Discussion of the Internet of Things tends to focus on the software: communications, data integration, and analysis. But all that data about the physical world must enter the IoT through sensors of various kinds, and discussion of these sensors tends to be a bit vague, restricted to how many of them there are (lots) and how much data they generate (lots and lots).
The limits of cost, sensitivity, ruggedness, and lifespan of sensors have a real effect on IoT implementation, and these limits are often inadequately appreciated. Upcoming advances in sensor technology will be transforming the way IoT works in almost every industry.
Fundamentally, sensors convert some measure of the physical environment into an electrical signal that lets an IoT network understand and present information about that environment. Sensors measure pressure, position and motion, vibration, temperature, humidity, chemical concentrations, electrical and magnetic signatures, and radiation, among other things.
Vast amounts of useful information can come from sensing something as simple as vibration. Specific changes in the vibration of a rotary motor, for example, can indicate shaft misalignment, load imbalances, damage to the bearings, looseness, and other faults that may lead to failure, and specific sensors can monitor all of these.
But the IoT can also incorporate much more abstruse measurements. In mining, sensors measure ambient carbon monoxide, propane and nitrogen dioxide in each part of the mine in order to control zone ventilation, while high-frequency electromagnetic spectroscopy, neutron activation, and high-speed X-ray fluorescent sensors measure and report ore grade in real time.
Corrosion sensing is another big area, working on a problem that faces all infrastructure, from aircraft turbine blades to the reinforced concrete in bridges. Corrosion sensors detect concentrations of ions such as hydrogen and chloride which serve as corrosion signatures, or listen for sounds that cracks make during pressure and temperature changes.
We just saw that there are already sensors that can pretty much sense any physical parameter you'd like to name. But many industries struggle with sensor implementation, and have to accept that certain important parts of their operations will remain partly or completely unmeasured.
There are three major holdups: power, cost, and ruggedness.
Power Aside from exceptions like piezoelectric sensors, every individual wireless sensor needs to have a source of power, usually in the form of a battery. And batteries, as we all know, eventually lose power.
This is a problem, particularly if you rely on large numbers of remote sensors. They will go dead, and paying for remote battery replacement can be prohibitive—one industry estimate is that unscheduled battery replacements can incur 10 times the initial cost of the original battery.
There are several ways manufacturers are getting around these limitations: decreasing the power demands of the sensor, specifying higher-performance batteries such as new generations of lithium batteries with long rated lives, and energy harvesting.
Of these, energy harvesting seems to show the best balance of cost and life. This method uses transducers that can convert vibration, light, temperature differences, or transmitted RF signals into energy usable by the sensor. These enable the sensor to recharge a battery from the environment around it, meaning that, potentially, the sensor's power would last for the life of the device.
Cost Sensors range in price from a few cents to tens or hundreds of dollars, and many of them are necessary for full coverage. For example, the railcar industry would love to have less than a dozen sensors on each railcar, to measure things like pressure to detect toxic gas leaks or door position. But at current costs, it estimates that it would cost around $500 per railcar to do that—and calculates that $200 is about the limit for sensor use to be economic. Marginal costs make or break any business providing a continuous service, and for many industries, the benefits of sensor in many applications do not yet outweigh their costs.
Ruggedness Sensors are often exposed to extremely harsh environments, in factories, mines, and outdoors, facing such hazards as vibration, temperature extremes, moisture, dust, corrosive chemicals, impacts, and radiation. They need to endure these for years while continuing to report accurately and consistently.
Given the immense potential benefits, there is large investment in sensor improvement. Prices will continue to drop, power consumption will continue to drop, sensitivity and accuracy will continue to go up, and new technologies will continue to emerge.
At some point, a difference in degree will become a difference in kind. Once the price/benefit curve drops below a certain level, a vast number of new sensors can be embedded in all remote and hazardous environments. These environments will no longer be hard to perceive. The operational state of every rotating motor, bridge span, and heart valve will be as easily perceived as the time of day or the outdoor temperature.