Cyclostationary spectrum sensing based fisher analyzer under stochastic geometric network model

Abstract

Although spectrum sensing has been extensively researched, most of the existing works either assume that there is only one primary user (PU) or do not consider the topology of PU network at all. In this paper, a novel stochastic geometric network model is proposed to model the practical circumstances. In this model, the PU network is modeled as a random geometric network that following a Poisson point process (PPP) which can better describe small-scale mobile PUs. Different from the classical Shannon theorem that only considers noise, the aggregated interference caused by PUs located outside the sensing range is taken into account in this paper. The interference known as spatial alarm (SFA) can decrease the second user's (SU's) medium access probability. In this paper, cyclostationary spectrum sensing based partial QR decomposition (CSS-PQR) is adopted in each SU, and to obtain more effective and more accurate detection performance, a location-aware cooperative sensing algorithm that linearly combine multiple sensing results is used. Particularly the Fisher Analyzer (FA) is utilized to determine the linear coefficients. It can be proved that the proposed model is more accordant with practical circumstances, and the simulation results show that the proposed algorithm performs much better than the traditional ones, such as MAJ-OC based, ML-OC based CSS algorithms, in terms of false-alarm probabilities and detection probabilities.