To successfully deliver the IPTV content to end-users, robust and resource-efficient error recovery techniques are required that can overcome the problems encountered during the end-to-end delivery process. To achieve this objective by also enabling quick error recovery, one preferred approach is to install a dedicated server, referred to as the error recovery server, between the head-end office and the end-users, to handle the repair packet requests coming from the end-users. However, as the number of users connected to the error recovery server increases, the probability of the server being overloaded by too many requests also increases, leading to scalability problems and increased response times. As a result, we cannot anymore guarantee the timely delivery of the repair packets, which would lead to unacceptable performance degradations at the end users. To prevent this from happening, we can minimize the recovery load at the error recovery server by using a peer-based error recovery approach in conjunction with a server-based error recovery approach. In this paper, our objective is to develop an effective joint server-based and peer-based error recovery solution that minimizes the probability of the server entering a non-responsive state by pushing the error recovery load towards the end-users, whenever needed and possible. The information on the perceived network characteristics is used to make resource-optimal decisions to find the recovery-peers. By taking into account the impact of latency on the peer-selection policy, we improve the fairness in the network and avoid over-utilizing the resources at the error recovery server. Through simulations, we show the significant performance improvements that can be achieved with the proposed joint error recovery framework when compared to a server-based error recovery framework.
展开▼