup, Delhi is showing an AQI (air quality index) of between 500 and 1000 at various points in the city, well above ‘extreme hazard’ levels; Beijing had declared an emergency and shut down schools and offices at an AQI of 250-300.

The IndiaSpend pollution sensor project goes beyond data journalism into ‘sensor journalism’, using sensors to create and then analyse data (as against data journalism, which uses existing data). The air-quality monitors it’s using have been developed by the startup, to be relatively cheap, focused on PM2.5 and PM10 levels, and are networked, using a GPRS module and SIM cards.

Air-quality sensors alone haven’t fully answered the question of what exactly is causing Delhi’s intense pollution and what’s the impact of, for instance, an “odd-even cars” experiment. That answer likely requires a lot more data from many locations, something industrial IoT is perfect for, with advanced analytics.

In a recent deal with the Delhi Dialogue Commission, IBM will “leverage IoT and machine learning, with analytics from cognitive computing and statistical modeling.” A key question: how Delhi’s 7.4 million vehicles contribute to air pollution. It hopes to provide insights and recommendations to improve Delhi’s air quality.

This is part of IBM’s Green Horizons project, which uses IoT sensors and machine learning to analyse big data, and discover where pollution is coming from. Using these resources, IBM says the Beijing government was able reduce ultra-fine particulate matter by 20% in the first three quarters of 2015.

Smartness as a Service — This is among the most mature of IoT applications. Dumb utility meters which had to be read by meter-men peering at the meters gave way over a decade ago to automatic meter reading: dumb meters with a cellular module, which could send an SMS with a reading once a day, or more or less frequently. The SMS gave away to GPRS data, and then to smart meters, with two-way connection to a network. A power utility company can talk back to such a meter, asking it to reduce load, or disconnect and reconnect remotely, inform the consumer of an impending power outage, or dynamically change loads to changing tariffs.

Smart meters are expensive and complex to manage, and Ericsson says it has the answer. It’s launching SMaaS, or Smart Metering as a Service. The company will own and maintain the meters, and manage the process including data collection and analysis, on behalf of utilities, with service-level agreements. SMaaS has been tested out effectively with utilities in northern Europe, according to Ericsson sources, before taking it global, which is slated for Q2 of 2016.

Why the Telcos Like IIoT— A lot of the data from industrial IoT networks first comes to cellular operators, who are getting very interested in IoT—not just in the GPRS/3G data revenue from millions of devices, but in adding value (and making money) beyond that.

Ericsson’s User and IoT Data Analytics, which it showcased, curiously, in the consumer show CES at Las Vegas last month and which will be commercially available in Q2 this year, gives operators a “real-time analytics engine embedded in the subscriber database’. This extends Ericsson user data consolidation service, which already manages subscriber data in the billions, according to the company, giving telcos useful macro views, analytics and relationships that they can use for increased business.

From the telcos’ viewpoint, IoT also needs tech different from current systems and base stations that are designed for a few hundred or thousand users per cell site, and vendors are on it. In the US, AT&T is now deploying the first low-power wide-area (LPWA) system, supplied by Ericsson, which will support millions of IoT connections per cell site. At CES last month, Ericsson also demonstrated software that supports massive IoT deployment on current LTE network infrastructure.

Coming up in the years ahead to 2020: the first ‘smart cities’, overhaul of highways and train systems, overhauled mass transit in major cities, enormously networked healthcare systems with data sourced from millions on wearables.

All of these will depend completely and deeply on large IIoT networks linking millions of sensors and connected devices, backed by software, processing, analytics, intelligence. And, of course, many new startups across the ecosystem. But that is another story.

Disclaimer: The views expressed in the article above are those of the authors' and do not necessarily represent or reflect the views of this publishing house. Unless otherwise noted, the author is writing in his/her personal capacity. They are not intended and should not be thought to represent official ideas, attitudes, or policies of any agency or institution.

Prasanto K. Roy

The author is Head (Media) at Trivone

More From The Author >>