SNC: Scalable NDN-based conferencing architecture

Abstract

We present a scalable NDN-based conferencing architecture for conference control and management. Conference control functions are divided into individual service modules and are loaded into different control servers. Control servers are added into the control cluster as the number of parcitipants increases. Participants are allowed to register and receive variety of conference control services from any control servers within the cluster. Participant registration and access records are maintained locally by specific control servers. This work presents essential building blocks using (1) light-weighted metadata for effective participant record management, (2) distributed and self-organized storage and replication of participant records, (3) efficient, secure and distributed conference access mechanisms, and (4) dynamic and effective control server discovery. This system enables quick and efficient cluster access by supporting seamless participant mobility and automatic server load balancing. The control cluster is self-organized and self-healing and able to scale to large size of participants. It can also be extended to host multiple conferences. Using CCNx toolkits, we have demonstrated how SNC can automatically (seamlessly) select the closest (small delay) server avoiding congested (under-attack) and far away servers. Using simulation experiments, we have also shown the effectiveness of interest response suppression mechanisms which SNC can employ.