MobiCom'24 Mambas

In this repository, you can find two components of codes in Mambas:

a) the 5G PHY implementation on the COSMOS testbed (labeled as [Testbed] in paper),

b) the ray-tracing based simulations (labeled as [Simulation] in the paper).

Prerequisite

Python

1. numpy: e.g., conda install anaconda::numpy
2. tqdm: e.g., conda install conda-forge::tqdm
3. scipy: e.g., conda install anaconda::scipy
4. matplotlib: e.g., conda install conda-forge::matplotlib
5. casadi: e.g., conda install conda-forge::casadi

MATLAB

1. 5G Toolbox

UHD

UHD 4.5.0 is recommended, and in this case, you can skip Step 1.

5G PHY Implementation

This 5G PHY implementation measures the over-the-air metrics, SNR, EVM, BLER, BER and etc, under an over-the-air USRP setting.

You can find the codes in the folder 5g-phy-implementation, and eventually, you can implement the SNR-MCS curves as shown in Fig. 8 in the paper (shown below).

Ray-Tracing based Simulations

There are three steps in the simulations:

1. Generate the Ray-Tracing dataset in MATLAB
2. Perform Mambas or other baselines on the generated Ray-Tracing dataset
3. Plot the simulation results

You can find the codes in the folder ray-tracing-based-simulations, and eventually, you can implement the simulation experiments as shown in Fig. 13-14 in the paper (shown below).

Reference

If you find our work useful in your research, please consider citing our paper:

@inproceedings{gao2024mambas,
  title={Mambas: Maneuvering analog multi-user beamforming using an array of subarrays in mmWave networks},
  author={Gao, Zhihui and Qi, Zhenzhou and Chen, Tingjun},
  booktitle={Proc. ACM MobiCom'24},
  year={2024}
}

If you have any further questions, please feel free to contact us at :D

zhihui.gao@duke.edu