A survey of recent results in FPGA security and intellectual property protection

Abstract

Field programmable gate arrays (FPGAs) are reconfigurable devices which have emerged as an interesting trade-off between the efficiency of application-specific integrated circuits (ASICs) and the versatility of standard microprocessors [81]. Progresses over the last 10 years have improved their capabilities to the point where they can hold a complete system on a chip (SoC) and thus become an attractive platform for an increasing number of applications (e.g., signal processing, image processing, aerospace, etc.). In view of the important data manipulated by these devices, but also of the high amount of intellectual property (IP) they may contain, security-related questions have arisen. First, can we use FPGAs as security devices for example, securely and efficiently encrypting sensitive data (in particular when compared to software solutions)? Second, how can we guarantee that the IP corresponding to FPGA designs is protected (i.e., cannot be easily counterfeited)? Such questions have been the target of a large number of papers in the literature, including several surveys, example [13, 71, 83]. In this chapter, we take another look at them and review a number of important recent results related to security IPs and IP security in modern reconfigurable devices. The chapter is structured in three main sections. First, we briefly describe the structure of recent FPGAs. Next, we discuss security IPs in FPGAs, taking the example of symmetric encryption with the AES Rijndael, and including their performance evaluations and resistance against physical attacks. Finally, we emphasize recent trends for improving IP security in FPGAs, including bitstream security, the use of code watermarking techniques and the exploitation of physically unclonable functions (PUFs).