چكيده به لاتين
Software-defined networks, by decoupling the control plane from the data plane, have provided significant support for network equipment management, cost reduction, and service quality assurance. Nevertheless, due to the continuous development of applications and diverse services in computer network infrastructures, the demand for data processing, maintenance, and analysis has grown exponentially. In this regard, due to limitations in processing power, storage space, and battery capacity, Internet of Things (IoT) devices cannot adequately meet the requirements of applications in terms of latency and energy consumption. To address this challenge, computational offloading onto Mobile Edge Computing (MEC) and Mobile Edge Computing (Fog Computing) platforms has been considered as alternative solutions to cloud infrastructure. Computational offloading involves transferring computations, functions, and executable code from a resource-constrained device to one with sufficient resources to execute them. A critical issue in the computational offloading process is decision-making, which includes minimizing delay, energy consumption, selecting the offloading infrastructure, and choosing the destination node. Users, in order to perform computational offloading and utilize the processing resources of edge nodes or cloud infrastructure, are required to pay fees determined by service providers. Due to the existence of a controller in software-defined networks, the network structure and the status of existing nodes in terms of processing power are continuously monitored, leading to valuable information extraction. Having comprehensive controller knowledge can assist users in making the best decisions regarding offloading destination selection.
Most of the existing work in the field of computational offloading in software-defined networks does not take pricing into account. Furthermore, these works primarily focus on processing resources and neglect communication resources. Our research aims to provide a computational offloading method in software-defined networks while considering pricing. In this method, a greedy algorithm from game theory is used to find the optimal offloading destination, aiming to enhance user efficiency while adhering to delay thresholds and precise calculation of queue delay. Moreover, communication resources, in addition to processing resources, play a crucial role in achieving this goal. This approach has been implemented in a real testbed in combination with the Ryu-based software-defined controller. evaluation results produced by this environment demonstrate the superior performance of this method in the aforementioned objectives compared to a similar approach, with a calculated efficiency improvement of up to tenfold.
