In this paper we shall present the physical realization of a neural decision engine which is part of an automatic coin recognizer to be included in commercial vending machines. The neural decision engine is based on two successive processing stages: classification and validation. Following the results provided by previous analysis [1], we have decided to implement both stages by means of the Multilayer Perceptron neural model. The physical VLSI realization of the system is motivated by the need to improve the confidentiality on the final product, as well as to provide some features (especially, processing speed) which are not affordable in a software implementation. The design alternatives used in the implementation have been carefully chosen, so as to provide a proper balance between the throughput of the system and its cost (in terms of occupied area). As the reults show, the proposed implementation represents a competitive alternative, both in terms of cost and processing speed, for the commercial application considered.
JavaScript jest wyłączony w Twojej przeglądarce internetowej. Włącz go, a następnie odśwież stronę, aby móc w pełni z niej korzystać.