[toc]

###### tags: `Reading sessions`

# 2023

<https://eurosp2023.ieee-security.org/accepted_and_awards.html>

## [Multi-Factor Credential Hashing for Asymmetric Brute-Force Attack Resistance](https://arxiv.org/pdf/2306.08169.pdf)
* Vivek Nair and Dawn Song

* [SS] In a typical password hashing deployment, a user reg- isters an account with a server by providing a password, a hash of which is stored by the server in a database. Later, the user initiates a login process with a server by providing the password, which the server hashes and compares to the stored hash to authenticate the request.
Password based hasing suffers from brute force attack. Current practice is to add salt to increase the brute force attack time. There are some purpose built hashes such as Bcrypt, which they call adaptive password hashing, that increases resistance against the brute force attack (increased attack time) however, that slows down the verification time. Such functions where verification time scales with brute force attack time are called to have symmetric resistance.
Another way to increse the attack time is to apply multi factor authentications — for example, secondary authentication factors are independently verified.
In this paper, authors present a multi-factor credential hashing approach that differs from typical password hashing by atomically verifying all of a user’s credentials with a single hash.
One of the goals is to have the Asymmetric Resistance — for a given fixed verification time, a multi-factor hash should be significantly more difficult to brute-force than a standard password hash. Unlike symmetric resistance, this property says that the verification will not grow in a similar fashion as the attack time grows.
In their constrction, they use two hash functions inner and outer. The inner one (H1) is usual adaptive password hashing such as Argon2 and outer one (H2) is SHA256. H1 takes input passowrd, salt, password and 2FA (OTP) as input and the output goes through H2 to produce the final output. The OTP key blinded as bld = key + H1_output which is stored along with H2 output, salt and counter. While login, user can be verified having the password and the OTP. So, this construction simultaneously verifying password and OTP, where OTP is enhancing the entropy of the crednetial. One can increase the difficulty of brute force attack by increasing the OTP length, however, this will not affect the verification time at the same scale.

* [BB] Short rview by BB

###### tags: `` ``

## [Understanding the Security Risks of Decentralized Exchanges by Uncovering Unfair Trades in the Wild](https://ieeexplore.ieee.org/document/10190515)
* By authors
* [AA] Short review by AA
* [BB] Short rview by BB

###### tags: `` ``

## [When the Curious Abandon Honesty: Federated Learning Is Not Private](https://link)
* By authors
* [AA] Short review by AA
* [BB] Short rview by BB

###### tags: `` ``

## [Been Here Already? Detecting Synchronized Browsers in the Wild](https://link)
* By authors
* [AA] Short review by AA
* [BB] Short rview by BB

###### tags: `` ``

## [Android, Notify Me When It Is Time To Go Phishing](https://link)

###### tags: `` ``

## [Reconstructing Individual Data Points in Federated Learning Hardened with Differential Privacy and Secure Aggregation](https://arxiv.org/pdf/2301.04017.pdf)

###### tags: `` ``

---