Spontaneous Device Association Using Inaudible Audio Signatures.

We describe a system that uses high-frequency audio signatures for spontaneous device association between mobile devices and physical end-points in smart environments. The system utilizes a method for extracting signatures from recorded audio signals, which we evaluated in two conditions: indoor, where background noise consists of workplace sounds, and urban, where the system is tested in a noisy outdoor setting. The system performed reliably both indoors (100% at distances up to 5 m) and outdoors (97.3% -- 100% at distances up to 5 m), and a slight performance drop was measured when the mobile device was allowed to move freely (89.2% indoors, 84.1% outdoors). Further, we evaluated the system's performance with an occluded receiver, and measured the battery consumption on the mobile device. Our system provides a practical and reliable means for ad-hoc service and object discovery in urban computing settings.