Abstract:
Edge computing is a broad concept covering a variety of computing technologies such as cloudlet-based computing, Fog Computing, Mobile Cloud Computing (MCC), and Multi-Access Edge Computing (MEC). All these computing paradigms have a common approach of bringing the cloud-computing capabilities to the edge of the network. The term edge used in this area refers to a location close to the end-users. Computation (task) o oading is an essential feature of edge computing. It reduces the battery consumption and makes it possible to execute applications that are unable to be executed on mobile devices due to their insu cient processing power, memory, and unpredictable network connectivity. In this regard, edge computing can enable new applications and use cases for the environments where there are many mobile users, such as university campuses, airports, and smart roads. The increasing number of smart devices that are connected to the Internet brings new challenges in terms of high communication delay and computational resources congestion. Although edge computing overcomes these challenges by bringing cloud computing capabilities to close proximity of the end-user, it presents a very dynamic and exible environment where both computational and networking resources are utilized in real-time in accordance with the requirements. Hence, e ciently managing and orchestrating di erent types of resources become crucial issues. To overcome these challenges, we introduced novel edge orchestrators. Firstly, we developed an edge computing simulator, namely EgeCloudSim, to evaluate the performance of the proposed orchestration algorithms. Secondly, we proposed a fuzzy logic-based workload orchestrator for multi-tier mobile edge computing systems. As a nal contribution, we presented a machine learning (ML) based workload orchestrator for multi-tier multi-access vehicular edge computing (VEC) environments.
