CPU/GPU Hybrid Detection for Malware Signatures
Radu Velea, Ştefan Drăgan · 2017
Malware detection is an important aspect of cyber security. The process of identifying malicious code in files or network traffic is very complex and requires a lot of computational resources. Most security solutions that deal with malware detection implement advanced string matching algorithms or look for certain behavioral patterns during program execution. These methods of detection can cause significant performance penalties for real-time applications, can limit the scan surface or degrade user experience. In this paper we discuss a hybrid approach that leverages CPU and GPU compute capabilities in order to accelerate pattern matching for malware signatures. The solution presented focuses on improving performance and reducing power consumption of string matching algorithms on devices such as ultrabooks and laptops.