A fuzzy neural approach for dynamic spectrum allocation in cognitive radio networks

In this paper, decision making scheme in cognitive radio is proposed by using fuzzy neural system, due to which secondary users can utilizes the spectrum effectively with seamless communication between cognitive radio and primary users. The spectrum sensing performance is enhanced by using either multiple antennas or multistage spectrum sensing. Due to multiple antennas at sensing node causes more equipment cost in spectrum sensing technique, therefore two stage spectrum sensing scheme is introduced. The proposed fuzzy neural decision making technique include two stage spectrum sensing schemes for identifying available spectrum. In first stage, three parameters such as spectrum utilization efficiency, degree of mobility and distance to the primary user of cognitive radio network are considered as inputs to fuzzy logic decision making process, while output of that process gives spectrum access decision, based on linguistic knowledge of 27 rules. Feedback neural network configuration included in second stage of spectrum sensing, which is trained with the help of generalized delta learning rule. Neural network has two input parameters such as output of fuzzy logic based spectrum sensing and desired values. The reference signal of neural network is obtain from transformation of output membership function in first stage to their mid singleton values. Simulation results shows significant improvement in sensing accuracy by exhibiting higher probability of detection.