The rise of the Internet of Things (IoT) is driving a significant increase in the volume of data generated at the network edge, where requirements for low latency and efficient resource management are critical. Edge computing emerges as a promising response to the limitations of the centralized cloud model, by bringing computing resources closer to the data sources. In this context, message brokers play a key role, but existing solutions are rarely adapted to edge environments and mostly focus on the publish/subscribe model, without providing a unified solution. This work introduces LuffyMQ, a geo-distributed messaging system built on RabbitMQ and extended with specific modules to address edge constraints. Our approach supports both major asynchronous models : publish/subscribe and point-to-point with message queues (work queue). Intelligent strategies for load balancing and synchronization node placement are proposed. Experiments conducted in a geo-distributed environment demonstrate the effectiveness of LuffyMQ. In the point-to-point model, the system reduces average latency by up to 75% compared to RabbitMQ’s native strategy and centralized configurations. In the publish/subscribe model, it maintains latency close to the lower bound while improving scalability, achieving up to 200% more messages processed under heavy load.
| Date | 4 Nov 2025 |
|---|
| Original language | French |
|---|
| Awarding Institution | - École de technologie supérieure
|
|---|
| Supervisor | Julien Gascon-Samson (Supervisor) |
|---|
Snene, A. (Author),
Gascon-Samson (Supervisor),
4 Nov 2025Student thesis: Master's thesis › Master in Engineering: Engineering