No description or website provided.
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
CaseStudy
DataHex3
.gitignore
BlockageSimFn.m
BlockageSimFnNLOS.m
Generate_Mobility.m
HPC_Matlab.md
README.md
SimulationLOS.m
SimulationNLOS.m
TheoryLOS.m
TheoryNLOS.m
TheoryNLOS2.m
find_blockage_distance.m
hexagonal.m
mybatch.sbatch
plotResults.m
processData8.m
processData9.m
processDataHex.m
staticBlockage.m
test_Animate.m
test_Execute.m

README.md

Millimeter Wave Blockage Modeling

Work towards my thesis at NYU. This work is submitted to JSAC, please find the paper here.

We present a simplified model for the key QoS parameters such as blockage probability, frequency, and duration in mmWave cellular systems. Our model considers an open park-like area with dynamic blockage due to mobile blockers and self-blockage due to user's own body. A typical user is at the center and the BSs are distributed uniformly around the user. The user is considered blocked when all potential BSs around the UE are blocked simultaneously.

Understand the Code

SimulationLOS.m

The MATLAB simulations considers random waypoint mobility model for blockers. The main file is SimulationLOS.m.

It takes around 1 hour to run for a single iteration. We obtained results for 10,000 iterations using NYU HPC (high performance computing)

Note: SimulationLOS.m uses the following the functions- Generate_Mobility.m (copied from here), BlockageSimFn.m, find_blockage_distance.m.

The results are analysed using processData8.m

TheoryLOS.m

The TheoryLOS.m implements the theoretical results of LOS blockage. Finally, plotResults.m takes the data from csv files and plots the nice figures comparing theory and simulation :)

Results

Our results tentatively show that the density of BS required to provide acceptable quality of experience to AR/VR applications is much higher than that obtained by capacity requirements alone. This suggests that the mmWave cellular networks may be blockage limited instead of capacity limited.

Also, the optimal height of BSs would be lower as compared to microwave BSs.

Future Work

The following extensions are planned for future work

  • Generalized blockage model: We can add a simple model for static blockage in our analysis of dynamic and self-blockage.
  • Capacity analysis: The data rates of a typical user can be evaluated using the generalized blockage model. We are interested in evaluating whether 5G mmWave is capacity limited or blockage limited.
  • Fallback to 4G LTE: We plan to explore the potential solution to blockages as switching to 4G LTE. Whether 4G would be able to handle the huge intermittent 5G traffic.
  • Deterministic networks: We have considered a random deployment of BSs in our analysis. However, in most cases, the deployments of BSs are based on a deterministic hexagonal grid. Therefore, a blockage model for the deterministic networks is more practical.
  • Backhoul capacity analysis: With the UEs switching between BSs in case of blockage events, the backhoul capacity requirements for the BS may have high fluctuations. It is interesting to study that random process.
  • Correlated blockage: We assumed the dynamic blockages of all the BSs are independent (So the overall blockage probability is the product of the blockage probabilities of all the BS-UE links). However, when blockage of multiple BSs are correlated (depends on blocker location;) we might get a higher blockage probability.

Note our recent work include a hexagonal cell analysis for open park scenario and a NLOS model for urban scenario. The details can be found in our paper. The corresponding MATLAB codes will be updated soon.