Time-Space Consistency In Large-Scale Distributed Virtual Environment

Time-space consistency is one of the fundamental requirements of the distributed virtual environment. With the boosting of the relevant techniques, more and more distributed virtual environments are employed. Now, a large-scale simulation system contains great numbers of simulation entities, which are more lifelike in virtual environment. Consequently, time-space inconsistency becomes one of the most prominent problems that are calling for solutions in large-scale distributed virtual environment. Thus, it is necessary to study time-space consistency of virtual environment in depth, and put forward efficient mechanism to ensure the time-space consistency of the distributed virtual environment.This dissertation first analyzes the causes of time-space inconsistency, and then studies the effects of the time-space inconsistency on the distributed virtual environment. While considering the time-space consistency in the large-scale distributed virtual environment, this dissertation summarizes the relevant research works on time-space consistency in the large-scale distributed virtual environment. The research works in this dissertation emphasis on the time-space consistency model, entity utility model, and time-space consistency control method. Based on the research works, this dissertation also research on the implementation technology of time-space consistency controlling prototype based on HLA/RTI, and other relevant technologies. The main contributions of the dissertation are as follows.(1) Discussed and studied the theory about time-space consistency models. Then, this dissertation presented an entity state computation model and a time-space causal consistency model in the distributed virtual environment. The entity state computation model describes causal relationship among entity states and events. The time-space causal consistency model describes the main requirement for time-space consistency in distributed virtual environment. This dissertation also presents a method to compute the grade of time-space causal consistency. The theory base is provides for the research on time-space causal consistency methods.(2) The entity utility model available can't provide enough information for time-space consistency control because it describes entity utility just with a numerical value. Based on fuzzy description logic, REL-UF, a new entity utility model is presented. This model provided the mechanism to describe entity utility from multi-views. This model also defined the entity fuzzy categories and entities fuzzy relations, satisfying the information demand of adaptive time-space consistency control. The results of experimentation shows that compared with the available entity utility model, this model can provide better and more enough adaptive information for the time-space consistency control.(3) To guarantee time-space causal consistency, an adaptive utility driven consistency mechanism (UDCM) is presented which can choose control technology or adjust control parameters according to entity utility. There are a few new technologies and improved technologies are integrated in UDCM. These technologies mainly include group synchronization method, local lag, entity migration method, and time warp method.Group synchronization is presented. This method can dynamically create or destroy synchronization group, and limit synchronization among a few entities and a period of simulate time. It can improve time-space consistency while guarantee the real-time demand of system.Improved local lag method. Lag time is decided by entity utility and network delay. Test result proves that compared with the local lag available, this method can improve system responsibility while guarantee the same time-space consistency.Devise entity migration method based on expectation. This dissertation adopts the distance between entities and the type of entities to decide whether to migrate an entity. And devise the entity state update method based on priority. With this migration method, a lot of migration of no use is avoided, and the process migration is accelerated.Improved time warp technology. Time warp is both very time consuming and memory consuming. In this dissertation, current entity utilities and history entity utilities are adopted to decide which entity states are included in the current snapshot. The entity states that are less possible to affect simulation correct will not be recorded in snapshot. Thus, memory and computation resource consumption can be effectively reduced.(4) Relevance filter method can reduce the network load. Presented a relevance filter method, which is based on multi-level grid. By this method, a region is described with three grids; each of the grids has the different cell size. The method uses large cells to describe the center of a region, and the small cells are used to describe the margin of a region. Thus, compared with even grid, the method can describe a region with better precision. The method also reduces the execution time consumption also.Based on the above research, a time-space consistency control prototype system based on HLA/RTI, C-Controller, is developed. The time-space evaluation system based on HLA/RTI is also developed, which can test the effect of C-Controller in distributed virtual environments. The experimentation result shows that C-Controller can get factory time-space consistency.