changed 3 years ago  
Published Linked with GitHub

EPF update 2

Going from the last update, I was focusing on gaining more knowledge on the beacon chain and the algorithms that are in use to justify and finalize blocks (casper FFG and LMD GHOST) from a security standpoint.

Since validators are central to the security layer of the protocol (due to their stake and duties that they perform), I decided to do a deep dive into validator economics to understand how the network behaves under various actions validators can take as well network conditions.

In order to achieve this I used radCad, which is a dynamic system engineering software tool which can be used to model the Ethereum network. The software package is written in python and allows you to run parametrised models to conduct custom analysis of the POS network.

The main two ways you can run analysis and simulations are:

  • Time-based analysis: A simple example of this could be validator revenue and over time given their hardware setup
  • Phase space analysis: A phase space is a space where all possible states in dynamical systems are represented in a snapshot. Each state corresponds to a single unique instance in the phase space. An example experiment would be assessing validator yield depending on eth staked and eth price, irrespective of the time dimension. This is in order to see at which Eth price (in terms of USD) validator yield starts to deteriorate fast which can pose an issue to the overall security of the network.

The majority of the experiments that come baked in with the software tool are based on Hoban and Bergers Economic model which is highly comprehensive. To give a brief overview there are over 45 state variables and 35 system parameters that are modifiable when setting up an experiment analysis hence there is a lot of room to conduct very granular analysis. I have to say that covering all these variables also enabled me to get a better overview of the economics of the consensus layer in more detail and begin to grasp areas when the overall network security may start to deteriorate.

I highly recommend going through the radcad course if you want to quickly build a mental model of the intricacies of the CL from a cryptoeconomics standpoint and to learn to build comprehensive Ethereum economic models with speed. The course comes with a number of modelling questions and exercises for which you will be awarded a certificate if you manage to complete all of them.
After going over the full course, I am certain the toolset will come in handy in the near future whenever I need to validate security-related technical ideas and experiments.

Some resources I found to be useful in conjunction to the course include:

Image Not Showing Possible Reasons
  • The image file may be corrupted
  • The server hosting the image is unavailable
  • The image path is incorrect
  • The image format is not supported
Learn More →

Fig1: Certificate of completion

Initial project idea

As my interest continued to grow in topics at the intersection of CL security and cryptoeconomics, I reached out to Fredrik Svantes (EF Security Research) realising that his team was drafting up an intradisciplinary project which caught my attention. During our initial conversation, we discussed the potential benefits of using automated systems to track and detect anomalies in the Ethereum network as well as on-chain activity, to improve overall security and efficiency. We also discussed the use of machine learning to carry out automated detection using unsupervised learning techniques.

Since the project is in its initial phase and has not been started, I thought it would be a good challenge for me to focus my energy on getting the project off the ground in the upcoming months.

Next steps

  • Look into finding good data providers for network beaconchain data
  • Do some preliminary analysis using network data and beacon chain
  • Do some initial research on big data platforms to house all the netowrk data to perform analysis
Select a repo