Reduced Complexity Channel Models for IMT-Advanced Evaluation

Yu Zhang, Jianhua Zhang, Peter J. Smith, Mansoor Shafi, Ping Zhang

Introduction

The design and evaluation of next-generation wireless networks, such as IMT-Advanced (4G and 5G), rely heavily on accurate channel models. These models simulate how radio waves propagate through complex environments, interacting with obstacles like buildings and foliage. Engineers typically choose between simple correlation-based models, which are computationally efficient but lack physical realism, and Geometry-Based Stochastic Models (GBSMs), which offer high accuracy at the cost of significant computational overhead.

As systems evolve toward wideband MIMO (Multiple Input Multiple Output) with massive antenna arrays, the complexity of GBSMs becomes a major bottleneck. Simulating a full network environment requires thousands of "drops" or channel snapshots, each involving intensive convolutions. This paper proposes a suite of simplification techniques to reduce this computational burden while maintaining high simulation accuracy.

The Geometry-Based Stochastic Model

GBSMs characterize the radio channel by defining the positions of scatterers (clusters) and the paths (rays) between the transmitter and receiver.

Geometry-based stochastic model (GBSM) illustrating the complex relationship between base station (BS) and mobile station (MS) arrays via multiple signal clusters and rays.

As shown in the figure above, each signal path is defined by specific angles of departure and arrival, delays, and powers. The sheer number of these parameters, especially in wideband scenarios with many clusters, leads to the high computational cost addressed by the authors.

Proposed Simplification Methods

The researchers investigated five specific simplification strategies (SM-A through SM-E) applied to the standard WINNER channel model:

  1. Cluster Clipping (SM-A): This method removes signal clusters with power levels significantly lower (e.g., 25 dB) than the strongest path. To maintain the total received power, the energy from clipped clusters is redistributed to the nearest remaining cluster.
  2. Fixed Delay Spread (SM-B): Instead of treating the Root Mean Square (RMS) Delay Spread as a random variable for every snapshot, it is fixed to its statistical mean.
  3. Fixed Cross-Polarization Ratio (SM-C): The XPR, which describes signal leakage between different polarizations, is fixed to its mean value.
  4. Uncorrelated Large-Scale Parameters (SM-D): Mathematical correlations between variables like delay spread, angular spread, and shadow fading are removed to simplify the generation of random variables.
  5. The "All-In" Model (SM-E): This comprehensive model combines all the above simplifications for maximum efficiency.

Results and Performance Trade-offs

The study validated these simplifications using both indoor measurements at 5.25 GHz and synthetic outdoor data for urban macrocell scenarios. The primary goal was to find the optimal "clipping threshold" that balances speed and accuracy.

Impact of cluster clipping threshold on normalized computational time (NCT) and mean relative error (MRE) of delay spread.

As illustrated in the results, increasing the clipping threshold significantly reduces the Normalized Computational Time (NCT). For example, the "SM-E" model achieved a speedup of up to 57% in indoor scenarios. Crucially, the Mean Relative Error (MRE) of the delay spread remains low at moderate thresholds. The researchers found that the relative deviation of the 5% outage capacity—a key metric for network performance—remained within a tight 5% margin.

Conclusion

By strategically reducing the complexity of GBSMs, specifically through cluster clipping and fixing certain large-scale parameters to their means, simulation times can be nearly halved with negligible loss in accuracy. These simplified models provide a practical tool for the large-scale system-level evaluations required for modern wireless standards. Future work will extend these techniques to three-dimensional models to account for elevation spread, further refining the balance between realism and computational efficiency.

Summary updated

BibTeX
@article{zhang2009reduced,
  author = {Yu Zhang and Jianhua Zhang and Peter J. Smith and Mansoor Shafi and Ping Zhang},
  title = {{Reduced Complexity Channel Models for IMT-Advanced Evaluation}},
  journal = {EURASIP J. Wirel. Commun. Netw.},
  year = {2009},
  doi = {10.1155/2009/195480},
}
Figure