# EPF Cohort 5 - Week 12 Date: 02/09/2024 ### Updates This week was rather interesting. Why? 1. Two more people joined us for working on the rated list dht [1], Lucas Aschenbach and Kilian Glas. 2. I attended the DAS collaborators meet where Leonardo Bautista from Codex presented some interesting insights on sampling and reconstruction based on simulations. * If we break down the ethereum network as is into levels based on a node's view (children, children of children etc.) much like the rated list. Then, through simulations it is shown that most of the samples needed would be reached through immediate peers only. This is a really strong insight of a DAS world. * In essence, he presented simulations with parameters varying over custody counts of individual nodes and node degrees of individual nodes. As part of their simulations they also scraped peering information of the entire ethereum network. ***This would prove to be very useful for rated list simulation*** 3. We started work on the simulator for the rated list. Kilian suggested the use of PeerSim as a simulator but we decided against it since that project is years old. * One of the main features of PeerSim is that it has a object oriented design (since it is written in java). It also clones the implementation of a p2p node and runs many instances in parallel to mimic actual behaviour. * The above feature is also why we rejected the idea, because with this project we want test against the security of the protocol more than we want to test against concurrency (many peers running at the same time). I want to ellaborate a little on the design of the simulator to highlight how we plan on simulating different conditions 1. Through dependency injection the simulator class accepts an already constructed graph and an implementation of the rated list node. 2. It then allows to bind "Node Profiles" to each vertex of the graph. * Node profiles are just a set of rules that define the behaviour of a node. There can be profiles defined for offline nodes, naively malicious nodes, adaptively malicious nodes etc. * This allows us to eliminate the need to clone instances of nodes and run them parallely as in peersim and at the same time test the security of protocol. That's it folks. > You can find the specs of rated list [here](https://github.com/dankrad/rated-list-specs) > You can find the code of rated list [here](https://github.com/hopinheimer/rated-list-research) ### Next Steps 1. Finish up on the sample gossip part of the spec. 2. Continue with the simulator design 3. Since the four of us decided to have weekly calls, organize them and have fun discussions about the ethereum network and peerdas. <- Exciting ### Personal Notes [1] The best part is that we were/are a part of TUM(Technische Universistat Munchen). I have attended Kilian's lectures on P2P security as a part of my masters.