Non-Asymptotic Analysis of Scaled Largest Eigenvalue Based Spectrum Sensing

In this paper, we analyze the non-asymptotic performance of scaled largest eigenvalue based detection, which is an optimal detector in the presence of a single primary user. Exact distributions of the test statistics have been derived, which lead to finite-dimensional characterizations of the false alarm probability. These results are obtained by taking advantage of the properties of the Mellin transform for products of independent random variables. Simulations are provided to verify the derived results, and to compare with the asymptotic result in literature.