In this paper, we predict gain and excess noise factor for different avalanche photodetectors using multilayer perceptron neural network. We assume the most important factors such as bias voltage, length, type and doping concentration of absorption and multiplication regions, as the inputs of network. Also, gain and excess noise are outputs. After training, we apply the test patterns to network and compare outputs with experimental results. Finally, we consider the behavior of excess noise for different values in input parameters, such as length of absorption and multiplication regions.

Full text: IJEIR_1897_Final.pdf