I’m listening to music on my headphones. This is an easy process because of Bluetooth* standards. This easy format lets me connect any of my headphone sets into a phone, PC, or smartwatch anywhere I go. That’s the beauty of open technologies.
As we discuss in our paper “Reduce IoT Cost and Enable Scaling Through Open Wireless Sensor Networks,” manufacturing environments do not enjoy these benefits with their sensor networks. A century or so of submersion in proprietary sensor technologies has left today’s enterprises with solutions incapable of easy integration with the organization’s modern IT systems. In some cases, this integration can be done, but only within the confines and considerable expense of the solution provider’s own ecosystem.
As a sizable manufacturing company, Intel is no stranger to these issues. We need better sensor approaches as much as anyone else, and we believed that conducting a successful proof of concept (PoC) within our own walls could help other interested parties see how to follow suit and help grow demand for low-cost scalable plug-and-play sensor networks. Overall, we were committed to finding scalable solutions so that the next wave of developers could enjoy affordable, compatible approaches to breaking open today’s sensor bottlenecks.
Our paper describes how we were successful in this pursuit. We deployed over 100 Bluetooth Low Energy sensors; achieved long distances of 30 feet and had over 99% reliability in a harsh and challenging production environment; and used readily available, off-the-shelf components throughout the solution.
In the event that anyone thinks that our results are exclusive to Intel or our fab environment, I wanted to offer a few broader use case examples to serve as inspiration.
Monitoring Exhaust Laterals
This was the application in our PoC because it was a simple way to start, and most tools have exhaust. Exhaust requires monitoring in order to ensure that machines are healthy and outputting within specification. Wireless sensors enable pressure to be monitored at practically any point in the exhaust system for fractional costs. Of course, the same is also true of wired sensors—especially if those sensors were placed at the time of original fixture construction or if we spend large amounts of money to install cable to all the places we want to monitor.
If the exhaust configuration changes, or if the company wishes to change or add sensors over time, then the cost to run new infrastructure to additional wired monitoring points is about 20x more expensive than installing wireless alternatives. Wired approaches also tend to require more gateways because of the 1:1 nature of their connection to gateways that will bridge them to the internet. Those gateways will be part of the proprietary solution and are likely to be more expensive to install and support/maintain. Not least of all, the cost of wiring and the labor required to install that wiring need to be considered.
With wireless, a sensor may be based on a platform such as the Arduino Primo Core. Wireless sensor costs could easily be half that of a proprietary wired equivalent. Obviously, there are no wires to worry about. In our PoC, we connected to Intel IoT Gateways. We chose to standardize the wireless upstream data being sent to a Message Queuing Telemetry Transport (MQTT) broker. Unlike the data publication used in proprietary solutions, MQTT is open and allows nodes to connect as subscribers, publishers, or both. You can think of it as blog content, such as this article. Intel posts the article into a general content engine, and users who subscribe to the blog can filter by metadata tags to see only the articles that interest them. With MQTT, our wireless sensor data can be published to any subscriber in the system that wishes to receive a live data stream from, in our case, exhaust laterals. This is a key feature in allowing IT analytics systems to use of manufacturing floor data—which was previously unavailable—for higher productivity, safety, and tool longevity.
Liquids are necessary in almost any manufacturing operation, and companies need to monitor the temperature and levels of these liquids. Often, chemicals move around on factory floors, such as when they’re taken from storage, placed near a machine, and then moved back into storage, potentially in a different location than where they started. This mobility obviously makes fixed, wired sensors problematic if not impossible, depending on the liquids and environment. Nevertheless, monitoring remains essential. Many chemicals are expensive. Some must be exceptionally pure. If they fall outside of their allowable thresholds (think of someone accidentally leaving a batch on a dock), replacing them might cost tens of thousands of dollars. They might even damage their host machines.
Building wireless temperature and level sensors is not the toughest of engineering feats. Imagine the millions of dollars such a simple tool will save manufacturers everywhere in the near future.
That really is the point. Companies have been using proprietary legacy manufacturing sensor technologies for so long, and the habit of using them is so ingrained throughout industries, that it can be hard to see that there are lower cost methods of connecting sensors. Our PoC shows that the door to an enterprise scalable, plug-and-play wireless sensor network can be unlocked, opening the way for connecting new sensors and publishing data to be as simple and transparent as, say, connecting a Bluetooth headset to an audio device.
Read the IT@Intel Solution Brief, “Reduce IoT Cost and Enable Scaling Through Open Wireless Sensor Networks,” to find out more about our wireless sensor tests and the benefits that such approaches can bring to your manufacturing environment.