TL;DR
This paper tackles the critical issue of how physical obstructions (blockages) in mmWave cellular networks, which are often correlated, impact network performance. By developing analytical models for both 1D and 2D networks, the authors demonstrate that ignoring this correlation leads to inaccurate predictions, especially for larger blockages, and highlight the need for more sophisticated modeling.
What Problem Does This Paper Address?
Millimeter-wave (mmWave) cellular networks, crucial for future high-speed wireless communication, are highly susceptible to signal degradation caused by physical blockages like buildings, foliage, and even people. Traditional models often treat these blocking events as independent. However, a single large object can obstruct signals from multiple base stations (BSs) to a user, creating a spatial correlation in blocking probabilities. This paper addresses the gap in understanding how this blocking correlation specifically affects the signal-to-interference-plus-noise ratio (SINR) coverage and overall performance of mmWave networks, arguing that ignoring it leads to inaccurate system analysis.
Methods
The authors employ a combination of analytical modeling and simulation. They start with an exact analysis of a 1D network (e.g., a street or highway) to gain insights into the impact of blocking correlation. This involves modeling blockages as a Poisson Boolean process and deriving expressions for SINR coverage probability that explicitly account for the correlated blocking events. Leveraging these insights, they then develop an approximate analytical framework for a more general 2D network. This 2D analysis considers first-order blocking correlation, characterizing the interference at the user by accounting for the correlation between the blocking of serving and interfering links. The paper also defines and uses specific path-loss models (Bounded Power Law Path-Loss and LOS Association Only Path-Loss) and a cell-association model based on maximum received power.
Datasets
Not discussed in the paper. The paper focuses on analytical modeling and simulation-based validation rather than training on specific datasets.
Key Insights and Ideas
- Correlation is Crucial: The core insight is that blocking events are not independent in mmWave networks, and this correlation significantly impacts performance metrics like SINR coverage.
- Size Matters: The impact of blocking correlation is particularly pronounced when blockages are large, as a single large object can simultaneously affect multiple links.
- 1D vs. 2D Complexity: Analyzing 1D networks allows for exact solutions, providing a foundation for understanding the correlation's effects. However, extending this to 2D networks becomes analytically intractable, necessitating approximations.
- Independent Blocking is Misleading: The common assumption of independent blocking can lead to overestimation of coverage probability, especially in scenarios with significant blockage correlation.
- Blockages Aren't Always Bad: In some cases, blockages can actually improve SINR coverage by reducing interference from Line-of-Sight (LoS) interferers, though this effect is also influenced by correlation.
Key Results
- The paper derives exact SINR coverage probability expressions for a 1D mmWave network that explicitly include blocking correlation.
- An approximate analytical framework for 2D networks is developed, providing compact expressions for SINR coverage probability, including simplified bounds.
- Numerical results (Figures 1 and 2) demonstrate that considering blocking correlation leads to significantly different coverage probabilities compared to the independent blocking assumption, especially for low to moderate SINR thresholds and varying blockage densities.
- Figure 3 shows that the derived analytical model for the 2D serving BS distance distribution closely matches simulation results, even with a first-order correlation assumption, and that the independent blocking assumption overestimates coverage.
- Figure 4 illustrates that the LoS association probability decreases with increasing blockage correlation, and the independent blocking assumption is only accurate for very small blockages.
- Figure 5 shows that blocking correlation improves the accuracy of SINR coverage predictions in 2D networks.
Limitations and Future Work
- 2D Exact Analysis: The paper acknowledges that an exact analysis for 2D networks is intractable due to the complexity of conditioning on blockage locations in all directions. The 2D analysis presented is approximate.
- First-Order Correlation: The 2D analysis primarily considers first-order correlation between the serving link and interfering links, ignoring correlations among interfering links themselves. While shown to be reasonably accurate, higher-order correlations could be explored.
- Future Work: The authors suggest several extensions:
- Studying other association rules and channel models.
- Utilizing the framework to model correlation in object detection for sensing and radar applications.
- Extending the analysis to 3D deployments of BSs, which are becoming more common in dense urban environments.
- Investigating other path-loss models and blockage shapes.
Summary updated