Investigation Of Traffic Load And Task Scheduling In Secure Cloud Computing

Cloud Computing is a steadily maturing large-scale model for providing on-demand IT resources(e.g.compute,storage,networks,platforms and applications)as a service over the internet.Concerning the three common service categories,i.e.Software-as-a–Service(Saa S),Platform-as-a-Service(Paa S),and Infrastructure-as-a-Service(Iaa S),there are still many challenges and issues to be addressed for both cloud providers and consumers,such as data traffic,privacy,encryption,computation cost,complexity and resources utilization.This thesis is an endeavor to design several schemes in order to improve the performance of Paa S to serve Saa S,and to provide more facilities to Iaa S.The findings of this thesis provide several solutions that address above-mentioned cloud computing challenges,especially traffic load and task scheduling in secure cloud computing.The main contributions of this thesis are briefly described below.To address the traffic load and queries processing for secure cloud computing,a novel search algorithm called Multi-broadcast Searchable Keywords Encryption(MBSKE)is proposed to process queries having a set of keywords.This set of keywords is sent from the users to the cloud server in an encrypted form,thus hiding all information about the user or the content of the queries from the cloud server.The proposed method uses caching algorithm and provides an improvement of 40% in terms of runtime and trapdoor.In addition,the method minimizes computational costs,complexity,and maximizes throughput,in the cloud environment,whilst maintaining privacy and confidentiality of both the user and the cloud.The cloud returns encrypted query results to the user,where data is decrypted using the users' private keys.With the development of cloud computing,sensitive information of outsourced data is at the risk of unauthorized accesses and the cost of implementation is high.Several approaches have been provided to enable searching the encrypted data to protect data privacy,but the cost of traffic load and complexity is high.To combat this issue,this thesis presents a cache algorithm for query in the user side to reduce communication cost between the user and cloud provider.In addition,the thesis proposes a parallel searching algorithm to reduce the searching time and traffic overload in cloud server.By adopting this scheme,cloud servers can be optimally utilized with the possibility to reduce the number of the cloud resources for the same task.Furthermore,it is noted that the use of cache has contributed positively in the process of sending the query.In cloud computing and heterogeneous computing systems,the scheduling algorithms are important for achieving high performance.In this thesis,a new scheduling algorithm is proposed for a bounded number of fully connected graph based on Improve Mi M-Ma M scheduling task(I-MMST)in cloud computing.Besides,significant makespan improvement can be achieved by introducing a look-ahead feature without increasing the time complexity associated with computation cost based on the principle of components analysis algorithm(PCA).Our analysis and experiments based on randomly generated graphs with various characteristics,show that the proposed scheduling algorithm significantly surpass previous approaches in term of makespan,speed,and efficiency.Although cloud computing has been used extensively in various applications,the task and resource scheduling is still an area that is worth further investigation.In a heterogeneous computing system,tasks scheduling algorithms,which allocate incoming tasks to machines,are necessary to reduce makespan and maximizes resource utilization.The proposed algorithm provides a significant improvement with respect to the makespan and reduces the complexity by employing the Principle Components Analysis(PCA)algorithm thus,reducing the expected time to compute(ETC)a matrix.The simulation results confirm the superior performance of our proposed algorithm for heterogeneous systems in terms of efficiency,speedup and schedule length ratio when compared to the state-of-the-art Min-Min,Max-Min,Qo S-Guide and Mi M-Ma M scheduling algorithms.As the traffic in cloud server has been increasing constantly in the past years,it is almost impossible for one server to handle all the requests coming from the various clients.Hence the solution is to balance the heavy load by transferring the traffic to the underutilized switches.Traditional load balancers use very expensive and inflexible hardware.An alternative of these hardware based load balancers is to use Software Defined Networks(SDN).These SDN load balancers do not require costly hardware and can be programmed,which makes it easier to implement user-defined algorithms and load balancing strategies.In this thesis,the control traffic balancing problem is first formulated to find the optimal control traffic forwarding paths for each switch in such a way that the average control traffic delay in the whole network is minimized.This problem is critical in SDNs because the timely delivery of control traffic initiated by Open Flow switches directly impacts the effectiveness of the routing strategies.In this thesis,different algorithms are implemented and compared to transfer the data among the host's based load balancing technique using Open Flow versus switches connected to the controller.