Consistency in replicated continuous interactive media

Abstract

In this paper we investigate how consistency can be ensured for replicated continuous interactive media, i.e., replicated media which change their state in reaction to user initiated operations as well as because of the passing of time. Typical examples for this media class are networked computer games and distributed VR applications. Existing approaches to reach consistency for replicated discrete interactive media are briefly outlined and it is shown that these fail in the continuous domain. In order to allow a thorough discussion of the problem, a formal definition of the term consistency in the continuous domain is given. Based on this definition we show that an important tradeoff relationship exists between the responsiveness of the medium and the appearance of short-term inconsistencies. Until now this tradeoff was not taken into consideration for consistency in the continuous domain, thereby severely limiting the consistency related fidelity for a large number of applications. We show that for those applications the fidelity can be significantly raised by voluntarily decreasing the responsiveness of the medium. This concept is called local lag. It enables the distribution of continuous interactive media that are more vulnerable to short-term inconsistencies than, e.g., battlefield simulations. We prove that the concept of local lag is valid by describing how local lag was successfully used to ensure consistency in a 3D telecooperation application.