Topic-Oriented Bucket-Based Fast Multicast Routing in SDN-Like Publish/Subscribe Middleware

IEEE Access

卷:8

Jonathon Wong¹  Hans-Arno Jacobsen¹  Yang Zhang²  Yulong Shi²  Junliang Chen² 

[1] Middleware Systems Research Group, University of Toronto, Toronto, ON, Canada;

[2] State Key Laboratory of Networking and Switching Technology, Beijing University of Posts and Telecommunications, Beijing, China;

关键词: IoT;    publish/subscribe;    middleware;    SDN;    bucket;    multicast;   

DOI  :  10.1109/ACCESS.2020.2994268

来源: DOAJ

【 摘 要 】

In traditional IP-based publish/subscribe middlewares, a detour to overlay network is demanded to match events with defined filters, which introduces more latency overhead for delivering events from publishers to subscribers. The emerging Software Defined Networking (SDN) creates boundless possibilities to improve the efficiency of event delivery because of its centralized control and customized programmability. In this paper, we propose a topic-oriented bucket-based fast multicast routing to improve the efficiency of delivering events in SDN-like publish/subscribe middlewares. First, we design an SDN-like publish/subscribe middleware hierarchical architecture with an implementation framework in SDN controller to deliver events. Topic encoding, topic prefix matching, and the subscription coverage relationships between topics are considered to reduce the number of flow entries and improve the matching abilities of SDN switches. Then, a topic-oriented multicast tree construction algorithm is proposed to build publish/subscribe overlay networks with the minimal total link delay of event transmission and minimal switches in SDN controllers to deliver events fast. A topic-oriented bucket-based multicast forwarding algorithm is designed to achieve efficient multicast forwarding in SDN switches. Finally, experiments are conducted to verify that our construction algorithm has the minimal total delay of event transmission and our bucket-based multicast forwarding algorithm is effective.

【 授权许可】

Unknown