Threshold-Learning in Local Spectrum Sensing of Cognitive Radio

Spectrum sensing is important for cognitive radios to utilize the idle spectrum opportunities, and recently cooperation schemes have been introduced to enhance spectrum sensing in specific areas. However, when a mobile cognitive node roams among heterogenous wireless network, it will be difficult to catch the changes of primary user´s behavior, or to setup the cooperation relationship with local network nodes in a short time. In this paper, an self-learning spectrum sensing framework is proposed, which can enable the single mobile cognitive node to work in unknown wireless environment. When the wireless environment changes, the main sensing parameters (such as decision threshold, sampling frequency) could be adapted to optimum in the self- earning process. One adaptive algorithm is proposed to find the optimal decision threshold in energy detection sensing method. Simulation results show that, the proposed scheme could converge to optimal sensing parameters in spatial and temporal varying environment.