Authenticating the TV user by biometrics

No comments

ABSTRACT

This paper presents results from a feasibility study investigating biometric continuous measurements to identify the user holding a remote control. The study investigated a combination of accelerometers, temperature and humidity sensors and measuring skin conductance/resistance to investigate if a simple pick-up of a remote control can be sufficient to identify users in a household.

Results show that eight repetitions of the movement of pick-up are enough to identify five users in a household with the necessary recognition rate validity of close to 100%. To investigate the overall user experience and acceptance of such an approach, a user study with twelve participants was performed, with an UI simulating the authentication and the personalisation of the content.

The hedonic quality is improved by such a method and pragmatic quality is reasonable compared to other remote controls and overall results indicate that users in this study accept the solution. Based on this study we developed a set of guidelines that help to implement such a solution in the user interface to support the necessary user experience dimensions of identity, stimulation, perception of privacy and overall user acceptance.

INTRODUCTION

We are living in an era of recommendations, customised services and tailored content. In order to provide content that is interesting for a (single) user, the system must identify the user – either by having the user identify herself in front of the TV by entering a pin-code, or using alternative means of identification via connected devices.

At this identification step, any modern remote control including an orientation sensor can help: it can provide the information who currently controls the TV, replacing tedious log-in procedures, by identifying the user based on biometric continuous measurements.

Identification of users based on biometrics characteristics is nothing new in the living room. Innovative projects like Bernhaupt et al (1) and Wilfinger et al (2) using fingerprint have been already conducted in 2010 and 2009.

The goal of this research was to investigate the feasibility of a remote control (currently already applied in the market) that identifies the user via simple biometric data within a few seconds after being picked up. Building on findings from the biometrics research field that showed that bio-feedback obtained by measuring skin conductance or resistance, heart beat, temperature or humidity can be used to identify a set of users, this study explored the combination of a 9-axis sensor in conjunction with Galvanic Skin Response sensors (GSR) for user identification, limited to the pick-up gesture of the remote control, either with the left or the right hand.

The main goals of this feasibility study were to investigate:

The number of different users that can be distinguished using the sensor data
Which part of the sensor data or combination of data from the sensors allows user identification with a precision of 95% or higher
The number of different users that can be identified within 3 seconds after picking up the remote control