ODDMS: Online Distributed Dynamic Meeting Scheduler

With advent of social media, need for scheduling very large meetings while proving a degree of privacy to the participants has become an important problem. Existing solutions based on a global calendar expose individuals data to the calendar provider and thus are unsuitable for open meetings with quorum constraints. We propose an online distributed dynamic meeting scheduler (ODDMS). It is able to efficiently schedule meetings involving a large number of participants, without having complete knowledge of individual participants and their preferences, thus preserving privacy. The algorithm uses a modified negotiation-based distributed schedule that resolves the problem of deadlock and contention using hidden naive Bayes learning method. We compare our work with a baseline centralized algorithm and two existing algorithms based on voting mechanism and naive Bayesian methods. Simulation studies show that ODDMS performs similar to baseline centralized algorithm under light load condition and significantly outperforms the existing distributed algorithms under heavy load condition.