The implementation and efficient control of smart home applications face technical challenges in particular with regards to the coordination of the three platforms: IoT nodes, mobile devices and containers on the cloud with heterogeneous components. Each component has a set of constraints in terms of resources and battery capacity. Microservice architecture improves application performance by distributing data collection and processing tasks across multiple distributed components. Since task processing consumes energy, it is essential to define a mechanism that optimizes the distribution of tasks on the various components of the system and adapts to the network change to minimize energy consumption, meet the processing delay and avoid overloading resources.
In this work, we formulated the problem of distributing tasks in a heterogeneous system composed of IoT nodes, cloud-based containers, and mobile devices (DT_IoTMC) using a mixed-integer linear programming model. We proposed, designed and implemented a solution for the optimal distribution of tasks, which consists of two modules: a profiler that collects information about the system components and an optimizer that computes the optimal distribution of tasks and activates the appropriate microservices on the appropriate equipment. Two algorithms have been proposed to solve this problem: DT_IoTMC based on hybrid data offloading and DT_IoTM based on a two-level data offloading. These algorithms yield promising results showing the performance of our solution in minimizing the total energy consumption of the system.
| Date | 20 Dec 2019 |
|---|
| Original language | French |
|---|
| Awarding Institution | - École de technologie supérieure
|
|---|
| Supervisor | Kim Khoa Nguyen (Supervisor) |
|---|
Douma, R. (Author),
Nguyen (Supervisor),
20 Dec 2019Student thesis: Master's thesis › Master in Engineering: Information Technology Engineering