The Capacity of Known Interference Channel

Shengli Zhang, Soung Chang Liew, Jinyuan ChenIEEE JSAC 2015 · DOI

TL;DR

This paper investigates the capacity of a wireless communication channel where the receiver knows the interfering signal's data but not its channel gain. The authors propose a novel Blind Known Interference Cancellation (BKIC) scheme that achieves near-optimal performance, significantly outperforming traditional methods, especially in high signal-to-noise ratio (SNR) scenarios.

What Problem Does This Paper Address?

In many real-world wireless communication systems, a receiver often needs to decode a desired signal while simultaneously dealing with interference from other transmitters. A common scenario is when the receiver has access to the data of the interfering signal but not the exact channel gain through which this interference arrives. This is termed the "known interference channel." Traditional methods for dealing with this often rely on estimating the interference channel, which can be complex and prone to errors. This paper aims to characterize the theoretical capacity of such a channel and propose an efficient practical scheme to approach it, particularly focusing on scenarios where interference is significant.

Methods

The core of the paper lies in two main contributions:

  • Deriving a Tight Upper Bound for Capacity: The authors first establish a theoretical upper limit on the achievable data rate for the known interference channel. This bound is derived by considering the worst-case interference distribution and is shown to be influenced by the block length of the transmitted data.
  • Proposing a Blind Known Interference Cancellation (BKIC) Scheme: The paper introduces a novel BKIC scheme. Unlike traditional methods, BKIC does not require explicit estimation of the interference channel gain. Instead, it cleverly reformulates the problem into a standard MIMO (Multiple-Input Multiple-Output) framework, allowing the use of well-established MIMO signal processing techniques. This reformulation enables the receiver to effectively cancel the interference and recover the desired signal. The paper then derives a closed-form achievable rate for this BKIC scheme.

Datasets

Not discussed in the paper.

Key Insights and Ideas

  • Blind Cancellation is Powerful: The most significant insight is that effective interference cancellation can be achieved without knowing the exact interference channel gain. This "blind" approach simplifies the receiver's task and can lead to superior performance.
  • MIMO Reformulation is Key: Transforming the known interference channel problem into a standard MIMO problem is a crucial step. This allows leveraging decades of research and established algorithms in MIMO signal processing for interference management.
  • High SNR Behavior is Distinct: The paper highlights a critical difference between BKIC and traditional methods at high SNRs. BKIC's achievable rate continues to increase with SNR (non-zero degree of freedom), while traditional methods plateau at a fixed bound (zero degree of freedom). This means BKIC scales much better in challenging, high-interference environments.
  • Worst-Case Interference is Counter-Intuitive: The paper notes that the worst-case interference distribution for capacity is not necessarily Gaussian, but rather any distribution with a constant total power for each block.

Key Results

  • BKIC Achieves Near-Optimal Rates: The proposed BKIC scheme's achievable rate is shown to closely approach the derived theoretical upper bound, especially in high SNR and high interference regimes.
  • BKIC Significantly Outperforms Traditional KIC: The achievable rate of BKIC is consistently higher than that of traditional interference cancellation schemes, particularly as SNR increases. The gap between BKIC and traditional methods grows with SNR.
  • Block Length Matters: The achievable rate and capacity are influenced by the block length (T). A longer block length generally leads to better performance, acting as a pre-log factor in the capacity.
  • High Interference Regime: In scenarios with very high interference power relative to signal power, BKIC's performance gap to the upper bound vanishes.

Limitations and Future Work

  • Block Fading Assumption: The current analysis primarily assumes block fading, where the channel coefficient is constant within a block but can change between blocks. Extending the analysis to continuous fading channels, where the channel changes more dynamically, is suggested as future work.
  • MIMO Extension: The paper notes that extending the BKIC framework to multiple-input multiple-output (MIMO) channels, which are prevalent in modern wireless systems, is of significant interest. This would involve tackling the added complexity of multiple interference channels in a MIMO setting.
  • Joint Decoding and Recovery: The paper mentions that with channel coding, the problem might become underdetermined. Designing joint channel decoding and signal recovery algorithms is proposed as a future research direction.
  • Suboptimal Schemes: While BKIC offers near-optimal performance, the paper also discusses suboptimal recovery schemes (like BKIC-ZF) that are simpler to implement and equivalent to traditional KIC. Further exploration of advanced BKIC algorithms and their trade-offs is also suggested.

Summary updated

BibTeX
@article{zhang2015capacity,
  author = {Shengli Zhang and Soung Chang Liew and Jinyuan Chen},
  title = {{The Capacity of Known Interference Channel}},
  journal = {IEEE J. Sel. Areas Commun.},
  year = {2015},
  doi = {10.1109/jsac.2015.2417015},
}
Figure