Sonar generated acoustic signals transmitted in underwater channel for distant communications are affected by numerous factors like ambient noise, making them nonlinear and non-stationary in nature. In recent years, the application of Empirical Mode Decomposition(EMD) technique to analyze nonlinear and non-stationary signals has gained much attention. It is an empirical approach to decompose a signal into a set of oscillatory modes known as intrinsic mode functions(IMFs). In general, Hilbert transform is used in EMD for the identification of oscillatory signals. In this paper a new EMD algorithm is proposed using FFT to identify and extract the acoustic signals available in the underwater channel that are corrupted due to various ambient noises over a range of 100 Hz to 10 kHz in a shallow water region. Data for analysis are collected at a depth of 5 m and 10 m offshore Chennai at the Bay of Bengal. The algorithm is validated for different sets of known and unknown reference signals. It is observed that the proposed EMD algorithm identifies and extracts the reference signals against various ambient noises. Significant SNR improvement is also achieved for underwater acoustic signals. Sonar generated acoustic signals transmitted in underwater channels for distant communications are affected by numerous factors like ambient noise, making them nonlinear and non-stationary in nature. In recent years, the application of Empirical Mode Decomposition (EMD) technique to analyze nonlinear and non- stationary signals hasfted much attention. It is an empirical approach to decompose a signal into a set of oscillatory modes known as intrinsic mode functions (IMFs). In general, Hilbert transform is used in EMD for the identification of oscillatory signals. a new EMD algorithm is proposed using FFT to identify and extract the acoustic signals available in the underwater channel that are corrupted due to various ambient noises over a range of 100 Hz to 10 kHz in a shallow water region. Data for analysis are collected at a depth of 5 m and 10 m offshore Chennai at the Bay of Bengal. The algorithm is validated for different sets of known and unknown reference signals. I t is observed that the proposed EMD algorithm identifies and extracts the reference signals against various ambient noises. Significant SNR improvement is also achieved for underwater acoustic signals.