A unified framework for network bandwidth and link latency detector based on cloud computing

Abstract

In general, users often do not know which organizations or services obtain the right to use and may store, utilize, or redistribute their data when sensitive data have been released to a cloud service. The research field of usage control deals with such troubles by enforcing constraints on the usage of data after it has been revealed and is therefore principally important in the cloud environment. Usually, existing solutions apply cryptographic methods to maintain sensitive user data confidential against untrusted servers, by disclosing data decryption keys particularly to the authorized persons. In doing so, on the other hand, these solutions unavoidably bring in heavy computation overhead on the data owner for key distribution and data management when fine-grained data access control is needed and as a result do not perform well. To avoid these problems in this work, we propose customized bivariate parametric detection mechanism (cbPDM) that utilizes a sequential probability ratio test, permitting for control over the false-positive rate while examining the trade-off between detection time and strength of an anomaly and also the packet delivery rate. The method is examined using the bitrate signal-to-noise ratio (SNR) metric, which is an effective metric for anomaly detection. This enhanced detection method does not need or try to model the full traffic patterns. First, the anomaly detection controls aggregate traffic, devoid of flow separation or deep packet inspection. After that, unlike prior anomaly detection approaches, our method computerizes training and does not require hand-tuned or hard-coded parameters. After that, we make use of both the packet rate and the sample entropy of the packet size distribution guides to guarantee robustness against false positives, consequently overcoming one of the traditional problems of anomaly detection methods. From the comparison results we can see the proposed method, which is better than existing methods due to itsrate of packet delivery and bit-rate values.