Optical and Acoustical Underwater Sensor Network
Iuliu Vasilescu, Keith Kotay & Daniela Rus
The application of wireless sensor networks to the underwater domain has huge potential for monitoring the health of river and marine environments. The oceans alone cover 70% of our planet and along with rivers and lakes are critical to our well-being. Monitoring these environments is difficult and costly for humans: divers are regulated in the hours and depths at which they can work, and require a boat on the surface that is costly to operate and subject to weather conditions. A sensor network deployed underwater could monitor physical variables such as water temperature and pressure as well as variables such as conductivity, turbidity and certain pollutants. The network could track plumes of silt due to dredging operations or pollutants flowing in from land, and it could monitor and model the behavior of underwater ecosystems.
However a number of problems confront us in achieving this goal. Some such as power efficiency, deployment and repair are common to wireless sensor network deployments on land, though perhaps more difficult in the underwater environment. Other issues, however, render the problem radically different. A key issue is communications. Current terrestrial wireless sensor network applications to date has used radio. At all frequencies radio waves are attenuated so strongly in salt water that radio communications is impractical. From the electromagnetic spectrum the visible light is much less attenuated especially at blue-green side of the spectrum. Acoustic communication is the other feasible method of communication in underwater environment. While the data rate of optical communications can be significant the range is limited by the the water turbidity and attenuation to meters.
Our sensor network contains static and mobile nodes. All nodes contain flash memory for Each static nodes is build around a CPU unit developed by CSIRO called a Fleck, based on the Atmega128 processor, with 128kbyte of program flash memory, 4kbyte of RAM, and 512kbyte of flash memory for data logging/storage. The Fleck is interfaced to a special optical communications board through 2 digital IO pins. One of these pins is used to turn an LED on or off, while the other is used to sense the output from a matched photodiode. All the analog electronics (e.g., amplifiers etc) are on the interface board. The Fleck is also interfaced with a sensor board. All these boards are connected together in a stack using stack-through connectors. The underwater sensor node is contained in a yellow watertight Otter box which measures 170 x 100 x 90mm and has been modified to incorporate the sensing and communication hardware. The Otter box is guaranteed to be watertight up to a depth of 30 meters. Each box has a high speed optical communication module that uses 532nm light, and is capable of a range of 2.2m a maximum data rate of 320kbits/s. Additionally, there is a acoustic communication module using 30kHz FSK modulation with a range of 20m omnidirectional, and a data rate of 50bit/s. The same module is also used for ranging (Two sensor nodes can determine the distance between them using time of flight of the sound waves). For sensing, each node has a pressure sensor, temperature sensor, and a CMUCam camera capable of color pictures with a 255 x 143 resolution. The top side of the box contains a 170 mm rod with an LED beacon, which an AUV can use to locate the box, dock, and pick it up. Future versions will contain a XENON flash tube for increasing the distance for reliable node location to about 20 meters.
We constructed 20 static sensor nodes (Aquaflecks) in colaboration with the CSIRO team in Australia. We also equiped two different AUVs (the CSIRO Starbug and our AMOUR) with optical and acoustical communication to be part as the mobile sensor nodes. We conducted experiments in CSIRO pool in Brisbane Australia as well as in MIT tow tank and Charles river in Boston. We demonstrated optical communications achieving 320kbits/s data rate over 2.2 meters range in clean water. We demonstrated our low cost acoustical communications achieving 50bits/s at 15 meters, as well as the distance measuring capability obtaining less the 2% error up to 15 meters.
Using the Starbug mobile node we demonstrated autonomuous data muling from a sensor network consisting of 6 sensor nodes placed in a grid. Starbug visited one by one the nodes, and colecting their data. Navigation between the nodes was basted on the magnetic compass.
Using AMOUR mobile node we demonstrated the ability to autonomuously dock with a static node, to pick it up and relocated it. Amour can locate and dock with a static node from a distance of 4 meters with a cone of 90 degrees.
We developed a TDMA based algorithm for acoustic communications in order to optimize the access to the limited bandwith available. The time slots are negotiated and dynamically atributed to nodes, in a completely distributed way.
As future work we want to work towards a full deployment of an underwater sensor network in the ocean environment, that will enable a full test of our concepts in a real environment. This will imply the AUVs, fully autonomuosly, deploying of the static nodes precisely in a specified locations. The nodes will remain there and collect data, while our acoustic communication and localization protocols can be tested. The mobile nodes (AUVs) will conduct experiments like collecting data from an interesting spot detected by the sensor network. The sensor network will also provide navigational aid to the AUVs throughout the experiment using their acoustic self localization and acoustic localization of the mobile nodes. The AuV will collect the data from the static nodes, using the high speed optical comms. When necessary the AUVs will replace the defective and the low battery nodes in an autonomuous fashion.
 Iuliu Vasilescu, Paulina Varshavskaya, Keith Kotay, Daniela Rus Autonomuous Undewater Optical Modular Robot - Design, Prototype and feasibility study. In Proceedings of the IEEE International Conference on Robotics and Automation, 2005.