Innovative Multicriteria Decision Making Approaches To Cloud Service Ranking And Selection

Nowadays swift technological developments in communication and computing technologies have dramatically transformed the information technology world.This has led to the emergence of a new computing paradigm namely cloud computing.While cloud computing offers prodigious opportunities,it also poses sundry challenges to the Organizational Managers(OMs)and Decision-Makers(DMs).The foremost challenge that the OMs/DMs face while moving to cloud computing is the selection of appropriate cloud service(s)that best meet their organizational requirements.It is a paramount decision that entails far-reaching implications for any organization.Previously,many authors have proposed various solutions to the Cloud Service Ranking and Selection(CSRS)problem.However,the existing CSRS solutions have issues related to consistency,reliability,complexity and so on.Keeping in sight the convolution of the CSRS problem,and the shortcomings of the existing approaches,this dissertation proposes innovative Multicriteria Decision Making(MCDM)solutions(methods/frameworks)to help OMs/DMs make well-informed CSRS decisions under crisp and fuzzy environments.The conspicuous and novel contributions of this dissertation to CSRS research are as follows.First,we propose a novel cloud brokerage framework namely Service Selection and Recommendation Framework(SSRF)for CSRS.Unlike all existing approaches/frameworks,SSRF covers the complete lifecycle of the CSRS problem,minimizes the dependencies on third-parties,and provides a mechanism to incorporate Quality of Experience(QoE)along with Quality of Service(QoS)in CSRS decision-making.To implement the service evaluation/ranking module of SSRF,we propose a novel integrated MCDM approach for CSRS aimed at resolving consistency,reliability and complexity issues of existing approaches.Further,we perform a case study to appreciate the performance of the proposed approach.Moreover,we perform a comprehensive analysis of the proposed approach considering a comparative analysis and sensitivity analysis.The results from the case study and comprehensive analysis favor the proposed approach.Second,we extend and improve the integrated MCDM approach discussed above to settle the limitations of the existing research related to(i)lack of incorporation of both QoS and QoE in decision making,(ii)lack of prequalification process to reduce search space and(iii)lack of a comprehensive mechanism to select cloud services with consensus.To this end,we propose a novel QoS/QoE aware MCDM-based Methodology for Optimal Service Selection(MOSS).Contrary to existing approaches,MOSS provides a prequalification mechanism using Pareto-optimality,evaluates services from QoS/QoE perspectives using Best Worst Method(BWM),ranks the services using a multi-MCDM approach based on QoS/QoE criteria,and enables CSRS with consensus.To judge the efficacy/practicability of MOSS,we implement MOSS in the context of an e-commerce company facing a CSRS decision using real-world QoS/QoE data.Moreover,we perform a thorough analysis of the MOSS.The results of the case study and thorough analysis authenticate the practicality and usability of the MOSS.To resolve the above-mentioned issues more fittingly,we further improve/enhance the consensual mechanism proposed in MOSS to develop a broader consensus on the ranks of cloud services.For this purpose,we propose a more compact/refined unified MCDM framework namely Cloud Service Scrutinization and Selection Framework(C3SF)for CSRS with a broader consensus.Moreover,besides proposing a novel MCDM method namely Modified BWM(MBWM)to enable OMs/DMs to develop an early-stage consensus on decision criteria and its evaluation,we also propose a two-step novel approach to develop a broader consensus on the ranks of cloud services.We implement C3SF for CSRS using a case study and perform an in-depth analysis to appreciate its performance.The results show that C3SF is practical and provides precise and consensual recommendations for CSRS.Last but not least,although BWM used in contribution 1&2,and MBWM proposed as part of C3SF in contribution 2 perform better than existing methods,these methods lack the ability to handle fuzzy information.To address the limitations of existing methods,we propose another novel MCDM method called Fuzzy Linear Best Worst Method(FLBWM)for precise decision making under a fuzzy environment.Contrary to existing MCDM methods,the FLBWM is robust,requires less data,gives more consistent results,and effectively handles imprecise information.Next,to further improve the accuracy,credibility,and transparency of CSRS and provide an easy,pervasive,transparent,cost-effective and user-friendly way for CSRS,we propose a novel framework called Cloud Service Selection as a Service(CSSaaS).The CSSaaS framework is unique and first of its kind that not only gives a new direction to CSRS research but also paves the way towards offering CSRS as a service just like any other service in the cloud.It consists of multiple interdependent services/components.We implement the ranking/recommendation service of the CSSaaS framework using FLBWM.Moreover,to simplify the programming of the ranking/recommendation service,we propose an FLBWM based robust algorithm namely RecServ.We thoroughly and exhaustively assess,evaluate,and validate the FLBWM using illustrative applications.We also perform an in-depth analysis of the FLBWM from diverse perspectives.The results show that FLBWM outperforms existing approaches for CSRS.All the CSRS solutions proposed in this dissertation are novel,unique and different from existing approaches.The proposed CSRS solutions can be utilized to make precise CSRS decisions irrespective of the type/category of the cloud service.The managerial implications of the proposed solutions lie in providing comprehensive decision support to OMs/DMs.The proposed solutions have been thoroughly and exhaustively evaluated,compared and analyzed from diverse perspectives using illustrative applications,case studies,and theoretical analysis.The results have shown that the proposed solutions outperform existing methods/approaches to CSRS.