Mobility and Fault Tolerance-Based IoT Task Offloading and Scheduling for Intelligent Transportation Systems Using a Fog–Cloud Hybrid Environment

Abstract Intelligent Transportation Systems (ITS) uses IoT devices to communicate with home servers. With increasing number of IoT devices the ITS need to offload and schedule the huge amount of data cloud and fog. The data are divided into delay sensitive and computation intensive moods. Due to the vast variety of data and the importance of the data the researchers introduced the IoT task of offloading and scheduling to meet certain QoS. Due to various factors such as vehicles mobility, directions, node energy, and processing requirement the system becomes NP hard. To address the above limitations, we have proposed a fog–cloud collaboration system that reduce the above limitations. In the proposed model we used the Analytic Hierarchy Process (AHP) to organize tasks based on different factors, such as their size, the amount of processing power they need, and the priority of tasks. The proposed model decides whether to execute the tasks locally or to offload to cloud or fog nodes. The proposed model consists of three main algorithms, The first algorithm is used for offloading decisions while considering vehicles mobility, task size, processing power required and other factors. The second algorithm is used to prioritize tasks as delay sensitive and computation intensive by using Analytical Hierarchy Process (AHP) while considering threshold values. Algorithm 2 decide and provide two queues list one is for offloading fog nodes while the queue list for offloading to cloud nodes. The third algorithm is used to allocate tasks to the most efficient fog nodes and cloud nodes based on the received queues. We also prioritize fog nodes based on node energy, ram size, processing power, and current load. After successful allocation to fog node the proposed system, also uses a fault tolerance mechanism to check and monitor the task execution. This framework can assign backup nodes in situations where failures are more common and can replicate important tasks because of this clustering. To test the framework under various conditions, including different workloads, node density, and failure scenarios, we performed extensive simulations using OMNeT++. The proposed model reduced the tasks delays by up to 28. 4% and energy usage by about 31. 6% compared to other models like DCP, FADEC, and a few others. Also, it was found that when the system is under heavy load, the success rate of tasks increased by 9–12%. Based on these results, the model is well-suited for situations where reliability and fast response times are important, such as in smart healthcare, transportation, and industrial settings.