This repository represents my work for the final project of NU CE 334 Blockchains and Decentralization.

This is the code for analyzing and simulating the private mining with reset scenario. See the paper for details on reducing the private mining scenario to a form that can be analyzed by code.

Files

src/reset_policy.rs

A reset policy, as defined by the paper, is characterized by a threshold function $r : \mathbb{N} \rightarrow \mathbb{N}$ which maps a value of $h_A$ to the corresponding value of $h'$ above which the adversary should reset. As described in the paper, $r$ is constant once you pass $h_A \geq k$. As such, only the first $k + 1$ values of the function are required to completely describe a reset policy.

Therefore, throughout the code, the [i64] type is used to describe a reset policy, where policy[i] takes the value $r(i)$. This means that the first value of the [i64] array is always 0.

The file src/reset_policy.rs provides utilities for creating reset policies. The function linear_reset_policy creates a reset policy for the given confirmation depth where $r$ takes on max_tolerable_height_difference for $h_A \geq k$, and linearly interpolates the points for $0 < h_A < k$ PolicyIterator is an iterator that, for a given confirmation depth enumerates every single valid reset policy with thresholds below some maximum value; i.e. every non-decreasing sequence of size size, starting at 0 and not exceeding max_value. The size argument taken by PolicyIterator::new is one more than the confirmation depth (to account for the required $0$ at the beginning).

src/reset_policy.rs

This file follows the process described in the paper to create a transition matrix out of the private mining with reset scenario, and then solves for the expected value to reach the absorbing win state.

src/main.rs

This file defines a parameter space over $\alpha$, $k$, $\lambda_B\Delta$, and all reset policies below a certain threshold, and then calculates the expected number of steps for each set of parameters. The results are printed to standard output.

src/simulation.rs

This file contains functions for calculating the expected number of steps using simulations. simulate_until_win simulates one trial given $\alpha$ and the reset policy, while calculate_expected_steps simulates the given number of trials for the same given parameters and returns the average.

These functions are not used in the actual code; they only exist for testing purposes to verify the analytic solutions. These functions also do not account for possible propagation delay (because when I got to that point I was already confident in the analytic solutions' correctness).

plot_results.py

Given a file named results.txt with the standard output of the Rust program, this file creates the plots seen in the paper. It was run in Google Colab, and might not work as a headless script.