Industries are constantly searching for new approaches to advance operational efficiency and product quality. With unplanned downtime estimated to cost manufacturers $50 billion each year, improving asset health and availability is a top priority.

In the past, preventive maintenance and routine inspection were relied on to reduce outages. Today, equipment condition monitoring fueled by the Industrial Internet of Things (IIoT) is giving rise to a predictive maintenance strategy.

A Primer on IIoT-Enabled Condition Monitoring

Figure 1. Predictive maintenance strategy aligns maintenance activities with actual asset conditions to maximize uptime and Overall Equipment Effectiveness (OEE).

Condition monitoring solutions give a voice to not only equipment, but also its environment, to continuously report on various operational and health parameters. With this data, manufacturers can identify symptoms and conditions that could eventually lead to machine breakdowns that will disrupt production. This data enables predictive maintenance, which is performed based on actual asset conditions rather than a pre-planned schedule. This eliminates redundant planned downtime with its associated costs and lost productivity.

Industrial condition monitoring takes many forms, depending on the asset to be monitored. For example, vibration analysis is commonly used for rotating equipment. Abnormal motion implies structural issues like loose bearings or misaligned shafts that require corrective actions. Not only vibration, but also temperature, moisture content, pressure, input current, and sound level, are important indicators of wear and tear in cross-industry assets — from motors, valves, and pumps to tanks and pipelines.

Optimal environmental conditions also play a role in improving equipment lifetime. Excessive atmospheric humidity, for example, can cause condensation and alter conductivity, leading to damage and malfunctioning of electronic components — and it can also accelerate corrosion. On the other hand, an overly arid atmosphere results in friction and electrostatic charge as well as making device components brittle.

In data centers, where intensive heat is constantly emitted from servers, deploying temperature sensors on a rack level helps detect problematic hot spots. This enables execution of precision cooling to ensure safe and efficient operation of equipment.

Overall, environmental data provides valuable contextual information to diagnose hidden factors that contribute to asset failures.

Connectivity Challenges for Industrial Condition Monitoring

Data for condition monitoring can generally be gathered in two ways: having a technician walk around with a handheld troubleshooting device and/or deploying a factory-wide sensor network for automatic data collection. Different from process automation and control tasks, condition monitoring does not necessarily require high-bandwidth and millisecond-latency communications. In fact, it mostly uses small bursts of telemetry data on a periodic basis, every few seconds or minutes. What matters more is reliable and scalable data transfer from thousands, if not tens of thousands, of granular data acquisition points, whether embedded in assets or installed as stand-alone sensing devices.

Most of the time, hard-wiring is not feasible due to the capital- and labor-intensive installation process and potential downtime. A wired infrastructure also makes future expansion and modifications difficult. Wireless solutions provide significant advantages in terms of flexibility and cost-effectiveness. Nevertheless, condition monitoring in industrial environments poses major technical challenges not all radio technologies can address.

Long-Lasting Operation with Independent Batteries

As many industrial assets are located nowhere near a power supply, sensing devices must be able to operate on batteries or by using self-powering techniques such as energy-harvesting. Low power consumption of the radio link is critical to minimize the cost and hassle of battery replacement and/or recharge, ensuring scalability of the condition monitoring network. The power equation entails a number of components including network topology, duty cycle operations, and synchronous/asynchronous communication.

Reliable and Extensive Coverage

Figure 2. Five requirements of industrial condition monitoring networks.

Extremely high “carrier-grade” reliability is a prerequisite for all industrial networks. The problem is that manufacturing plants present a hostile environment for radio propagation due to rebar structures and a high density of large mechanical and electronic equipment. Industrial campuses like open-pit and underground mines additionally entail challenging topography characterized by hills, extreme depths, or severe bends and corners. These physical obstructions greatly attenuate radio signals and inhibit successful data reception. On top of that, a large number of industrial facilities are located at remote, inaccessible locations where public communications infrastructure is not available.

Immunity Against In-Band or Adjacent-Channel Interference

Many existing wireless technologies operate in the license-free Industrial, Scientific, and Medical (ISM) frequency bands. To ensure network reliability, immunity against co-channel interference (crosstalk) is a must. Interference resilience can be obtained through a number of techniques, such as short on-air time, frequency hopping, and channel coding. The operating frequency should also be factored in. For example, the 2.4 GHz ISM band is often overcrowded with legacy industrial and communications systems, and thus may not be recommended for IoT deployments.

Scaling with Massive, Growing Data Points

To effectively scale with high numbers of data points, an RF system must also have a mitigation mechanism against intra-system interference. Conventional wireless solutions employ synchronous (time-synchronized) communication, wherein timeslots are scheduled for each device to avoid simultaneous transmissions that can result in packet collision. However, a large number of newer low-power RF technologies resort to asynchronous communication to minimize message overhead and associated energy consumption. Since asynchronous communication does not entail transmission coordination among end devices, other offset measures must be in place to circumvent self-interference.

Spectral efficiency, which refers to the optimized use of the bandwidth resource, is a major benchmark for the capacity and scalability of an RF system. A wireless solution designed for scalability should use fewer base stations to effectively handle very high data traffic, thereby reducing capital expense. At the same time, it should allow for seamless integration of new devices into the same network infrastructure over time, without risking a higher packet-error-rate that degrades system performance.

Support for Moving End Devices

Portable troubleshooting devices carried by a technician or mounted on mobile machinery like forklifts, excavators, or haul trucks require a communications technology that is resistant to Doppler shifts (the change in frequency or wavelength of a wave in relation to an observer who is moving relative to the wave source).

Navigating the Wireless Technology Landscape

Traditionally, the industrial wireless landscape has been dominated by high-throughput solutions like WLAN and cellular. While serving control-centric, localized tasks well, these networks are too expensive and power-hungry to connect massive, battery-operated end-points. Paying for a cellular data plan with each sensor node is not economically viable in the long run. Furthermore, both WLAN and cellular have their limits in network coverage. WLAN has a range of only a few tens of meters and very weak penetration capability, while cellular connections are often absent in remote locations.

In comparison with cellular and WLAN, low-power solutions based on the IEEE 802.15.4 standard (e.g., Wireless HART, ISA-100-11a, Zigbee, etc.) offer a more feasible solution. These networks reduce the data rate to 250 Kbit/s in order to reduce power consumption and adopt mesh topology to offset their short physical range. However, the mesh topology itself, wherein a signal needs to pass through multiple nodes to reach a destination, is power-inefficient and entails major engineering effort. Instead of falling into sleep mode to minimize energy usage, a device must stay awake most of the time to listen and relay messages. In addition, effective signal propagation in mesh networks requires a distribution of nodes in close proximity with each other. Redundant repeaters often need to be added to ensure reliability, which complicates network configuration and management, while increasing costs. The common use of 2.4 GHz operations and limited mobility support of the IEEE 802.15.4-based solutions are other factors to consider.

The emerging Low Power Wide Area Network (LPWAN) technology introduces a power-optimized design for wireless sensor networks. Ultra-low power consumption in LPWAN is attained through a combination of deep sleep mode, star topology, and a very lightweight Medium Access Control (MAC) protocol thanks to the asynchronous communication. In most LPWA networks, a sensor device sends a message directly to the base station whenever needed, without employing a strict synchronization scheme.

By using the unlicensed sub-GHz frequency band and very low data rates, LPWAN provides much higher receiver sensitivity than traditional wireless technologies. Sub-GHz signals are less susceptible to reflection and path loss while bending farther around obstacles compared to the 2.4 GHz signals. This translates into the excellent range and building penetration capability of LPWAN solutions.

Thanks to the low data rates and lightweight MAC protocols, the simple LPWAN waveforms demand much less transceiver memory and computing power. Also, LPWAN technologies leverage either the license-free spectrum or remaining resources in currently owned licensed bands to reduce operating costs. They can therefore be deployed at a fraction of the capital and operating expenses of other wireless alternatives.

These range, power, and cost advantages make LPWAN an ideal solution for condition monitoring, battery-operated, sensor networks. Nevertheless, it is equally important not to overlook the reliability, scalability, and interoperability issues in many LPWAN technologies. Asynchronous communication combined with conventional Ultra Narrow Band (UNB) and Spread Spectrum techniques intensify the interference problems in the license-free spectrum. In addition, many LPWAN technologies are proprietary, meaning they are either tied to a specific chipset vendor or a third-party managed network.

The best way to avoid these problems is to opt for an open-standard, software-driven solution certified by standards development organizations. The new Telegram Splitting Ultra-Narrow Band (TS UNB) protocol, for example, is an LPWAN technology recognized by the European Telecommunications Standards Institute (ETSI) for production-level interference resilience and network capacity in the license-free spectrum. The very short on-air time of each data packet, coupled with extremely small bandwidth usage for optimum spectral efficiency, minimizes both inter- and intra-system data collision.

An open, software-driven approach additionally provides enhanced interoperability with existing and future devices, as well as easy integration into the user’s preferred IT infrastructure. This helps streamline complexity and costs while fostering security and data privacy.


With the advent of IIoT, industrial condition monitoring and advanced predictive maintenance provide transparency over asset operations and conditions. Collecting sensor data at a granular level requires a power-efficient, long-range, and scalable communication infrastructure. LPWAN is a solution that offers a viable means for implementing a versatile, factory-wide condition monitoring network. Given the various LPWAN solutions available today, you should be able to create a highly robust, integrable, and interoperable IIoT architecture for aggregating data from a large number of independent battery-powered sensors.

This article was written by Albert Behr, CEO, BehrTech (Toronto, ON, Canada). For more information, contact Mr. Behr at This email address is being protected from spambots. You need JavaScript enabled to view it. or visit here .