A Complete Guide to Smart Wireless Sensor Design
This article provides an overview of several wireless standards and evaluates the suitability of Bluetooth® Low Energy ( BLE), SmartMesh (6LoWPAN based on IEEE 802.15.4e), and Thread/Zigbee (6LoWPAN based on IEEE 802.15.4) for use in harsh industrial RF environments, providing several comparative metrics including power consumption, reliability, security, and total cost of ownership. SmartMesh time synchronization consumes less power, and SmartMesh and BLE channel hopping capabilities provide higher reliability. A SmartMesh case study concluded that reliability reached 99.999996%. This article introduces BLE and SmartMesh wireless condition monitoring sensors from Analog Devices, including a new wireless sensor with edge artificial intelligence (AI) that can extend the battery life of constrained edge sensor nodes.
Sales of the smart sensors market for motor drive systems are expected to more than double (to $906 million) between 2022 and 2024. Within the smart sensor space, wireless and portable devices are expected to be the main growth drivers. Monitoring industrial machines using wireless environmental sensors (temperature, vibration) has a clear goal: to detect if the monitored equipment deviates from a healthy operating state.
For industrial wireless sensor applications, low power consumption, reliability, and security are always the most important requirements. Other requirements include low total cost of ownership (minimum gateway and maintenance effort), short-range communication, and a protocol that supports mesh networks in factory environments with a lot of metal obstacles (mesh networks help mitigate possible signal path shielding and reflections).
Figure 1 provides an overview of several wireless standards, and Table 1 compares and evaluates selected wireless standards based on key industrial requirements. Clearly, BLE and SmartMesh (6LoWPAN based on IEEE 802.15.4e) offer excellent comprehensive performance for industrial applications in terms of low power consumption, reliability, and security. Thread and Zigbee have low power consumption and implement secure meshes, but their reliability is relatively low.
Table 1. Matching of wireless standards with industrial application requirements
Table 2 provides more details about the Zigbee/Thread, SmartMesh, and BLE mesh standards. SmartMesh includes a time synchronized channel hopping (TSCH) protocol, where all nodes in the network are synchronized and communications are coordinated by a predetermined schedule. Time synchronization consumes low power and channel hopping provides high reliability. The BLE standard also includes channel hopping, but has some limitations compared to SmartMesh, such as line-powered routing nodes (which increases system cost and power consumption) and no support for TSCH. As mentioned earlier, Zigbee/Thread has relatively low reliability and does not offer many advantages over BLE.
Table 2. Key wireless standards and features for industrial applications
This article will focus on SmartMesh and BLE mesh as the most suitable wireless standards for industrial condition monitoring sensors.
Table 3 summarizes the Analog Devices Voyager 3 wireless vibration monitoring platform and next-generation wireless condition monitoring sensors. The Voyager 3 uses a SmartMesh module (LTP5901-IPC). The AI-enabled vibration sensor (still under development) uses a BLE microcontroller (MAX32666). Both sensors include temperature and battery state of health (SOH) sensors. The Voyager 3 and AI versions of the sensor use Analog Devices MEMS accelerometers (ADXL356, ADXL359) to measure vibration amplitude and frequency in industrial equipment. The FFT spectrum can identify increases in vibration amplitude and frequency, which can be a sign of faults such as motor imbalance, misalignment, and bearing damage.
Table 3. ADI Wireless Industrial Sensor Prototypes
Figure 2 outlines typical operation of the Voyager 3 and AI-enabled vibration sensors. As with many industrial sensors, the duty cycle is 1%; most of the time, the sensor is in a low-power mode. The sensor wakes up periodically to collect batches of data (or wakes up in the event of a high vibration amplitude shock), or to send a status update to the user. The user is usually notified via a flag that the monitored machine is healthy and that the user has an opportunity to collect more data.
SmartMesh IP networks employ multiple layers of security, which can be summarized as confidentiality, integrity, and authenticity. Figure 3 summarizes SmartMesh security. AES-128-bit end-to-end encryption ensures confidentiality even when there are multiple mesh nodes in the network. Transmitted data is protected by a message authentication code (message integrity check or MIC) to ensure that the data has not been tampered with. This prevents man-in-the-middle (MITM) attacks, as shown in Figure 3. SmartMesh supports multi-level device authentication to prevent unauthorized sensors from being added to the system.
Devices using BLE standard versions 4.0 and 4.1 have security vulnerabilities, but versions 4.2 and later have enhanced security (as shown in Figure 3). The MAX32666 from Analog Devices complies with BLE standard 5.0. This version introduces the P-256 elliptic curve Diffie-Hellman key exchange for pairing. In this protocol, the public keys of two devices are used to create a shared key between the two devices, the long-term key (LTK). This shared key is used for authentication and to generate keys to encrypt all communications to prevent MITM attacks.
Low power consumption
The sensors shown in Table 3 were run at a 1% duty cycle with a maximum payload of 90 bytes for the Voyager 3 and 510 bytes for the AI version. Figure 4 (adapted from Shahzad and Oelmann3) shows that for payloads between 500 bytes and 1000 bytes, BLE consumes less energy than Zigbee and Wi-Fi. Therefore, BLE is well suited for AI use cases. SmartMesh has very low power consumption, especially for payloads of 90 bytes or less, as used in the Voyager 3 sensors. The SmartMesh Power and Performance Estimation Tool available on the website can be used to estimate SmartMesh energy consumption. The SmartMesh Power Estimation Tool has been experimentally verified to be 87% to 99% accurate, depending on whether the sensor is a routing node or a leaf node.
In addition to the radio transmit power consumption, the total system power budget and total cost of ownership must also be considered. As shown in Table 2, both BLE and Zigbee operate using a single gateway. However, both also require line power to power the routing nodes. This increases the power budget and system total cost of ownership. In contrast, SmartMesh routing nodes only require an average of 50 µA of current, and the entire network can operate using a single gateway. SmartM esh is clearly a more energy-efficient implementation.
As mentioned earlier, SmartMesh uses TSCH, which has the following features:
-
All nodes in the network are synchronized.
-
Communications take place according to a communications schedule.
-
Time synchronization results in low power consumption.
-
Channel hopping brings high reliability.
-
The planned nature of communication brings a high degree of certainty.
The full network synchronization accuracy is less than 15µs. This high level of synchronization allows for significantly reduced power consumption. Average current consumption is 50 µA, and more than 99% of the time current consumption is 1.4 µA.
Table 4 lists some key application challenges and explains how SmartMesh and BLE mesh address them.
Table 4. Key challenges of wireless networks in industrial applications and BLE/SmartMesh performance
SmartMesh performs better in dense networks with a large number of nodes. Both BLE and SmartMesh perform well in dynamic industrial environments.
The reliability of SmartMesh was tested in ADI's wafer fab, which has a harsh RF environment full of metal and concrete. Thirty-two wireless sensor nodes are distributed in a Mesh network, and the farthest sensor node is four hops away from the gateway. Each sensor node sends four data packets every 30 seconds. Over a period of 83 days, the sensors sent 26,137,382 data packets and received 26,137,381 data packets, with a reliability of 99.999996%.
Next-generation wireless sensors include the MAX78000 microcontroller with an AI hardware accelerator that significantly reduces data movement and exploits parallelism to optimize energy use and throughput.
Current commercially available wireless industrial sensors typically operate at a very low duty cycle. The user sets the length of time the sensor sleeps, after which the sensor wakes up and measures temperature and vibration, then transmits the data back to the user's data aggregator via the radio. Commercially available sensors typically claim a battery life of 5 years, based on capturing data once every 24 hours, or once every 4 hours. The next generation of sensors will operate in a similar manner, but will limit the use of the radio using edge AI anomaly detection. When the sensor wakes up and measures data, it will only transmit the data back to the user if a vibration anomaly is detected. In this way, battery life can be extended by at least 20%.
For AI model training, the sensor collects the health data of the machine and then sends it wirelessly to the user for AI model development. The AI model is synthesized into C code using the MAX78000 tool and then transmitted back to the wireless sensor and placed in memory. After the code is deployed, the wireless sensor wakes up at predefined time intervals or when a high-g shock event occurs. An FFT is generated after the data is collected. Using the FFT, the MAX78000 makes inferences based on the data. If no anomaly is detected, the sensor returns to sleep. If an anomaly is detected, the user is notified. The user can then request the FFT or raw time domain data of the measured anomaly, which can be used for fault classification.
This article provides an overview of several wireless standards and evaluates the suitability of BLE, SmartMesh (6LoWPAN based on IEEE 802.15.4e), and Thread/Zigbee (IEEE 802.15.4) for use in harsh industrial RF environments. SmartMesh has superior reliability and low-power operation compared to BLE and Thread/Zigbee. For networks that require 500 to 1000 bytes of data transmission, BLE can operate more reliably and consume less power than Zigbee and Thread. Microcontrollers with embedded AI hardware accelerators can improve the decision-making capabilities of wireless sensor nodes and extend their battery life.
????Click here to explore the ADI "chip" world