استخدام الحوسبة العنقودية في شبكات الحساسات اللاسلكية من أجل تجميع الإشارة الصوتية

Abstract

Wireless sensor networks do not rely on infrastructure compared to other wired network models, which makes them suffer from limited storage, processing and communication resources.  We used hierarchical structure in WSN to improve its performance, special for multimedia applications that face many obstacles in terms of high computational density and requirements for excessive consumption of limited network resources, so it was necessary to develop appropriate technologies to overcome this.

In our research, clustering computing is applying to the multi-layer hierarchical structure in WSN with the aim of expanding the range of samples of the received audio signal. Then grouping them more comprehensively based on the spatial variation of the receiving points of the source signal, taking into account the fair distribution of computational tasks among the slave nodes and the leader node (Cluster head).

Based on the appropriate design of the filter bank of two branches (in the case of three nodes). One for passing low frequencies and the other for passing high frequencies. The reduction in the sampling rate in each branch of the filter bank (using Noble identity which is used in multirate signal processing) in order to reduce the consumed bandwidth within the framework of applications, for classifying the audio signal into a speech signal and a music signal.  The complete reconstruction of the audio signal is achieving at the leader of the computing cluster before applying the classification using appropriate algorithms.