Robert Alexandru Dobre

Modelling, simulation and testing of an autonomous embedded system supplied by a photovoltaic panel

The concept of energy harvesting, to produce electrical energy from other existing forms of energy from the environment, has become very popular for a range of low power applications, like wireless sensor networks, data-loggers or identification devices. Solar energy is the best type of energy harvesting in terms of produced power, but has a major drawback - no available energy during the night. A solution to compensate this problem is to store electrical energy in the daytime and to offer it in the night; there are many implementations, involving use of different type of accumulators (lead-acid, NiMH, LiIon) or supercapacitors. Presented paper focuses on modelling and simulation of an autonomous embedded system that uses supercapacitor for energy storage and photovoltaic panel for power generation. Design, implementation and testing of the embedded system are also presented.

New insights on the convergence analysis of the affine projection algorithm for system identification

The convergence analysis of the affine projection algorithm (APA) is far to be an easy or trivial problem. In order to have a tractable analytical approach, a set of assumptions must be accepted. As a consequence, important differences appear between the analytical model and the simulations. In this paper the APA convergence is studied for two levels of approximation. A compact equation for a correction term is derived to refine the analysis. This result illustrates a better matching between the theoretical findings and simulation, in terms of the asymptotic behavior of the algorithm.

Performance studies of electrochromic displays

The idea of having flexible, very thin, light, low power and even low cost display devices implemented using new materials and technologies is very exciting. Nowadays we can talk about more than just concepts, such devices exist, and they are part of an emerging concept: FOLAE (Flexible Organic and Large Area Electronics). Among the advantages of electrochromic devices are the low power consumption (they are non-emissive, i.e. passive) and the aspect like ink on paper with good viewing angle. Some studies are still necessary for further development, before proper performances are met and the functional behavior can be predicted. This paper presents the results of the research activity conducted to develop electric characterization platform for the organic electronics display devices, especially electrochromic displays, to permit a thorough study. The hardware part of platform permits the measuring of different electric and optical parameters. Charging/discharging a display element presents high interest for optimal driving circuitry. In this sense, the corresponding waveforms are presented. The contrast of the display is also measured for different operation conditions as driving voltage levels and duration. The effect of temperature on electrical and optical parameters (contrast) of the display will be also presented.

Improved convergence model of the affine projection algorithm for system identification

The convergence analysis of the affine projection algorithm (APA) was intensively studied, however, it is far to be a trivial task. In this context, a set of assumptions have to be considered, in order to allow a tractable analytical approach. As a consequence, notable differences occur between the simulation results and the analytical model. In this paper, the convergence of APA is studied and a refined approximation is proposed, named “the third level of approximation.” The results show a near perfect matching between the theoretical model and simulation results, in terms of the asymptotic behavior and convergence speed of the studied algorithm.

Improved regularization for a low-complexity RLS algorithm

The regularization of the recursive least-squares (RLS) adaptive algorithm is a subject rarely approached in literature. The classical regularization method is implemented in the initial stages of the algorithm and has a time-limited effect on the corresponding correlation matrix. This paper proposes a new regularization technique for the RLS — dichotomous coordinate descent (DCD) algorithm, which is applicable for every filter iteration and improves the performance of the algorithm in low signal-to-noise situations. The functionality of the method is demonstrated in acoustic echo cancellation scenarios.

High order sliding mode observers and singular value decomposition in communications

An important issue in communications, in general, and cryptography, in particular, is recovering the information of interest at the reception end. Recent techniques make use of sliding mode observers to achieve this. Another useful method to reconstruct the state space of the transmitter at the receiver end, is the singular value decomposition. Targeting secure communications, the aim of the present paper is to compare the efficiency of the two techniques, when applied to the estimation of the state space of the transmitter at the reception end. For numerical exemplification, a Sprotts jerk-like system was used. The example is taken without loss of generality, being applicable to all differential models expressing the temporal evolution of diode or bipolar-junction transistor based circuits, like the well-known Colpitts oscillator.

Robust audio forensic software for recovering speech signals drowned in loud music

Audio evidence, when accepted by the court, can decide the final verdict in a trial. In order to be evaluated, these materials must be authenticated, but also the intelligibility of the message must be undoubtable. Two main categories of multimedia forensics solve these problems: content authentication and noise reduction. The application presented in this paper is part of the latter category. In order to conceal a conversation, the first action that comes into mind is also the easiest one: turn loud a nearby audio source. Since the most available audio sources play musical materials, if a microphone was placed in the room, it would record the speech signal heavily masked by music. A classical adaptive filtering method could be applied to recover the speech only if the speakers and the musical source remain perfectly still or, in other words, the acoustic environment does not change in time. This ideal situation is not to be found very often in real situations. This paper presents a method for recovering speech signals masked by loud music that is robust to acoustic environment variations. The method is thoroughly described, tested, and compared with a solution based on the recursive least-squares (RLS) adaptive algorithm using a variable forgetting factor.

Automatic music transcription software based on constant Q transform

The paper presents an automatic music transcription software which uses a constant Q transform as a time-frequency analysis tool showing why this transform is more adequate than the discrete Fourier transform for the presented application. The software can transcribe melodic structures played in any musical key by any equal tempered instrument. The input of the software is represented by an audio file containing the recording to be transcribed and the output is the musical score with standard notation as a PDF file. MIDI (Musical Instrument Digital Interface) files can also be exported which are helpful in the musical production processes.

Search based applications for speech processing

Nowadays, the demand for human-to-machine communication systems by using human voice is constantly growing. These systems allow the user to get access to information, fundament decisions and solve daily life or business issues in a timely and natural manner, with as little effort as possible. The complexity of the user input requested by many front-end speech recognition and back-end processing systems leads to imminent barriers when it comes to fulfilling a request. The main contribution of this paper consists in evaluating speech recognition solutions that support search engine integration. Furthermore, we propose an architecture for an intelligent speech-to-process platform using semantic analysis of natural language for processing the back-end orientations that can greatly benefit from search engine technologies.

Analysis of power supply circuits for electroluminescent panels

The electroluminescent panel is a light emitting device that requires for normal operations alternative voltages with peak to peak amplitudes in 100V… 300V range and frequencies in 100Hz … 2 kHz range. Its advantages, when compared with standard light sources like incandescent lamps, gas-discharge lamps or light emitting diodes (LEDs), are lower power consumption, flexible substrate and uniform light without observable luminous points. One disadvantage of electroluminescent panels is the complex power supply required to drive them, but the continuous improvement in passive and active integrated devices for switched mode power supplies will eventually solve this issue. The present paper studies different topologies for these power supplies and the effect of the electric parameters like the amplitude, frequency, waveform of the supplying voltage on the light emission and on power consumption for electroluminescent panels with different size and colors.