Optimal Discrete Power Control in Poisson-Clustered Ad Hoc Networks

Beiyu Rong, Chun-Hung Liu, Shuguang Cui

Abstract

This paper investigates the optimization of $N$-layer discrete power control (DPC) schemes within wireless ad hoc networks modeled as Poisson Cluster Processes (PCP). While continuous power control is theoretically ideal, hardware constraints often necessitate discrete power levels. The authors demonstrate that a properly designed DPC scheme, utilizing a specific power scaling law derived through stochastic geometry, can significantly outperform constant power control. By balancing the interference "budget" across the network, the proposed scheme achieves substantial gains in transmission capacity and spatial reuse, proving that discrete power levels can be leveraged as a strategic advantage rather than a limitation.

Introduction

In large-scale wireless ad hoc networks, interference management is critical for performance. Traditional models often assume continuous power tuning, yet real-world digital handsets operate at discrete levels. Furthermore, standard Poisson Point Process (PPP) models fail to capture the clustered nature of real-world node deployments, such as users in hotspots. This paper addresses these gaps by developing an $N$-layer DPC scheme tailored for Poisson Cluster Processes, ensuring that the stepped nature of power levels enhances network efficiency.

Methodology

The authors utilize a Poisson Cluster Process where transmitters and receivers are distributed in circular clusters. The core of the proposed $N$-layer DPC scheme involves three main components:

  1. Tessellation and Selection: Each cluster is divided into $N$ concentric annuli. A transmitter selects one of $N$ discrete power levels ($P_i$) based on the ring occupied by its intended receiver.
  2. Stochastic Geometry Analysis: Using the "Geometric Conservation Property," the authors analyze the interaction of multiple concurrent power levels, treating them as thinned versions of the original process.
  3. Optimization and Scaling: A key insight is the "Interference Balancing" law: $P_i \in \Theta(\eta_i^{-\alpha/2})$, where $\eta_i$ is the selection probability and $\alpha$ is the path loss exponent. This law dictates that high transmit power should be used only when the probability of selection is low, effectively balancing the interference footprint.
Available region for power ratios and selection probabilities in a 2-layer DPC scheme.

The optimization process involves identifying the feasible regions for these parameters to ensure network stability and performance, as illustrated in the figure above for a simplified case.

Results and Discussion

The $N$-layer DPC scheme was evaluated using Monte Carlo simulations in Rayleigh fading environments. Key findings include:

  • Transmission Capacity (TC): The optimized DPC achieved nearly 3x the TC of networks with no power control, outperforming channel inversion and fractional power control.
  • Spatial Reuse: The outage-free spatial reuse factor was strictly higher, allowing more simultaneous transmissions per unit area.
  • Granularity: While increasing $N$ improves performance, simulations suggest that $N=15$ to $20$ layers are sufficient to capture the majority of available gains.
  • Fairness: By balancing outage probabilities across layers, the scheme ensures that remote receivers are not disproportionately affected by interference.

Conclusion

Discrete power control, when optimized for clustered geometries, provides a significant "location-dependent" advantage for spatial reuse. While the current model assumes uniform receiver distribution and focuses on interference-limited scenarios, it establishes a robust framework for future extensions. Potential research directions include non-circular clusters, mobile environments, and joint power-rate control optimization.

Summary updated

BibTeX
@article{rong2015optimal,
  author = {Beiyu Rong and Chun-Hung Liu and Shuguang Cui},
  title = {{Optimal Discrete Power Control in Poisson-Clustered Ad Hoc Networks}},
  journal = {IEEE Trans. Wireless Communications},
  year = {2015},
  doi = {10.1109/twc.2014.2334330},
}
Figure