On the Capacity of a Class of Cognitive Z-interference Channels

Jinhua Jiang, Ivana Maric, Andrea J. Goldsmith, Shlomo Shamai, Shuguang Cui

Abstract

This paper investigates the capacity of the Cognitive Z-interference Channel (C-ZIC), where the cognitive receiver is immune to primary user interference. By simplifying general coding schemes and deriving new outer bounds, the authors establish the exact capacity for a specific high-interference regime. Numerical results highlight the performance of different coding strategies, showing that superposition coding can outperform Dirty Paper Coding in high-interference scenarios.

Introduction

In cognitive radio, a secondary (cognitive) user shares spectrum with a primary user. A key assumption in theoretical models is that the cognitive transmitter possesses non-causal knowledge of the primary user's message. While general capacity formulas exist, they are often computationally intractable due to the use of complex auxiliary random variables. This paper addresses the Cognitive Z-interference Channel, a sub-class where the primary transmitter's signal does not reach the cognitive receiver due to physical distance or shadowing.

The Cognitive Z-interference Channel model, illustrating the Z-topology where the cognitive receiver $Y_1$ is shielded from primary user interference.

Methodology

The authors utilize classical information-theoretic tools to bound performance and simplify the rate regions:

  • Rate Splitting and Binning: Adapting Marton’s region for the specific Z-topology to remove redundant components.
  • Gel’fand-Pinsker Coding (Dirty Paper Coding): Used by the cognitive user to "pre-cancel" the interference it causes at the primary receiver.
  • Superposition Coding: A layered messaging technique that proves surprisingly effective in high-interference scenarios.
  • Outer Bounding: Deriving a tighter upper limit by considering the system as a broadcast channel with degraded message sets.

Theoretical models were validated using synthetic Gaussian channel simulations, varying power constraints ($P_1, P_2$) and the interference link gain ($b$).

Key Insights and Results

The study provides simplified, computable rate regions ($G_1, G_2, G_3$). A major result is the identification of a "Capacity Breakthrough": for interference gain $b$ within the range $1 \le b \le \sqrt{\frac{P_1+P_2+1}{P_1+1}}$, the achievable rate region matches the outer bound exactly, defining the true capacity for this regime.

Comparison of achievable rate regions ($G_1, G_2, G_3$) and the outer bound for different interference gains $b$.

Numerical simulations (Figure 3) show that while DPC-based schemes ($G_3$) are effective for moderate interference, superposition coding ($G_2$) becomes superior as the interference link gain $b$ increases. This challenges the common assumption that DPC is always the optimal strategy for cognitive interference management.

Conclusion

The authors successfully characterize the capacity of the Gaussian C-ZIC in a specific high-interference range and provide a tighter outer bound ($C^o_2$) than previously established. While a gap remains for very high interference levels, the work identifies critical thresholds where simpler superposition coding becomes the dominant strategy, providing valuable guidance for designing efficient cognitive radio systems.

Summary updated

BibTeX
@inproceedings{jiang2011capacity,
  author = {Jinhua Jiang and Ivana Maric and Andrea J. Goldsmith and Shlomo Shamai and Shuguang Cui},
  title = {{On the Capacity of a Class of Cognitive Z-interference Channels}},
  booktitle = {ICC},
  year = {2011},
  doi = {10.1109/icc.2011.5963016},
}
Figure