Security Analysis of Distributed Ledgers and Blockchains through\n Agent-based Simulation

Abstract

In this paper we describe LUNES-Blockchain, an agent-based simulator of\nblockchains that relies on Parallel and Distributed Simulation (PADS)\ntechniques to obtain high scalability. The software is organized as a\nmulti-level simulator that permits to simulate a virtual environment, made of\nmany nodes running the protocol of a specific Distributed Ledger Technology\n(DLT), such as the Bitcoin or the Ethereum blockchains. This virtual\nenvironment is executed on top of a lower-level Peer-to-Peer (P2P) network\noverlay, which can be structured based on different topologies and with a given\nnumber of nodes and edges. Functionalities at different levels of abstraction\nare managed separately, by different software modules and with different time\ngranularity. This allows for accurate simulations, where (and when) it is\nneeded, and enhances the simulation performance. Using LUNES-Blockchain, it is\npossible to simulate different types of attacks on the DLT. In this paper, we\nspecifically focus on the P2P layer, considering the selfish mining, the 51%\nattack and the Sybil attack. For which concerns selfish mining and the 51%\nattack, our aim is to understand how much the hash-rate (i.e. a general measure\nof the processing power in the blockchain network) of the attacker can\ninfluence the outcome of the misbehaviour. On the other hand, in the filtering\ndenial of service (i.e. Sybil Attack), we investigate which dissemination\nprotocol in the underlying P2P network makes the system more resilient to a\nvarying number of nodes that drop the messages. The results confirm the\nviability of the simulation-based techniques for the investigation of security\naspects of DLTs.\n