Summary: | 碩士 === 逢甲大學 === 資訊工程所 === 100 === The telephone companies usually can’t detect and deal with anomalous communication of telephone in real time, such as communication fraud or phone failure. The result is that the user is forced to pay the major phone bill and the image of the company is damaged. When the internet technology becomes more mature, the traditional PSTN (Public Switched Telephone Network) becomes VoIP (Voice over Internet Protocol), like Skype, Google talker and MSN. The hackers use the free phone calls to make international calls by fraud malware. It is important to detect and avoid communication fraud.
Entropy analyzes the behavior of information patterns, e.g. the rain gauge and earthquake prediction. This paper uses entropy to detect the anomalous communication by analyzing the call behavior. It sets the threshold range from historical call data and finds the patterns of anomalous communication. We also use sliding window entropy to find the change of call behavior. After locating the anomalous calls, we send the information back to call policy management to regulate the policy. This paper shows that we can identify the possible anomalous communications and the associated policies, therefore reducing communication fraud.
|