Recommendation of Cloud Services based on QoS(Quality of Service)

Abstract

In the era of Service Oriented Computing, Cloud Computing and IoT which has transformed the computing and made it an essential part of our daily life. Due to these computing paradigms, computing is not just related to office, studies or work, as well as not limited to gaming or entertainment purposes. Computing has become part of our daily life and our lifestyle has changed due to technology and advanced computing paradigms. Also, no one can ignore the importance of web services, after all, these computing paradigms are built on top of web services. Web Services are software applications which are loosely coupled, build using open protocols andcan interact with applications developed using different programming languages which makes web services important but selecting the right services for the job is also important. There are number of web services available online which can performs same functions. However, quality becomes the differentiation point between them. Quality of Service (QoS) can be determined by nonfunctional attributes of web services such as: response time, throughput, accuracy, resources utilization etc. On the basis of QoS, any suitable web servicecan be selected by the users for a specific purpose, though it is difficult for web services‟ user to use and check each web service available. To overcome this challenge, web service recommendation algorithms were introduced and many are now available like, Reputation Aware, Region Aware, Credibility Aware and Trust Aware, however, the accuracy to predict web service can still be improved. In this research, a web service recommendation algorithm is introduced which takes information ofusers and services to predict a web service. For the user side, users are initially clustered to identify reliable users, then reputation of users and similarity between different users is calculated. Later, based on these information QoS value of web service is predicted. For service side, services are clustered with respect to each user and services are identified which were grouped together on reliable and reputed users. Services similarity was calculated based on these groups and QoS value is predicted based on service similarity. Dynamic weight factor is introduced as „‟ to get final QoS value prediction by combining the user and service predictions.