Zurück zur Liste

Self-Localization of Large Aperture Microphone Arrays

AutorInnen Wilding, T.
Jahr 2015
Art der Arbeit Toningenieur-Projekt
Abstract When using microphone arrays in for example hands-free telephoning, they are usually rather small, due to the space reserved for the application. Thus it is easy to find the positions of the used microphone with the accuracy needed for the application when measuring by hand. When using a large scale microphone array, it can be assumed that measuring the positions of each microphone can be a complicated and time consuming procedure for a single person to undertake, either using a tape-measure or an equivalent device. In some applications, the microphones or sensors may even change their positions over time, which would include the time as another degree of freedom, which will not be treated in this project. Furthermore, microphone arrays are usually deployed in different environments, which also makes it important to know the locations of the microphones. The whole process could be greatly simplified, if there existed a simple method which could achieve the same, or at least sufficiently accurate results. If not measured by hand, usually the time-of-flight (TOF) from the sources to the sensors (I.e. microphones), which can be reformulated as a cost function and solved using mathematical solvers, is used for finding the distances between microphones and sources. The problem that arises here is, that if the problem is not initialized carefully, it can easily get stuck in a local, rather than the intended global minimum. The initialization problem is usually solved by introducing additional constraints, such as using sub-arrays which have a certain geometrical structure (linear,...) and reduce the number of unknowns, or by positioning all sources on a plane, or using a pyramid shaped structure for the sources surrounding the sensors. When using subarrays, another possibility is to separate the computation process and estimate the shape of the subarrays in the first step, and then use the results obtained by these subarrays to estimate the positions of the sources.
URL http://phaidra.kug.ac.at/o:16233
BetreuerInnen Schörkhuber, C.