Feature ArticleAcoustic Tomography With An Underwater Sensor Network
By Andrew Goodney • Young H. Cho
Node locations in the Marina Del Rey test bed, aerial and schematic views.
As more and more scientific and industrial research and exploration is done in and around the world’s oceans, there is an increased need to create long-term monitoring and data collection technology. Data such as temperature, salinity, oxygen concentration and current flow, among others, need to be collected and transmitted to engineers and scientists for analysis and action. For many applications, the data must be of high spatial resolution, necessitating the deployment of many sensors. When real-time collection and analysis are required, the sensors must be connected to form an underwater sensor network.
Generally, underwater sensor networks are formed using acoustic communication links due to the favorable propagation properties of an acoustic signal underwater. However, using acoustic signals for data transmission provides a unique opportunity to layer on a sensing component to the underwater sensor network. Acoustic tomography uses the travel time for acoustic signals to infer data about the properties of the water through which the signal travels, typically temperature and/or current flow. The distance between nodes in an underwater sensor network typically ranges from tens to hundreds of meters. Thus, this technique is called small-scale acoustic tomography to differentiate it from existing, large-scale ocean acoustic tomography techniques.
Performing acoustic tomography between the nodes of an underwater sensor network can be a stand-alone application for an underwater sensor network, or it can enhance the utility of such a network for many ecological or defense applications. From long-term continuous temperature monitoring of sensitive habitats to detecting the heat signatures from submarines, small-scale acoustic tomography is a new and promising underwater sensing technique.
Underwater Sensor Networks
A sensor network is a network of nodes built to perform environmental sensing. They are typically built with inexpensive, commodity parts, but by using sophisticated algorithms a sensor network can often outperform more expensive single-sited sensing equipment by combining readings across time and space. Building an underwater sensor network can be a relatively inexpensive way to collect quality data about the underwater environment. A large challenge faced when deploying underwater sensor networks is the acoustic communications equipment required. There are no inexpensive, short- to medium-range underwater modems and/or hydrophones available in the market today. Several research projects at universities and their industrial partners aim to change this, but the technology has not become widely available. Another challenge for those that deploy underwater sensor networks are the algorithms and protocols required to convey data across the network. Since the propagation properties of acoustic signals underwater differ so drastically from radio frequency (RF) in the air, sensor network algorithms and protocols must be redesigned to take into account the unique properties of underwater acoustic communications.
Underwater Acoustic Sensor Nodes
Sensor network nodes are typically built with low-cost embedded systems. However, for the deployment discussed in this article, we used a low-cost, small-form-factor PC. In doing so, we were able to rapidly prototype and test our acoustic and signal-processing algorithms.
This platform is made up of several off-the-shelf components: a small-form-factor PC 96-kilohertz audio recording and playback capabilities, hydrophones for sending and receiving acoustic signals, and a GPS receiver with GPS pulse-per-second signal.
The current cost of such a setup is approximately $500. However, a large portion of that cost is the hydrophone. Unfortunately, this is a limiting factor to wider deployment of underwater sensor networks until hydrophone manufacturers develop and market a sub-$20 hydrophone capable of sending and receiving signals at distances on the order of 1 kilometer. For this deployment, we used Aquarian Audio Products (Anacortes, Washington) hydrophones for receiving and Benthowave Instrument Inc. (Collingwood, Canada) hydrophones for transmit.
Using a general purpose PC allows for rapid development and experimentation in underwater signal processing and underwater sensor network algorithms. The CPU can be used for digital signal-like processing, without the difficulty of the digital signal processing (DSP) or field-programmable gate array (FPGA) development cycle. Once successfully prototyped, these algorithms can be ported to an embedded DSP or FPGA.
The built-in sound card provides the ability to collect high-quality acoustic data from the water. If the final application requires ultrasonic signals, the system can be prototyped and tested at lower frequencies before developing and testing specialized hardware that supports such frequencies.
Our prototype system uses receive and transmit hydrophones, instead of a more typical switched TX/RX (transmitting/receiving) single hydrophone. Doing so has several advantages during research and development deployments. We can loop back the local acoustic signal with the receive hydrophone to monitor the quality and timing of a transmitted signal.
Providing a GPS receiver with pulse per second allows the operating system clock on the small-form-factor PC to be synchronized globally with high accuracy, as well as providing a way to time-stamp acoustic signals precisely. These features are required to properly time-stamp collected environmental data and perform acoustic tomography. To continue this article please click here.
Andrew Goodney is a graduate research assistant at the University of Southern California’s Information Sciences Institute. His research interests are sensing and sensor networks, health and wellness technology (mobile), motion picture/VFX/multimedia technology, computer network interconnects and performance, software-defined networking, SAN disk architectures, distributed and cloud computing, database performance and scaling, and computer science/electrical engineering education.
Young H. Cho is an electrical engineer/computer scientist at the University of Southern California’s (USC) Information Sciences Institute (ISI). In 2008, he started his position as a research scientist at ISI to continue his research in computer networks, wireless sensor networks and other distributed computing systems. He also has a joint appointment as research assistant professor in the Department of Computer Science and the Department of Electrical Engineering at USC.