Architecture of an intelligent system Sensor box Sensor box GPS box
Sensor box
Control box
User end
Managing system faults
One of the hallmarks of IoT solutions is the use of multiple, inexpensive sensors. The Texas A&M system capitalizes on having multiple sensors in a way that improves resilience: it is capable of fault detection and fault tolerance. Even though soil moisture levels at separate sensor sites are often different, the correlation depends on the distance between them. The correlation among data from multiple sensors naturally provides a means for checking each other.
Usually, we can infer the data at site A from looking at data from two other sites (B and C). Even though the inferred result is normally inaccurate (otherwise the sensor at site A becomes unnecessary), it helps to monitor if there is large data deviation at site A. In other words, a large discrepancy between the inferred data and the actual data implies a fault. If the fault is transient (e.g., communication fault due to interference), the system temporarily uses the inferred data. By doing so, occasional faults can be tolerated by the overall system. When the system detects a fault (e.g., sensor battery depletion), the system sends the user a message through an internet-based interface such as a phone app. The fault detection and fault tolerance calls for human intervention only when it is very necessary.
On-farm IoT communication
Communication is an essential part of every IoT solution, and low-power wireless communications are another hallmark of IoT. A great variety of technologies exist both for device-to-device and device-to- internet connections. In the Texas A&M prototype, two subnetworks compose the communication of the system: one is the wireless sensor network that connects to the irrigation controller, and the other is a local area network that connects to the internet. The wireless sensor network uses ZigBee technology so that sensors communicate directly with the irrigation controller (i.e., Intel Edison IoT processor), where the machine learning-based control software executes.
The sensors communicate with the controller using industry standard formats (e.g., SDI-12 and NMEA-0183). The connection between the controller and the internet is through Wi-Fi, which also allows a remote user interface. This on-farm IoT processor is different from other IoT systems in one significant way. By having the machine learning algorithms run on- farm (instead of online), the system can withstand loss of communication and thus is more fault tolerant.
The system produced at Texas A&M incorporates each of the key features of an IoT solution: sensing, communication and analytics. These key features are the appeal of IoT. Combining sensing, control and analysis make smart automated systems possible.
Jiang Hu, PhD, is a professor and co-director of graduate programs for the electrical and computer engineering
department at Texas A&M University. His research is focused on energy-efficient computing systems, including power
management of multi-core processors and data centers and low power computing.
Charles Hillyer, PhD, is the assistant professor and extension
specialist of irrigation engineering at the University of Texas A&M AgriLife Extension Service at Amarillo. He serves as a member of the IA Editorial Committee.
irrigationtoday.org 15
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40
