Adaptive application offloading for QoS maximization in cloud-fog environment with delay-constraint

The IoT devices with advanced computational powers supported by Artificial Intelligence (AI) in diversified application domains require infrastructures that can deliver resources and computational services based on application needs. Fog computing has emerged as a platform for applications that can provide flexible and shared computational capabilities to such delay-sensitive services along with maintaining network availability, minimum latency, and Quality of Service (QoS) maximization. The proposed model offers the adaptive offloading of the applications while achieving the maximum QoS in a delay-constrained service environment. Moreover, to minimize the resource starvation condition, a probabilistic application task scheduler algorithm is presented, which considers the task’s initial priority and deadline. The tasks within waiting queues are assigned new priorities and used to select the most relevant task, reducing the response time and failure rate. Besides, a cost parameter is included in the proposed model that chooses the cost-efficient option for offloading the applications to the available servers. This work aims to maximize the overall QoS and optimize the service cost, which is demonstrated through simulation results. Our proposed model adaptive application offloading for QoS maximization with delay constraint (AAOQM-DC) yields a performance improvement by getting 0.11-2.49 sec. downfall in overall response time, 0.12- 1.38 sec. decrement in network delay, and enhanced QoS by achieving 0.07-5.12% lower failure rate in comparison to state-of-the-art approaches.