WearID: A Wrist-worn Backscatter Reader to Measure Interactions with the Physical World

Overview

The ability to monitor tactile interactions between people and objects is important for a range of applications including home automation, health and wellness, smart spaces, augmented reality, and tele-rehabilitation. Perhaps the most promising way to monitor such interaction is by using passive UHF RFID tags that can be cheaply attached to objects. Recent work on interactive RFID systems has shown that such tags can be used not only for identification, but also for sensing the type of interaction by analyzing low-level channel parameters like phase and RSSI. This makes it possible to detect whether the interaction involves touching a tag, blocking a tag, or moving a tagged object,

In this work, we explore the challenges in designing a practical wearable backscatter reader that operates within the form-factor and power constraints of a smartwatch-class device, and the opportunities that such a device presents for monitoring interactions. Our work is inspired by early work on the design of glove and bracelet-sized NFC readers for activity recognition iBracelet. However, an NFC-based approach has two key limitations: a) NFC operates at much lower frequency than UHF RFID and therefore does not provide useful phase information that is crucial for a range of interactive RFID-based applications, and b) NFC operates at short ranges of a few centimeters with smartwatch form-factor loop antennas, and is therefore more suited for intentional interactions (e.g. credit card payment). In contrast, UHF RFID readers operate at a wavelength of roughly 30cm and does not have the inherent range limitations of NFC, which makes it is more suitable for measuring tactile interactions with tagged objects. We refer to our design as WearID.

Hardware

The first key consideration in the design of WearID is optimizing power consumption. Here is how the hardware looks like. The form factor and weight make the system suitable to be worn on wrist.

The receiver is mainly passive components, which reduce sensitivity of the system and as a resut reduce the reading range of the system. We also reduced the RF output power to further reduce power consumption, while losing more range, but all in all, the range loss comes in handy in detection of objects in close proximity of user's hand.

Hardware design files can be found here.

Body adjacency effects

To design a wearable device requires studying the effects of body on the system carefully. Body mostly affects the RF channel and impedance of RF components which results in reduction of output power. In following figures, we have measured spatial locations where RF power generated by the reader drops down to -25 dBm. Figure on the right shows the coverage for WearID in absence of hand, and in left figure, the hand is presnet inside the watch and you can see how much attenuation it causes.

Feature extraction

Figures below show an example of segmentation, digital filtering, envelope detection, and feature extraction.

System is able to distinguihs between grabbing a tagged object, releasing a tagged object, touching a tag, and passing by a tag, with 86% accuracy.

People