Detection of Android malware security on system calls

In order to detect the security of Android software which isn't installed efficiently and accurately, this paper proposes a new malware detection method based on system calls frequency. Software samples are pretreated by Classification to refine and distinguish characteristics of various types of Android software; Restrictions and normalization are applied to process system call frequency information to improve the detection accuracy. Combined with the characteristics of random forest algorithms and data obtained, optimal training model is set up, and ultimately Android software security is detected effectively. Through experiments, the detection accuracy rate of Android tool software and financial software more than 93 percent, with higher accuracy.