Advancements in oceanic research have resulted in a plethora of activities such as undersea oil/gas exploration, environmental monitoring, SONAR-based coastal surveillance, which all have increased the acoustic noise levels in the ocean, raising concerns in the scientific community. Knowledge about the statistical characteristics of noise sources and their spatial distribution is important for understanding the impact on marine life as well as for regulating and policing such activities. Furthermore, as studies have shown, assuming the underwater noise to be Gaussian is seldom valid; hence, online profiling of the sources forming the ambient noise is also essential to increase the performance of acoustic communication systems in the harsh underwater environment. In this paper, real-time separation of underwater acoustic noise sources via Blind Source Separation (BSS) in the presence of various degrees of multipath as well as their classification based on the coherence of their Power Spectral Density (PSD) and the PSD of known noise sources are studied via simulations. Work is currently being conducted to validate the results and to localize the sources using real data collected from underwater communication testbeds.