🐼 PANDAS: Simulator and Prototype Experiments

---

📘 Table of Contents

1. Introduction
2. Hardware
3. Simulator
   • Setup and Build
   • Running Experiments
   • Plotting Results
4. Prototype
   • Setup and Build
   • Running Experiments
   • Plotting Results
5. Repository Structure

---

🧩 Introduction

The artifacts of this work are available at:
👉 CloudLargeScale-UCLouvain/PANDAS

The repository includes:

• The source code of PANDAS
• The scripts to run the experiments
• The plotting script to plot experiments results

Both the simulator and the prototype implementations were run on Linux-based operating systems.

---

Hardware

We run the simulator and a local version of the prototype on a server with 18-core Intel Xeon Gold 5220 CPU and 96 GB of RAM.

---

📁 Repository Structure

PANDAS/
├── simulator/                      # Source code for large-scale simulator experiments
│   ├── configs/                    # Configuration files for network simulations
│   │    ├── configPaper/           # Configuration files used in the paper for perf: Figure 12 13 and 14 
│   │    ├── configAdversary/       # Configuration files for resilience against dead nodes: Figure 15
│   │    ├── configPaper1K/         # Configuration files for test of redundancy of samples: Figure 9 10 11 12
│   │    ├── configReel/            # Configuration files for validation with the prototype and validation of strategy
│   │    └── configPaper/           # Configuration files for validation with the prototype and validation of strategy
│   ├── src/                        # Code of the simulator
│   │    └── kademlia/das           # Folder containing PANDAS code used in the sumlator
│   └── python/                     # Plotting scripts
└── prototype/                      # Source code for real-world prototype experiments
│   ├── Latency/                    # Latency used for the prototype
│   ├── Launching_scripts/          # Script used to install and launch experiment on grid5000
│   ├── results/                    # Prototype experiment results
│   └── python/                     # Log processing and plotting

---

🧪 Simulator

The simulator is implemented in Java 17 using Maven as a build tool.
It is built on top of Peersim and based on the Kademlia implementation by Daniele Furlan and Maurizio Bonani:
👉 http://peersim.sourceforge.net/code/kademlia.zip

🔧 Setup and Build

To install all requirements and build the simulator:

cd simulator/
sudo chmod +x install.sh && ./install.sh

This script will:

• Install Java 17 and Maven
• Fetch all dependencies
• Build the simulator

After this, the simulator is ready to run experiments.

---

▶️ Running Experiments

The simulator uses configuration files located in simulator/configs/.
Each configuration file corresponds to a specific network size (e.g., 1k.cfg simulates a network of 1,000 nodes).

To run an experiment:

cd simulator/
./run.sh <config_file>

🗂 Results will be stored in:

• simulator/Results/ → Experiment results
• simulator/logs/ → Detailed logs

---

📊 Plotting Results

To plot the experiment results:

cd simulator/python
python3 plot_results.py

The generated plots will be saved in simulator/python/plots/

---

🌐 Prototype

The prototype is implemented in Go and built on top of libp2p.

For the paper, we run the prototype on Grid5000, a French distributed cluster for experiment-driven research in computer science.
It can also be executed locally at a smaller scale.

---

🔧 Setup and Build

To install requirements and build the prototype:

cd prototype/
sudo chmod +x setup.sh && ./setup.sh

This script installs:

• The latest version of Go
• All required dependencies
• The prototype binaries

After setup, the prototype is ready to run experiments.

---

▶️ Running Experiments

1. Create a Topology File

First, generate a nodes.csv file defining the network topology (list of nodes and their unique peer IDs):

./create_topo.sh <number_of_nodes>

2. Run the Prototype

Run the network using the topology file:

./run_local.sh <topology_file>

🗂 Results will be stored in:

• prototype/results/ → Experiment results
• prototype/logs/ → Detailed logs

---

📊 Plotting Results

To process logs and plot results:

cd prototype/python
python3 process_logs.py <results_folder>

This generates visual analyses and metrics from the prototype experiments.

---