ELASTIC- Enabling Massive-Antenna for Joint Spectrum Sensing and Sharing: How Many Antennas Do We Need?

Abstract

Massive Antenna and Cognitive Radio (CR) technologies have attracted many research interests due to the additional resources offered in striving against the spectrum crisis. In this paper, we propose a general framework to EnlabLe mASsive anTenna exploItation for spectrum sensing and sharing in CR (ELASTIC). Using random matrix theory and moment matching method, we derived a simple approximation of the distributions of three eigen-value based detectors namely the Largest Eigenvalue (LE), the Scaled LE (SLE), and Standard Condition Number (SCN). This has led to a simple analytical formulation and optimization of the number of antennas and the number of samples to reach a target performance. To exploit the large number of antennas, we proposed two sensing scenarios. The first is based on full antenna exploitation and guarantees an optimal performance despite the transmission conditions. The second is based on partial antenna exploitation, which determines the exact number of antennas and samples required to reach the target performance. We have shown that the framework offers an additional degree of freedom in the selection of the optimal system parameters, namely the number of antennas, while the remaining antennas are exploited for sharing in other dimensions of the spectrum hypercube.

Enhanced Detection Algorithms Based on Eigenvalues and Energy in Random Matrix Theory Paradigm

“…The commonly used data fusion method is the weighted sum of data, in a similar vein, we firstly consider the weighted average of the MED statistic in (8) and ED statistic in (12), which is expressed as bellow…”

“…These algorithms mostly consider the statistical distribution of the eigenvalues by exploiting the recent results from random matrix theory (RMT). For simplicity, they are collectively called eigenvalue-based detectors (EBD), which rely on the utilization of RMT and different eigenvalue properties of the sample covariance matrix in decision-making process [8]. These EBD techniques can be categorized into the maximum eigenvalue detector (MED) [9]- [14], the maximum eigenvalue to trace (MET) (also called as scaled largest eigenvalue detector) [15]- [20], and the maximum-minimum eigenvalue detector (MME) (also called standard condition number detector ) [12], [14], [21], [22].…”