Fluid Mechanics and Acoustics Laboratory - UMR 5509

LMFA - UMR 5509
Laboratoire de Mécanique des Fluides et d’Acoustique

Supervisory authorities

Our partners

Home > News > Thesis defense > Thèses soutenues 2020

PhD defence ECL

Adrien Dagallier

Friday November 27th 2020, 15h, visioconference with registration

Adrien Dagallier

Modeling acoustic impulse arrivals for shot localization in complex environments

Composition du jury
- M. J. WHITE, PhD, US Army ERDC/CERL (Illinois), rapporteur
- J. PICAUT, Directeur de Recherche HDR, UMRAE, Nantes, rapporteur
- F. COULOUVRAT, Directeur de Recherche CNRS, Institut Jean le Rond d’Alembert, examinateur
- B. NICOLAS, Directrice de Recherche CNRS, CREATIS, examinatrice
- S. CHEINET, PhD HDR, ISL (Saint-Louis), Co-directeur de thèse
- D. JUVE, Professeur, LMFA, Directeur de thèse

Battlefield acoustics sensing systems have been used since the early 20th century for detection and localization of threats. Artillery and gun shots emit loud sounds (muzzle blast upon firing, ballistic wave emitted by the supersonic projectile, possible impact burst) which propagate at long ranges. These sounds may be recorded at low-cost, passive, all-weather, omnidirectional sensors, usually distributed over the monitored area. Sensor data are then fused, using localization algorithms and propagation models to relate observed features, e.g. times of arrival (TOAs) or spectra, to a plausible source position.
The originality of the team’s approach, through the Matching method, consists in factoring in the physics of propagation: wind and temperature effects, obstacles... A database of virtual sources acoustic features is numerically predicted at a set of sensors. Upon detection of an event, observed features are evaluated against the database. The estimated sound source position is that of the closest match. In practice, TOAs of signals at synchronous, distributed sensors are sufficient for localization of e.g., sniper shots in urban areas.
The database may be generated in advance, while the Matching is potentially real-time. Localization is robust to noise, sensor positioning, calibration, or environment data errors. However, building the database is computer-intensive, and handling of non-trivial geometries or sources is challenging. Integration of environment data, feasibility of artillery shot localization and of Matching multiple arrivals, are open questions.
The rationale of the present work is to develop a modeling suite, from procurement of terrain and atmosphere data, to shot ballistics and acoustic propagation, to compute TOAs of the acoustic emissions of supersonic shots in a consistent and physics-based fashion. Each time, limiting factors (sensor position error, atmospheric data accuracy, ballistic dispersion...) are determined, and all models are consequently refined, or simplified, to the befitting level of detail for the Matching phase. More specifically, a Fast-Marching acoustic propagation model is derived and implemented (IFM). IFM retains the physical generality of 3D+time solvers, while computing only TOAs and thus being much faster. IFM handles urban geometries with unstructured meshes, and long range propagation with terrain-following grids. Coupling to a ballistic model accounts for sound emissions of supersonic shots. Bullet hits in building façades or the ground and 3D aerodynamic effects for large caliber projectiles are considered. IFM is then coupled to computational fluid dynamics or meso-scale numerical weather prediction models to determine relevant atmospheric inputs in support or replacement of on-site measurements.
Two measurement campaigns were conducted for evaluation of the approach in built-up areas, including supersonic weapons and actual live ammunition.
Point source localization performance is state-of-the-art with down to 4 sensors. Sniper localization performs well with down to 6 sensors, including fully non-line-of-sight sensors configurations - which is to our knowledge a first for countersniper systems. Localization of artillery shots is demonstrated on the multiple arrivals of measured artillery signals, from a small baseline array, with little influence of the array geometry on the sensing performance, thanks to the accuracy of the predicted muzzle blast, ballistic wave and impact burst TOAs. Again, this is to our knowledge a first.
The modeling suite developed in this work may readily assess the performance of any synchronous, TOA-based sensing system in realistic scenarios, in arbitrarily complex, non-line-of-sight environments - with a common framework for both counter-sniper and counter-artillery systems. It could also be used as a decision aid, to choose the most fitting sensor configuration for surveillance of a given area, in a given scenario.

To obtain the connexion informaiton for the visioconference, please send an email to mailto:adrien.dagallier@isl.eu


  • Friday 27 November 2020 15:00-17:00 -

    Soutenance de thèse : Adrien Dagallier

Ajouter un événement iCal