Much of the discussion about the IoT is about communications and data analysis. But, increasingly, the real holdup in implementation can be found in physical objects, the “Things” of the IoT, and the way they are sensed and acted upon.
In order to perceive the state of an IoT-connected device, whether it’s a mining haul truck or a refrigerator, you need sensor data.
Sensors convert some measurement of the physical environment into an electrical signal, which can then be processed, analyzed, and used to make decisions. Those physical measurements include pressure, position and motion, vibration, temperature, humidity, ion concentration, magnetic signatures, and radiation, as well as imagery.
While most IoT articles seem to be about “smart” home applications of various sorts, the real push for hardware development is going on in industrial and commercial applications. Deloitte predicts that 60 percent of all wireless IoT devices will be purchased for use by enterprises and industries. Industry demands for low cost, ruggedness, and long operating life will spill over into the consumer market.
Most commercial applications require significant numbers of sensors, and one obstacle to wider implementation of IoT has been, simply, sensor cost. For example, the railcar industry would love to detect toxic gas leaks or know whether a door is properly shut on various railcars. But even the mere dozen or so sensors per railcar that would be necessary would cost around $500 per railcar, and the industry calculates that, given their margins, anything higher than $200 would simply not be economic. And there are around 1.3 million railcars of various types in use in the US.
And most industries are looking at many more sensors than that. A vehicle or piece of industrial machinery can use hundreds or thousands of sensors to be fully integrated. The benefits of IoT have to outweigh its costs, and the costs are too high.
But sensors have fallen in price by half over the last ten years, and this decline is continuing steadily. Increasingly, IoT-focused devices are becoming a larger part of the semiconductor industry’s production. According to Goldman Sachs, sensors and microcontroller sales are significantly outpacing the rest of the semiconductor market.
But sensors and other hardware face another problem: they need to be powered. Here, again, industrial and commercial applications are where the resistance is highest, and the promise greatest.
Mines, railways, pipeline networks, highways, and other large pieces of infrastructure require sensors that operate without access to power supplies. To function sensors must be consistently supplied with power, or they will bring down a system that depends on their data.
So manufacturers are reducing sensor power demands, implementing new generations of lithium batteries with long-rated lives, and finding ways to harvest energy from the environment.
That last, energy harvesting, is perhaps the most promising. In energy harvesting, transducers convert vibration, light, heat, or transmitted RF signals into energy usable by the sensor. The sensor can operate or recharge its battery, potentially allowing it to have power until the end of its useful life.
Falling prices and rising performance in sensors will transform different industries, depending on their margins, the number of sensors they need, and the value of the information those sensors generate. But the changeover will happen before anyone outside that industry knows it. It won’t be the same Year of the IoT in every industry. For some, it may well be this year.
Catch up on the State of the IoT series: The Open Connectivity Foundation and Why It Matters