Dynamically configurable online statistical flow feature extractor on FPGA

Statistical information of network traffic flows is essential for many network management and security applications in the Internet and data centers. In this work, we propose an architecture for a dynamically configurable online statistical flow feature extractor on FPGA. The proposed architecture computes a set of widely used statistical features of the network traffic flows on-the-fly. We design an application specific data forwarding mechanism to handle data hazards without stalling the pipeline. We prove that our architecture can correctly process any sequence of packets. Users can dynamically configure the feature extractor through run-time parameter. The post place-and-route results on a state-of-the-art FPGA device show that the feature extractor can achieve a throughput of 96 Gbps for supporting 64 K concurrent flows.