贝叶斯网络进行实时任务协调的多无人机系统自适应通信模式选择的概率模型.zip

# Probabilistic Model for Communication Mode Selection for Autonomous Multi-UAV Systems This project focuses on developing a probabilistic model for selecting communication modes in autonomous multi-UAV (Unmanned Aerial Vehicle) systems. It uses Bayesian Networks to make decisions based on real-time data in dynamic environments, enhancing collaboration and mission success rates in multi-UAV operations. ## Table of Contents - [Overview](#overview) - [Installation](#installation) - [Results](#results) - [Submission](#submission) - [Contributing](#contributing) - [License](#license) ## Overview The project addresses the challenge of communication mode selection in multi-UAV (Unmanned Aerial Vehicle) systems, which operate in unpredictable and dynamic environments. By utilizing a probabilistic model based on Bayesian Networks (BN), UAVs can autonomously and adaptively select between two communication networks: a local UAV-to-UAV network and a cloud-based network. This decision is made in real-time based on mission requirements, UAV network state, and environmental data, ensuring effective coordination and mission success. The model demonstrates high efficiency, with rapid decision-making, making it suitable for real-time multi-UAV operations. ## Installation ### System Requirements - MATLAB (compatible with the BNT toolbox) - The BNT library from [https://github.com/bayesnet/bnt](https://github.com/bayesnet/bnt) ### Getting Started This project depends on the [Bayesian Network Toolbox (BNT)](https://github.com/bayesnet/bnt). Follow these steps to set up the BNT library in MATLAB: 1. Clone the BNT repository: ```bash git clone https://github.com/bayesnet/bnt.git ``` 2. Clone this project repository: ```bash git clone https://github.com/MohandHAMADOUCHE/Probabilistic-uav-comm.git cd Probabilistic-uav-comm ``` 3. Open MATLAB and modify the variable `myPath` in the `add_bnt_paths()` function to match the path to your extracted BNT directory: ```matlab myPath = genpathKPM('C:\Users\info\Downloads\bnt\bnt'); % Modify this path as needed ``` 4. Execute the `main()` function in MATLAB to run the project: ```matlab main() ``` ## Results ### Validation with GeNIe The validation version using the **GeNIe** tool can be found in the folder `GeNie model`, in the file `Communication_network_selection_BN.xdsl`. GeNIe is a tool for modeling and inference of Bayesian Networks. You can download the GeNIe tool from their [official website](https://www.bayesfusion.com/genie/). #### Resulting Bayesian Network <p align="center"> <img src="./GeNie model/BN_picture.PNG" alt="Resulting Bayesian Network" width="400"/> </p> *Figure 1: The resulting Bayesian Network after validation in GeNIe.* ### Validation with Matlab I performed multiple runs to determine the **average CPU time** for both creating the Bayesian Network and performing the inference. - **Average Bayesian Network Creation Time**: Approximately **1 millisecond**. - **Average Inference Time**: Approximately **6 milliseconds**. These averages were obtained after several test executions, ensuring consistency in performance across different runs. #### Average CPU time images <p align="center"> <img src="./Execution%20results/mini_avg_creation_cpu_seconds.png" alt="Average Creation CPU Time" width="300"/> <img src="./Execution%20results/mini_avg_inference_cpu_ms.png" alt="Average Inference CPU Time" width="300"/> </p> *Figure 2 (left): Average CPU time required to create the Bayesian Network. Figure 3 (right): Average CPU time required to perform the inference.* #### Potential Speedup On their [website](https://www.cs.ubc.ca/~murphyk/Software/BNT/usage_02nov13.html#installC), they mention that using a C implementation could provide a **5 to 10 times speedup** in performance. This could significantly reduce the execution time for both the creation of the Bayesian Network and the inference process, making it even more suitable for real-time applications. ## Submission This work is currently under submission for a conference. Once the article is published, a link to the final version of the paper will be provided here. <!--This work has been published. You can access the article [here](https://link-to-your-article.com). --> ## Contributing We welcome contributions! If you are interested in improving the model or adding features, please follow these steps: 1. Fork the repository. 2. Create a new branch for your feature or bug fix. 3. Submit a pull request with a detailed explanation of your changes. ### Reporting Issues If you encounter any bugs or have suggestions, feel free to open an issue in the repository. We appreciate your feedback! ## License This project is licensed under the MIT License - see the [LICENSE](./LICENSE) file for details. The project is open source and free to use. Please cite this work in your publications: ```bibtex @misc{myModel, author = {Hamadouche, Mohand}, title = {Probabilistic Model for Communication Mode Selection for Autonomous Multi-UAV Systems}, year = {2024}, url = {https://github.com/MohandHAMADOUCHE/Probabilistic-uav-comm}, note = {GitHub repository, accessed September 18, 2024}, }