A doctoral thesis at the University of Kufa discusses the design of a system for scheduling tasks and allocating resources in cloud and fog computing for the Internet of Things and mobile crowd sensing environment

The Faculty of Computer Science and Mathematics at the University of Kufa discussed the doctoral thesis entitled “Efficient Resource Management Mechanism for Internet of Things and Mobile Crowd Sensing” by student Abbas Muhammad Ali Hussein Hashem. The thesis aims to schedule tasks and allocate resources to balance the load. The first work presents a Deadline and Budget Multi-Objective Dynamic Scheduling (DB-MODS) approach to execute user tasks on virtual machines within QoS limits in order to finish the task within the deadline and budget constraints and at four levels: First: Using K-Means machine learning technique to cluster tasks based on task length and deadline. Second: Developing an algorithm to classify virtual machines based on their capacity using thresholds. Third: Design a dual-fitness function that serves the goals of the user and the cloud system. Fourth: The system actively monitors task execution and dynamically manages resource usage during runtime.
The second work introduces modifications to the Particle swarm optimization algorithm is called PSO Optimized Leader (PSO-OL) in the fog-cloud system and these modifications are at four levels: First, Prevent Particle index Similarity (PPIS) method is proposed by evaluating the similarity between particles and adjusting their positions to obtain a wider range of solutions and preventing the algorithm avoids falling into the trap of local optima. Secondly, a strategy is proposed to select the farthest-best particle from the swarm in each iteration, which takes into account the current fitness value of the particle and the distance where a new dynamic adjustment factor is designed to control the particle selection. Third: Choose the second-best particle in addition to the best leader. Finally, a new crossover operator is proposed to obtain a new Optimized leader in each iteration based on the farthest-best and second-best particles to obtain an optimized leader and thus obtain the best solution. The results showed that it outperforms other algorithms.
The PhD student has published three papers, the first of which was published in different highly indexed journals, where the thesis was accepted with excellent.