A PCI Express Based Data Acquisition System for the Monitoring of Code Traces of RISC-V Microprocessors

In the digital world cyberattacks represent a serious threat that can compromise vulnerable systems. Microprocessors, in particular, are exposed to a wide number of cyberattacks, such as Spectre and Meltdown based attacks, as well as Code-Reuse Attacks. In this paper, a PCI express (PCIe) based data acquisition (DAQ) system was developed to acquire the code traces of a RISC-V microprocessor. The code traces dataset will be used to design a machine learning based system on FPGA for on-line detection of cyberattacks in RISC-V microprocessors. The DAQ system was designed in Verilog HDL for Xilinx FPGA platforms and interfaced with a PC running Linux to acquire the code traces and to save them on the hard disk. In the current stage of development, the DAQ system performance was evaluated by generating controlled code traces of 512 bit with frequency 50 MHz using an ad-hoc developed digital circuit. Tests were carried out on two different FPGA platforms and for three different sizes of the SRAM memory buffer (64 kB, 512 kB and 4 MB). The results have shown that the latency introduced by the PC operating system can seriously impact the data transfer rate, and the maximum achieved data throughput was found 577.17 MB/s in the case of SRAM memory size of 4 MB.