Protection technologies, security, situational awareness
Acoustic sensing systems, in the form of sensor arrays, are used for surveillance purposes in underwater and aerial applications. At the ranges of interest to sensing, the turbulence-induced fluctuations (in the ocean or in the atmosphere) tend to decorrelate the signatures collected at the sensors of the array, which is detrimental to the sensing performance. Such propagation-induced modulations must be accounted for in order to design more robust and efficient sensing approaches.
The ISL team on “sound propagation & sensing in complex environments” conducts a long-term research effort on the aerial aspect of the issue. Among others, it has developed state-of-the-art tools (numerical models, experimental databases) in order to document the atmospheric modulations of propagated sounds.
In parallel, the ENSTA Bretagne team on “signal and image for the observation systems” is interested in the design of algorithms for acoustic source localization in fluctuating oceans.
Recently, a partnership has been built on these topics between ISL and ENSTA Bretagne, gaining from the cross-over between aerial and underwater researches on the topic.
The proposed PhD study is within this partnership. It is to investigate the sensing methods capable of dealing with the signature disturbances induced by turbulence, focusing on narrow-band signals such as motor noises. The envisioned program includes two major steps:
- Analysis of sounds propagated through turbulence. The purpose of this step is to construct a robust, documented a-priori knowledge on the signal’s fluctuations caused by (oceanic and aerial) turbulence. Existing experimental databases of the ISL and of the French DGA may be used. Experimental or numerical investigations may also be conducted at ISL. An attempt will be made to address the question: to what extent are the perturbations of the acoustic signatures comparable in the air and in the ocean?
- Investigate sensing approaches capable of dealing with the signature disturbances induced by turbulence. The aim is here to design robust algorithms for localizing acoustic sources in presence of such disturbances. The a-priori knowledge obtained and validated in 1 will be accounted for through a Bayesian modelling, in line with some recent works conducted at ENSTA Bretagne. As a second step, optimization of the antenna geometry (and sensor placement) may also be investigated, inspired by theoretical contributions.
Localization: ISL (Institut franco-allemand de recherches de SaintLouis, Alsace). A period at ENSTA Bretagne (in Brest, for 6-12 months) is to be planned.