Secure timestamping

Just stumbled upon the project while looking for something. Is that something of interest to you? I work in related areas. Happy to collaborate.

2 Likes

Yes, it surely is. But can time-sync be done securely in byzantine environments with granularity in millisecs?

I am working on an extension of this paper: https://www.usenix.org/system/files/cset19-paper_anwar.pdf.

Primarily, the problems to be solved are:

  1. Restricting software access to the timer
  2. Ensuring that the OS scheduler cant prevent time-update related processes from advancing
  3. Ensuring the OS or routers can’t delay network packets

Interesting! Will go through the linked paper. Essentially, we had arrived at a similar conclusion when we had gone through SGX based approaches primarily the ability of a malicious actor to add arbitrary delays before passing network packets to the TEE negating the benefits of the latter (which seemed unsolvable in our opinion).

@Prateek was once looking at it.

There are some relevant thread on ethresearch that you might want to check:

  1. https://ethresear.ch/t/time-as-a-public-service-in-byzantine-context/6895
  2. https://ethresear.ch/t/time-attacks-and-security-models/6936

Thanks for sharing!

@prateek @vunguyen The service provider model mentioned in the first link seems interesting for the cluster-based approach.

Might as well add https://ethresear.ch/t/sensor-fusion-for-bft-clock-sync/7142