David Sodoyer | Institut français des scineces et technologies des transports, de l'aménagement et des réseaux (original) (raw)
Papers by David Sodoyer
HAL (Le Centre pour la Communication Scientifique Directe), 2014
Global System for Mobile communications-Railways (GSM-R) is now widely used for train-to-track ra... more Global System for Mobile communications-Railways (GSM-R) is now widely used for train-to-track radio communication. As for any radio communication system, non-intentional and intentional ElectroMagnetic Interference (EMI) can affect this track-to-train radio link. This paper studies the impact of low-power electromagnetic jammers on this communication. Firstly, experiments performed on a dedicated test bench are presented to evaluate the impact of selected jamming signals on a GSM-R radio communication. In certain jamming conditions, loss of connection can result. Then, a method to detect these jamming signals is presented. We considered and modelled a simplified GSM-R communication chain using a GMSK modulation. This was used to evaluate the impact of the jammer on the IQ constellation representation. As for the experimental part, the same interfering signals are applied. We then use the Error Vector Magnitude parameter to detect the presence of these jamming signals. Depending on the parameters of the modulation chain, we adapt our processing to detect the disturbance. A good detection of the jammed signal is obtained by analysing the EVM parameter while trying to detect these signals in an AWGN channel.
Bookmarks Related papers MentionsView impact
2022 30th European Signal Processing Conference (EUSIPCO)
Bookmarks Related papers MentionsView impact
2015 1st URSI Atlantic Radio Science Conference (URSI AT-RASC), 2015
Bookmarks Related papers MentionsView impact
Seventh International Symposium on Signal Processing and Its Applications, 2003. Proceedings., 2003
We present a new approach to the source separation problem for multiple speech signals. Using the... more We present a new approach to the source separation problem for multiple speech signals. Using the extra visual information of the speaker's face, the method aims to extract an acoustic speech signal from other acoustic signals by exploiting its coherence with the speaker's lip movements. We define a statistical model of the joint probability of visual and spectral audio input for quantifying the audio-visual coherence. Then, separation can be achieved by maximising this joint probability. Experiments on additive mixtures of 2, 3 and 5 sources show that the algorithm performs well, and systematically better than the classical BSS algorithm JADE.
Bookmarks Related papers MentionsView impact
Speech Communication, 2004
Bookmarks Related papers MentionsView impact
EURASIP Journal on Advances in Signal Processing, 2002
Bookmarks Related papers MentionsView impact
2006 IEEE International Conference on Acoustics Speed and Signal Processing Proceedings
Bookmarks Related papers MentionsView impact
Running title: Voice activity detection based on lip movements This paper presents a quantitative... more Running title: Voice activity detection based on lip movements This paper presents a quantitative and comprehensive study of the lip movements of a given speaker in different speech / non speech contexts, with a particular focus on silences (i.e., when no sound is produced by the speaker). The aim is to characterize the relationship between "lip activity" and "speech activity", and then to use visual speech information as a Voice Activity Detector (VAD). To this aim, an original audio-visual corpus was recorded with two speakers involved in a face-to-face spontaneous dialog, although being in separate rooms. Each speaker communicated with the other using a microphone, a camera, a screen, and headphones. This system was used to capture separate audio stimuli for each speaker and to monitor each speaker's lip movements in synchrony with the recorded sound. A comprehensive analysis was carried out on the lip shapes and lip movements corresponding to either sile...
Bookmarks Related papers MentionsView impact
GSM-Railway is the ground to train radio communication system currently under deployment along th... more GSM-Railway is the ground to train radio communication system currently under deployment along the European railway infrastructure. 150.000 km of railway track are foreseen to be covered in Europe in the coming years. The system is designed to ensure different communication services between ground and trains and is replacing progressively various old analogue radio equipment. It is one of the components of the “Euroradio” system, managing voice and signalling data. As a component of a critical infrastructure, GSM-R is a potential target for intentional electromagnetic interference. Indeed, malicious attacks can generate and propagate electromagnetic energy into the GSM-R frequency bands with the objective to impact the whole railway system. This paper presents a methodology based on the error vector magnitude parameter enabling to detect the presence of jamming signals at the receiver. This early detection can then be used by the railway system to appropriately react.
Bookmarks Related papers MentionsView impact
In this paper, we investigate the impact of electromagnetic interference (EMI) on the quality of ... more In this paper, we investigate the impact of electromagnetic interference (EMI) on the quality of a Global System for Mobile communications - Railways (GSM-R) ground to train radio link. Scenarios of radiated electromagnetic (EM) disturbances applied to a Gaussian minimum shift keying (GMSK) based receiver, potentially impairing the communication quality are assessed. The objective is to establish a link between the disturbance characteristics of the interference and the received communication quality; the physical layer is considered. A GSM-R communication chain is modeled. Radiated EMI perturbations are injected to study, at the receiver, the bit error rate (BER). The signal plus interference characteristics are analyzed to detect the interference. In a second step, this work will be used to identify suitable criteria to implement an automatic EMI detection and identification system.
Bookmarks Related papers MentionsView impact
Transportation research procedia, 2014
En raison de son deploiement tres vaste et de sa relative facilite d'acces, l'infrastruct... more En raison de son deploiement tres vaste et de sa relative facilite d'acces, l'infrastructure ferroviaire apparait susceptible de subir d’eventuelles attaques. Celles-ci pourraient avoir des consequences significatives sur l’economie ainsi qu’en termes de securite. Ces dernieres annees, des efforts considerables ont ete deployes en vue de developper un systeme de gestion du trafic ferroviaire paneuropeen interoperable. Ce systemes sont de ce fait bien documentes et leurs caracteristiques sont egalement largement diffusees et disponibles. Des communications radio de differentes natures sont largement employees pour gerer ce trafic ferroviaire. Dans ce contexte, le projet europeen "SECRET" vise a developper des solutions innovantes afin d’ameliorer la resilience electromagnetique des systemes radio ferroviaires. Cet article explore differents aspects de recherche du projet. Une premiere partie analyse l'infrastructure ferroviaire et les risques d'attaques elec...
Bookmarks Related papers MentionsView impact
Looking at the speaker’s face seems useful to better hear a speech signal and extract it from com... more Looking at the speaker’s face seems useful to better hear a speech signal and extract it from competing sources before identification. This might result in elaborating new speech enhancement or extraction techniques exploiting the audio-visual coherence of speech stimuli. In this paper, we present a set of experiments on a novel algorithm plugging audio-visual coherence estimated by statistical tools, on classical blind source separation algorithms. We show in the case of additive mixtures that this algorithm performs better than classical blind tools both when there are as many sensors as sources, and when there are less sensors than sources. Audiovisual coherence enables to focus on the speech source to extract. It may also be used at the output of a classical source separation algorithm, to select the “best” sensor in reference to a target source.
Bookmarks Related papers MentionsView impact
more and more domains use wireless communications solutions to perform remote control actions or ... more more and more domains use wireless communications solutions to perform remote control actions or remote monitoring. However, in some cases, these solutions may be vulnerable to jamming devices readily available on the internet. In order to provide resiliency solutions against these jamming devices, we have investigated methods for detecting jamming signals. We were particularly interested in the European rail industry. Indeed, the European railway network is significantly evolving in order to increase its performances and to permit border crossing by the trains. For this purpose, a new and unique communication system is under deployment in Europe for ground-to-train radio operating, the Global System for Mobile communications - Railways (GSM-R). However, under specific conditions, this telecommunication system can be affected by certain jamming devices and in the case of absence of communication links for a given duration, the trains are forced to stop to guarantee the safety of the...
Bookmarks Related papers MentionsView impact
With the deployment of the European Rail Traffic Management System (ERTMS), most of the existing ... more With the deployment of the European Rail Traffic Management System (ERTMS), most of the existing groundbased signaling equipment operated along railway tracks will gradually disappear. The signaling information and command/control signals will become exchanged over Global System for Mobile communications-Railways (GSM-R). Consequently, railway signaling resilience should be re-evaluated against Intentional ElectroMagnetic Interference (IEMI) threats. In this paper we investigate GSM-R vulnerability to IEMI. These IEMIs consist in known interferers and their corresponding interference characteristics. The paper starts by describing the GSM-R communication model. Then, the interference characteristics are analyzed and progressively modified in order to study the corresponding impact on the communication. Results are presented according to these interferer characteristics and associated communication bit error rate (BER).
Bookmarks Related papers MentionsView impact
In this paper, a method to address the automatic instrument recognition in polyphonic music is in... more In this paper, a method to address the automatic instrument recognition in polyphonic music is introduced. It is based on the decomposition of the music signal with instrument-specific harmonic atoms, yielding to an approximate object representation of the signal. A post-processing is then applied to exhibit ensemble saliences that give clues about the number of instrument playing and the instrument labels. After a parameter optimization on a development set, the whole algorithm is applied on artificial mixes of solo performances. The identification of the number of instrument reaches 73 % and the identification of the ensemble label without prior knowledge on the number of instruments 17 %.
Bookmarks Related papers MentionsView impact
Ensuring passengers’ safety is one of the daily concerns of railway operators. To do this, variou... more Ensuring passengers’ safety is one of the daily concerns of railway operators. To do this, various image and sound processing techniques have been proposed in the scientific community. Since the beginning of the 2010s, the development of deep learning made it possible to develop these research areas in the railway field included. Thus, this article deals with the audio events detection task (screams, glass breaks, gunshots, sprays) using deep learning techniques. It describes the methodology for designing a deep learning architecture that is both suitable for audio detection and optimised for embedded railway systems. We will describe how we designed from scratch two CRNN (Convolutional Recurrent Neural Network) for the detection task. And since the creation of a large and varied training database is one of the challenges of deep learning, this article also deals with the innovative methodology used to build a database of audio events in the railway environment. Finally, we will sho...
Bookmarks Related papers MentionsView impact
Bookmarks Related papers MentionsView impact
Expert Systems with Applications
Bookmarks Related papers MentionsView impact
International Journal of Speech Technology
Bookmarks Related papers MentionsView impact
2016 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2016
Deep Neural Networks (DNNs) have recently become a popular technique for regression and classific... more Deep Neural Networks (DNNs) have recently become a popular technique for regression and classification problems. Their capacity to learn high-order correlations between input and output data proves to be very powerful for automatic speech recognition. In this paper we investigate the use of DNNs for automatic scream and shouted speech detection, within the framework of surveillance systems in public transportation. We recorded a database of sounds occurring in subway trains in real conditions of exploitation and used DNNs to classify the sounds into screams, shouts and other categories. We report encouraging results, given the difficulty of the task, especially when a high level of surrounding noise is present.
Bookmarks Related papers MentionsView impact
HAL (Le Centre pour la Communication Scientifique Directe), 2014
Global System for Mobile communications-Railways (GSM-R) is now widely used for train-to-track ra... more Global System for Mobile communications-Railways (GSM-R) is now widely used for train-to-track radio communication. As for any radio communication system, non-intentional and intentional ElectroMagnetic Interference (EMI) can affect this track-to-train radio link. This paper studies the impact of low-power electromagnetic jammers on this communication. Firstly, experiments performed on a dedicated test bench are presented to evaluate the impact of selected jamming signals on a GSM-R radio communication. In certain jamming conditions, loss of connection can result. Then, a method to detect these jamming signals is presented. We considered and modelled a simplified GSM-R communication chain using a GMSK modulation. This was used to evaluate the impact of the jammer on the IQ constellation representation. As for the experimental part, the same interfering signals are applied. We then use the Error Vector Magnitude parameter to detect the presence of these jamming signals. Depending on the parameters of the modulation chain, we adapt our processing to detect the disturbance. A good detection of the jammed signal is obtained by analysing the EVM parameter while trying to detect these signals in an AWGN channel.
Bookmarks Related papers MentionsView impact
2022 30th European Signal Processing Conference (EUSIPCO)
Bookmarks Related papers MentionsView impact
2015 1st URSI Atlantic Radio Science Conference (URSI AT-RASC), 2015
Bookmarks Related papers MentionsView impact
Seventh International Symposium on Signal Processing and Its Applications, 2003. Proceedings., 2003
We present a new approach to the source separation problem for multiple speech signals. Using the... more We present a new approach to the source separation problem for multiple speech signals. Using the extra visual information of the speaker's face, the method aims to extract an acoustic speech signal from other acoustic signals by exploiting its coherence with the speaker's lip movements. We define a statistical model of the joint probability of visual and spectral audio input for quantifying the audio-visual coherence. Then, separation can be achieved by maximising this joint probability. Experiments on additive mixtures of 2, 3 and 5 sources show that the algorithm performs well, and systematically better than the classical BSS algorithm JADE.
Bookmarks Related papers MentionsView impact
Speech Communication, 2004
Bookmarks Related papers MentionsView impact
EURASIP Journal on Advances in Signal Processing, 2002
Bookmarks Related papers MentionsView impact
2006 IEEE International Conference on Acoustics Speed and Signal Processing Proceedings
Bookmarks Related papers MentionsView impact
Running title: Voice activity detection based on lip movements This paper presents a quantitative... more Running title: Voice activity detection based on lip movements This paper presents a quantitative and comprehensive study of the lip movements of a given speaker in different speech / non speech contexts, with a particular focus on silences (i.e., when no sound is produced by the speaker). The aim is to characterize the relationship between "lip activity" and "speech activity", and then to use visual speech information as a Voice Activity Detector (VAD). To this aim, an original audio-visual corpus was recorded with two speakers involved in a face-to-face spontaneous dialog, although being in separate rooms. Each speaker communicated with the other using a microphone, a camera, a screen, and headphones. This system was used to capture separate audio stimuli for each speaker and to monitor each speaker's lip movements in synchrony with the recorded sound. A comprehensive analysis was carried out on the lip shapes and lip movements corresponding to either sile...
Bookmarks Related papers MentionsView impact
GSM-Railway is the ground to train radio communication system currently under deployment along th... more GSM-Railway is the ground to train radio communication system currently under deployment along the European railway infrastructure. 150.000 km of railway track are foreseen to be covered in Europe in the coming years. The system is designed to ensure different communication services between ground and trains and is replacing progressively various old analogue radio equipment. It is one of the components of the “Euroradio” system, managing voice and signalling data. As a component of a critical infrastructure, GSM-R is a potential target for intentional electromagnetic interference. Indeed, malicious attacks can generate and propagate electromagnetic energy into the GSM-R frequency bands with the objective to impact the whole railway system. This paper presents a methodology based on the error vector magnitude parameter enabling to detect the presence of jamming signals at the receiver. This early detection can then be used by the railway system to appropriately react.
Bookmarks Related papers MentionsView impact
In this paper, we investigate the impact of electromagnetic interference (EMI) on the quality of ... more In this paper, we investigate the impact of electromagnetic interference (EMI) on the quality of a Global System for Mobile communications - Railways (GSM-R) ground to train radio link. Scenarios of radiated electromagnetic (EM) disturbances applied to a Gaussian minimum shift keying (GMSK) based receiver, potentially impairing the communication quality are assessed. The objective is to establish a link between the disturbance characteristics of the interference and the received communication quality; the physical layer is considered. A GSM-R communication chain is modeled. Radiated EMI perturbations are injected to study, at the receiver, the bit error rate (BER). The signal plus interference characteristics are analyzed to detect the interference. In a second step, this work will be used to identify suitable criteria to implement an automatic EMI detection and identification system.
Bookmarks Related papers MentionsView impact
Transportation research procedia, 2014
En raison de son deploiement tres vaste et de sa relative facilite d'acces, l'infrastruct... more En raison de son deploiement tres vaste et de sa relative facilite d'acces, l'infrastructure ferroviaire apparait susceptible de subir d’eventuelles attaques. Celles-ci pourraient avoir des consequences significatives sur l’economie ainsi qu’en termes de securite. Ces dernieres annees, des efforts considerables ont ete deployes en vue de developper un systeme de gestion du trafic ferroviaire paneuropeen interoperable. Ce systemes sont de ce fait bien documentes et leurs caracteristiques sont egalement largement diffusees et disponibles. Des communications radio de differentes natures sont largement employees pour gerer ce trafic ferroviaire. Dans ce contexte, le projet europeen "SECRET" vise a developper des solutions innovantes afin d’ameliorer la resilience electromagnetique des systemes radio ferroviaires. Cet article explore differents aspects de recherche du projet. Une premiere partie analyse l'infrastructure ferroviaire et les risques d'attaques elec...
Bookmarks Related papers MentionsView impact
Looking at the speaker’s face seems useful to better hear a speech signal and extract it from com... more Looking at the speaker’s face seems useful to better hear a speech signal and extract it from competing sources before identification. This might result in elaborating new speech enhancement or extraction techniques exploiting the audio-visual coherence of speech stimuli. In this paper, we present a set of experiments on a novel algorithm plugging audio-visual coherence estimated by statistical tools, on classical blind source separation algorithms. We show in the case of additive mixtures that this algorithm performs better than classical blind tools both when there are as many sensors as sources, and when there are less sensors than sources. Audiovisual coherence enables to focus on the speech source to extract. It may also be used at the output of a classical source separation algorithm, to select the “best” sensor in reference to a target source.
Bookmarks Related papers MentionsView impact
more and more domains use wireless communications solutions to perform remote control actions or ... more more and more domains use wireless communications solutions to perform remote control actions or remote monitoring. However, in some cases, these solutions may be vulnerable to jamming devices readily available on the internet. In order to provide resiliency solutions against these jamming devices, we have investigated methods for detecting jamming signals. We were particularly interested in the European rail industry. Indeed, the European railway network is significantly evolving in order to increase its performances and to permit border crossing by the trains. For this purpose, a new and unique communication system is under deployment in Europe for ground-to-train radio operating, the Global System for Mobile communications - Railways (GSM-R). However, under specific conditions, this telecommunication system can be affected by certain jamming devices and in the case of absence of communication links for a given duration, the trains are forced to stop to guarantee the safety of the...
Bookmarks Related papers MentionsView impact
With the deployment of the European Rail Traffic Management System (ERTMS), most of the existing ... more With the deployment of the European Rail Traffic Management System (ERTMS), most of the existing groundbased signaling equipment operated along railway tracks will gradually disappear. The signaling information and command/control signals will become exchanged over Global System for Mobile communications-Railways (GSM-R). Consequently, railway signaling resilience should be re-evaluated against Intentional ElectroMagnetic Interference (IEMI) threats. In this paper we investigate GSM-R vulnerability to IEMI. These IEMIs consist in known interferers and their corresponding interference characteristics. The paper starts by describing the GSM-R communication model. Then, the interference characteristics are analyzed and progressively modified in order to study the corresponding impact on the communication. Results are presented according to these interferer characteristics and associated communication bit error rate (BER).
Bookmarks Related papers MentionsView impact
In this paper, a method to address the automatic instrument recognition in polyphonic music is in... more In this paper, a method to address the automatic instrument recognition in polyphonic music is introduced. It is based on the decomposition of the music signal with instrument-specific harmonic atoms, yielding to an approximate object representation of the signal. A post-processing is then applied to exhibit ensemble saliences that give clues about the number of instrument playing and the instrument labels. After a parameter optimization on a development set, the whole algorithm is applied on artificial mixes of solo performances. The identification of the number of instrument reaches 73 % and the identification of the ensemble label without prior knowledge on the number of instruments 17 %.
Bookmarks Related papers MentionsView impact
Ensuring passengers’ safety is one of the daily concerns of railway operators. To do this, variou... more Ensuring passengers’ safety is one of the daily concerns of railway operators. To do this, various image and sound processing techniques have been proposed in the scientific community. Since the beginning of the 2010s, the development of deep learning made it possible to develop these research areas in the railway field included. Thus, this article deals with the audio events detection task (screams, glass breaks, gunshots, sprays) using deep learning techniques. It describes the methodology for designing a deep learning architecture that is both suitable for audio detection and optimised for embedded railway systems. We will describe how we designed from scratch two CRNN (Convolutional Recurrent Neural Network) for the detection task. And since the creation of a large and varied training database is one of the challenges of deep learning, this article also deals with the innovative methodology used to build a database of audio events in the railway environment. Finally, we will sho...
Bookmarks Related papers MentionsView impact
Bookmarks Related papers MentionsView impact
Expert Systems with Applications
Bookmarks Related papers MentionsView impact
International Journal of Speech Technology
Bookmarks Related papers MentionsView impact
2016 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2016
Deep Neural Networks (DNNs) have recently become a popular technique for regression and classific... more Deep Neural Networks (DNNs) have recently become a popular technique for regression and classification problems. Their capacity to learn high-order correlations between input and output data proves to be very powerful for automatic speech recognition. In this paper we investigate the use of DNNs for automatic scream and shouted speech detection, within the framework of surveillance systems in public transportation. We recorded a database of sounds occurring in subway trains in real conditions of exploitation and used DNNs to classify the sounds into screams, shouts and other categories. We report encouraging results, given the difficulty of the task, especially when a high level of surrounding noise is present.
Bookmarks Related papers MentionsView impact