Log In
Sign Up
Romania
Citizenship:
Romania
Ph.D. degree award:
2003
Mr.
Constantin
Paleologu
Dr.
Professor
-
UNIVERSITATEA NAȚIONALĂ DE ȘTIINȚĂ ȘI TEHNOLOGIE POLITEHNICA BUCUREȘTI
Researcher | Teaching staff | Scientific reviewer | Consultant
>20
years
Personal public profile link.
Curriculum Vitae (14/05/2024)
Expertise & keywords
Adaptive filtering
Adaptive algorithms
Acoustic signal processing
Sparse systems
Echo cancellation
Projects
Publications & Patents
Entrepreneurship
Reviewer section
Efficient algorithms for reducing sound artifacts in hearing aids
Call name:
P 4 - Proiecte de cercetare exploratorie - PCE-2021
PN-III-P4-PCE-2021-0780
2022
-
2024
Role in this project:
Coordinating institution:
UNIVERSITATEA "VALAHIA" TARGOVISTE
Project partners:
UNIVERSITATEA "VALAHIA" TARGOVISTE (RO)
Affiliation:
Project website:
https://www.earned.valahia.ro/
Abstract:
The quality of life of human beings is seriously impacted by the loss of hearing and this is an increasing medical problem in Romania and the EU. While significant advances have been achieved in hearing aid design, there is still user dissatisfaction with commercially available hearing aids, especially regarding howling, occlusion, and entrainment effects. This project proposes an original approach to reduce these annoying sounds by combining active noise control (ANC), acoustic feedback cancellation (AFC), and machine learning (ML) approaches. Multiple and novel ways the machine learning-based block can influence the parameters of the linear/nonlinear adaptive filters will be investigated by using the team's extensive experience in the ANC and AFC domains. Up to 20 dB improvement on maximum stable gain (MSG), added stable gain (ASG) is expected, increased robustness to noise, and under-modeling of the feedback path for both real speech and music signals. The proposed algorithms will be implemented on DSPs and FPGAs. It is expected that the proposed algorithms could have potential applications in telecommunications (e.g. stereo acoustic echo cancellation, mobile communications), consumer electronics (e.g. the ANC systems for motorcycle helmets), etc.
Read more
Novel sparsity-aware adaptive algorithms for acoustic applications
Call name:
P 4 - Proiecte de cercetare exploratorie - PCE-2021
PN-III-P4-PCE-2021-0438
2022
-
2024
Role in this project:
Coordinating institution:
UNIVERSITATEA POLITEHNICA DIN BUCURESTI
Project partners:
UNIVERSITATEA NAŢIONALĂ DE ŞTIINŢĂ ŞI TEHNOLOGIE POLITEHNICA BUCUREŞTI (RO)
Affiliation:
Project website:
http://www.comm.pub.ro/naralan
Abstract:
In many applications related to the acoustic devices, we need to estimate an acoustic impulse response, due to the acoustic coupling between loudspeaker and microphone. The practical solution is to use an adaptive filter working in a system identification scenario. In this project, we will develop new adaptive algorithms for acoustic applications, by following two main ideas. (i) First, the sparse nature of the acoustic impulse responses is exploited. Most of the state-of-art algorithms perform well for very sparse impulse responses, but they usually fail when the number of non-zero coefficients increases (like in acoustic scenarios). The project will provide a novel framework to design sparsity-aware algorithms, by integrating an individual control approach based on specific optimization criteria (so that even the small coefficients could contribute to the overall convergence) together with the model uncertainties (related to the time-variant characteristic of the system). (ii) Second, a critical issue is related to the long length of the acoustic impulse response (hundreds or even thousands of coefficients). To this purpose, we will follow a novel approach, by exploiting tensor-based decomposition techniques, together with low-rank approximation methods. The goal is to reformulate a high-dimension system identification problem based on low-dimension problems (i.e., shorter filters) that are combined together. The gain is twofold, in terms of both performance and complexity.
Read more
Adaptive Algorithms for Multilinear System Identification Using Tensor Modelling
Call name:
P 1 - SP 1.1 - Proiecte de cercetare Postdoctorală
PN-III-P1-1.1-PD-2019-0340
2021
-
2022
Role in this project:
Key expert
Coordinating institution:
UNIVERSITATEA POLITEHNICA DIN BUCURESTI
Project partners:
UNIVERSITATEA POLITEHNICA DIN BUCURESTI (RO)
Affiliation:
UNIVERSITATEA POLITEHNICA DIN BUCURESTI (RO)
Project website:
https://lauramariadogariu.github.io/adamsystem/
Abstract:
In many applications related to system identification, the unknown system can be modeled as a finite impulse response filter with a large number of coefficients, which raises challenges for the adaptive filter used for this purpose. In this context, the system can be regarded as a multidimensional structure, mathematically modeled by a tensor, which may be decomposed into shorter, easier to identify impulse responses. In addition, specific properties of the systems, such as the sparsity, could be used to simplify the problem. However, solutions have been proposed so far only for second and third order tensors. Motivated by the good performance of these solutions (developed during the PhD studies of the project leader), the project aims to provide a general framework for multilinear system identification, by using tensor modelling and decompositions. The solution will be applied first in the context of the Wiener filter, then for the main categories of adaptive filters, i.e., the least-mean-square (LMS) algorithm, the affine projection algorithm (APA), and a Kalman filter (KF) will also be developed. Convergence analysis and low-complexity versions of these algorithms will also be derived, as well as combinations of adaptive filters (e.g., combination scheme of LMS and Kalman-based algorithms, a combination scheme of APAs using different projection orders), to be used in certain applications. Software simulations will be conducted to test the performance of all the proposed algorithms and, based on the results, the best performing solutions will be implemented on fixed-point DSP or FPGA, targeted for specific applications. The results of the project will be disseminated in top ISI journals (e.g., IEEE Signal Processing Letters, Signal Processing, etc.) and conferences (e.g., ICASSP, EUSIPCO, etc.).
Read more
New Decomposition-Based Algorithms for Sparse System Identification
Call name:
P 1 - SP 1.1 - Proiecte de cercetare pentru stimularea tinerelor echipe independente
PN-III-P1-1.1-TE-2019-0529
2020
-
2022
Role in this project:
Coordinating institution:
UNIVERSITATEA POLITEHNICA DIN BUCURESTI
Project partners:
UNIVERSITATEA POLITEHNICA DIN BUCURESTI (RO)
Affiliation:
UNIVERSITATEA POLITEHNICA DIN BUCURESTI (RO)
Project website:
http://www.comm.pub.ro/dasti/
Abstract:
Sparsity represents a key feature in system identification scenarios where only a small percentage of the impulse response components have a significant magnitude. Many important applications can be formulated in terms of sparse system identification problems, e.g., network/acoustic echo cancellation, satellite-linked communications, radar systems, underwater communications, microphone arrays, etc. In these frameworks, adaptive filters represent some of the most popular solutions for real-world/real-time applications. The current state-of-art algorithms used in sparse system identification scenarios are mainly based on the proportionate approach, i.e., each coefficient of the filter is updated independently of the others (in proportion to its magnitude). However, in most applications, a major limitation is the high length of the impulse response (e.g., hundreds/thousands of coefficients), which poses significant challenges in terms of complexity, convergence, and accuracy of the solution. In this project, we focus on a new approach that exploits the impulse response decomposition based on the nearest Kronecker product and low-rank approximations, which fits very well for sparse system identification problems. The basic idea is to transform a high-dimension system identification problem into smaller problems (i.e., shorter filters) that are connected to match the original purpose. The gain will be twofold, in terms of both performance and complexity. The new family of decomposition-based adaptive algorithms will involve fast converging techniques, like least-squares methods, which are usually prohibitive due to their high complexity. Also, we will exploit these decomposition-based structures in the context of multichannel and multidimensional adaptive algorithms, which broaden the applicability of the developed solutions from a system identification perspective. The algorithms will be implemented in fixed-point arithmetic on DSP/FPGA and tested in specific scenarios.
Read more
A multimodal video accessibility enhancement system dedicated to deaf and hearing impaired users
Call name:
P 1 - SP 1.1 - Proiecte de cercetare pentru stimularea tinerelor echipe independente
PN-III-P1-1.1-TE-2019-0420
2020
-
2022
Role in this project:
Coordinating institution:
UNIVERSITATEA POLITEHNICA DIN BUCURESTI
Project partners:
UNIVERSITATEA POLITEHNICA DIN BUCURESTI (RO)
Affiliation:
UNIVERSITATEA POLITEHNICA DIN BUCURESTI (RO)
Project website:
http://smartviewproject.go.ro/
Abstract:
The main objective of the SMART-VIEW research project is to enhance the quality of life of hearing impaired people by increasing their accessibility and comprehension over multimedia documents. The specific objectives imposed to the project are: (1). A novel methodology for automatic synchronization of the close captions/subtitles with the audio-video streams especially for live generated products; (2). A novel framework for robust active speaker detection, based on a multimodal fusion of text, audio and visual information; (3). An efficient subtitle positioning optimization strategy that makes it possible to appropriately place the subtitle on the screen, by taking into account both the visual content and the detected active speaker. The SMART-VIEW project will have a significant impact over the scientific community through the new set of concepts proposed in the area of computer vision technologies specifically designed and tuned in order to achieve real-time processing capabilities when implemented on GPU. At the social level, the project will have a direct and immediate impact over the deaf and hearing impaired community. The SMART-VIEW system will increase the user autonomy and will enhance their perception of the multimedia content. At the economic level the SMART-VIEW system could be included into any national or international television system that will broadcast multimedia services dedicated to hearing impaired users or to non-native speakers.
Read more
Contributions to REversible data hiding in the Encrypted Domain
Call name:
P 1 - SP 1.1 - Proiecte de cercetare Postdoctorală
PN-III-P1-1.1-PD-2019-1165
2020
-
2022
Role in this project:
Coordinating institution:
UNIVERSITATEA "VALAHIA" TARGOVISTE
Project partners:
UNIVERSITATEA "VALAHIA" TARGOVISTE (RO)
Affiliation:
UNIVERSITATEA "VALAHIA" TARGOVISTE (RO)
Project website:
https://creed.valahia.ro/
Abstract:
Reversible data hiding/watermarking (RDH) inserts hidden data into a host signal. This process must be performed in an imperceptible manner that allows for the exact recovery of both the host and the hidden data. Specialized RDH approaches were developed based on the nature of the host file (image, audio or video and more recently plaintext or encrypted).
Recently, the increased interest in Cloud computing platforms and privacy preserving applications has fueled the development of new signal processing algorithms for the encrypted domain. Such an approach, data hiding in the encrypted domain has received renewed interest from the signal processing research community. Reversible data hiding schemes in the encrypted domain (RDH-ED) aim to insert additional information into an encrypted host file without revealing its plaintext content and to exactly recover the original host by extracting the hidden data. Current RDH-ED schemes can be classified into two distinct categories, namely the ones based on vacating room before encryption (VRBE) and those based on vacating room after encryption (VRAE).
The main objectives of the CREED project are to:
- further improve existing VRAE based RDH-ED and develop new VRAE approaches;
- develop a VRAE inspired VRBE RDH-ED scheme that maintains the good performance of VRBE and the standard decryption of VRAE (and other new VRBE approaches);
- develop a generalized RDH-ED framework for multiple file types.
Read more
Robust Multi-Microphone Speech Enhancement Algorithms for Automobile Communications Systems
Call name:
Projects for Young Research Teams - RUTE -2014 call
PN-II-RU-TE-2014-4-1880
2015
-
2017
Role in this project:
Key expert
Coordinating institution:
UNIVERSITATEA POLITEHNICA DIN BUCURESTI
Project partners:
UNIVERSITATEA POLITEHNICA DIN BUCURESTI (RO)
Affiliation:
UNIVERSITATEA POLITEHNICA DIN BUCURESTI (RO)
Project website:
http://www.comm.pub.ro/spenhaut/
Abstract:
For audio systems integrated into a car there are many applications that have not been sufficiently exploited and prospects of development of research. An important application for hands-free communication systems or voice commands recognition is to separate speech signal and reduce the noise signal to obtain an enhanced undistorted and intelligible voice signal. Another future application for a car audio system is amplifying the voice signal coming from the driver or front passenger to be heard over the rear passengers. Voice driver amplification system, however, can get into audio feedback with the rear speakers sound system, therefore it is necessary to cancel the acoustic feedback. This project aims at the development of robust systems to improve speech signals processed by audio communication systems of the automobile to increase comfort and reduce the risk of accidents. There will be studied multi-microphone noise reduction techniques, for which both spectral and spatial characteristics of the signal sources can be used. Also there will be studied adaptive algorithms for noise reduction using multi-microphones and for acoustic feedback cancellation. The developed methods will be real-time implemented and tested on hardware platforms with DSP and FPGA in order to be able to integrate it into the car audio system.
Read more
Advanced Adaptive Algorithms for Digital Hearing Aids
Call name:
Exploratory Research Projects - PCE-2011 call
PN-II-ID-PCE-2011-3-0097
2011
-
2016
Role in this project:
Key expert
Coordinating institution:
Universitatea VALAHIA din Targoviste
Project partners:
Universitatea VALAHIA din Targoviste (RO)
Affiliation:
Project website:
http://adaptive.valahia.ro
Abstract:
Despite the advanced technology used by commercially available digital hearing aids, the user dissatisfaction is still high. Advanced adaptive filtering techniques can improve the performance of hearing aids in cases of whistling/buzzing, wind noise and intelligibility of the speech signals in noisy situations. Different time domain or frequency domain adaptive algorithms adapted for feedback cancellation using dichotomous coordinate descent iterations, variable step size, variable filter length, subband and filtered-x versions are proposed. Also, a new widely linear perspective is expected to lead to new noise reduction filters that preserve the spatial information without affecting the intelligibility of speech signals heard by the hearing impaired people. Algorithms that react fast at the onset of the wind noise and slower when the power of the wind is reduced, are expected to reduce the annoying wind noise problem. We propose to recast several fast algorithms in high dimensional reproducing kernel Hilbert spaces to yield powerful nonlinear extensions. Their connections with neural networks and machine learning will be exploited in order to assess the environment to identify the annoying interference signals. Real time implementations of the best algorithms on FPGA are expected to confirm these improvements. These adaptive algorithms developed for hearing aids will be of great use in other areas such as teleconferencing, consumer electronics and wireless communication.
Read more
NEW ADAPTIVE ALGORITHMS WITH VARIABLE CONVERGENCE
Call name:
Projects for Young Research Teams - TE-2010 call
PN-II-RU-TE-2010-0050
2010
-
2013
Role in this project:
Project coordinator
Coordinating institution:
UNIVERSITATEA POLITEHNICA DIN BUCURESTI
Project partners:
UNIVERSITATEA POLITEHNICA DIN BUCURESTI (RO)
Affiliation:
UNIVERSITATEA POLITEHNICA DIN BUCURESTI (RO)
Project website:
http://www.comm.pub.ro/te-vss
Abstract:
ADAPTIVE SIGNAL PROCESSING TECHNIQUES ARE FREQUENTLY INVOLVED IN MODERN COMMUNICATIONS SYSTEMS. IN THIS CONTEXT, THE KEY BLOCK OF ANY ADAPTIVE SYSTEM CONSISTS OF THE ADAPTIVE ALGORITHM. THIS INFLUENCES THE SYSTEM PERFORMANCE IN TERMS OF CONVERGENCE RATE, MISADJUSTMENT, AND STABILITY. THESE PERFORMANCE CRITERIA DEPEND ON THE ADAPTATION PARAMETERS [I.E., THE STEP-SIZE IN CASE OF THE LMS (LEAST-MEAN-SQUARE) ALGORITHMS, AND THE FORGETTING FACTOR IN CASE OF THE RLS (RECURSIVE LEAST-SQUARES) ALGORITHMS]. USING CONSTANT VALUES FOR THESE PARAMETERS LEAD TO A COMPROMISE BETWEEN THE PERFORMANCES CRITERIA. A SOLUTION IS TO ANALYZE THE CONVERGENCE STATE OF THE ALGORITHMS AND TO USE VARIABLE ADAPTATION PARAMETERS, THUS RESULTING ADAPTIVE ALGORITHMS WITH VARIABLE CONVERGENCE. NEVERTHELESS, THE EXISTING SOLUTIONS DO NOT OFFER A PROPER ACCURACY, SINCE THEIR CONVERGENCE CRITERIA (ERROR ENERGY OR MEAN SQUARE DEVIATION) ARE NOT VERY PRECISELY IN DETECTING THE CONVERGENCE STATE OF THE ALGORITHMS. THE MAIN GOAL OF THIS RESEARCH PROJECT IS TO PROPOSE NEW ADAPTIVE ALGORITHMS WITH VARIABLE CONVERGENCE, BASED ON NEW MODELS FOR ANALYZING THE CONVERGENCE STATE. THE BASIC IDEA IS TO SEPARATE THE CONVERGENCE AND MISADJUSTMENT COMPONENTS OF THE ALGORITHMS, AND TO INCORPORATE THEM INTO SPECIFIC COST FUNCTIONS. THE PROCEDURE WILL BE APPLIED FOR BOTH THE LMS AND RLS FAMILY OF ALGORITHMS, RESULTING NEW VSS-LMS (VARIABLE STEP-SIZE LMS), RESPECTIVELY NEW VFF-RLS (VARIABLE FORGETTING FACTOR RLS) ALGORITHMS. ALSO, WE AIM TO REDUCE THE COMPUTATIONAL COMPLEXITY OF THE PROPOSED VFF-RLS ALGORITHMS BY USING THE DCD (DICHOTOMOUS COORDINATE DESCENT) METHODS. THE PROPOSED ALGORITHMS WILL BE IMPLEMENTED ON FPGA AND TESTED IN AN ACOUSTIC ECHO CANCELLATION SCENARIO.
Read more
FILE DESCRIPTION
DOCUMENT
List of research grants as project coordinator
Download (11.64 kb) 17/04/2016
List of research grants as partner team leader
List of research grants as project coordinator or partner team leader
Significant R&D projects for enterprises, as project manager
R&D activities in enterprises
Peer-review activity for international programs/projects
[T: 0.431, O: 246]