With the penetration of wireless devices and broadband access network, interactive multimedia streaming to handhelds has become a reality. However, it is still challenging to cost-effectively offer such services to large number of users. With the advances in capabilities of mobile devices (in terms of processing, storage and battery lifetime), we study mobile peer-to-peer (P2P) streaming for scalable interactive multimedia applications. In the network, videos are divided into segments. These segments are collaboratively and distributively cached in the mobile devices which have heterogeneous caching capacities. The segments are searched and accessed among the mobiles with different probabilities, using interactive DVR functionalities such as random seek, pause, restart, etc. The major c...[ Read more ]

With the penetration of wireless devices and broadband access network, interactive multimedia streaming to handhelds has become a reality. However, it is still challenging to cost-effectively offer such services to large number of users. With the advances in capabilities of mobile devices (in terms of processing, storage and battery lifetime), we study mobile peer-to-peer (P2P) streaming for scalable interactive multimedia applications. In the network, videos are divided into segments. These segments are collaboratively and distributively cached in the mobile devices which have heterogeneous caching capacities. The segments are searched and accessed among the mobiles with different probabilities, using interactive DVR functionalities such as random seek, pause, restart, etc. The major challenge is to design distributed algorithm to decide on which segment to cache at each mobile to achieve overall low segment access cost.

We first formulate the problem of segment caching to minimize segment access cost, and show that the problem is NP-hard. We then present OPSEC (Optimized Segment Caching), a distributed algorithm which achieves collaborative and efficient segment caching. Using simulation, we show that OPSEC indeed achieves much lower segment access cost as compared with some recent schemes. It also has low control overhead and light server load. It caches segments effectively according to their access probability, and is adaptive to network dynamics with fast convergence time.