Asymptotic Performance Analysis of Generalized User Selection for Interference-Limited Multiuser Secondary Networks

Yazan H. Al-Badarneh, Costas N. Georghiades, Mohamed-Slim Alouini

TL;DR

This paper tackles the challenge of optimizing performance in cognitive radio networks where secondary users (SUs) share spectrum with primary users (PUs). By employing extreme value theory, the authors derive accurate analytical expressions for key performance metrics like throughput and error rates for a generalized user selection strategy, specifically selecting the k-th best user, which offers more flexibility than simply picking the absolute best.

What Problem Does This Paper Address?

Cognitive radio (CR) aims to improve spectrum utilization by allowing secondary users (SUs) to opportunistically access licensed spectrum. A crucial constraint is that SUs must not cause excessive interference to primary users (PUs). This paper focuses on "underlay" CR systems, where SUs transmit simultaneously with PUs, and the system is "interference-limited," meaning interference from the PU is the dominant factor, not noise at the SU receiver.

A common technique to boost performance in multi-user systems is "multiuser diversity," where the system selects users with the best channel conditions. However, simply selecting the absolute best user (k=1) might not always be optimal due to factors like user availability or traffic conditions. This paper addresses this by analyzing a generalized user selection scheme: selecting the k-th best user based on their Signal-to-Interference Ratio (SIR). The core challenge is to derive accurate performance metrics for this k-th best user selection in an interference-limited CR system, especially when the secondary transmitter (ST) dynamically adjusts its power to meet interference constraints.

Methods

The paper's primary methodological contribution is the application of Extreme Value Theory (EVT). Specifically, the authors leverage EVT to analyze the behavior of the k-th largest order statistic of the Signal-to-Interference Ratio (SIR) among a large number of secondary users. They demonstrate that this k-th highest SIR converges in distribution to an inverse gamma random variable as the number of secondary users approaches infinity.

This theoretical result is then used to derive asymptotic expressions for several key performance metrics:

  • Average Throughput: The expected data rate achievable.
  • Effective Throughput: Throughput considering a statistical Quality of Service (QoS) constraint (delay exponent).
  • Average Bit Error Rate (BER): The probability of incorrect symbol detection.
  • Outage Probability: The probability that the system performance falls below a certain threshold.

The analysis considers a continuous power adaptation policy at the secondary transmitter (ST). This policy dynamically adjusts the ST's transmit power to ensure the interference to the primary receiver (PR) stays below a predefined threshold, while also considering a maximum transmit power limit. The paper also briefly touches upon the impact of imperfect channel state information (CSI).

Datasets

Not discussed in the paper. The evaluation relies on numerical simulations to validate the derived analytical expressions.

Key Insights and Ideas

  • EVT is powerful for analyzing complex systems: The paper showcases how EVT can simplify the analysis of complex wireless systems with many users, providing tractable and accurate asymptotic results.
  • Generalized user selection offers flexibility: Selecting the k-th best user (not just the absolute best) is shown to be a practical and analytically tractable approach for improving performance in CR networks.
  • Interference-limited systems behave differently: The analysis specifically accounts for the dominant role of interference from primary users, leading to SIR-based analysis rather than the more common SNR-based approaches.
  • Asymptotic analysis is accurate for realistic scenarios: The derived expressions, based on a large number of secondary users, are shown to be accurate even for moderately sized user populations, making them valuable for practical system design.
  • Power adaptation is crucial: The continuous power adaptation strategy plays a vital role in satisfying interference constraints while maximizing performance.

Key Results

The paper derives novel closed-form asymptotic expressions for the average throughput, effective throughput, average BER, and outage probability for the k-th best SU in an interference-limited underlay CR network with continuous power adaptation.

  • Accuracy of Asymptotic Expressions: Numerical simulations validate the derived asymptotic expressions, showing good agreement with Monte Carlo simulations for various system parameters (number of users N, interference level T, transmit power Ps, and user selection parameter k). The accuracy generally improves with increasing N.
  • Impact of Parameters:
    • Number of Users (N): Increasing N generally improves throughput and reduces BER/outage probability, as expected from multiuser diversity.
  • Interference Level (T): Higher interference levels (larger T) tend to decrease throughput and increase outage probability, especially for limited ST power.
  • ST Transmit Power (Ps): Higher ST transmit power generally leads to better throughput, but the gains saturate, particularly when interference is high. The unlimited power case provides an upper bound.
  • Selection Parameter (k): Increasing k (selecting a user with a lower SIR rank) generally leads to lower throughput and higher BER/outage probability.

Limitations and Future Work

  • Perfect CSI Assumption: While the paper briefly discusses imperfect CSI, the primary analysis assumes perfect CSI for the ST-to-PR channel. The impact of imperfect CSI on the secondary network channels is also considered.
  • Block Fading: The analysis assumes block fading, where channel conditions remain constant within a block but change between blocks.
  • Specific Channel Models: The analysis uses specific fading models (Rayleigh for PT-PR, Gamma/Nakagami-m for SU channels).
  • Future Work: The paper suggests exploring the impact of imperfect CSI more deeply and potentially extending the analysis to other system configurations or more complex fading environments. The authors also note that their work provides a foundation for analyzing more advanced scheduling and power control schemes in CR networks.

Summary updated

Abstract

We analyze the asymptotic performance of a generalized multiuser diversity scheme for an interference-limited secondary multiuser network of underlay cognitive radio systems. Assuming a large number of secondary users (SUs) and that the noise at each SU's receiver is negligible compared to the interference from the primary transmitter, the secondary transmitter (ST) transmits information to the kth best SU, namely, the one with the kth highest signal-to-interference ratio (SIR). We use extreme value theory to show that the kth highest SIR converges uniformly in distribution to an inverse gamma random variable for a fixed k and large number of SUs. We use this result to derive asymptotic expressions for the average throughput, effective throughput, average bit error rate and outage probability of the kth best SU under continuous power adaptation at the ST, which ensures satisfaction of the instantaneous interference constraint at the primary receiver caused by the ST. Numerical simulations show that our derived asymptotic expressions are accurate for different values of system parameters.
BibTeX
@article{albadarneh2019asymptotic,
  author = {Yazan H. Al-Badarneh and Costas N. Georghiades and Mohamed-Slim Alouini},
  title = {{Asymptotic Performance Analysis of Generalized User Selection for Interference-Limited Multiuser Secondary Networks}},
  journal = {IEEE Trans. Cogn. Commun. Netw.},
  year = {2019},
  url = {https://openreview.net/forum?id=VeDv2d9ldiN},
}
Figure