Abstract
The joint statistics of partial sums of ordered random variables (RVs) are essential for accurately characterizing the performance of diverse wireless communication systems. While previous research established a unified analytical framework for independent and identically distributed (i.i.d.) RVs, the assumption of identical distribution often fails in real-world scenarios. This paper addresses this gap by introducing a generalized MGF-based framework for independent but non-identically distributed (i.n.d.) RVs. We provide a mathematical formalism to derive joint statistics and illustrate its utility through the performance analysis of Generalized Selection Combining (GSC)-based RAKE receivers in frequency-selective fading channels with non-uniform power delay profiles.
Introduction
In wireless communication engineering, "order statistics"—the mathematical ranking of random variables from largest to smallest—is a fundamental tool. It is used to analyze systems that select or combine signal paths based on their strength, such as RAKE receivers or multi-user scheduling algorithms.
Historically, performance analysis has relied heavily on the simplifying assumption that signal paths are independent and identically distributed (i.i.d.). However, physical reality is rarely so uniform. In practical environments, signal paths often exhibit different average power levels due to varying distances and obstacles, necessitating a framework for independent but non-identically distributed (i.n.d.) variables. This report presents a unified framework to solve the combinatorial complexity inherent in i.n.d. order statistics.
The Challenge: Beyond the i.i.d. Assumption
When analyzing diversity techniques like Generalized Selection Combining (GSC), engineers must determine the probability density function (PDF) of the sum of the $m$ strongest signals out of $N$ available paths. The mathematical difficulty arises because, although the original signals may be independent, the process of sorting them introduces statistical dependence.
In the i.i.d. case, the symmetry of the variables simplifies the math. In the i.n.d. case, however, one must account for every possible permutation of which specific variable (each with its own unique distribution) ends up at which rank. As $N$ increases, this leads to a "combinatorial explosion" that makes traditional direct integration methods computationally prohibitive.
The Unified MGF-Based Framework
The authors propose a systematic two-step approach using the Moment Generating Function (MGF) to bypass the complexities of direct PDF derivation:
Step I: MGF Derivation via Power Sets
Instead of tackling the PDF directly, the framework derives the joint MGF. To manage the complexity of non-identical distributions, the authors introduce "Common Functions"—mathematical mixtures of Cumulative Distribution Functions (CDFs) and MGFs. They utilize power set notation to represent the various combinations of variables compactly, allowing for a structured representation of the joint statistics of any subset of ordered variables.
Step II: Inverse Laplace Transform
Once the joint MGF is expressed in a manageable form, an inverse Laplace transform is applied to obtain the joint PDF. The framework provides specific "simplifying relationships" (Integrals $J_m$, $J'_m$, and $J''_{a,b}$) that allow researchers to avoid the multi-fold integrations typically required in these problems.
Key Insights and Results
1. Flexibility and Generality
Unlike previous methodologies that focused solely on the sum of the $m$ largest variables, this framework is flexible. It can determine the joint statistics of any partial sum (e.g., the sum of the 2nd and 5th strongest paths) or the joint PDF of a specific ordered variable and a partial sum.
2. Closed-Form Solutions for Rayleigh Fading
For the exponential distribution—which models the common Rayleigh fading environment—the authors provide ready-to-use closed-form expressions. This eliminates the need for engineers to perform slow or unstable numerical simulations for standard fading scenarios.
3. Practical Application: GSC RAKE Receivers
The framework was successfully applied to calculate the "capture probability" of GSC RAKE receivers operating over frequency-selective channels with non-uniform power delay profiles. It also proved effective in analyzing finger replacement schemes during soft handovers, where non-identical fading is the standard condition.
Limitations and Future Directions
While the framework significantly simplifies the analysis of i.n.d. variables, it is not without limitations. As the number of paths $N$ grows very large, the number of terms in the power set summations increases, which may lead to higher computational costs. Additionally, for non-exponential distributions, the inverse Laplace transform may not always yield a closed-form solution, requiring numerical contour integration.
Future work may extend this framework to more complex scheduling in multi-user MIMO systems, where users possess highly heterogeneous channel statistics.
Reference: Nam, S. S., Yang, H.-C., Alouini, M.-S., & Kim, D. I. (2013). Technical Report: An MGF-based Unified Framework to Determine the Joint Statistics of Partial Sums of Ordered i.n.d. Random Variables. arXiv:1307.8199v2 [cs.IT].
Summary updated