Optical neural network with autonulling bridge cells using a nearest-neighbor interconnection scheme

A cellular neural network (CNN) using autonulling bridge networks as the unit cells is discussed. This system uses phototransistors as optically controlled transducers in the unknown branch of the bridge. When incorporated either as an input to a CNN or as an integral part of a CNN, the phototransistor-bridge cell (PBC) offers a means to detect small signal changes on large background levels in real-time. It also enables the reduction of noise using the interconnection techniques established for CNNs. The inherent properties of the bridge network also contribute to noise reduction. The stability of the stand-alone PBC is discussed, and PSPICE simulations are conducted for a single PBC and a 4×4 array of PBCs with nearest-neighbor interconnection