Analog VLSI CMOS implementation of beta neuron in a low power low voltage environment

摘要

A new CMOS analog programmable beta function for Beta Basis Function Neural Networks is proposed. Based on current mode operations and using devices operating in the subthreshold region, low power low voltage operations are carried out. The supply voltage is 1.3V and the quiescent power consumption is 4??W. The proposed circuit makes use of the weak inversion MOS model. It carried out a compact implementation of a current mode logarithmic converter, a one quadrant programmable multiplier and a current exponentiator. The beta exponents of the design are tuneable by means of control sources. Design modularity concerns also the maximum beta function value. Pspice simulations were done for a 0,35??m CMOS process of AMS. Correct operations of the different blocks were observed. The relative error between theory and simulations of the beta function is less than 2%.