hw9

Hardware-Oriented Security and Trust ECE 4156 HST / ECE 6156 HST Spring 2023 Assoc. Prof. Vincent John Mooney III Georgia Institute of Technology Homework 9, 100 pts. (ECE 4156) 125 pts. (ECE 6156) Due Friday April 14 prior to 11:55pm 1) (40 pts.) Consider an instance of the HELP PUF with 2 23 hazard-free paths. Assume that these hazard-free paths consist of 2 22 rising (zero to one transition) paths and 2 22 falling (one to zero transition) paths. a. (5 pts.) Describe the physical sources of entropy (randomness) in the HELP PUF. b. (10 pts.) Consider two specific paths whose delays are compared to produce a response bit. Discuss the entropy (i) on each path and (ii) in the comparison bit result. In particular, consider two PUF instances HELP 1 and HELP 2 . Consider the same two specific paths in each HELP instance. Discuss the entropy in these paths for both HELP 1 and HELP 2 . c. (10 pts.) Continuing in the previous question, discuss how the entropy on each path can change the comparison bit result. In particular, consider how the path lengths in HELP 1 and HELP 2 may be related (correlated) to each other due to the use of Hardware Description Language (HDL) synthesis tools. d. (15 pts.) Is it possible to build a model of the HELP PUF? In particular, assuming you are the adversary and you have access to the HELP PUF for a month, how would you propose to model build the HELP PUF? Make sure to discuss in your answer the fact that the HELP PUF you are trying to clone uses 2 22 rising (zero to one transition) paths and 2 22 falling (one to zero transition) paths. 2) (15 pts.) Consider a PUF which is enrolled and then tested a week later at all possible combinations of 0ºC, 25ºC, 85ºC and supply voltage (nominal), supply voltage (5% more than nominal) and supply voltage (5% less than nominal). Consider the case of a specific challenge-response pair. One approach to measure HD intra is to compare the value of the response from enrollment versus the response from each of the combinations (nine comparisons total). A second approach is to compare each response with each other (45 comparisons in total). In your opinion, which approach is more appropriate for the authentication use case for the PUF (explain why)? Which approach to measure HD intra is more appropriate for encryption (please explain why)?