Azzeddine Oudjehane, SAIT Polytechnic, Canada

Recent studies have focused on developing scalable architectures for social web platforms, with a particular emphasis on providing real-time access to fresh data. These architectures utilize a content-based distributed publish/subscribe (P/S) infrastructure that incorporates stable brokers to facilitate efficient communication. While P/S systems that employ gossip protocols are advantageous due to their ability to scale easily in both vertical and horizontal directions, they face significant challenges when it comes to managing end-to-end message delays and ensuring order-based consistency of messages. In response to these challenges, this paper presents a novel deadline-constrained causal order protocol designed to operate within a scalable P/S architecture. This protocol aims to guarantee that messages are delivered according to their causal relationships while minimizing the communication overhead involved in the process. To achieve this, each broker within the system maintains a two-dimensional vector that captures its knowledge of the most recent message sent by every other broker, indexed by a specific time reference (t). Furthermore, each broker is responsible for disseminating a one-dimensional vector that indicates the maximum number of gossip rounds permissible within a given timeframe, referred to as the deadline. This innovative approach not only enhances the efficiency of message delivery but also ensures that the delivery order reflects the causal dependencies inherent in the message exchanges, thus contributing to a more reliable and coherent communication framework within social web platforms.

APA:
Oudjehane, A. (2023). A Causal Order Multicast Protocol Leveraging P/S Techniques to Optimize Delivery Time Efficiency. Journal of Science and Engineering Management, 3(2), 27-36. https://doi.org/10.33832/jsem.2022.3.2.03