NeTS: Small: Collaborative Research: Logical Localization for Mobile Devices through Ambience Sensing

University Of Wisconsin-Madison, Madison WI

Investigators

Abstract

The notion of location is broad, ranging from physical coordinates (latitude/longitude) to logical locations (like Starbucks, WalMart, museums). Logical locations are gaining importance because they express the context of the user, allowing applications to carry out context-specific interaction. For instance, a person entering Starbucks may receive an advertisement for purchasing coffee. Evidently, such applications expect to learn the existence of the user within the confines of a logical place (like Starbucks). The rich body of localization literature based on GPS/WiFi/GSM is dominantly physical in nature, and translating them to logical locations is error prone. This is because a dividing wall may separate two logical locations and an error margin of few meters may place the mobile device on the incorrect side of this wall. This proposal proposes a unique system called SurroundSense that breaks away from physical localization and investigates direct methods of recognizing logical locations. The main idea in SurroundSense exploits the observation that a logical place has a distinct ambience in terms of its background sound, light, color, RF signals, and layout. The attributes of the ambience can be sensed through mobile phone sensors and suitably combined to form a fingerprint of that place. This fingerprint can then be used to identify the logical location of the phone by comparing it against an existing database of location-fingerprints. As an example, a Starbucks may include sound signatures from coffee machines and microwaves that are different from the clinking of forks and spoons in restaurants. Similarly, signature colors in the decor may vary from store to store, lighting styles and store layouts may be different, and so will the WiFi SSIDs audible at that location. The combination of all these attributes, a fingerprint, is likely to exhibit reasonable diversity for high quality, logical localization.