Adaptive Localized Curve Tracing Methodology for Efficient Transmission of Audio Streams over Noisy ATM Networks

Abstract

To combat channel impairments during transmission of high fidelity audio streams over noisy ATM networks in addition to providing security and high compression to the audio data have been challenging tasks for the researchers in the field. This paper proposes a multi functionality model employed at the transmitting end that provides the advantages of multi level compression, error recovery mechanism and the security for the audio data in ATM networks. The proposed model introduces the modulo coefficient booster transform (MCBT) that translates the relevant frequency domain coefficients of the audio stream filtered from a human acoustic modeler into strictly increasing/decreasing booster coefficients (employing coefficient localization, burrow wheeler transform and mathematical curve tracing). The booster coefficients obtained are fit to corresponding modulo mathematical functions, which are encoded and transmitted after incorporating security and error recovery mechanisms. Employing the PRATM model to provide error recovery for the transmitted data in the noisy channel provides reliability to the data. The proposed transform can be completely invertible (being a function of set lapse time and error deviation threshold) and provides additional security by means of localizations of the functions generated based on hamming distance computations. The model was tested for various bit error rates of high fidelity audio streams simulated in noisy network environments. The PSNRs , security and the overall reliability of the proposed model encourage the model to be preferred for transmissions of audio streams over noisy ATM networks