Yu, WeiGe, Linqiang2016-10-242016-10-242016-10-242016-05DSP2016Gehttp://hdl.handle.net/11603/3288(D. Sc.) -- Towson University, 2016With the popularity of mobile networks, it has become a burgeoning target for cyber-attacks. For example, malware has proven to be a serious problem for the mobile platform because malicious applications can be distributed to mobile devices through an application market. From the defender's perspective, how to effectively detect threats and enhance the cognitive performance of mobile networks becomes a challenging issue. In addition, mobile networks have limited network resources and mobile devices are characterized by limited storage capacity, constraint battery life time, and limited computation resources so that developing a scalable, reliable and robust cyber threat defense system is challenging . To address those challenges, in this dissertation we develop effective schemes to efficiently conduct threat detection in mobile networks. First, we develop an Artificial Neural Network (ANN)-based malware detection scheme to detect unknown malware on mobile devices. Second, to enable the effective detection and desirable impact on the performance of mobile networks, we develop both sampling and aggregation techniques to achieve desirable tradeoffs between the detection accuracy and the use for network resources. Third, we develop MapReduce-based Machine Learning (MML) schemes with the goal of rapidly and accurately detecting and processing of malicious traffic in a cloud environment.Behavior-based malware detection approach on mobile devices -- Effective sampling and data aggregation techniques in host-based intrusion detection -- MapReduce based machine learning techniques for processing massive network threat monitoring dataapplication/pdfxiii, 126 pagesen-USText