Welcome to CS 6831, Designing Secure Cryptography. We will be learning about the modern theory and practice of the cryptographic tools deployed to protect contemporary computing systems. In addition to learning more about the types of cryptographic tools currently used, we will be focusing on understanding the processes that, ideally, lead to secure tools. Topics include formulation of security definitions, reduction-based concrete security analysis, evaluation of real-world implementations, common pitfalls, and attacks. Active research topics will be highlighted throughout.
Instructor: Tom Ristenpart (https://rist.tech.cornell.edu)
Time: 1:55pm-3:10pm M/W
NYC classroom: Bloomberg 091
Ithaca classroom: Gates 405
Video recordings: Available here (Cornell login required)
By the end of the class you should have the ability to design and argue the security of new cryptographic tools, at the level of quality of new publishable research results. This will be a fast-paced course.
Students will be expected to have taken courses or otherwise have substantial background on computer security; have knowledge of basic cryptographic concepts (encryption, hashing); and experience doing formal math (proofs, etc.).
The class is for PhD students, and will assume a lot of shared background material. Cornell Tech masters students should only consider the class if they are comfortable with probability theory, discrete mathematics, computational complexity, and computer security. You might refer to the Bellare-Rogaway notes, Section 1.5. For a taste of what they mean by mathematical maturity, refer to the Boneh-Shoup book Section 2.4. If that material doesn't look familiar, or you spent a lot of time struggling to understand it, then that may suggest you'll have a hard time getting much out of the class. The first time I took a PhD class of this form, I spent innumerable hours a week on it, and had the freedom to do so as a PhD student with few other responsibilities.
It is really important to note that this does not mean you are not capable of mastering cryptography, and I don't want to discourage you from the field. Rather, there's a bit of a jargon and concept barrier that must be filled in, and doing so would unfairly sacrifice content from the PhD students. Usually we teach a masters level course that covers a subset of this material at a more suitable pace, but unfortunately that's not possible this year.
With that set of warnings in place, Cornell Tech students should talk to me if they want to take the class after the first lecture. Be prepared to answer some questions about your understanding of Boneh-Shoup Section 2.4.
The class will involve two main deliverables. One will be scribe notes. The goal will be to have a complete set of notes for future reference. We will figure out early on the scribing schedule. Each student will be responsible for delivering one or more complete writeups of lecture topic(s). This will notably include finished, precise proofs that may be only covered at a high level in class, accompanied by beautiful English prose. See new section added below.
Second will be a semester project, the aspirational goal being a publication-worthy result in applied cryptography. The project deliverable will be a short report together with a one-on-one oral discussion (possibly over Zoom) with the me about the result. A paper worked on with others can substitute for the short report, though you will need to identify your specific contributions. The one-on-one discussion will allow you to gain practice describing your work in a professional setting, allow me to assess what you've learned from the class, and help determine your final grade. More details will be forthcoming.
I may also point out problems or omitted proofs that you will want to work on to solidify your understanding of material. These will be ungraded, but I can give you feedback if you desire.
Grades will be assigned according to the following breakdown:
- Participation: 20%
- Scribe notes : 40%
- Project: 40%
I will send out a spreadsheet with a tentativie scribing schedule. PhD students will be doing scribing. There will be one scribe and two proofreaders for each lecture. The scribe will write up an initial draft of the notes for the lecture, and the proofreaders will provide technical and editorial feedback before I take a look. The scribe and proofreaders are, together, also responsible for coming up with three self-study questions (i.e., homework problems) that speak to the material. These should not be questions that require regurgitating something, but rather that make people think deeply about the material. Feel free to borrow problems from public sources (e.g., Boneh-Shoup book), but be very clear about attribution (even if it was just used for inspiration) and include the attribution in the notes. Initial draft is due a week after lecture, and proof-read finald raft 1.5 weeks after lecture. If there's going to be some reason for a delay, let me know.
The masters students will be responsible for testing out (a reasonable subset) of these questions by working out candidate solutions individually. We'll group things into ~6-lecture chunks, and assign a problem from each. So masters students can expect 3 problem sets, finishing 3/4 of the way through the class to minimize problem set and end-of-term project overlaps. If the problems are too hard, we'll adjust accordingly how many students need to solve. These will be peer-graded by the PhD student(s) who first conceived the problem with my oversight and final approval of grading. The involved PhD students and masters students will determine the worst solution, and the responsible student will get the honor of typesetting a correct solution for inclusion in an appendix in the notes. They can use as reference all the solutions, and work with whomever can help them out to do so. For the masters students the 40% of class grade will be associated to performance on problem sets. Of course, MS students can help out with proof-reading and embellishments to scribe notes via pull requests (see below), and this will positively count towards your participation and scribe grades.
I have added the notes to the git repository. You can therefore fork this repository, work in a local copy, and then submit a pull request to me once the notes are ready for my review. They should already be proofread by both proofreaders before a PR comes in. Let me know if this workflow doesn't work out for you.
The end result should be an open-source set of notes that others can use as they explore this topic. Everyone who finishes the class will get appropriate credit for contributing.
All this will require some good will and coordination among students. If anyone has any concerns or problems, don't hesitate to talk to me.
We will partition semester project into two milestones. The first is a proposal, due on March 6, via email. It should be no more than one page, and be a brief description of the proposed project. It is ok to have groups, but the level of work should scale in the size of the group. The final deliverable will be a project report due at end of semester, exact date to be determined.
The project should have a well-defined goal that is on the cutting edge of cryptography, the litmus being some publishable result (not necessarily a whole paper, but a result within a publishable paper would be fine). You can propose an ongoing research project as your class project, but make sure I know who is involved, what your contribution will be. Similarly the writeup can be dovetailed with a research paper in preparation, with the same requirement that your writing contribution should be identifiable.
Time allowing we'll have either group presentations at the end of the semester, or meetings with just me to go over the project and other topics.
One of the goals of the course will be to produce lecture notes on this content useful for others. The following excellent references will be useful throughout the course, particularly the Boneh-Shoup book and Bellare-Rogaway notes.
-
A Graduate Course in Cryptography. Boneh-Shoup book. A mixture of asymptotic and concrete security.
-
Bellare and Rogaway lecture notes. A somewhat dated set of lecture notes. Focuses on concrete security approach.
-
Bellare lecture slides. Slides from Mihir Bellare's graduate course in cryptography, based in large part on the notes.
-
Modern Cryptography by Katz and Lindell. Katz-Lindell book focuses on asymptotic definitions.
A very preliminary and aspirational schedule is below to give a taste of the scope of what we're hoping to cover.
| Date | Topic | Slides | Notes |
|---|---|---|---|
| Jan 23 | Intro | Slides | |
| Jan 28 | Block ciphers, PRPs, PRFs | Slides | |
| Jan 30 | PRP/PRF Switching lemma, Luby-Rackoff, Shuffling | Slides | |
| Feb 4 | Block ciphers, cryptanalysis | Slides | |
| Feb 6 | Frequency analysis, tweakable block ciphers | Slides | |
| Feb 11 | Tweakable block ciphers, randomized encryption | Slides | |
| Feb 13 | Randomized encryption | Slides | |
| Feb 18 | Authenticated encryption | Slides | |
| Feb 20 | Message authentication | Slides | |
| Feb 25 | No Lecture (February break) | ||
| Feb 27 | AE in practice and theory | Slides | |
| Mar 4 | Class cancelled | ||
| Mar 6 | Hash functions | Slides | |
| Mar 11 | Hash functions 2 | Slides | |
| Mar 13 | Hash functions 3 | Slides | |
| Mar 18 | Computational number theory basics, RSA | ||
| Mar 20 | Public-key encryption | ||
| Mar 27 | Diffie-Hellman | ||
| Apr 1 | No lecture (Spring Break) | ||
| Apr 3 | No lecture (Spring Break) | ||
| Apr 8 | Digital signatures | ||
| Apr 10 | TLS 1.3 | ||
| Apr 15 | Signal | ||
| Apr 17 | Group messaging | ||
| Apr 22 | Identification protocols and Fiat-Shamir | ||
| Apr 24 | Zero-knowledge proofs | ||
| Apr 29 | TBA | ||
| May 1 | TBA | ||
| May 6 | TBA |