| Summary: | 碩士 === 大葉大學 === 資訊管理學系碩士班 === 98 === The various portable mobile devices have recently become increasingly popular, especially Smartphone. In fact, hundreds of cell phone viruses have emerged in the past two years. The cell phone viruses targets industrial secrets and had caused the leakage of user privacy and enterprise’s secrets. Thus, mobile devices security becomes an important issue. The virus detection needs to take the quite huge resources. However, the system resources are limited in a mobile platform. How to use the least system resources to gain the detection effect is a big challenge.
Presently most of researches about cell phone viruses are focus on virus detection techniques development and the studies considerate the detection capability of cell phone viruses and resources consumption are less simultaneously in a mobile platform.
In this study, we conduct a set of experiments to evaluate and analyze performance of the virus detection techniques (Bayesian network (BN), detection tree C5.0 and neural network) by utilizing five kinds of evaluation indexes. The five kinds of evaluation indexes are detection rate, false positive rate, overall accuracy rate, resources consumption and energy consumption. We collect 40 types of cell phone viruses (79 cell phone viruses) that are used to evaluate the detection capability of these techniques. In order to find out the best detection technique, we analysis the performance of these techniques in a mobile platform through detection rate, false positive rate, overall accuracy rate, and Receiver Operating Characteristic (ROC) curve. The detection tree C5.0 is better than other methods. Thus, we development a viruses filtering system base on detection tree C5.0 in the HTC HD2 cell phone with windows mobile 6.5. By experiment design, the 27 cell phone viruses that propagate with MMS are used to test the detection capability and the resources consumption and energy consumption in the detection process.
Our experimental results give some practical and useful guidelines to mobile security researchers, so that they can acquire insight to apply these techniques to the area of cell phone virus detection in the mobile platform and devise more effective virus detection models.
|
|---|
