A Neural Network-Based Cooperative Spectrum Sensing Scheme for Cognitive Radio Systems

* Final gross prices may vary according to local VAT.

Get Access

Abstract

In the single spectrum sensing, it is difficult to overcome obstacles, path loss, deep noise and fading in the network. Cooperative spectrum sensing (CSS) has been proposed as one of solutions to overcome the drawback of the single spectrum sensing. In the CSS, a fusion center collects the sensing information from all secondary users and makes a final decision. Even though CSS can provide better sensing performance than the single spectrum sensing, the problems above mentioned still remain. Thus effective decision method is needed for more adaptive to communication environment. In the paper, we propose a cooperative spectrum sensing utilizing neural network for cognitive radio systems. In the proposed scheme, weight factors of the neural network are trained by using historical sensing information stored on buffer. After that, a final decision is made by using current sensing information as input of neural network. Through the simulation result, we can find the proposed scheme has better performance than other comparison schemes such as AND, OR and half voting rules.