Intelligent hybrid cooperative spectrum sensing: A multi-stage decision fusion approach

摘要

Cooperative spectrum sensing is a powerful sensing approach which is based on sharing information about channel activities among secondary users (SUs). Cooperative spectrum sensing aims to overcome hidden node problem, shadowing and fading problems, it also enhances sensing accuracy. However, sensing accuracy may degrade due to various reasons: if environmental properties are poor or intra-node characteristics are continuously altering. Thus, this paper proposes a novel approach of a multi-stage hybrid cooperative spectrum sensing model. The first stage, integrates a fuzzy logic system for local fusion center, whereby SU-mobility and its environmental properties, and its neighbors' environmental properties are included in local sensing decision process. Second stage proposes a neural network, based on backpropagation learning algorithm, for global fusion center. All SUs transmit their sensing information, local decision, and their intra-node characteristics to be augmented for an optimized global sensing decision. Neural network is trained based on a real-world measured power dataset. Extensive simulations on the proposed multi-stage model were performed. The results showed high robustness against instantaneous changes in SU-mobility levels and good detection performance at very low signal-to-noise ratio (SNR) levels. The proposed prediction model outperformed the state-of-the-art work with high detection accuracy at very poor environmental conditions and at different speeds levels.