Trust Evaluation Models for the Security Problems in Multi Tenant Cloud Computing

Abstract

Importance of digital world increases after pandemic, as most of service get online and need services. A large number of digital platforms rely on cloud infrastructure to provide services to their customers. Due to the fact that these services require machines to execute user requests, some of the work is outsourced to tenants. Inner security is also a major concern to monitor. Cloud infrastructure requirement for management of various services indirectly depends on virtual machine acting as tenants. It is necessary to monitor cloud service tenants in order to improve the privacy and security of cloud services. Multi-tenant cloud architecture need two inner and outer side security. In order for cloud users to evaluate the trustworthiness of cloud service providers when constructing or migrating their distributed systems to cloud data centers, the proposed scheme must be implemented. In this work, a trust evaluation scheme based on a fuzzy logic system is discussed in greater detail. The trust-based tenant machine evaluation method proposed in this paper has been implemented. Proposed model has created a virtual window to evaluate trust value of each tenant present in cloud. The Sorensen trust value was estimated by the model based on the behavior of machines in the network. Proposed model utilizes mutual trust value of the nodes by using Sorensen function. To increase the detection performance individual machine utilization parameter was used. HITS algorithm finds the hub value of the work for node class detection into legitimate and malicious nodes. One more method of trust evaluation was proposed is Leicht Holme Newman trust in cloud. The extracted feature values were then used to train the Adaptive neural fuzzy interference system mathematical model, which was then used to train the model. This behavior of tenants in virtual window to identify malicious nodes. The trained ANFIS model divides the machine into two categories: the first is a malicious node, and the second is a legitimate node. The experiments were carried out under a variety of cloud environmental conditions. The implementation was carried out using the MATLAB software. The results show that different evaluation parameter values were optimized in the work. Result shows that proposed LHN-ANFIS has improved the recall value by 2.17% as compared to SHCTM and 3.23% as compared to TMM model. Further it was obtained that FNR of the model was also reduced by 33.74% as compared to SHCTM model. Use of social trust and ANFIS model for detection of malicious node increases the work performance. newline