SAW injection locked oscillators: dynamic behaviour and application to neural networks

The characteristics of an injection-locked oscillator are examined in two parts. Part A covers the dynamic behaviour of a surface acoustic wave injection-locked oscillator and its similarity near lock-in to that of an acousto-optic modulator. It is shown that a reflective wave component travels between the injected source and the oscillator resulting in periodic locking and unlocking of tile system. These reflected waves are responsible for the classical modulation sidebands near lock-in. In part B a SAW based neural network that can associate, categorize and store input stimuli is presented. The network can be modelled as a modified Kohonen self-organizing feature map