Fault avoidance and recovery for distributed multimedia multicast

We present a fault avoidance and recovery algorithm designed for distributed multimedia multicast. Our algorithm is focused on fault avoidance to provide heterogeneous quality of service (QoS) guarantees to applications. We adopt a receiver-oriented approach to fault avoidance and recovery to achieve scalability. The receiver-oriented approach also addresses the heterogeneity of distributed multicast. The QoS of an application is used as an important parameter for fault detection. Both the fault avoidance and fault recovery aspects of the algorithm are based on the estimate of the QoS parameter. To suppress the overhead of multicast, a two-phase commit protocol is used in the algorithm. The key contribution of the algorithm lies in combining proactive and reactive control: (1) to avoid congestion in the network prior to the occurrence of error; (2) to provide heterogeneous QoS to applications by QoS-based fault detection and control; (3) to achieve scalability of distributed multimedia multicasts by receiver-oriented fault control and message suppression. Our algorithm makes very few assumptions about the underlying network, hence it can be easily integrated with popular multicast protocols.