Skip to content

Latest commit

 

History

History
52 lines (45 loc) · 3.48 KB

README.md

File metadata and controls

52 lines (45 loc) · 3.48 KB

Data-Driven Beamforming Codebook Design to Improve Coverage in Millimeter Wave Networks

M. F. Ozkoc, C. Tunc, and S. Panwar, "Data-Driven Beamforming Codebook Design to Improve Coverage in Millimeter Wave Networks," 2022 IEEE 95th Vehicular Technology Conference (VTC2022-Spring)

Reproducing our results

  1. Download path information that we generated using Ray Tracing simulator, Remcom Wireless InSite. This includes BS locations, UE locations, AoA, AoD, power, and delays of each path.
  2. Using the ray tracing information we can generate channel data for each transmitter antenna. rayTracingToChannels script provided for a simple 32 element ULA array. The antenna array can be changed by adjusting the TX elements location matrix.
  3. Generate codebooks with proposed and baseline algorithms using the Main script.
  4. We used NYU HPC for running the design algorithms in every parameter scenario, I can share the final data and HPC scripts with interested readers. Please contact Mustafa through the email given in the paper.

Abstract:

In 5G systems, a predefined codebook with a limited number of beams is used during the initial access and beam management procedures to establish and maintain the connection between the users and the network. At 5G millimeter wave (mmWave) frequencies, due to the very narrow and directional beams obtained by beamforming, intelligently designing a codebook with a limited number of beams is crucial to avoid coverage holes. We formulate an optimization problem for the beam-codebook design to maximize the coverage probability, which is a quadratically-constrained mixed-integer problem. We propose a set of data-driven codebook design algorithms to solve the optimization problem, which, for a given codebook size constraint, adapts the codebook to the deployment scenario using the provided input channel data. For a sample deployment scenario, we show that as the codebook size increases, the proposed algorithms converge to the upper bound in terms of the coverage probability much faster than several benchmark algorithms. Hence, the proposed algorithms can achieve the coverage levels of benchmark algorithms with a much smaller codebook size. This can significantly reduce the initial access, beam management, and handover delays, which in turn provide higher data rates, lower latency, and lower interruption times.

Main Contributions

Our main contributions can be summarized as follows:

  • We formulate a data-driven beamforming codebook design problem to improve the coverage performance of a network by intelligently designing the beamforming codebook for a given codebook size constraint.

  • We propose a set of heuristic algorithms to solve the coverage-optimal codebook design problem, which outperforms the benchmark codebook design algorithms in terms of the coverage/outage probability.

  • The proposed receiver channel-based codebook selection algorithm, outperforms other benchmark and heuristic algorithms for all scenarios we considered. Moreover, the proposed cluster-based proportional beam waterfilling algorithm, performs well especially for low levels of signal-to-noise-ratio (SNR), which shows its effectiveness to improve the coverage of users with poor channel conditions.