Based OMAP3530 Microphone Array Life Detector Design And Implementation

Life detection is the key point in earthquake rescue, which is a difficult one because of the complicated environment after earthquake. The traditional life detection equipments are restricted by their sensors and their costs so they can not been widely used. A new type of life detection equipment is badly needed. One of the development directions is considering the voice location in search since voice signal is one of the most obvious feature of life. This life-detector excels in its small size, low power consumption, little cost and sensor free. It can detect source correctly in real time and will be widely used in near future.The localization algorithm and hardware platform are discussed, the software implementation of algorithm on this hardware platform is especially detailed. The voice signal is received by a cross microphone array. The method is based on near-field microphone speech propagation model. Modified MUSIC algorithm and TI OMAP3530 dual core processor are adopted as the localization method and hardware platform in this system.This theme focuses on the software implementation.It is implemented on OMAP3530EVM platform, which consists mainly of three parts, the localization program on DSP core, the management thread program on ARM core and the graphic user interface thread on ARM core.By means of tasks assignment, DSP core focuses on signal processing and ARM core focuses on system level management. The comm optimization unication between two cores'software is built by TI Codec Engine, which is based on remote progress call principle. In this framework, the DSP software must be programmed according to XDM standard then it can provide APIs interface to user layer through intermediate layer, finally the user program can simply call the VISA APIs to locate the source. The uppermost layer of this software is the graphic user interface programmed by QT/Embedded.The dual core framework, XDM standard, rules of Codec Engine call and QT call, software optimization and trade-offs are detailed in this theme. The testing results in realtime speech environment has showed that the life detection equipment exhibits a stable status, and it can get the correct source position in real time.The drawbacks of this system and their causes have been explained in the theme.And it also proposes the corresponding methods to improve this system.