Introduction
In modern wireless communication systems, diversity combining techniques are essential for mitigating the effects of fading. Among these, Maximum Ratio Combining (MRC) and Equal Gain Combining (EGC) are the most prominent. While MRC is theoretically optimal, EGC is frequently preferred in practical hardware implementations because it does not require the estimation of individual channel gains, significantly reducing complexity and cost. However, the performance analysis of EGC receivers—specifically calculating the Outage Probability (OP)—is notoriously difficult.
The Challenge of Rare Event Estimation
The Outage Probability of an $L$-branch EGC receiver is defined by the Cumulative Distribution Function (CDF) of the sum of $L$ channel envelopes. Unlike MRC, which involves the sum of squared envelopes and often yields closed-form solutions, the sum of envelopes in EGC lacks a general closed-form expression for most fading models, such as Rayleigh or Rice.
As the industry moves toward 5G and 6G standards, there is an increasing demand for "ultra-reliable" communications with target outage probabilities as low as $10^{-9}$. Standard Monte Carlo (MC) simulations are computationally prohibitive for such rare events; estimating a $10^{-9}$ probability with reasonable accuracy would require trillions of samples.
Proposed Importance Sampling Framework
This paper introduces a novel Importance Sampling (IS) estimator based on a sample rejection scheme. The core innovation lies in a geometric observation: the region of space where an EGC outage occurs ($\sum R_i \leq \gamma_0$) is a subset of the region where an MRC outage occurs ($\sum R_i^2 \leq \gamma_0^2$).
By leveraging the fact that closed-form solutions for MRC outage probabilities are available for many fading distributions, the authors use the truncated MRC distribution as a "proposal distribution." The algorithm samples from this "sphere" and applies a rejection step to identify samples that also fall within the EGC "simplex."
Key Features:
- Bounded Relative Error: The authors mathematically prove that the estimator's relative error remains bounded even as the outage probability approaches zero. This ensures that the number of required samples does not explode for rare events.
- Versatility: The method is successfully applied to independent Rayleigh, correlated Rayleigh, and i.i.d. Rice fading models, as well as Generalized Selection Combining (GSC) scenarios.
- Ease of Implementation: Unlike previous methods like Hazard Rate Twisting (HRT), which can be complex to implement for correlated channels, the sample rejection approach remains straightforward.
Performance and Results
Numerical experiments demonstrate a massive gain in efficiency. To achieve a 5% relative error for an outage probability of $10^{-9}$, the proposed IS estimator requires only $1.5 \times 10^5$ samples, compared to the $1.5 \times 10^{12}$ samples required by naive Monte Carlo—a seven-order-of-magnitude speedup.
Furthermore, the estimator maintains a constant number of samples across varying probability levels, validating the theoretical "bounded relative error" property. While HRT might offer slightly higher efficiency in specific independent cases, the proposed method's robustness in correlated and Rice fading scenarios makes it a superior general-purpose tool.
Conclusion
The proposed sample rejection estimator provides a highly efficient and mathematically rigorous solution for evaluating EGC performance in ultra-reliable wireless networks. By bridging the gap between the analytically tractable MRC and the practically desirable EGC, this work enables engineers to accurately simulate and design next-generation communication links. Future work may extend this framework to other fading distributions like Nakagami-$m$, $\kappa-\mu$, and $\eta-\mu$, where MRC solutions are also known.
Reference: N. Ben Rached, A. Kammoun, M.-S. Alouini, and R. Tempone, "An Accurate Sample Rejection Estimator of the Outage Probability With Equal Gain Combining," IEEE Open Journal of the Communications Society, vol. 1, 2020.
Summary updated