Mihaela Cirlugea

An Evolutionary Multi-objective Optimization Approach to Design the Sound Processor of a Hearing Aid

Hearing impairment is one of the most common problems nowadays which affects the quality of life in terms of speech understanding and communication abilities. Therefore, untreated hearing loss may negatively impact social life. Mild to moderate hearing loss may be compensated by employing a hearing aid. The hearing aid performs sub-band division of the auditory signal, and then amplifies each sub-band in order to correct the patient’s audiogram. Designing the hearing aid sound processor for optimal audiogram correction requires the determination of the number of sub-bands, filter order and sub-band gains respectively. Considering the conflicting nature between solution quality and hardware resources, designing the sound processor is a matter of compromise. This article proposes an evolutionary approach to designing the sound processor, which treats the design as a multi-objective optimization (MOO) problem. Thus, the sub-band number, the sub-band filter order and the audiogram correction function are simultaneously optimized. The determination of the subband gains, for a given configuration of sub-band number and filter order, is treated as a single-objective optimization (SOO) problem. Dedicated genetic algorithms (GA) are used to solve both MOO and SOO problems. The MOO was handled with the Pareto frontier method. The functionality of the proposed design algorithm is illustrated on two conclusive design examples for optimal audiogram correction.

Analog Multiplying/Weighting VLSI Cells for SVM Classifiers

VLSI support vector machine classifiers require a large amount of calculations, therefore their implementation needs high density, high speed and low power circuits. In a SVM architecture based on a multiplying law the main building blocks are multipliers. We propose in this paper multiplying and weighting cells, developed by using a model consisting of a compound of two inverse non-linear functions. This procedure is suitable for VLSI implementation because it permits the use of simple nonlinearized standard DA cells that compensate each other nonlinearities to obtain an extended domain of operation. The resulted weighting/multiplying cells were analyzed and tested by simulations.

Modular Log-domain Multipliers for SVM Classifiers

We propose modular log-domain structures for implementing analog multipliers intended to be used in SVM (support vector machines) classifiers and neural network structures

Measuring method used for DC-DC converters

We created fast and precise behavioral blocks used for the measurement of power, average and root mean square values in electronic circuits. The obtained results are better than the ones offered by the Pspice functions. The method was tested on some filters and illustrated on some DC-DC converters, where the switch period is of /spl mu/seconds and it is hard to obtain accurate data. The signals can be seen through an adjustable window.

An electronically programmable current balancing instrumentation amplifier for biomedical monitoring

Instrumentation amplifiers (IA) used in wearable biomedical monitoring systems have to satisfy several performance requirements, such as low-voltage supply and low-power consumption, enabling increased portability. In the same time, a very important feature is wide range parameter programmability. This paper proposes a solution to implement electronic gain programming into an IA. The proposed solution is developed around the current balancing concept. An electronically tunable transconductance amplifier is used for adjusting the IA voltage gain as a function of the transconductor bias currents. Additionally, the proposed IA outputs a differential signal, useful for latter differential analog signal processing. Simulation results validate the proposed IA.

Log-Domain Binary SVM Image Classifier

We propose in this paper a binary log-domain Support Vector Machine classifier based on a polynomial decision function. To implement such a classifier log-domain multipliers proposed by the authors are used. For the parallel-serial implementation a log-domain summing amplifier and a current mode comparator are also needed. Current mode log-domain design is used for its low voltage, low power and high frequency characteristics. The resulted classifier is simulated taking into account real parameters of transistors in BiCMOS technology.

Improving Measuring Procedures in DC-DC Converters

The PSpice functions used for measuring the power, average and root mean square are not accurate enough because of using specific techniques like products or filters. We created fast and precise specialised behavioural blocks that obtain the power, average and root mean square values of the signals by applying an integral relation. The computation is done after the transitory regime is finished and so the values are constant and easy to be read. The method was used for measurements in DC-DC converters, where the switching period is very small (10-20 museconds).

Current-mode Configurable Analog Block

The paper presents a novel ELIN analog configurable block (CAB). Using this CAB one can implement first and second order transfer functions such as lossy and lossless integrator, low-pass, high-pass, band-pass and band-stop biquads. For implementing the building blocks, we used log-domain cells. Due to their simple structures and current-mode operation this circuits are suitable for low-voltage, low-power applications such as field programmable analog arrays. The circuits were implemented using a generic 180 nm BiCMOS process. SPICE analyses were performed to validate the design.

ABM blocks for precise measurements in DC-DC converters

This paper presents some fast and precise behavioral blocks used for the measurement of power, average and root mean square values in electronic circuits. They are fast and precise and the obtained results are better than the ones obtained with the Pspice functions. The method was tested on some filters, like LPF or HPF, and applied on DC-DC converters. We choosed to apply the method over these converters because the switch period is of the order of microseconds and accurate data is hard to be obtained. For the view of the measured signals is used an adjustable window.

Log-Domain Linear SVM Classifier

We develop in this paper support vector machine (SVM) parallel classifier for image classifications. The SVM classifier basic modules are new modular log-domain four-quadrant current multipliers. The circuit was simulated taking into account real parameters of transistors in BiCMOS technology. There are many advantages of using such modular log-domain structures in high frequency large dimension circuits and designs.