|Field of activity:||Multidimensional Signals and Systems Theory|
|Research topic:||Multidimensional Signals and Systems|
In the context of the SCENIC project we have investigated the joint solution of two problems with strong relevance in multidimensional signal theory and array signal processing, i.e.,
Although these two problems are inherently connected, in most cases they are addressed separately by a two-step approach:
This two-step approach is not always favorable as in some applications the characteristics of the propagation medium might be unknown and additional invasive measurements undesired.
Thus, techniques have been proposed in order to solve both problems jointly by relying merely on the time delays observed at the sensor array, i.e., Time Of Arrival (TOA) or, in absence of source-receiver synchronization, Time Difference Of Arrivals (TDOA) .
In the following we give a brief description of the major results and achievements yielded by our research:
The dependence of the speed of sound on the air temperature is well known in acoustics.
TDOA-based acoustic source localization requires accurate knowledge about speed of sound in order to convert time differences into range differences related by geometric constraints.
In uncontrolled scenarios, e.g., due to seasonal temperature variations or temperature variations in a room caused by waste heat from equipment or presence of people, the standard value for the speed of sound at 20 °C degrees might lead to unsatisfactory localization results.
In this case the deployment of extra temperature sensors is not always favorable and therefore the direct estimation of the actual speed of sound from TDOAs becomes a crucial part of the localization algorithm.
LMS has studied the effects of temperature variations on the localization accuracy.
It turned out that undesired temperature variations yield an error in the source distance estimation. The magnitude of the error depends on the array geometry as well as on the localization algorithm; the pictures below show a square microphone array and the corresponding 2D error distribution in cm.
In order to compensate for this error we proposed an efficient TDOA-based speed estimation algorithm that enables high localization accuracy even if temperature variations occur.
The effectiveness of the proposed speed estimation algorithm has been shown by inferring the average air temperature of a room from the so-obtained speed of sound estimates. A demonstration of this technique was given during the Industry Day organized in Milan in the context of the SCENIC project.
Speed of sound estimates were also used in combination with a reflector localization technique developed by the Speech and Audio Processing Group of the Imperial College (London). The goal was to perform accurate room geometry inference regardless of the actual air temperature. The following pictures show the block diagram of the proposed approach and the obtained results for real measurements carried out in our laboratory.
||Ladan Zamaninezhad, P. Annibale, R. Rabenstein
Localization of Environmental Reflectors from a Single Measured Transfer Function
IEEE Int. Symp. Communications, Control, and Signal Processing (ISCCSP), Pages: 157-160, Athens, Greece, May 2014
||P. Annibale, R. Rabenstein
Closed-Form Estimation of the Speed of Propagating Waves from Time Measurements
Springer Journal on Multidimensional Systems and Signal Processing (MDSSP) Vol. 25, Num. 2, Pages: 361-378, 2014
||P. Annibale, R. Rabenstein, Martin Kreissig, Bin Yang
Joint consistent graph synthesis and speed of sound estimation for acoustic localization in multi-source reverberant environments
8th International Workshop on Multidimensional Systems (nDS 2013), Pages: 1-6, Erlangen, Germany, Sep. 2013
||P. Annibale, J. Filos, P. A. Naylor, R. Rabenstein
TDOA-based Speed of Sound Estimation for Air Temperature and Room Geometry Inference
IEEE Transactions on Audio, Speech and Language Processing (IEEE TASLP) Vol. 21, Num. 2, Pages: 234 - 246, Feb. 2013