Min-Cost Matchmaker Problem in Distributed Publish/Subscribe Infrastructures

The publish/subscribe (pub/sub) paradigm provides content-oriented data dissemination, where communication channels between content providers and content consumers are set up on the basis of interest matches between content provided by the publishers and content requested by the subscribers. In this paper, we study a distributed matchmaker system which resides on the data dissemination path, intercepts both publication announcements and subscription requests so as to match publishers to interested subscribers. We formalize an optimization problem, namely the MinCost Matchmaker problem, whose objective is to minimize the number of publication/subscription messages being maintained and transmitted throughout the system. We study the complexity of the Min-Cost Matchmaker problem and propose algorithms to solve the problem efficiently. Our simulation results show that a system implementing such a Min-Cost Matchmaker structure can significantly reduce the number of overall publication/subscription messages maintained/transmitted compared to existing systems where either publication announcements or subscription requests are broadcast. These results cast light on a new direction of designing a more efficient matchmaker structure for publish/subscribe systems.