Input Device for a Biomorphic Neuroprocessor Based on a Memristor-Diode Crossbar for the Pulse Coding of Information

An input device for a biomorphic neuroprocessor is developed. It is designed to encode incoming information into biomorphic impulses and then transfer it to a storage matrix. The electrical circuit of the encoder is built using extra-large logic matrices based on a memristor-diode crossbar. Using SPICE simulation, the operability of the electrical circuit of the encoder is shown in the mode of the simultaneous population coding of a binary number into frequency and delay. This coding mode, which has the highest fault tolerance, transmits more information than other modes, since it takes into account the value of the encoded number together with its derivatives in space and time. The proposed hardware implementation of the encoder has a high integration of elements and significantly lower power consumption than hardware implementations based on transistor programmable logic and very large integrated circuits (VLICs).