A 360° video is a medium that offers panoramic immersion but requires high resolution and significant computational power, especially in multi-user contexts. To improve processing and near real-time transmission, techniques such as tiling are used. These tiles, encoded at different qualities depending on the user’s gaze, allow for bandwidth reduction by transmitting only the relevant areas.
The pipeline, often sequential, is usually accelerated by grouping steps on a single powerful machine. This generates monolithic architectures that lead to resource waste. Containerization can help adjust resources according to needs and enable better scalability.
Each processing step, often independent, can be transformed into a microservice, and then deployed as a container. These containers can be distributed geographically to reduce latency by bringing processing closer to users, following the edge computing principle.
This thesis proposes a processing pipeline adapted to 360° videos, decomposed into containerized microservices. Various deployment strategies inspired by the literature are evaluated in terms of resource consumption and performance in multi-user environments. Under our test conditions, experimental results indicate that, in terms of delivery time, a monolithic deployment appears to offer the shortest delays when located close to the client. However, this comes at the cost of increased resource usage, unlike more distributed deployments. Indeed, distributed microservice deployments can deliver 360° video within very reasonable delay while consuming fewer resources and being spread across geographically distant machines.
| Date | 7 Nov 2025 |
|---|
| Original language | French |
|---|
| Awarding Institution | - École de technologie supérieure
|
|---|
| Supervisor | Aris Leivadeas (Supervisor) & Stéphane Coulombe (Co-supervisor) |
|---|
Rusu, B. (Author),
Leivadeas (Supervisor) &
Coulombe (Co-supervisor),
7 Nov 2025Student thesis: Master's thesis › Master in Engineering: Information Technology Engineering