LONG BASELINE ACOUSTIC POSITIONING WITH COMPENSATED TIME-VARYING CLOCK-OFFSET AND PSEUDORANGE ESTIMATION BASED ON A RAYTRACING MODEL AS A REFERENCE FOR INERTIAL NAVIGATION SYSTEM IN A SHALLOW WATER

• Main Author: S M Simamora, Yohannes
• Format: Dissertations
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/64659
• Institution: Institut Teknologi Bandung

This research was motivated by the importance of building under water navigation systems in a shallow water. The availability of such navigation system is fundamentally required for underwater vessels including autonomous underwater vehicles (AUVs). Similar to the case in terrestrial and aerial applications, pairing inertial navigation system (INS) with a reference system in underwater applications is necessary. This is due to an inherent disadvantage of an INS, i.e. to accumulate errors over time. The availability of a position reference would enable an INS to correct and compensate its errors. For terrestrial and aerial applications, a position reference is commonly provided by a satellite-based navigation system such as the global positioning system (GPS). The GPS works based on time-of-flight (ToF) measurement principle, i.e. range measurement based on the time required by an electromagnetic wave to travel from one point to another. Based on this principle, a navigation target is able to estimate its own position based on its ranges towards the GPS satellites that are involved in the ToFs. The measurement principle also applies in range measurement using underwater acoustic waves. However, as an electromagnetic based system, the GPS is incapable of providing references in the water. This is because the waves would suffer from rapid attenuation in the medium. Therefore, an INS/GPS scheme could not be applied for underwater application such as for AUV navigation. As an alternative solution, AUV may use an integrating the INS with a long baseline positioning system or LBL system. This INS/LBL scheme is worth to consider since a LBL system is capable of providing navigation with good accuracy. Besides, a LBL system shares similarities with the GPS in terms of configuration and working principles. Therefore, the trilateration principles can also be applied in LBL Nonetheless, there are several sources of uncertainties in LBL positioning that would affect the navigation accuracy. First one is the target motion during the positioning. Since the propagation speed of underwater acoustic wave is considerable low (? 1500 m/s), it is probable that the target has moved from the position in question. In a similar situation, GPS could assume that the target does not move as electromagnetic wave propagates in a very high speed, i.e. 3×108 m/s. Second uncertainty is the clock-offset between a LBL transponder and the target. This discrepancy results in a biased traveling time computation in the ToF measurements. This situation becomes a time-varying case when the LBL has been operated for a considerable time. In this case, the transponder clock would drift from the actual clock due to aging and operating environment. Recalling that the wave speed is the multiplying factor in a ToF measurement, the presence of clockoffset in a milisecond would produce range bias in meters. Since LBL positioning is based on range of the target against each LBL transponder, the pseudorange, i.e. range plus bias, would be introduced to the position estimation. Third uncertainty is the wave trajectory, i.e. the raytrace in a ToF measurement that is not necessarily a straight line (line-of-sight–LoS). When a ToF measurement does not meet the LoS condition, the distance traveled by the wave is not equivalent to the distance between the transponder and the target. There are two type non-Los in ToF measurement: multipath and bending raytrace. The multipath phenomenon is also a problem found in a GPS application. On the other hand, a bending raytrace would only the case in underwater acoustic based applications such as LBL system. This is because the propagation speed of underwater acoustic wave is a function of depth, temperature, and salinity. Therefore, the speed may vary in accordance with a certain sound speed profile (SSP) under the guidance of the Snell’s law. In a shallow water application, there is certain depth where it is not adequate to approach propagation speed of the underwater acoustic wave neither as a constant nor as an isogradient SSP. This research addresses uncertainties in LBL positioning. The LBL acts as a reference in INS correction and compensation mechanism. The reasearch’s main contributions are the compensation of time-varying clock-offset and pseudorange estimation between an AUV a transponder based on raytracing model between the nodes. Specifically, the dynamics of time-varying clock-offset is modeled as an autoregressive (AR) and is incorporated to the kinematics model of the AUV. On the other hand, pseudorange estimation is yielded by exploiting the Snell parameter that exists in both pair of the incremental raytracing equations, i.e. propagation time and horizontal range (from the source). In this case, Snell parameter is estimated for each propagation time, i.e. the ToF. Its estimated value is then applied to compute the horizontal range between the transponder and the AUV. Furthermore, horizontal pseudoranges of the AUV towards each LBL transponder are used to estimate the AUV position. The estimation uses trilateration by finding least square solution for the collected horizontal pseudoranges. Finally, position provided by the LBL is used as the reference for INS correction and compensation. For several scenarios, it was shown by simulation that the LBL is capable to compensate the time-varing clock-offset while provides position accuracy around 0.5×0.5×0.5 m3. On the other hand, the INS is capable of correcting and compensating its errors with good accordance with the reference from the LBL. At the final part of the research, the performance of INS/LBL was obtained by conducting posterior Cramér-Rao Bound (PCRB) on the proposed LBL system. It was shown that the LBL system manages to achieve the aforementioned performance while the estimator variances submit to the PCRB. positioning.