A framework for network traffic analysis using GPUs
Tutor / director / evaluatorGelado Fernandez, Isaac
Document typeMaster thesis (pre-Bologna period)
Rights accessOpen Access
During the last years the computer networks have become an important part of our society. Networks have kept growing in size and complexity, making more complex its management and traffic monitoring and analysis processes, due to the huge amount of data and calculations involved. In the last decade, several researchers found effective to use graphics processing units (GPUs) rather than a traditional processors (CPU) to boost the execution of some algorithms not related to graphics (GPGPU). In 2006 the GPU chip manufacturer NVIDIA launched CUDA, a library that allows software developers to use their GPUs to perform general purpose algorithm calculations, using the C programming language. This thesis presents a framework which tries to simplify the task of programming network traffic analysis with CUDA to software developers. The objectives of the framework have been abstracting the task of obtaining network packets, simplify the task of creating network analysis programs using CUDA and offering an easy way to reuse the analysis code. Several network traffic analysis have also been developed.