Pseudo-Noise Based Time of Arrival Estimation for Underwater Acoustic Sensor Localization Huai Huang, Yahong Rosa Zheng, Weimin Duan Dept. of Electrical & Computer Engineering, Missouri University of Science & Technology, Rolla, MO 65409 Abstract—This paper compares the performance of single pseudo-noise (PN) and dual PN (DPN) sequences for time of ar- rival (ToA) estimation in underwater acoustic (UWA) localization. The single PN scheme uses the correlation of a local PN sequence and the received PN signal to estimate the ToA. The DPN scheme utilizes the cross correlation of the two received PN segments in one signal frame to calculate the ToA. Both simulation and field test results show that the DPN design outperforms the single PN scheme, as the DPN scheme is robust to the severe underwater acoustic channel dispersion and the high carrier frequency offset (CFO) in low-cost hardware systems where the atomic clock is unavailable. I. INTRODUCTION Underwater wireless sensor networks (UWSN) have found important applications in ocean exploration, critical structure monitoring, coastal surveillance, and disaster mitigation. In these applications, sensing information is often tagged with time and locations that can be used for tracking nodes and coordinating motion [1]. For example, bridge scour monitoring is made possible with “smart rocks” that are equipped with acoustic transceivers and sensors, and anchor nodes near the river banks utilize the acoustic communication signals to locate the sensor nodes in water [2]. The exiting localization schemes in underwater wireless sensor networks are usually classified into two categories: range-based schemes and range-free schemes [3]. Since range- free schemes can only obtain coarse localization, range-based schemes are widely used. In the range-based approaches, the distance is measured by several schemes: received signal strength indicator (RSSI), time difference of arrival (TDoA) and time of arrival (ToA) [4]. Most range-based localization schemes use ToA or TDoA due to the slow sound propagation in underwater ( 1500 m/s) and the ToA and TDoA schemes can achieve better accuracy than the RSSI schemes [3]. In this paper, we investigate low cost ToA estimation methods using pseudo noise (PN) sequences. We identify two challenges when these PN based ToA estimation methods are applied in practical underwater localization system. First, the ToA estimation is sensitive to the dispersion in un- derwater acoustic (UWA) channels. Typically, UWA channels exhibit severe multipath fading and Doppler spread that arises from nodes or water motion [5]. For example, the multipath delay in the shallow water channel is typically at the level of several tens milliseconds. Moreover, the direct path may not exhibit the strongest energy, which results in the ambiguity for the arrival time estimation. The experimental results in [2] have shown that ToA estimations fluctuated due to the severe multipath spread in the UWA channels. Second, the high carrier frequency offset (CFO) in the low cost hardware system may greatly lower the accuracy of PN based ToA estimation. The single PN based design was initially implemented on a DSP platform with high performance piezo-electrical crystal oscillator [2]. To further lower the hardware cost and power consumption, we inves- tigate a new PN based localization system with cheap and low power consumption MCU (Micro-controller Unit) and oscillator. We identify a major challenge in the low cost ToA estimation design: the cheap oscillator in the low cost design has unstable carrier frequency, which results in high carrier frequency offset (CFO), up to 1800 PPM (Parts Per Million). Through simulation and field test, we find that the high CFO greatly degrades the accuracy of the traditional PN based ToA estimation. We evaluate two PN based schemes for the ToA estimation under these two challenges: dual PN (DPN) scheme and single PN scheme. The DPN signal frame consists of two identical PN sequences separated by some gaps. We calculate the cross correlation of the two PN segments in the received DPN frame, rather than the correlation of the received single PN and a local PN, which is used in the single PN scheme. Therefore, the dual PN based approach has two advantages over the single PN method. First, the cross correlation operation in the dual PN method could focus the multipath signal, which is capable of combatting the multipath effect in the UWA channels. Second, since the cross correlation is operated with two segments of the received PN signals, the high CFO has no influence on the ToA estimation accuracy in the low cost transceiver system where cheap system clocks are used. Both simulation and field test show that the dual PN based localization scheme is robust in the low cost localization design and in UWA channels. II. PN BASED TOA ESTIMATION SCHEME We consider two PN based approaches in ToA estimation: single PN scheme and DPN scheme. Both single PN and DPN schemes estimate the ToA by detecting the peak index of the correlation output. In this section, we briefly review these two PN based ToA estimation methods. A. Single PN Scheme The transmitted signal in the single PN scheme is depicted in Fig. 1. A PN sequence of length N S1 is added before