Distributed discovery using preference for query hits

Abstract

We consider peer-to-peer and distributed networks where a large number of objects (such as users, information and applications) are distributed and they move dynamically. In such a distributed network environment, the ability to discover objects that match given query criteria, such as a set of keywords, is highly desirable. We propose a discovery mechanism that is scalable, adaptable and robust to a dynamic network environment. The proposed discovery mechanism is fully distributed (without using any centralized entity in the network) and includes novel features that utilize information about the preference of the discovery originator in order to provide higher quality hits to discovery originators selectively, resulting in less traffic than an existing broadcast based discovery mechanism. Our study methodology of the proposed discovery mechanism involves extensive simulations that compared our discovery mechanism with the performance of an existing broadcast based discovery mechanism.