Scalable service discovery in ubiquitous and pervasive computing architectures: A percolation-driven approach

Abstract

Listen

Translate article

The diffusion of ubiquitous and pervasive computing applications, providing advanced services anytime and everywhere across heterogeneous mobile networks, is fostering the development of new architectures and models, based on the emerging peer-to-peer paradigm, to allow the effective use of multiple replicated resources and services without the scalability and adaptiveness limitations characterizing traditional client–server organizations. In this scenario, effective service discovery facilities are needed to minimize the administrative overhead and increase the overall usability and perceived service quality. However, most of the currently available implementations lack efficiency due to the highly dynamic and hierarchically flat nature of the underlying wireless mobile networking environment, that make any traditional solution based on centralized schemes and statically defined roles practically unfeasible. Accordingly we proposed a novel service discovery approach based on a fully distributed and parallel search model, that does not require any centralized intelligence, fixed roles and stable communication infrastructure. It exploits, in a dynamic scale-free scenario characterized by competition for links and balanced node insertion and removal rates, a widely known random walk-based search paradigm benefitting from bond percolation in power law organizations to automatically limit the search space (by shielding low connectivity nodes from search traffic) and drastically reduce the total control overhead. We analyzed the effectiveness and performance of the proposed solution by using discrete event-driven simulation, whose results were satisfactory also in highly mobile ad-hoc environments. This indicates that the proposed ideas are promising and deserve further exploitation by practical application solutions designers.