Nature-Inspired Interference Management in Smart Peer Groups

Abstract

Applications in intelligent environments are context-aware and context-altering in order to support users in their everyday tasks. In multi-user environments with shared context, interferences are likely to occur. An interference is an application-induced context that forces other applications to react. In our COMITY project, we developed interference detection and resolution algorithms. However, especially the constraint satisfaction-based resolution algorithm is computationally expensive. It requires a full-fledged machine in order to achieve runtimes suitable for interactive systems. In spontaneously formed smart peer groups, such infrastructure is not given. In this paper, we present an approach for detecting and resolving interferences in smart peer groups. The approach is inspired by two coordination mechanisms from nature. We map the problems of interference detection and resolution to the rules of the flocking mechanism, show how to implement a flocking-based interference management, and present a local inhibition-based leader election for smart peer groups in order to elect the coordinating entity. Finally, we show the feasibility of our approach by evaluating our prototype.