Charles Tatkeu

Deep neural networks for automatic detection of screams and shouted speech in subway trains

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.

Phase Coded Costas Signals for Ambiguity Function Improvement and Grating Lobes Suppression

In this paper, a novel radar signal is proposed. It is based on a modification of Costas signals obtained by increasing the frequency separation of Costas frequency hops beyond the orthogonality condition. Indeed, Costas codes are well known for their good Range-Doppler properties. Their orthogonality condition states that frequency separation between frequency hops of the Costas code should be at most equal to the inverse of bit duration of the code. Increasing frequency separation beyond this condition is sometimes needed in radar applications. However, this leads to the appearance of grating lobes near the main temporal lobe of the autocorrelation function. The originality of the proposed approach, in this paper, consists in proposing a modification of Costas sub-pulses in order to overcome grating lobes problem. Indeed, we show that when introducing Phase coding in the sub-pulses, grating lobes are lowered significantly. The study is carried out for several phase codes. The proposed code design is validated through computer experiments.

Evaluation of coding's methods for the development of a radar sensor for localization and communication dedicated to guided transport

To explore possibilities of new exploitation modes of automatic guided transport, our laboratories designed formerly a sensor radar called DIREP (Detection et Identification des Rames En Panne: detection and identification of broken-down trains). This system is founded on the principle of a co-operative radar using a transponder inside targets. It is based on a numerical correlation receiver and has a very broad band from 50 to 100 MHz. which remains unexploited. Thus we propose to benefit from this band to establish high flow communication. The proposed system is made of microwave transmitting and receiving equipment fixed on each train, one ahead and other behind. The two trains exchange data and specific signals coded that make the system able: to deduce the distance between the trains; to identify the state of the train (broken-down or not); and to allow the transmission of pictures or audio-video records with a high data flow. The aim of this work is to propose technical solutions for multiplexing communication data and localization code in order to allow high data flow transfer. Simulations are computed to evaluate the systems performance: BER, computing time and complexity.

Methods of target recognition for UWB radar

With the growth of embedded components (GPS, radar, cameras…), road vehicle becomes more and more intelligent and reassuring.

Doubly coded costas signals for grating lobes mitigation

In this paper, a novel design scheme for radar signal is proposed. It is based on a modification of Costas codes design obtained by increasing the frequency separation of frequency hops beyond the orthogonality condition. This condition states that frequency separation between frequency hops should be at most equal to the inverse of bit duration of the code. Indeed, increasing frequency separation beyond this condition is sometimes needed in radar applications. However, this leads to the appearance of grating lobes near the main temporal lobe of the autocorrelation function. This paper proposes overcoming this problem using sub-pulse coding. The originality of the proposed approach consists in choosing for Costas sub-pulses other Costas codes instead of constant frequency ones in order to lower autocorrelation grating lobes that would normally appear when the orthogonality condition is not satisfied. Indeed, we show that, under some conditions on Costas signal parameters in the sub-pulse, good elimination of grating lobes is achieved.

Short Range Radar Based on UWB Technology

In this Chapter, a short range radar system based on ultra-wideband (UWB) technology is presented. The radar and its applications were reserved during a long time to national defence, air security or weather services domains. Since a few years, with the emergence of new technologies, the radar applications were developed and become known in many sectors of daily life. The arrival of a new technology called Ultra-Wideband (UWB) allows in particular the development of compact and low-cost radar with multiple fields of applications. UWB uses very short non-sinusoidal pulses that have widths less than 1.5 ns, so that the spectrum of the transmitted signals may spread over several Gigahertz. This radar offers a resolution in distance of about a few centimetres, for example 15 cm for a pulse width of 1ns, making this system very interesting in several short range applications. UWB radar has many applications in Medical, Building, Surveillance, Security and Monitoring applications [Dam2007] and will appear more and more in our daily life. For example, UWB radar systems for local monitoring allow creating dome radar surveillance around a sensitive object or subject. These compact systems contain a small UWB radar with a range of about 10 meters, a standard radio system for transmitting the alarm in case of intrusion and a GPS system for localisation function. These systems can be engaged in public safety functions in buildings, aircrafts or for artwork protection in a museum, but also as an alarm system around a house or near a swimming pool to avoid too frequent small children drowning. Thanks to their sensitivity the UWB radar can detect movement as slight as a heart beat or breathing rate. These systems is responsive enough to accurately depict a heart rate as compared to others existing systems. UWB radar could be used as ground penetrating radar (GPR). GPR systems can obtain very precise and detailed images sub-soil. UWB radar is moved along surface and sends electromagnetic pulses into the ground. The analysis of received echoes can produce a very specific profile of underground. The investigation depth varies, depending on the type of ground, from a few meters in asphalt or clay to more than a hundred meters in limestone or granite, or even several kilometres into the ice. Finally, UWB radar could be used for short range collision avoidance as mentioned in this paper. This collision avoidance system, 24 GHz UWB Short Range Radar (SRR), was developed principally by European car manufacturers. It is a combination of an UWB radar and a

Collision avoidance radar system using UWB waveforms signature for road applications

The concept of the intelligent car consists in equipping the vehicle with an optimum number of sensors such as radar, in order to perceive the environment which surrounds it, to ensure an automatic guidance control of the vehicle and to communicate with the external medium. In this paper, we propose to exploit the UWB technology for road radar applications, in order to profit from its inherent advantages (good range resolution, low power consumption, low price etc.). All exiting studies on UWB radar concerns essentially medical and military applications. The originality of this radar is its capacity to detect easily various obstacles and to give their signatures. This paper describes the used radar and presents an interesting study of the adapted UWB waveforms which allow location and signatures of different kind of obstacles detected in short ranges with a good resolution.

UWB Radar system for road anti-collision application

The inter-vehicles communication is becoming increasingly mandatory to improve the road safety and to allow a better transport management. The concept behind an intelligent vehicle consists in equipping this vehicle with an optimum number of sensors, in order to perceive the environment surrounding it and to ensure an automatic guidance control and the communication with the infrastructure and other vehicles. Among the most often used tools for this objective, we can cite the embedded radar.

Blind Channel Equalization Using Constrained Generalized Pattern Search Optimization and Reinitialization Strategy

We propose a global convergence baud-spaced blind equalization method in this paper. This method is based on the application of both generalized pattern optimization and channel surfing reinitialization. The potentially used unimodal cost function relies on higher- order statistics, and its optimization is achieved using a pattern search algorithm. Since the convergence to the global minimum is not unconditionally warranted, we make use of channel surfing reinitialization (CSR) strategy to find the right global minimum. The proposed algorithm is analyzed, and simulation results using a severe frequency selective propagation channel are given. Detailed comparisons with constant modulus algorithm (CMA) are highlighted. The proposed algorithm performances are evaluated in terms of intersymbol interference, normalized received signal constellations, and root mean square error vector magnitude. In case of nonconstant modulus input signals, our algorithm outperforms significantly CMA algorithm with full channel surfing reinitialization strategy. However, comparable performances are obtained for constant modulus signals.

A new approach to improve mobile localisation based on angular deviation measurements in urban areas

In urban areas, localisation systems based on the use of satellites such as GPS, GLONASS or the future EGNOS may not provide a sufficient availability and have to be complemented by other systems. Our study is focused on a localisation of terrestrial mobiles in urban areas using a radiogoniometer approach or direction finding of several broadcasting stations. The errors of these angular measurements are mainly due to the multipath and masking effects of buildings and trees. This paper presents theoretical and experimental results on angular deviation of direction finding obtained in simple street configurations in two frequency bands: 100 and 900 MHz. The theoretical analysis is based on the uniform theory of diffraction and Snell-Descartes reflection laws. Two specific radiogoniometers presenting specific radiation pattern were developed. Direction finding measurements performed with these sensors are compared using those obtained with a Doppler radiogoniometer. Measurements using existing broadcasting stations in urban areas are also presented and a simple solution to improve angular measurements accuracy is suggested.