On the Capacity of the Two-User Gaussian Causal Cognitive Interference Channel

Martina Cardone, Daniela Tuninetti, Raymond Knopp, Umer SalimIEEE Trans. Information Theory 2014 · OpenReview

TL;DR

This paper analyzes the capacity of a "causal cognitive interference channel" (CCIC), a more realistic model for cognitive radio networks where a secondary user (the "cognitive" user) can only learn about the primary user's transmissions through a noisy, causal link. The authors characterize the achievable communication rates (capacity) for various configurations of this channel, providing bounds that are very close to the theoretical maximums. This work is crucial for understanding the potential benefits of cognitive radio in practical scenarios.

What Problem Does This Paper Address?

In cognitive radio, a secondary user aims to opportunistically use licensed radio spectrum without causing harmful interference to the primary user. A common theoretical model assumes the secondary user has perfect, non-causal knowledge of the primary user's messages. However, in reality, this knowledge is imperfect and obtained through a noisy channel, and the secondary user can only access information that has already been transmitted (causal knowledge). This paper tackles the problem of determining the communication capacity of such a "causal cognitive interference channel" (CCIC), which is a more practical and challenging scenario. The authors aim to quantify the achievable data rates and understand how this causal, noisy information affects the overall system performance compared to simpler models.

Methods

The paper employs established techniques from information theory, specifically focusing on characterizing the "capacity region" of the communication channel. This involves finding both achievable communication rates (inner bounds) and theoretical upper limits on these rates (outer bounds). The core contribution lies in deriving these bounds for the CCIC, particularly for symmetric, Z-channel (no primary-to-cognitive interference), and S-channel (no cognitive-to-primary interference) configurations. A key metric used is the "generalized degrees of freedom" (gDoF), which provides insight into the high-SNR performance. The authors leverage existing outer bounds for bilateral cooperation and adapt them to the unilateral cooperation scenario of the CCIC. They also develop novel achievable schemes, often based on superposition coding and dirty paper coding (DPC), to demonstrate that the derived outer bounds are nearly tight, meaning the capacity is known "to within a constant gap."

Datasets

Not discussed in the paper.

Key Insights and Ideas

  • Realism Matters: The paper highlights that the "causal" nature of information acquisition in cognitive radio significantly impacts achievable rates, making it a more complex but realistic problem than models assuming perfect non-causal knowledge.
  • Cooperation Gains: Unilateral cooperation (where only one user cooperates) can indeed provide substantial gains in communication rates, especially in certain interference and cooperation strength regimes.
  • Channel Structure is Key: The specific structure of the interference channel (e.g., Z-channel vs. S-channel vs. fully connected) dramatically influences the achievable rates and the effectiveness of cooperation.
  • DPC for Weak Interference: Dirty Paper Coding (DPC) emerges as a crucial technique for achieving near-optimal performance in weak interference scenarios, allowing the cognitive user to effectively "cancel" known interference.
  • gDoF as a Performance Metric: The generalized degrees of freedom (gDoF) provides a powerful tool for comparing different channel models and understanding the fundamental performance limits under varying signal-to-noise ratios and interference levels.

Key Results

The paper provides characterizations of the capacity region for various CCIC configurations:

  • Symmetric CCIC: The capacity region is characterized to within 1 bit for most parameter regimes, and the sum-capacity to within 3.16 bits in some cases.
  • Z-channel and S-channel: The capacity regions are determined to within 2 bits.
  • General CCIC: The capacity region is characterized to within 2 bits for cases where interference is not weak at both receivers.
  • gDoF Analysis: The paper identifies specific parameter regimes where causal unilateral cooperation offers no gDoF gain over the noncooperative case, and other regimes where it matches the performance of non-causal cognitive radio or bilateral cooperation.

Limitations and Future Work

  • Constant Gap: While the paper characterizes the capacity to within a constant gap (e.g., 1 or 2 bits), it does not achieve exact capacity in all cases. Closing this gap further would require more advanced techniques.
  • Open Problems: The paper identifies the development of outer bounds for specific rate combinations (e.g., $2R_c + R_p$) in the weak interference regime as an important open problem for general ICs with source cooperation.
  • Practical Implementation: While the theoretical framework is established, the practical implementation of complex coding schemes like DPC in real-world cognitive radio systems remains a challenge.
  • Extensions: The authors suggest future work could explore extensions to more users, different channel models, or scenarios with feedback from destinations to sources.

Summary updated

Abstract

This paper considers the two-user Gaussian causal cognitive interference channel (GCCIC), which consists of two source-destination pairs that share the same channel and where one full-duplex cognitive source can causally learn the message of the primary source through a noisy link. The GCCIC is an interference channel with unilateral source cooperation that better models practical cognitive radio networks than the commonly used model which assumes that one source has perfect noncausal knowledge of the other source's message. First, the sum-capacity of the symmetric GCCIC is determined to within a constant gap. Then, the insights gained from the study of the symmetric GCCIC are extended to more general cases. In particular, the whole capacity region of the Gaussian Z-channel, i.e., when there is no interference from the primary user, and of the Gaussian S-channel, i.e., when there is no interference from the secondary user, are both characterized to within 2 bits. The fully connected general, i.e., no-symmetric, GCCIC is also considered and its capacity region is characterized to within 2 bits when, roughly speaking, the interference is not weak at both receivers. The parameter regimes where the GCCIC is equivalent, in terms of generalized degrees-of-freedom, to the noncooperative interference channel (i.e., unilateral causal cooperation is not useful), to the non-causal cognitive interference channel (i.e., causal cooperation attains the ultimate limit of cognitive radio technology), and to bilateral source cooperation are identified. These comparisons shed light into the parameter regimes and network topologies that in practice might provide an unbounded throughput gain compared to currently available (non cognitive) technologies.
BibTeX
@article{cardone2014capacity,
  author = {Martina Cardone and Daniela Tuninetti and Raymond Knopp and Umer Salim},
  title = {{On the Capacity of the Two-User Gaussian Causal Cognitive Interference Channel}},
  journal = {IEEE Trans. Information Theory},
  year = {2014},
  url = {https://openreview.net/forum?id=lBwmgyx1Hcw},
}
Figure