Progressive DVE Prefetching Mechanism Using Social Recommendation and Pull-Push Strategy

In order to improve prediction precision of the scene prefetching and reduce the scene transmission latency in the distributed virtual environment（DVE） based peer to peer network, we propose an efficient progressive prefetching mechanism, which combines the social recommendation with different Pull-Push strategies. First, we describe and quantify the avatar interests in the virtual environment to obtain the interest similarity between avatars and generate pre-recommend scene set. Then, we divide the avatar viewing area into the pull area and the push area, and provide different Pull-Push strategies for prefetching the scene in different area. Finally, we demonstrate the proposed prefetching mechanism using an open-source DVE platform called FLo D. Experiment results show that our approach improves the scene prediction precision, reduces the information exchange, and promotes the scene transmission efficiency and avatar roaming fluency.