PINCH: Self-Organized Context Neighborhoods for Smart Environments

Abstract

Today's "smart'" domains are driven by lightweight battery operated devices carried by people and embedded in environments. Many applications rely on continuous neighbor discovery, i.e., the ability to detect other nearby devices. Application uses for neighbor discovery are widely varying, but they all rely on a protocol in which devices exchange periodic beacons containing device identifiers. Many applications also ultimately involve assessing and adapting to context information sensed about the physical world and the device's situation in that world (e.g., its location or speed, the ambient temperature or sound, etc.). In this paper, we define Proactive Implicit Neighborhood Context Heuristics (PINCH), which leverages unused payload in periodic neighbor discovery beacons to opportunistically distribute context information in a local area. PINCH's self-organizing algorithms use limited local views of the state of a one-hop network neighborhood to determine the most useful type of context information for a device to sense and share. In this paper, we develop the algorithms, integrate an implementation of PINCHwith a smart city simulator, and benchmark the tradeoffs of self-organized local context sharing with 2.4GHz neighbor discovery beacons.