Kalman-based underwater tracking with unknown effective sound velocity

In underwater tracking based on range measurements, the transit time between a pinger and a receiver is often measured, then it is converted into a slant range by multiplying an effective sound velocity (ESV). Clearly, the accuracy of the calculated slant range depends largely on the accuracy of the ESV. Even with a small percentage ESV error, the resulting range measurement can have significant error particularly when the transit time is large. The ESV between a pinger and a receiver certainly is location dependent, and is very difficult to determine accurately because the ESV would be affected by many factors, such as water temperature, pressure and salinity. Even with a very good knowledge of the sound velocity profile (SVP), the calculated ESV based on the ray tracing theory, or an equivalent, might still involve non-negligible error. However, a common assumption adopted by many published papers related to Kalman-based underwater tracking is that the ESV is not only a known quantity, but also location independent. This paper proposes a Kalman-based underwater tracking model that treats the location-dependent ESV as a state variable. Instead of using the calculated range as measurement, it employs the transient time directly as measurement. Both simulation and filed data have been used to demonstrate the efficiency and superiority of the proposed method. It has been shown that while using the proposed model and approach, the tracking accuracy can be significantly improved and the estimated ESV agrees well with its true value