Online Resource Monitoring Model in Cloud TV

In order to solve the problem of the experience degradation of user caused by the resource competition among the applications in cloud TV, we present an approach of online resource monitoring model (ORMM), which is applied in the television service engine (TVSE) framework. Firstly, specific to the shared resource and the exclusive resource, an appropriate resource usage sampling method using system calls and service arbiter is introduced, which simultaneously monitors all types of the system resources. Secondly, to support the most important interactive video application, a Hidden Markov Model based application anomaly detection algorithm using the slide window is designed in consideration of the state transition of the interactive video application. Finally, based on the testing in Android cloud TV, a performance evaluation illustrates the parameters selection of the model, and the detection precision of our algorithm is superior to other popular anomaly detection algorithms about 20%.