High-Level Modeling and Simulation Tool for Sensor Conditioning Circuit Based on Artificial Neural Networks

Abstract

For current microelectronic integrated systems, the design methodology involves different steps that end up in the full system simulation by means of electrical and physical models prior to its manufacture. However, the higher the circuit complexity, the more time is required to complete these simulations, jeopardizing the convergence of the numerical methods and, hence, meaning that the reliability of the results are not guaranteed. This paper shows the use of a high-level tool based on Matlab to simulate the operation of an artificial neural network implemented in a mixed analog-digital CMOS process, intended for sensor calibration purposes. The proposed standard tool enables modification of the neural model architecture to adapt its characteristics to those of the electronic system, resulting in accurate behavioral models that predict the complete microelectronic IC system behavior under different operation conditions before its physical implementation with a simple, time-efficient, and reliable solution.

Microelectronic CMOS Implementation of a Machine Learning Technique for Sensor Calibration

“…It consists of three building blocks: the multipliers, which multiply either the input or the intermediate layer signals by a set of coefficients that weigh the contribution of each input in the processor output; the adder to sum the input weighted signals plus an additional input called bias, and the activation function (AF) circuit that implements the non-linear operation with the previous weighted sum and generates the neuron output. The electrical implementation of each processor building block in standard 0.18µm CMOS technology with 1.8V power supply has been studied in [17], where electrical simulations in Cadence were presented. More insight in the electrical characterization by considering experimental measurements with the integrated building blocks, is presented next.…”

“…The sum of both the weighted input and bias signals is also carried out in this layer, and the operation result is processed by a nonlinear operation in order to obtain the output of each processor in the hidden layer. A weighting of the neuron outputs from the hidden layer is carried out in the last layer, and an addition of the resulting signals with another weighted bias signal is made before providing the final system output [17].…”

“…A description of the proposed CMOS building blocks conforming the mixed-mode MLP processing unit is presented in Section II, where an experimental characterization of each circuit is also presented. Section III shows a summary of the methodology described in [17] to efficiently carry out the ML-based system high-level simulations. The proposed system electrical characterization considering a real example of non-linear sensor characteristic and a performance comparison with related and similar works are presented in Section IV.…”

Simulation paradigm to study circuit performance in presence of component level fault

“…Various faults and detecting techniques in power circuits of PV systems are described in (13) . MATLAB based ANN model is presented for complex mixed analog and digital CMOS circuit for prediction of accurate behavior prior to implementation, according to authors this is reliable and less time consuming solution (14) . The aim of this article is to create awareness among the beginners, practicing engineers regarding the use of simulation models to analyze the circuit behavior in presence of component level fault.…”

Mobile Phone Usage Detection by ANN Trained with a Metaheuristic Algorithm †

“…The application areas of ANN in nanotechnology involve classification, diagnosis, monitoring, process control, design, scheduling, and planning, and so on [3]. According to [4], ANN is the most powerful solution in sensors pre-processing information; these types of network can adapt their behavior without previous knowledge of a particular sensor response. For these cases, training algorithms are used to achieve the expected input–output relationship, by adjusting their weights in an optimal solution [4], where the input training/testing data comes from sensors that can be nano-sensors.…”