Developing a GUI Application: GPU-Accelerated Malicious Domain Detection

Our study uses the power of a graphics processing unit (GPU) to run malicious domain detection algorithms quickly and efficiently. We have developed a graphical user interface-based system that allows users to upload datasets (malicious domains) into a local database and then run tests with a list of domains to identify whether they are malicious. We have collected real malicious domain data from malicious domain websites and tested the five most widely used string-matching algorithms (Naïve, Levenshtein distance, Hamming distance, KMP and Rabin Karp), which allow users to compare the speeds of different string algorithms with varying time complexities against the number of domains both on the GPU (or the CPU) and our sample. On a CPU, this task becomes slower as our dataset grows. On a GPU, however, these algorithms can be run on any dataset size within the limit of the GPU's capacity with consistent performance.