Malware Variants Detection Using Density Based Spatial Clustering with Global Opcode Matrix

Over the past decades, the amount of malware has rapidly increased. Malware detection becomes one of most mission critical security problems as its threats spread from personal computers to cloud server. Some researchers have proposed machine learning methods which can detect malware variants by searching the similarities between malware and its variants. However, the large search space causes large time cost and memory space occupation. To reduce the search space while retaining the accuracy, we firstly propose to convert malware into global opcode matrix which is based on 2-tuple opcodes, and then cluster the opcode matrixes to patterns. We can easily recognize the malware variants by searching the similarities with the patterns. The experiments demonstrate that our approach is more efficient than the state-of-art approaches in time cost, memory space occupation and accuracy.