Cloud computing, a distributed computing paradigm offers unprecedented computing power, scalable, flexible resources over the internet in a pay-as-you-use basis from data centers. The rapid adoption of cloud computing has led to the emergence of various cloud vendors (cloud service providers) offering functionally-equivalent cloud services at different levels of abstraction, performance and pricing policies with different set of features. This dramatically creates a healthy competition for both the cloud vendors and cloud service users. However, the identification of Trustworthy Cloud Service Providers (TCSPs) who can satisfy their QoS requirements hardens for cloud service users as there endures a trade-off between functional and non-functional requirements. This poses a significant challenge to academicians and enterprises for cloud services evaluation and selection to identify the appropriate TCSPs. The existent of similar cloud services and the diverse set of user requirements complicates the process of cloud service selection.
Over the past few decades, trust based cloud service selection models attained huge attention from researchers and individuals since trustworthiness is a degree of compliance of a service provider and user requirements defined in Service Level Agreement (SLA) Sarbjeet Singh, 2017. In general, trustworthiness is measured using subjective and objective assessment based on user feedbacks and QoS monitoring respectively. The most existent approaches evaluate trust based on predefined features, since the user feedbacks reflects the biased values. The dynamic nature of cloud service makes the QoS experienced by the user different from the QoS claimed by the service provider.
The state-of-the trust based cloud service selection models employs different statistical methods to determine the trustworthiness of cloud service providers. Among these, Multi Criteria Decision Making approaches like Analytic Hierarchy Process (AHP), (ELECTRE), (VIKOR), (GRA), (TOPSIS), (PROMETHEE), etc. were found to be most appropriate for cloud service problem as it reveals the intrinsic relations among the multi criteria and decision . An extensive literature on the existing MADM reveals the prevalence of AHP-TOPSIS, AHP-FTOPSIS, Improved TOPSIS evaluate the alternatives (CSPs) based on distance relationship among data sequences that reflects the relationship among them. However, distance as measurement scale cannot reflect the dynamic relationship between alternatives. The gray correlation in GRA, considers as a measurement scale to find similarity between two CSPs. For instance, the alternatives indices have larger difference, the grey correlation degree among alternatives and the reliable ideal (best) solution may be equal. The above mentioned MADM approaches are not suitable to assess all kinds of alternatives since its measurement utilizes grey correlation degree or separation measures in TOPSIS. Hence, this paper presents EGT, a novel hybrid MADM approach based on Shannon entropy, integrates GRA and TOPSIS to evaluate the performance of service providers to obtain accurate service ranking to determine the trustworthy cloud service providers.