An Artificial Neural Network Based Attenuation Tomography in Free Space Optical Network

The free space optical communication has attracted more attention nowadays. The attenuation coefficient is an important parameter to be monitored. However, the traditional internal measurement has many difficulties in FSO environment. So in this paper, network tomography, which is a typical external measurement technique is employed for attenuation coefficient monitoring in FSO. The artificial neural network is employed for estimating the link attenuation with the measured route level data. Besides, a heuristic algorithm is proposed for route set generation. An estimation MSE of 2.82*10 ⁻⁷ and mean error rate of 0.0027% are achieved through training of artificial neural networks.