net-VE에서 EKF-DR 경로예측을 이용한 동적데이터 공유 메카니즘 구현

Abstract

net-VE(Network Virtual Environment) is a system that links the distributed network to the virtual reality technology and offers 3 dimensional world to help distributed multi users mutually interact through real time networking. In the net-VE system, distributed network technology, such as network broadband, fairness, distributed mutual interaction, synchronization, resource management and expansion, and other different technologies become organic within a single system. Additionally, synchronization resulting from time delay requires technologies regarding collision check at a natural reality, accord and resolution. Consistency at a distributed virtual environment of the client-server structure is maintained by endless exchange of static information among distributed clients. The regular transfer of static information brings traffic overhead of network. The precise way for network users to know the others’ static is to transfer the packet by hand shaking for each frame, which takes a big burden of the synchronization and reduces the velocity. As the current static transferring value, , does not arrive at the time, , without packet delay time, a reconciliation between the scope setting permitting errors and broadband for updates is needed during a cycle interval of the synchronization. There are two approaches solving the bandwidth problem within the net-VE system where you can lessen the occurrence of dynamic data or minimize the received messages and decrease the bandwidth of the overall virtual world. The main technique that minimizes the received traffic of dynamic data and lessens the overall bandwidth is the Dead-reckoning algorithm. This is first of all a method that transfers current coordinates and speeds constantly and synchronizes them, then estimates the location during the synchronization for each client. From the recipient, they can take this value and apply Dead-reckoning and estimate the observation location till the next synchronization. While the movement becomes unnatural when the synchronization time becomes long. As the movement locus becomes an aggregation of straight lines, the merit is that the error does not accumulate as it lessens the errors whenever the synchronization occurs. Secondly, as the Dead-reckoning of DIS(Distributed Interactive Simulation) calculates the location of its own object, it transfers and performs synchronization when the error becomes bigger than the limit value, so that while there is less delay when the dynamic data does not change directions aggressively, efficiency is significantly decreased. As mentioned above, although the Dead-reckoning method is the most used and efficient method, it is difficult to perform estimations for single-shot and unpredictable events. So, when a new participant is linked to the 3D scene of the network’s virtual world, and as the frequent broadcast that maintains consistency and has mutual reactions with other participants and sharing initiates jerks that are caused by network delay, and collision and late information transfer, it is important to reflect the shared event among multi users through location reinforcement by prompt path estimation. A Kalman filter measures the status by repeating the estimation and filtering every level; as this filter uses the most recent estimated status information, it can calculate the best estimation possible for future statuses. This study will design a path estimation model that uses EKF-DR(Extended Kalman Filter) which measures the status by repeating the estimation and filtering for each level and will estimate the path of the dynamic data at the client cache. Also, it will decrease the estimation error and increase the path estimation efficiency by using the actual measured value of the movement locus and the measured error value information. It will finally simulate the path estimation that uses the Dead-reckoning algorithm of the DIS and EKF-DR and compare the results. When a new client is linked to a 3D scene of the network virtual world, it interpolates the previous location with the Dead-reckoning path prediction algorithm of DIS to maintain consistency and shows the natural dynamic data sharing scene of the 3D virtual world. The shared dynamic data of the 3D virtual environment is implementation using the DIS/X3D/Java EAI.