MatthiasLohr / bfebench

mirror repository for https://gitlab.com/MatthiasLohr/bfebench for anonymization via https://anonymous.4open.science

BFEBench - Blockchain-based Fair Exchange Benchmark Tool

The BFEbench provides a benchmark framework for blockchain based fair exchange protocol. For the main project goals see the [base milestone](%"Comparison Fairswap vs. StateChannelFairswap")

Supported protocols:

• Fairswap
• FairswapReusable
• StateChannelFileSale

Usage

• Download and install bfebench (e.g., by cloning the git repository and executing pip install -e .). We suggest using a virtual environment for that.
• Select a Ethereum compatible blockchain platform to run bfebench on. We provide a couple of test networks in the blockchain-networks directory, which are preconfigured and can be started using docker-compose.
• Create an environments configuration file for the blockchain environment you selected. For the test networks in blockchain-networks, we provide a ready-to-use environments configuration file (named bfebench-environments.yaml).
• Select a protocol for benchmarking. You can list all available protocols by executing bfebench list-protocols.
• Select the strategies to be simulated for seller and buyer. You can list all available strategies for the selected protocol by executing bfebench list-strategies <protocol>.
• Start the simulation by executing

  bfebench run <protocol> <seller strategy> <buyer strategy> <file to be exchanged> -e <environments configuration>
  

For a list of all supported commands and options run bfebench -h.

Evaluation

• Exp1: Runtime Performance
  • Only faithful exchange
  • Compare all protocol impls (4 variants) with different data sizes
  • Output:
    • Protocol Comparision -- Chart: x=size, y=time (containing all 4 protocol impls)
      • Matrix:
      • Protocol: [Public,Public-SC,Permissioned,Permissioned-SC]
      • Data size: [1KB,2KB,3KB,...]
    • Effect of Chain -- 4 Charts (1 per protocol impl): x=percentage of execution time (CPU-seller/chain/CPU-buyer)
• Exp2: Costs
  • a) Protocol Comparision
    • only for Faithful
    • Costs in relation to exchanged asset size
      • Matrix:
      • Protocol: [Public,Public-SC,Permissioned,Permissioned-SC]
      • Data size: [1KB,2KB,3KB,...]
  • b) Cheating variant cost comparison: Max damage for faithful party
    • Fixed data size (log influence of size on cost in case of cheating)
    • b1) What is a baseline of max damage for faithful party? (not a key contrib of paper)
      • Matrix:
      • Protocol: [Public,Permissioned]
      • Data size: [1KB,2KB,3KB,...]
      • Strategy: [SellerFaithful/BuyerMaxDamage,SellerMaxDamage/BuyerFaithful]
    • b2) How does the cost with state channel behave in case of cheating and how des it relate to the baseline (b1)?
      • Matrix:
      • Protocol: [Public-SC,Permissioned-SC]
      • Data size: [1KB,2KB,3KB,...]
      • Strategy: [Cheating1,Cheating2,Cheating3,...]