Using User Perception to Determine Suitable Error Thresholds for Dead Reckoning in Distributed Interactive Applications

摘要

Entity state update mechanisms are readily employed in Distributed Interactive Applications (DIAs), particularly in networked games. These mechanisms use prediction techniques in order to reduce the number of update packets sent across the network, while maintaining a high level of consistency from the remote user's point of view. These mechanisms only send update packets when the local user's actual behaviour differs from the predictive behaviour by a certain value, often referred to as the error threshold. In practice, this value is arbitrarily chosen and typically reflects what 'appears' to be suitable. It has been illustrated in various other media that psycho-perceptual measures can be used to greatly improve compression techniques, while maintaining satisfactory end-user experience. The best example of this is the MP3 compression used in audio. This paper describes a preliminary study designed to collect information relating to a subject's perception of a networked computer game. The main motivation behind this work is to investigate if psycho-perceptual measures can be used to obtain appropriate error threshold measures for entity state prediction mechanisms. Here, we employ Dead Reckoning as it is the simplest and most commonly used of these mechanisms in distributed gaming. The experiment outlined in this paper attempts to determine if an error threshold can be chosen from the users' perception, where user feedback is determined via linguistic variables. Furthermore, the effects of convergence, the speed of the entity and the shape of the entity trajectory are also examined from a psycho-perceptual viewpoint. The results are presented within.