Channel hopping is one of the methods used in wireless communication to avoid interference and noise, especially in applications where relying on this for transferring information is critical. Industrial wireless communication is the best example here, where blocking important control signals is critical from the application’s point of view.
The usage of wireless communication for process control is something new in the industrial area, mainly from the lack of trust in its abilities. Until recent years, wireless has been considered lacking robustness. The current situation is slightly different, with the availability of new technologies and protocols for wireless communication like the IEEE 802.15.4 standard, which is being constantly updated, more powerful, less power consuming and secure micro-controllers based on the Arm Cortex-M family of processors. A channel hopping technique is one of the ways wireless communication can be made more robust and trustworthy for industrial applications.
With the latest developments and advancements in the semiconductor area, the amount of processing power available per module has increased significantly allowing the implementation of complex control systems, having multiple inputs and outputs paired to devices distributed in larger areas. In some cases, connecting these devices cannot be done with wires, either because they are too far apart, or, because it is not allowed. Some examples were wireless is used for process control include:
- In-satellite communication. By using wireless, the dry mass of the satellite is reduced, which would lead to a lower manufacturing cost. 
- Explosive atmospheres or environments (as defined by directives 99/92/EC and 94/9/EC).
- Sensors and actuators being hundreds of meters apart and placed in less accessible locations.
The implemented channel hopping technique was tested on a control system comprised of a controller and a simulated BLDC motor with the following continuous transfer function:
To simulate as best as possible the process, the transfer function (1) was discretized using the Tustin method with a minimum achievable 20-microsecond sampling time obtaining the discrete transfer function below:
To have good simulation of a real BLDC motor, we have implemented the discrete transfer function on the NUCLEO-F767ZI development board from STMicroelectronics shown in figure 1. It features the STM32F767ZI high performance micro-controller based on the Arm Cortex-M7 core, running at 216 MHz, having 2 MB of FLASH, 512 KB of RAM and various digital, analog and communication interfaces.
For the controller design, we have used the module criteria with the following controller transfer function:
and, by applying the Tustin method with a 15.625 millisecond sampling time, in discrete form:
A simple diagram depicting the control system is shown in the figure below. Two Andustria codename “Hydrogen” boards are used, one implementing the controller and the interface to the PC and the other, applying the command to the simulated process, which in our case, is the BLDC motor simulation running on the NUCLEO-F767ZI development board.
Although the transceiver used in this application transmits data using O-QPSK modulation, we have selected the channel definition and numbering for SUN FSK operating mode #1 and #2 as a reference, as it is defined in the IEEE 802.15.4-2015 standard. Therefore, we have defined 17 channels ranging from 868.2 MHz to 869.6 MHz with a 400 kHz channel space. The channel hopping sequence used to test the wireless control loop described in this paper is the one described in the IEEE 802.15.4-2015 standard in the TSCH (time slotted channel hopping) channel access method: the first 17 outputs modulo 17 of the 9-bit linear feedback shift register with the polynomial x9+x5+1 and a starting seed of 255. Using this method, we were able to randomly shuffle the defined channels and obtain the following hopping pattern:
In the above table, channel identifiers range from 0 to 16 having 863.2 MHz the central frequency of channel 0. For compliance with the ETSI EN 300 220 standard for wireless communication in the European Union, we have chosen the operational frequency bands covering the 863-870 MHz frequency range, namely band K, L, M, N and P. Also, we have set the output transmit power of the radio transceiver to be under 25 mW (14 dBm) and using the listen-before-talk algorithm or polite spectrum access, how it is specified by ETSI, to check for other ongoing transmissions before starting our own data transmission.
We have tested the communication between the two wireless nodes by placing them outdoors, 20 meters apart transmitting frames at the maximum allowed transmit power of 14 dBm. We have also placed at 10 meters apart of the wireless nodes, another wireless node in continuous transmission to observe the behavior of the wireless control loop when interference is present.
The figure below presents the output of the process in blue and the command, in red, as it is sent by the controller node. The set-point of the control loop is a rectangular signal with a 1000 RPM low, a 2000 RPM high and a period of one second. On the X axis we have the time in seconds and on the Y axis we have the output of the system in volts, as it is a simulated process. We have tested the resistance to interference by setting another Andustria Hydrogen board in continuous transmission mode on one of the operating channels at maximum power. The system was not affected by the interference and it performed normally in the expected parameters.
To observe the behavior of the wireless control loop when interference in present, we have placed another Andustria Hydrogen board in continuous transmission on the first channel. Also, the wireless control loop was configured to use only one channel for frame transmission to simulate the worst-case scenario. The output and command of the control loop are shown in the figure below.
As is can be seen, the overall performance of the wireless control loop is reduced, having a greater process overshoot caused by blocked command frames. We have observed that the control loop continues to work when the transmit power of the jamming wireless node is set to maximum -7 dBm. When configuring the jamming node with a greater transmit power, the communication in the wireless control loop is completely blocked.
Although throughout this study we obtained good results, the behavior of the wireless control network can be improved even further by implementing adaptive channel hopping mechanisms to blacklist occupied channels. Besides the advantages presented, using channel hopping has some disadvantages: for the solution to work properly, a time synchronization mechanism that requires precise and more expensive crystals must be implemented. This also implies using more computing power needed also when the hopping scheme is of high complexity.
The full study about using channel hopping in wireless process control has been submitted and accepted by the ICSTCC 2019 IEEE conference which will be held in Sinaia, Romania between the 9th and 11th of October 2019.
Conference webpage: http://icstcc2019.cs.upt.ro/