Sound Source Localization Algorithm Based On Signal Power Correction

Sound source localization technology is originally used to locate artillery firingposition and point of impact in the military. With the progress of science andtechnology, sound source localization technology is widely used and developedrapidly both in the civil and military. In civilian areas, it is an important componentof the applications of video conferencing, voice recognition and intelligence robots,etc. In military areas, it is the primary means of investigating sensitive military area,fired detection, wireless sensor network target detection in enemy-occupied area andcounter-terrorism operations and so on. Therefore, researching for faster, universal,more accurate and robust localization algorithm has very important theoretical andpractical significance for studying and realizing many kinds of applications.Industrial design and manufacturing capabilities have big progress with the fastdevelopment of science and technology. As a result of these improvements, soundsource localization technology develops in accordance with two different waysnamely bionics and microphone array methods.Because of the principle of bionics based sound source localization technologyis very complicated, and its realization requires higher performances of devicesmanufacturing and sensors, the implement faces many difficulties, therefore,although this localization method is very promising, its development is relativelyslow.Due to its relatively simple localization principle and the requirements ofequipment are not very high, it is more convenient to implement, and themicrophone array based sound source localization technology became the mainmethod in locating sound source.The microphone array based sound source localization technologies can bedivided into the following three categories by principle: one is the steerable beamforming method based on the maximum output power, one is the high resolutionspectrum estimation based method, and one is TDOA (Time Difference of Arrival)based method. Although the steerable beam forming method based on the maximum outputpower and the high resolution spectrum estimation based method have superiorperformances in some applications, they need much more calculations, too, and arenot fit for small-scale positioning system; TDOA based localization method issimple in principle and need less calculation, but because of that its accuracy relyheavily on the preset sound speed of the environment and require more accurateTDOA measurement, its portability isn’t very strong, and not suitable for thewireless sensor networks applications whose time synchronization can not beguaranteed.Another kind of sound source localization algorithm based on ILD (Inter-auralLevel Difference) is proposed to locate sound source. According to the inversesquare law of energy propagation of acoustic signal, this algorithm using the energyratio information of the same period of signal in a pair of microphones, combinedwith the relative position information of the microphone pair to locate the soundsource. Because of that it doesn’t use voice speed information or require strictsynchronization between microphones, its portability is strong and can be used inmuch more locating situations. However, these algorithms do not deal with noisesignal properly while calculating signal power, and make them only suitable for thespecial situation of White Gaussian Noise, but isn’t robust to the general backgroundnoise.To overcome the shortcomings of the localization algorithms above, a noiserobust sound source localization algorithm based on the signal energy correction isproposed. To locate the sound source, the first step is to estimate the parameters ofthe background noise, then using these estimated parameters to correct the signalenergy that the microphone receives and get the position relationship between thesound source and microphone array with the corrected signal power ratio.Using no speed or time delay information enhances the portability of thealgorithm, and energy correction method reduces the impact of general backgroundnoise on the locating accuracy. Theoretical derivation and experiment resultsverified its validity and robustness to the general noise. Other three work was done in this paper by studying of sound sourcelocalization algorithm based on the signal energy correction as follows:First, a parallel weak signal detection algorithm based on envelope analysis isproposed to detect the target signal. This algorithm overcomes the shortcomings ofexisting algorithms that can only detect whether an known periodical signal hasappeared or not, and can be used in many application conditions.The second is to use the “Two-step Signal Endpoint Location Method” todetermine the corresponding position of the same period signal that received bydifferent microphones. This method can reduce the positioning error that may beraised by common used sliding window method.The third work is to give a theoretical explanation of the phenomenon that,using microphone array to locate the near-field sound source is more valid than tolocate the far-field sound source. As an analysis result, far and near field conversionmethod is proposed and verified by experiments, which method can expand theapplication scope of microphone array based sound source localization.