A reliable and scalable overlay multicast architecture for large-scale video surveillance applications

In the paper, we propose a hybrid overlay multicast architecture (called RSOMA) to cascade existing surveillance information islands to solve video transmission and a large number of users accessing under heterogeneous network environment. Through connecting the pre-deployed streaming delivery servers in existing system and organizing them into a mesh-first overlay backbone service domain, video streaming are forwarded to a set of clients. Clients locate a nearby streaming delivery server and tap into one or more video streaming sessions via the server. Clients, who access a uniform video streaming will form hierarchical clusters and a peer-to-peer application-layer multicast protocol is used for the communication between the clustered end-users. We demonstrate the efficacy of our architecture via a set of simulation experiments. Results show that comparing with existing streaming delivery mode in surveillance systems, RSOMA is able to provide efficient resource share with less control overhead, especially for large-scale video surveillance applications. It alleviates the stress around streaming servers and provides scalability. The latency incurred and redundant packets duplicated in data transmission are low.