ray-tracing-based-simulations

Ray-Tracing based Simulations

In this code, you can generate your own Ray-Tracing dataset in MATLAB and perform the simulation of Mambas.

Step 1: Generate the Ray-Tracing Dataset

In MATLAB, open RayTracing.m.

You can customize the three parameters below:

1. ueNum: the number of users;
2. distBound: the min/max BS-user distance in the unit of m;
3. angleBound: the field of view (FoV) of the BS, or the min/max azimuth angle of the users, in the unit of degree.

Below is an example of the MATLAB Ray-Tracing interface.

Generating the whole dataset can take hours. To skip the process, you may unzip Dataset.zip, which is pre-generated under the given settings.

Step 2: Perform Mambas

In Python, open main.py.

You can customize the arguments below:

1. scheduler: the user selection algorithm, where
   • DP: the dynamic programming algorithm proposed in the paper;
   • TDMA: the baseline that only selects a single user using standard proportional fair (PF);
   • All: the baseline that always selects all the users;
   • Spacing: the baseline of S2-MAS that selects the subset of users with a minimum angular spacing;
2. optimizer: the beamforming optimization algorithm, where
   • Individual: the MAX-GAIN-ZF algorithm proposed in the paper;
   • Conjugate: the baseline of conjugate beamforming, discussed in Eq. (2) in the paper;
   • MU-ZF: the baseline of full digital multi-user beamforming with zero-forcing, discussed in Eq. (3) in the paper;
   • Nullify: the baseline of Nulli-Fi, which fine-tunes the beamforming weights from conjugate beamforming to minimize the cross-user interference;
3. bound: the field of view (FoV) of the BS, or the min/max azimuth angle of the users, in the unit of degree;
4. ueNum: the number of users, which is U in the paper;
5. elem: the number of elements in a subarray as an elemx1 geometry, which is N_s in the paper;
6. array: the number of subarrays in the whole array of subarray, which is M in the paper.

Note that ueNum or U in main.py can be smaller than the ueNum in RayTracing.m, where only the first U users will be considered.

Performing all the algorithms can take days. To skip the process, you may unzip Result.zip, which contains most of the combinations above.

Step 3: Plot the Results

In MATLAB, open Plot.m.

You can customize the parameters below to specify the results to be visualized:

1. scenarioShow: the scenario index to see the scheduling details;
2. slotNum: the number of time slots of the simulation, which should be lower than 20.

For the first figure (an example shown below), it shows the detailed scheduling of one simulated Ray-Tracing scenario (X-axis as the time slot index and Y-axis as the user index) over baselines and proposed Mambas, where each grid indicates the data rate (in Mbps) achieved given the time slot and user. Note that 0 means either the user is not activated or packet loss.

For the second figure (an example shown below), it shows the statistical results over all the Ray-Tracing scenarios w.r.t. the averaged data rates over users per scenario.