WIDE-BAND CHANNEL MODEL FOR UNDERWATER ACOUSTIC COMMUNICATION AND ITS APPLICATION TO KALMAN FILTER BASED CHANNEL ESTIMATOR

Abstract

Underwater acoustics communication system design is a challenging endeavor due to several technical as well as environmental parameters such as salinity of water, temperature, frequency and time selectivity of the channel, different sea bed surfaces, Doppler effect and more. The task of a channel model is to provide a software that determines the channel behavior without a need for practical deployment. In this thesis, unlike the already existing narrow-band communication channel models, a model is designed to take into account variation of the channel impulse response as a function of communication signal frequency. The proposed model utilizes a conceptually simple sum-of-sinusoids approach and incorporates frequency-dependency into the model output and is suitable for modelling multiple-input multiple-output (MIMO) channels. The model addresses multi-path fading effect and the effect of Doppler shift dependence on the signal angles of arrival. The capabilities of the proposed model are tested by comparing with data observed during sea-trials. In addition, an OFDM based underwater transmission systems with iterative data decoding and channel estimation is used on a narrow-band and the proposed wide-band model. It is demonstrated that auto-regression channel models used to operate channel estimators need to be adjusted to address frequency dependence showcased by the wide-band channel model.