Parallelism in Signature Based Virus Scanning with CUDA

Abstract

Information security is playing big role in the computer technologies. Its job is to detect unauthorized violation of the information integrity, secure it and also recover it, if the integrity was violated. One of the things that can alter an information are computer viruses. One of the task of the information security is also to detect these malicious applications and prevent their goal. This can be achieved in various techniques and one of them is signature based virus scanning. This technique uses a virus database (virus signatures) to detect if a file or application is infected with a specific virus. In this paper we are going to see in more details how is this implemented, which algorithm are mostly used and also try to improve its performance by parallelizing it on GPU by using CUDA. We are also going to see how CUDA utilizes large number of threads to solve a specific problem and use it to implement a parallel signature based virus scanner. Later we are going to see the performance benchmarks of the conducted experiments and discuss them and give a final conclusions for the usage of a GPU in signature based virus scanning.