راضيه روستائي

Nowadays, mobile devices (MDs) are involved in almost all aspects of human life, leading to expectations to employ them for various emerging computation-intensive applications. Despite these expectations, the deployment of these applications is challenged with the limitations of MDs in terms of battery, storage, and computation capacity. To overcome these challenges, users can use resources offered by the remote cloud, edge cloud, cloudlet, or a combination of all three. Although computation offloading can yield to extend MDs’ capabilities, the offloading efficiency is highly influenced by the challenges associated with the selected opportunity to offload. The rental unit price of resource providers regarding their market power, bandwidth cost and high latency devoted to offloading to the remote clouds, limited communication and computation edge resources as well as the competition arisen to access them, performance degradation considering untrusted participants of D2D cloudlet, lack of sufficient motivation in order to participate in the D2D cloudlet, MUs mobility, and the decision offloading complexity are some of these challenges. In this thesis, we propose an offloading middleware, which considers conditions of MUs, their applications and the opportunities to offload and offers services to deal with the associated concerns. All proposals are modeled with the aid of the game theory and simulated using designed scenarios. Finally, we eva‎luate them based on predefined performance metrics such as resource providers’ and MUs’ utilities, average allocated resources, number of components, the amount of workload and invocation data, number of D2D devices , and their capacity. Results showed that the proposed middleware has improved the offloading efficiency at least twice more than other alternatives.

ميان افزار تخليه بار , رايانش ابر راه دور , رايانش ابر حاشيه , رايانش ابر كوچك دستگاه به دستگاه , رايانش تركيبي , قيمت گذاري منابع , تخصيص منابع , تخليه بار , سربار تصميم گيري

Offloading Middleware , Cloud Computing , Edge Computing , D2D Cloudlet , Hybrid Computing , Pricing , Resource Allocation , Offloading , Decision Overhead

Razie Roostaei

Zeinab Movahedi