A Systematic Literature Review on Vulnerabilities, Mitigation Techniques, and Attacks in Field-Programmable Gate Arrays

This paper presents a systematic literature review (SLR) of the vulnerabilities of field-programmable gate arrays (FPGAs), based on 51 carefully selected articles from a pool of 271 unique publications sourced from various databases, including IEEE, ACM, and SpringerLink, covering the period from January 2012 to September 2022. The study identifies 22 distinct vulnerabilities and examines 27 types of attacks that exploit these vulnerabilities. We provide detailed descriptions of each vulnerability, including associated threat models, and review various mitigation techniques while highlighting the application of machine learning (ML) in both detection and defense whenever applicable. The significance of this work stems from its focus on a vulnerability-centric approach, where mitigation strategies directly target vulnerabilities rather than simply countering attacks. This approach highlights the critical role of safeguarding bitstream access, which can prevent a broad spectrum of attacks. Additionally, the study notes the absence of common vulnerabilities and exposures (CVE) datasets for the selected period, suggesting a gap in the documentation of FPGA-related vulnerabilities. The findings of this review have important implications for mitigating economic and security risks in FPGA applications, and we suggest potential future research directions, including further integration of ML techniques to enhance FPGA security.