Client-side architecture for mobile service QoS monitoring using Generalized Extreme Value theorem

The vast demand for mobile services is resulting in new challenges for traditional network management. This paper presents a new mobile service management architecture that solely relies on data collected from the mobile clients without the need to insert measurement probes in the core transport network. The proposed architecture is based on an algorithm that exploits Generalized Extreme Value models (GEV) and Joint Probability Distributions (JPD) to predict potential mobile service degradations. The extracted prediction models are based on collected extreme values of Quality of Service (QoS) parameters. Since the parameter monitoring effort is delegated to the mobile clients, our proposed architecture is more scalable and serves to offload the service from the monitoring burden inherent in traditional service management architectures. The performance analysis presented in this paper illustrates the efficacies of the proposed approach compared to traditional service management approaches.