Lieu : Arpajon, 91297 · Contrat : CDD · Rémunération : A négocier
Thesis subject: Detection, separation and localization of infrasound sources
Research field: signal processing, acoustic, antenna, detection, source separation, learning method, neural network.
Context: CEA-DAM is a research center which exploits and analyses the infrasound data in view of the development of the International Monitoring System (IMS) being set up by the Comprehensive Test-Ban Treaty (CTBT). So far, correlation-based method is used by the CTBT organization to analyze the signals recorded within the different stations of the IMS. The measurement of the angle of arrival and propagation velocity of coherent infrasound sources is achieved via propagation delays estimation (between different sensors) and by using an appropriate time frequency grid. In the frequency band of interest, the real data analysis within the current operational system has shown the existence of several interfering (non-desired) signals which have to be taken into account. It is observed that such interference signals can lead to erroneous detections and inaccurate estimations. The used detection method has been essentially developed to detect and localize a single coherent source signal within a given time-frequency cell. In order to overcome this limitation, it is needed to consider source separation algorithms to get rid or mitigate the impact of the interfering signals.
Thesis objectives: The main objective of this thesis is to elaborate new detection and localization methods, to validate and test them on both controlled real and simulated data. More precisely we will investigate the following items:
Consider high resolution methods like MUSIC (Multiple SIgnal Classification) to detect multiple narrow-band sources within the same time frequency cell.
Develop detection and estimation methods for the mitigation of spatially distributed wideband sources.
Develop statistical criteria for the estimation of the number of sources via a penalized maximum likelihood approach (such as AIC, BIC, MDL).
Consider classification methods based on learning approaches, such as neural network, to improve broadband DOA estimation.
For the performance assessment and validation of the different methods under investigation, we will develop and enrich databases of controlled real-life data or synthetic data that would be representative of the genuine conditions and different scenarios for infrasound source detection. In particular, these data should well represent the diversity and variability of the signals and noise/interference sources as well as different array configurations.
The outcome is to improve operational infrasound monitoring procedures by elaborating a cost-effective high-resolution detection algorithm, in order to characterize sources of interest in noisy measurements in the presence of interfering signals.
Master 2, Ingénieur