MWC 2021: Demonstrating accurate RF sensing with Qualcomm Wi-Fi Technology
At MWC 2021, we showcased some of the latest technologies and inventions from Qualcomm, in areas like 5G and Wi-Fi and beyond. These technologies are driving and scaling digital transformation.
In this Wi-Fi demo at MWC 2021, we highlighted one of the capabilities that can be enabled with Qualcomm Wi-Fi technologies. RF sensing uses radio signals like Wi-Fi on a smartphone to detect events of interest in different environments. In this demo, we showed how our proof-of-concept algorithm, based on ‘weakly supervised’ machine learning, can accurately detect the position of a moving person without that person carrying a device. This is particularly important because detection without needing a device could unlock new use cases for home automation, security, and healthcare applications.
For this demo we deployed multiple Wi-Fi access points to cover the areas of interest. Before using the system, the user does a quick walk-through of the areas to be monitored, providing a rough house map and an indication of which room they are walking in. After that, the system is ready for predicting the position, and we can demonstrate the following examples.
We show a typical U.S. home first. The person shown on screen represents the ground truth position, while a highlighted floor indicates the room in which the algorithm predicts the person to be. The prediction remains accurate as the person moves throughout the house, and the overall accuracy of the prediction versus ground truth position consistently remains over 94 percent.
The second scenario is an enterprise building. Once again, we distinguish between the ground truth and the predicted position. As the person walks around, the positioning error is consistently below 1 meter.
This demo shows how Qualcomm’s Wi-Fi Technology can enable advanced RF sensing algorithms, leading the way forward to many more use cases.
Get the latest updates from MWC 2021: https://www.qualcomm.com/company/even...
Learn more about Qualcomm Wi-Fi technology: https://www.qualcomm.com/products/wi-fi
Chinese, English, French, German, Japanese, Korean, Spanish, and Portuguese captions available.
In this Wi-Fi demo at MWC 2021, we highlighted one of the capabilities that can be enabled with Qualcomm Wi-Fi technologies. RF sensing uses radio signals like Wi-Fi on a smartphone to detect events of interest in different environments. In this demo, we showed how our proof-of-concept algorithm, based on ‘weakly supervised’ machine learning, can accurately detect the position of a moving person without that person carrying a device. This is particularly important because detection without needing a device could unlock new use cases for home automation, security, and healthcare applications.
For this demo we deployed multiple Wi-Fi access points to cover the areas of interest. Before using the system, the user does a quick walk-through of the areas to be monitored, providing a rough house map and an indication of which room they are walking in. After that, the system is ready for predicting the position, and we can demonstrate the following examples.
We show a typical U.S. home first. The person shown on screen represents the ground truth position, while a highlighted floor indicates the room in which the algorithm predicts the person to be. The prediction remains accurate as the person moves throughout the house, and the overall accuracy of the prediction versus ground truth position consistently remains over 94 percent.
The second scenario is an enterprise building. Once again, we distinguish between the ground truth and the predicted position. As the person walks around, the positioning error is consistently below 1 meter.
This demo shows how Qualcomm’s Wi-Fi Technology can enable advanced RF sensing algorithms, leading the way forward to many more use cases.
Get the latest updates from MWC 2021: https://www.qualcomm.com/company/even...
Learn more about Qualcomm Wi-Fi technology: https://www.qualcomm.com/products/wi-fi
Chinese, English, French, German, Japanese, Korean, Spanish, and Portuguese captions available.
همه توضیحات ...