A Consistency Model for Distributed Virtual Reality Systems

As a further development of works published in previous DepCoS-RELCOMEX conferences, this paper is devoted to one of the most important problems taking place in distributed virtual reality systems (DVR systems): maintenance of consistency among the processes within DVR system, which is required to ensure that all users have a sense of presence in a common shared virtual world. In this paper an event-based model of distributed computations in DVR systems is discussed. Then we consider different approaches to consistency definition, such as causal, observational and space-time consistency. In order to reach high space-time consistency in DVR system, large amounts of geographically distributed data should be replicated in real time. However, due to hardware limitations it is still impossible to achieve absolute consistency. Moreover, in most cases the absolute consistency is not necessary. In this work we propose a data consistency model based on selective consistency principle allowing to ensure consistency only on those parameters which are in most importance for specific task. Within the model the issues of data representation, storage and communication are discussed.