| Abstract [eng] |
Fog Computing conception was introduced by OpenFog Consortium which was established in 2015 by such companies as Cisco, Microsoft, Intel etc. This paradigm seeks to solve Cloud Computing issues. They are due to long transmission distances, higher data flow, data loss, latency, and energy consumption. Services have to be closer to end-users as a part of a solution. But, fog devices are known for being mobile and heterogenous. Their resources can be limited, and their availability can be constantly changing. A service placement optimization is needed to meet the QoS requirements. A service placement orchestration is proposed in this thesis, which functions as a multi-agent system. This two-step process is made of Multi-Objective Particle Swarm Optimization (IMOPSO) to generate particle sets as the first step and the Analytical Hierarchy Process (AHP) with a specific judgement matrix as the second step for service distribution priority decisions. This dynamic service orchestration method is important for its quality that it does not focus on a central control unit. It allows to make a service placement decision for any orchestrator which synchronizes its resource availabilities with other nodes within the fog network. It greatly improves scalability and resilience to fog node failures. It also supports mobility and adapts to resource availability changes. |
