Abstract
In underlay cognitive radio (CR) systems, secondary users (SUs) share spectrum with primary users (PUs) under strict interference constraints. Opportunistic scheduling is often employed to maximize system throughput by selecting the SU with the best channel conditions. However, the fairness of such scheduling—defined by the selection probability of each user—has remained largely unexplored analytically. This paper presents the first closed-form mathematical expressions for user selection probability in these systems, validated through extensive simulations. Our findings reveal that a user's chance of being selected is governed by the ratio of their distance to the cognitive base station versus their distance to the primary receiver.
Introduction
Cognitive Radio (CR) technology addresses spectrum scarcity by allowing unlicensed secondary users to access licensed bands. In the underlay paradigm, SUs transmit concurrently with PUs, provided the resulting interference at the primary receiver (PU-RX) stays below a predefined threshold. To optimize performance, opportunistic scheduling selects the "best" SU based on a metric that considers both the desired signal strength and the potential interference caused. While metrics like outage probability and throughput are well-documented, the "selection probability"—a critical measure of fairness—has lacked a rigorous analytical framework until now.
System Model and Methodology
The study considers a network with $K$ secondary transmitters, a cognitive base station (eNodeB), and a primary user pair. The channel between any two nodes is modeled using Rayleigh fading, representing typical non-line-of-sight urban environments.
The scheduling algorithm utilizes an optimal selection metric, $\beta^*$, defined as the ratio of the channel gain to the destination ($G_{SD}$) over the channel gain to the primary receiver ($G_{SP}$): $$\beta^* = \frac{G_{SD}}{G_{SP}}$$ The user with the highest $\beta^*$ is selected for transmission. The authors derive the Cumulative Distribution Function (CDF) and Probability Density Function (PDF) for this metric to obtain closed-form expressions for the selection probability of each user for $K=2$ and $K=3$. These analytical results are then validated against Monte Carlo simulations.
Impact of Geographical Distribution
The analytical results demonstrate that fairness is highly sensitive to the relative distances of SUs to the eNodeB and the PU-RX. The "Ratio Rule" derived by the authors suggests that fairness depends on the fraction of a user's distance to the cognitive receiver versus their distance to the primary receiver.
Location Advantage
When an SU is positioned significantly closer to the eNodeB than its peers, its selection probability increases dramatically. For instance, if one user has a geographical advantage, its selection probability can rise to approximately 69%, leaving other users with significantly reduced access to the channel. This occurs because proximity to the base station typically yields a higher $G_{SD}$, making it more likely to satisfy the $\beta^*$ criteria.
Location Penalty
Conversely, proximity to the primary receiver acts as a significant bottleneck. An SU near the PU-RX must maintain very low transmit power to avoid interference, resulting in a lower effective selection metric. As shown in the figure below, a user's selection probability can drop to as low as 8.7% if they are unfortunately close to the primary receiver, even if their distance to the base station is identical to their competitors.
Conclusion
This research provides a theoretical foundation for understanding fairness in underlay CR systems. The derived closed-form solutions allow network engineers to predict and mitigate unfairness in opportunistic scheduling without relying on computationally expensive simulations. The perfect match between analytical curves and simulated data confirms the accuracy of these new formulas. Future work will extend these models to more complex fading environments, such as Nakagami-$m$ or Rician fading, to account for a wider variety of real-world communication scenarios.
Summary updated