A hybrid priority-based video-on-demand resource sharing scheme

Video-on-demand (VoD) environments frequently batch video requests to decrease I/O demand and increase throughput. Since users may leave due to waiting too long, a good video scheduling policy has to consider not only the batch size, but also the user defection probabilities and waiting times. Moreover, a practical VoD resource sharing scheme should endeavor to provide some free streams to serve a high-priority clients requests immediately, since the high-priority clients might pay for the requested video. To tackle these problems, this work proposes a hybrid resource sharing model that integrates controlled multicasting and batching. The proposed hybrid model applies a bandwidth borrowing and reserving scheme to give high-priority clients a prompt service, while still providing low-priority clients with a reasonable service. Furthermore, a novel probability model-based scheduling policy is proposed to alleviate the user defection behavior and unfairness issue. Experimental results demonstrate that the proposed resource sharing scheme is effective and feasible in terms of blocking probability of high-priority clients, the defection probability, service delay time and fairness to low-priority users.