“This article reviews the latest technologies for connecting industrial sensor nodes in the Internet of Things using the ISM bands below 1GHz. From simple transceivers to microcontrollers with embedded RF, there are numerous options for implementing a low-cost, long-range, low-power link for a sensor node.”
Connecting devices is a fundamental requirement in the Internet of Things (IoT) and can be even more difficult in an industrial environment. Industrial IoT requires the reliable connection of many devices in physically and electromagnetically harsh environments. This can all be challenging for wireless systems, especially in the crowded 2,4 GHz band where Wi-Fi and Bluetooth work. The use of unlicensed bands below 1 GHz – 433 MHz, 868 MHz or 915 MHz – offers greater range than 2,4 GHz but with lower data rates. This is valid for industrial applications that do not need streaming video data but the transfer of data from sensors and diagnostic information from the equipment. This also makes it possible for developers to trade off data rate and range to reduce power consumption, allowing for longer battery life or resorting to energy capture-based power supplies. In this way, significant savings can be achieved by reducing operating costs by opting for batteries with replacement cycles of up to 20 years with the appropriate design. However, these links must be very reliable and this depends on the combination of the radio hardware of the transceiver, the microcontroller and the embedded protocol stacks. Having multiple options in these three areas gives system developers the flexibility to optimize their designs for each industrial implementation. Links below 1 GHz for narrowband can have a range of up to 10 km since there is less attenuation in walls or buildings, thus reducing or eliminating the need to install gateways or repeaters, which in turn reduces the cost of network implementation. Less crowded spectrum also means easier transmissions and fewer retries, so it's more efficient and saves battery power. Radio sensitivity is inversely proportional to channel bandwidth, so narrower bandwidth creates higher receiver sensitivity and allows efficient operation at lower transmission speeds. For example, at 300 MHz, if the error of the transmitter and receiver crystals are both 10 ppm (parts per million), the error is 3 kHz each. For the application to transmit and receive efficiently, the minimum channel bandwidth is double the error rate, or 6 kHz, which is ideal for narrowband applications. To achieve the same at 2,4 GHz requires a minimum channel bandwidth of 48 kHz which wastes bandwidth in narrowband applications and requires significantly higher operating power. One of the drawbacks may be that a larger antenna is needed for the node below 1 GHz, while at 433 MHz the antenna is 17 cm.
Silicon Labs' first wireless microcontroller (Figure 1) combines an ARM Cortex-M3 CPU, USB, and a sub-1GHz radio along with power consumption optimizations for IoT designs. The EZR32LG family of pin-compatible devices features 64/128/256 kB Flash and is compatible with Silicon Labs' EZRadio or EZRadioPRO transceivers, which cover the sub-1 GHz frequency bands between 142 and 1050 MHz. They reach a sensitivity of up to –133 dBm using EZRadioPro to provide maximum range or reduced to minimize consumption. This device covers all major frequency bands and achieves optimum phase noise, blocking, and selectivity for narrowband and licensed band applications such as FCC Part 90 and 169 MHz wireless Mbus. Adjacent channel selectivity of 69 dB with 12,5 kHz channel spacing ensures that there is minimal crosstalk between channels and any other electrical noise can be filtered out efficiently to prevent interference with the radio link. Communications between the radio and the microcontroller are carried out through USART, PRS and IRQ, for which it is necessary to configure the pins as indicated in Table 1. A key element in the EZR32LG is an AES accelerator that provides security to the wireless node. This is responsible for AES encryption and decryption with 128-bit or 256-bit keys. Encrypting or decrypting a 128-bit block of data takes 52 cycles with 128-bit keys and 75 cycles with 256-bit keys. The AES module is a slave AHB that allows efficient access to data and key records, so all write access to the AES block must be 32-bit operations.
For connection to sensors, or directly to the equipment, there are 38 GPIO (General Purpose Input/Output) pins divided into ports of up to 16 pins each. These pins can be individually configured as output or input, although there are also more advanced settings for each pin, such as open drain, filtering, and control level. The GPIO pins can also be controlled by pins on peripherals such as timers, PWM outputs, or communication via USART. GPIO allows up to 16 asynchronous external pin interrupts to perform interrupts from any pin on the device. Designers can also use Silicon Labs' Peripheral Reflex System, in which the input value of one pin can be routed through other peripherals. This offers the flexibility to pin-out so that changes can be easily made to the board during development or between product generations. Higher level support is also possible with an embedded device. The SimpleLink CC1310 controller developed by Texas Instruments combines the RF transceiver and a microcontroller optimized for a real-time operating system, as well as separate dedicated controllers for power management and sensors. The sub-1 GHz transceiver has its own Cortex-M0 microcontroller (Figure 2) to handle time-critical aspects of radio protocols without the need for the main processor, thus leaving more resources for the user application. It can support a wide variety of modulation formats, frequency bands, and accelerators, from 625 bit/s for long range and high robustness to data rates of up to 4 Mbps, all for a variety of modulation formats from FSK. and multilevel MSK up to OOK (On-Off Keying). The 3 MHz Cortex-M48 microcontroller supports multiple physical layers and RF standards and runs the real-time operating system that configures clock and power management.
The sensor controller is a separate processor that controls various peripherals in such a way that the main CPU does not have to be activated, for example to run an A/D converter sample or query a digital sensor via SPI, thus saving power and activation time. A separate transceiver may be advantageous. Microchip's MRF49XA (Figure 3) targets short-range, two-way wireless applications in the 434/868/915MHz ISM frequency band for frequency shift keying (FSK) modulation with frequency hopping spread spectrum (FHSS) capability similar to the used in the Bluetooth standard, and can be easily integrated with Microchip's 8, 16 or 32 bit PIC microcontrollers. FHSS offers the ability to move between different frequency bands in a set order to avoid noisy channels, but may interfere with adjacent DSSS (direct sequence spread spectrum) channels. This provides great resistance to adjacent channel interference; In addition, its bit error rate (Bit Error Rate, BER), greater communication coverage of lower frequencies, together with a higher output power, make it suitable for industrial IoT designs. To minimize total system cost, the transceiver uses a low-cost, general-purpose 10 MHz crystal, a pass filter, and an affordable microcontroller. The MRF49XA provides a clock signal to the microcontroller and avoids the need to add a second crystal to the circuit board and works with PIC microcontrollers via 4-wire SPI, with interrupt and reset. Figure 4 shows the interface between the microcontroller and MRF49XA. The MRF49XA RF transceiver is also supplied on a PICtail daughter card for integration with the PIC18 Explorer 8-bit and Explorer 16 16-bit modular microcontroller development boards.
Conclusion Designs below 1 GHz can greatly extend the battery life of a wireless sensor node in the Industrial IoT. When there are thousands of nodes in a network, the operational cost of battery replacement can be prohibitive, so reducing the replacement cycle to ten or even twenty years can make a significant difference. Finding the trade-off between narrowband data rates and range on the one hand, and battery life on the other in the bands below 1 GHz, either with an integrated device or with a separate transceiver and With a low-cost microcontroller, a developer can optimize the design of a node to achieve the best performance and supply critical equipment data to the IoT.
