TY  - JOUR
AU  - Bevilaqua, Bruno 
AU  - Spohn, Marco Aurélio 
PY  - 2023
TI  - Self-Managed Federation of MQTT Brokers with Dynamic Topology Control
JF  - Journal of Computer Science
VL  - 19
IS  - 11
DO  - 10.3844/jcssp.2023.1398.1409
UR  - https://thescipub.com/abstract/jcssp.2023.1398.1409
AB  - The Message Queuing Telemetry Transport (MQTT) protocol is most used in Internet of Things (IoT) applications. The protocol implements the Publish/Subscribe (P/S) communication model. Publishers are entities providing data to a server (broker), and subscribers are those showing interest in such data. The standard MQTT scenario relies on a single broker, a potential bottleneck, and a single point of failure. The best way to scale MQTT systems is through horizontal approaches like clustering and federation. In particular, this study focuses on improving the capabilities of a self-managed federation of brokers. We present the first solution to address the dynamic management of an overlay network for the federation of autonomous brokers. The system provides the primary mechanisms for building and self-healing the federation network. We develop a new variant for the original federation protocol integrating the dynamic topology management. We present a case study as a proof of concept, showing that all designed features work as expected.