Abstract—Cloud computing can be considered as one of the most promising environments for e-learning systems. Unfortunately, services provided by the cloud are not always free. Hence, the primary goal of this paper is to select learning resources that best fit learner’s budget dedicated for the e-learning class. In other words, learner gets the best e-learning curriculum that he can afford. The problem is modeled as a constrained optimization problem and a proposed solution based on genetic algorithms is introduced. Simulation study shows that the proposed solution provides results identical to optimal solutions in most cases. Index Terms—E-Learning systems, cloud computing, genetic algorithms, learning objects. I. INTRODUCTION Cloud computing [1]-[4] is introduced mainly to promote collaboration. It offers users three usage scenarios; infrastructure as a service (IaaS), platform as a service (PaaS), and software as a service (SaaS). These scenarios provide the required services to users with much lower cost in a pay-per-use basis compared with having the services locally. Cloud computing supports learning anywhere concept of the e-learning systems. In other words, it reduces the requirements at the learner’s side to just a web-browser on a computing device. Unfortunately, it imposes a cost of using computing resources on clients (educational institutes, individual learners … etc.). This cost mainly covers the licensed software cost that supports the e-learning system at the cloud in addition to the cost of utilizing the hardware resources of the cloud. Several research efforts have been accomplished on e-learning systems on cloud computing environments. The study in [5] concludes that e-learning on cloud computing environments can reduce the cost and enhance the learning process management. In contrast, the study in [6] verifies that cloud computing can be used to build the next generation platform-independent e-learning systems. Meanwhile, the study in [7] shows that the use of cloud computing improves periodical IT tasks. On the other hand, the study in [8] concludes that utilizing cloud computing as an e-learning ecosystem provides a fault tolerant system against hardware/software failures. Additionally, the study in [9] shows that e-learning systems on cloud computing environments could render the cost needed for building an e-learning system affordable. A major problem of using e-learning systems on cloud Manuscript received August 12, 2011; revised October 3, 2011. Gamal A. Ebrahim is with the Computer and Systems Engineering Department, Faculty of Engineering, Ain Shams University, Cairo, Egypt (e-mail: gamal.ebrahim@eng.asu.edu.eg). computing environments is to select the best curriculum that fits the client’s budget. This does not mean the cloud sacrifices the curriculum quality but means it selects the curriculum that utilizes certain licensed software with certain needs of processing and storage at the cloud to fit the client’s budget. However, the integrity and quality of the curriculum are maintained. This problem is tackled in this paper based on genetic algorithms. Mainly, because genetic algorithms have powerful capabilities to search in huge search spaces [10]-[13]. They have been adopted in the study in [14] to find the best set of curriculums that represents a learning path for the learner based on his pre-test score. However, this study does not consider the impact of cloud computing on e-learning systems. More importantly, it deals with the curriculum as a one atomic entity. In general, curriculums should be divided into a set of learning objects [15]. Then, the most appropriate learning objects for the learner should be selected. The rest of this paper is organized as follows: section II presents the mathematical model of the problem. Then, section III details the proposed algorithm followed by section IV that provides the results of an experimental study of the proposed algorithm. Finally, section V concludes this paper. II. MATHEMATICAL MODEL OF THE PROBLEM Assume there is a cost that will be incurred by the learner for his e-learning class. This cost corresponds to the services provided by the e-learning system for the content provider, the software licensing, and the running time and storage used by the learner at the cloud. Moreover, the learner seeks a specific set of learning objects that provides him the required curriculum corresponding to his learning study level. The problem of seeking a curriculum corresponding to learner’s study level has been studied in the past. However, the complexities of using e-learning systems on cloud computing environments impose several challenges. First, the cost of using the resources at the cloud must be paid. Second, there could be many contents that can be utilized to deliver certain curriculum with each one of them has its own cost, capabilities, benefits given to the learner … etc. Third, there should be a certain budget dedicated by the client to the e-learning class. Forth, there could be certain preferences in the delivered contents from the learner’s point of view. For example, some learners need to have much graphics and animations in their delivered contents. Finally, there could be several ways to provide the curriculum by composing them from their elementary learning objects. Unfortunately, learning objects may not be compatible from the learner’s perspective. For example, one learning object may Budget-Aware e-Learning Systems on Cloud Computing Environments: A Genetic Approach Gamal A. Ebrahim International Journal of Information and Education Technology, Vol. 2, No. 6, December 2012 580 DOI: 10.7763/IJIET.2012.V2.208