Reflections on IoT

There is no question about it that Internet of Things (IoT) devices are making a huge impact on the world around us. Everything from industry to agriculture, smart homes, and wearables has been transformed by the wide availability of inexpensive, powerful, connected compute systems. There are still some application areas that could greatly benefit from this technology, but in which there are limiting factors that prevent its practical application, however.

Widely distributed networks of sensors, for example, are challenging to implement. This is in large part due to the fact that the sensors will often be in remote areas where no connection to the power grid is readily available. Needing to rely on solar, wind, or other intermittently available sources of power, energy efficiency becomes a critical consideration. Advances in processing units and sensing systems have gone a long way toward keeping these IoT devices online, but when it comes to wireless communication, batteries can be drained faster than you can say Jack Robinson.

Backscatter communication technologies are promising candidates that could help to solve this present issue. These systems do not generate the communications signals themselves, rather they reflect signals that are sent by a remote transmitter. As the signals are reflected, they are also modulated. By changing the signal in this way the device can encode information in it before it is returned to the transmitter. The advantage of this slightly convoluted approach is that it requires very little energy on the part of the device using the backscatter transmitter.

However, determining the best way to reflect signals can be difficult because real-world results can differ from the simulations that are used to plan networks. This has limited the adoption of backscatter communication systems, but the work of a team led by researchers at Pusan National University may soon change that. In this work, they have developed a better way to model how devices should reflect signals. Additionally, they showed how multiple antennas can be used for simultaneous signal transmission and reception. Together, these innovations have increased transmission rates, reduced error rates, and also enhanced energy efficiency by 40 percent.

The key to better modeling signal reflection was the use of transfer learning. This is a machine learning technique that uses knowledge gained from one task to improve performance on another. The researchers first trained an artificial neural network with simulated data to understand how the devices behave under different conditions. They then trained it with real data to make better predictions.

This approach allowed them to predict signal reflections with high accuracy, leading to more efficient data transmission. They used 4-QAM and 16-QAM modulation schemes to optimize this process, keeping energy consumption very low (below 0.6 mW).

The team also designed a communication system with multiple antennas, enhancing signal reception and efficiency. Using dual-polarized Vivaldi antennas, they achieved high signal gain and good performance in suppressing unwanted signals.

The system was tested in the 5.725 GHz to 5.875 GHz C-band of the Industrial, Scientific, and Medical radio band and demonstrated high efficiency and reliability in data transmission. These findings hint that a more reliable and efficient backscatter communication system could soon be made available for a wide range of IoT use cases.