Sea Technology

JUL 2013

The industry's recognized authority for design, engineering and application of equipment and services in the global ocean community

Issue link:

Contents of this Issue


Page 35 of 71

(Left) Test bed temperature map after three hours of heating. (Right) Test bed temperature map after six hours of heating. we can measure exactly when a signal is transmitted into the water. Similarly, at a receiving node, we can measure exactly when it arrives. By subtracting these globally accurate (GPS-referenced) time stamps, we can determine the time of fight (and, thus, speed of sound). An acoustic tomography system that uses GPS is only possible at or near the surface. However, many deployment scenarios for acoustic tomography will be deep underwater where GPS is not available. To address this need, we have also developed and tested a timing method that does not rely on GPS, yet obtains similar precision to our GPS-based timing technique. For this technique to work, we must assume we know the node locations. Then, we essentially perform acoustic ranging at high precision. However, since we know the node distances, we can invert to obtain speed of sound. Signal Processing An acoustic tomography system needs to send an acoustic signal through the water. Design or selection of the acoustic signal is a critical step. The arrival time of the signal at a receiver needs to be determined with high accuracy and precision in the presence of a noisy underwater acoustic channel. We can invert to obtain time-of-fight. Also, it is desirable that the signal itself can be used to identify the sending node, as well as carry other data that might be of use to the sensor network. When looking for such a signal, one can use an existing acoustic signal such as is used by commercial modems, or a custom application specifc signal can be developed. We found that existing modem signals are designed with other goals in mind (e.g., high bandwidth, noise immunity), which do not allow for highly precise arrival-time detection. Thus, we have developed a custom acoustic signal that can carry low bandwidth data (e.g., node identifcation), while also allowing for good noise immunity with highly precise arrival time detection. We start with binary phase shift keying modulated pseudorandom bit sequences, also known as chirps. We group these chirps into classes or families that have good auto (ACF) and cross (CCF) correlation properties, which allow them to be distinguished from each other (low CCF), while also allowing for precise arrival time detection (narrow peak in the ACF). Pseudonoise-based chirps are also resilient in the presence of multipath. Such classes also form the basis for code division multiple access systems in RF. Using long sequences (thousands of bits in length) creates spreading gain when correlating at the receiver to detect an incoming signal. Longer signals have more spreading gain and can therefore be detected even in the case of a very high noise channel. Thus, we need a fexible way to generate classes of long pseudonoise sequences. We use a method based on Legendre sequences that creates sequences of length P = 3 mod 4, where P is a prime number. This construction allows for essentially arbitrary chirp length, with class size of (P−1)/2. We can then assign the (P−1)/2 codes to the nodes in such a way that they can transmit signals that identify the node, along with carrying small amounts of data. Marina Test Bed We developed a remotely accessible underwater sensor network research test bed at the Information Sciences Institute (ISI) in Marina Del Rey, California. ISI is part of the Viterbi School of Engineering at the University of Southern Cali- 36 st / July 2013

Articles in this issue

Links on this page

Archives of this issue

view archives of Sea Technology - JUL 2013