Selfish mining simulator

This is a selfish-mining attack simulator with interactive GUI, implemented as a project for Dr. Grunspan Cyril's course "Cryptocurrencies" at ESILV.
The purpose of this work is to compute the profitability of selfish-mining strategy against Bitcoin blockchain, and in particular the first-day of gain.
Selfish mining is a deviant mining strategy, where a major mining operator withholds mined blocks and releases them with a well timed strategy in order to invalidate the maximum number of blocks mined by the rest of the network.
Profitability of the attack is directly related to reduction of mining difficulty as consequense of slowering down of the network.
In this sense, for some ranges of hashrate and connection rate, selfish-mining attacker may have to stand a first period when his reward is lower than the honest counterpart and this could be inhibitory since mining costs are generally constant and the attack may not be convenient.

Simulator interface shows:

• attack reward over time, parametrized on the hashrate;
• difficulty adjustment over time, where 1 is the initial difficulty level;
• distribution of the starting winning day, i.e. how many days attacker needs to wait before his strategy starts to be more profitable than honest mining (warning: average value is more undestimated as more simulations lie under the honest reward level, and the attacker never starts to have an advantage - I will fix this bug as soon as possible);
• distribution of orphan blocks, i.e. honestly mined blocks that end up to be invalidated after the release of the attacker branch.

Control variables are:

• attacker hashrate, i.e. attacker computational mining power over the total of all the miners;
• attacker connection rate, i.e. percentage of the blockchain the attacker has an advantage while release his branch;
• time horizon, i.e. amount of days simulations are performed on;
• no. of simulation, i.e. quantity of reproduced single attacks (notice that low values, as 50, 100 or 200, are enough to obtain a good simulations, higher values are usefull to better visualize probability distributions).

Instruction to run

By Command Window:

CD [current path]
conda activate
bokeh serve --show simu_guy.py

Preview

Possible improvements

• fix bugs on average line in 'starting winning day' plot;
• show variance values of distributions;
• build online version of simulator interface.

Resources

Grunspan C. and Perez R. "On Profitability of Selfish Mining
https://webusers.imj-prg.fr/~ricardo.perez-marco/publications/articles/OnSelfishMining20.pdf