小波神经网络和B-QPSO算法在Ad Hoc异常检测中的应用

Network Anomaly Detection Based on Wavelet Neural Network and B-QPSO Algorithm in Mobile Ad Hoc

引入一种基于边界变异的QPSO算法,即B-QPSO算法.首先利用B-QPSO算法训练小波神经网络,再将经过B-QPSO算法训练后的小波神经网络应用于Ad Hoc异常检测中.仿真实验结果表明,该算法与梯度下降法、PSO算法和QPSO算法相比,收敛速度快,具有更好的全局收敛性,提高了异常检测的准确性,并且降低了对正常情况的误判率.

QPSO algorithm based on bounded mutation operator, named B-QPSO algorithm, is presented. The wavelet neural network is trained by B-QPSO algorithm, and the wavelet neural network trained by B-QPSO algorithm, is used in Ad Hoc anomaly detection. The simulation experimental result shows that this algorithm has more rapid convergence and better global convergence ability, compared with gradient descent algorithm, PSO algorithm and QPOS algorithm. The accuracy of anomaly detection is enhanced and the false positive rate for normal state is reduced.