Caching for data availability in mobile P2P streaming systems

Mobile peer to peer multimedia streaming is getting a lot attention, especially in instantaneous mobile networks in which users want to share their multimedia content as live streams from source peer to destination peers. The general public infatuation towards these systems is due to numerous advantages of the streaming mode: latency reduction, constant storage requirement at the receiver side, particularly suitable for low storage devises, copyright preservation, etc. Because of the temporal dimension of multimedia streams, such systems suffer however from the data unavailability problem i.e., streams that are currently viewed will not be consumed until the end because of the unpredictable source peer departure. One way to handle this problem is to use a cache to prefetch parts of the currently viewed stream so to ensure their availability. Nevertheless, when to start caching? What size of stream file need to be cached? remain open research issues that need to be addressed for maximum user satisfaction with limited cache size. This paper addresses these issues by proposing and using probabilistic users´ profile in the network to decide whether there is a need for pre-caching or not. It estimates the waiting time of senders and receivers as well as viewing duration of stream file for decision making. It demonstrates that the introduced preconditions have saved a considerable amount of caching resources.