الفهرس 
IntroductionOnly 14 pages are availabe for public view
 |  | from | 129 |  |  |
| | | from | 129 | | |
Abstract
Computing is a model of services that is delivered in a technique similar to utilities such as water, electricity, gas, and telephony. In computing, users utilize service that they need without considering where it is hosted. There are several computing paradigms such as parallel computing, distributed computing, and grid computing. The most recent emerging paradigm is cloud computing. It converts the vision of “computing utilities” into a reality. Cloud computing uses internet technologies to offer elastic services. Users in cloud computing able to dynamically acquire computing services needed at any time. They can access services from any device connected to Internet and pay for this service for as long as they need it. There are several problems in cloud computing. Among those problems, task scheduling is one of the major problems. The task scheduling plays the key role of the efficiency of the whole cloud computing facilities. Task scheduling means that allocate the best suitable resources to the task to be executed. In this work, there are two parts: First Part of this thesis contribution is based on developing new PSO algorithm in the static environment. The improvement resulted from merging mutation with standard PSO (LBMPSO). Three mathematical models based on the above proposed algorithms. The first model is to minimize execution time. The second model is to minimize transmission time. Third, consider expected Round trip time. LBMPSO can play a role in achieving reliability of cloud computing environment by considering the resources available and reschedule task that failure to allocate. Our approach LBMPSO compared with standard PSO, random algorithm, and Longest Cloudlet to Fastest Processor (LCFP) algorithm. Results show that LBMPSO can save in makespan, execution time, round trip time, transmission time and cost. Also, considers load balancing of the virtual machine.
iv
The second part of this thesis contribution considers the multi-objective problem in the dynamic environment. Jade is responsible for generating agent that make system dynamic. This part is also using the above-proposed algorithm. The proposed algorithm is adapted to be suitable for the multi-objective optimization problem. Multi-objective Load Balancing Mutation particle swarm optimization (MLBMPSO) is the algorithm that adapted to allocate tasks to the virtual machine. MLBMPSO considers two objective functions to minimize round trip time and total cost. MLBMPSO is efficient in allocating tasks to the available virtual machine in user level. MLBMPSO considers different parameters such as reliability, availability and load balancing of the virtual machine. Experimental results demonstrated that MLBMPSO outperformed the other algorithms in time and cost.