Testbed Architecture

The underwater acoustic networking testbed was designed to bridge the gap between experimentation and theoretical developments in underwater communications and networking, and is the result of a joint venture between the University at Buffalo and Teledyne Benthos. The objective of the project is to provide the research community with a versatile and shared reconfigurable platform to enable experimental evaluation of underwater communications and networking protocols.

The underwater acoustic networking testbed, which is being developed under sponsorship of the US National Science Foundation, is based on the Teledyne Benthos Telesonar SM-75 modem, which, in its many configurations, is also a key component in multiple U.S. Navy programs and of many wireless tsunami warning systems worldwide. The testbed consists of 11 Telesonar SM-75 modems, one sonar SM-975 modem, and one universal deckbox, UDB-9000, equipped with an acoustic transducer used for monitoring the underwater communications.

A number of protocols have already been developed using the underwater acoustic networking testbed. Some of the protocols are: The Internet Underwater: An IP-compatible Protocol Stack for Commercial Undersea Modems, Securing Underwater Acoustic Communications through Analog Network Coding, A Hybrid MAC Protocol with Channel-dependent Optimized Scheduling for Clustered Underwater Acoustic Sensor Networks (UW-ASNs), CDMA-based Analog Network Coding through Interference Cancellation for UW-ASNs.

External Controller for the SM-75 Acoustic Modem

The SM-75 modem is interfaced with an additional external controller on a daughterboard, a Gumstix (a small computer-on-module) residing on the Tobi expansion board.

The purpose of the Gumstix is to host the control logic in charge of implementing networking functionalities at all layers of the protocol stack by building on the physical/link layer application programming interface (API) exposed by the Benthos modem. Moreover, the Gumstix will add the capability to process one (or more) channels of analog signal or digital data for purposes defined by the software resident in the signal processing infrastructure. Most notably, the Gumstix will allow storing and processing data from (at least) two different channels, to allow MIMO and cooperative signal processing functionalities. The source of these data could be the modem's transducer used in parallel as a hydrophone or it could be an auxiliary hydrophone.

Developing Networking Protocols through the Modem Management Protocol (MMP)

The logic in control of the networking functionalities is implemented in the C language and housed on the Gumstix's processor. The Gumstix interfaces with the modem via the established API and may therefore be programmed by project participants. The modem's DSP will only implement physical layer and logical link control functionalities. The modem will offer an API that abstracts all the main physical and link layer functionalities to the Gumstix. In addition, it will offer access to a predefined set of status monitoring signals.

These signals in turn will provide information about the current state of the communication process at lower layers of the protocol stack, i.e., signal-to-noise ratio (SNR), channel impulse response (CIR) duration, round-trip time (RTT) and relative speeds of platforms among others, that is going to be leveraged by higher layers of the protocol stack. The means of passing information back and forth between a Teledyne Benthos acoustic modem and the Gumstix's processor is achieved through the use of a serial binary control protocol called Modem Management Protocol (MMP). The goal of MMP is to provide a static, unambiguous binary interface for machine-based command and control of the acoustic modem. The entirety of MMP may be used to control a vast array of modem functions and settings. Using the facilities of MMP the external processor on Gumstix, running a network state machine, can communicate with and control the lower protocol layers on the modem.

Controlling the Link Layer: Cross Layer Controller and Communication Architecture

A key architectural requirement in the design of communication protocols for underwater networking is to facilitate the use of cross-layer interactions. In underwater networks, the attainable capacity of each wireless link depends on the interference level perceived at the receiver. This, in turn, depends on the interaction of functionalities that are distributively handled by all network devices such as power control, routing, and rate policies. Hence, capacity and delay attainable at each link are location dependent, vary continuously, and may be bursty in nature.

Accordingly, making efficient utilization of network resources is a challenging task. Moreover, functionalities handled at different layers are inherently and strictly coupled due to the shared nature of the underwater acoustic channel. For this reason, to develop an optimized protocol, we need to make state information from lower layers of the protocol stack available to higher layers and vice versa, what is referred to as cross-layer information sharing. We accomplish this by designing an additional module, referred as cross-layer controller (XLC), in charge of controlling and regulating this information exchange among functionalities handled at different layers of the protocol stack.

Underwater Acoustic Channel Emulator

In underwater acoustic sensor networks (UW-ASNs) it is very difficult to conduct repeatable and realistic experiments through a reconfigurable experimental testbed alone; since the physical layer is strongly dependent on the UW-A channel environment and the exact conditions under which an experiment is conducted. Accordingly, we are developing an UW-A channel emulator that allows conducting laboratory controlled experiments.

The underwater acoustic channel emulator, residing in a personal computer (PC), is interfaced with the SM-75 modems through RS-232 serial port links that may control the modems to transmit and record custom-defined acoustic waveforms, as shown in Fig. 6. The transmitted acoustic signals are first captured by an audio input device and fed to the channel emulator. The captured signals are signal processed in Matlab to account for path loss, noise and multipath spread of a real underwater acoustic channel. In our design of the channel emulator we apply widely used channel models and allow the user to select the parameters that have an influence on those factors. Then, the modified acoustic signal is played by an audio output device and recorded by the receiver modem, which in turn attempts to decode the original transmitted information bits.

Graphical User Interface (GUI)

We have developed a GUI to accompany the underwater acoustic channel emulator. The user may use the environment setting to emulate several scenarios. Moreover, the user may select the location of the modems.

The proposal is based on an adaptation layer located between the data link layer and the network layer, such that the original TCP/IP network and transport layers are preserved unaltered to the maximum extent. The adaptation layer performs header compression and data fragmentation to guarantee energy efficiency. Furthermore, the proposed architecture includes mechanisms for auto-configuration based on router proxies that can avoid human-in-the-loop and save energy when broadcast is needed. The proposed architectural framework was implemented as a Linux device driver for a commercial underwater network modem SM-75 by Teledyne Benthos. Testing and simulation results illustrate that the architecture efficiently provides interoperability with TCP/IP.

The Internet Underwater architecture was tested extensively at Lake Erie, a few miles south of downtown Buffalo. We were able to demonstrated an IP-level connectivity by successfully sending and receiving a message from the deployed underwater sensor node to the boarder router using acoustic signals, which was relayed through a 3G/4G broadband Internet to a laptop and a smart-phone, connected to the traditional Internet.

Taking the Internet Underwater

Semi-permanent underwater Internet testbed deployment near to the small boat harbor in Lake Erie, Buffalo NY.

Testing Facilities

Experimentations in the diving pool of the Alumni Arena and in the water tank of the Underwater Acoustic Communications Lab at the University at Buffalo (UB).

Testing Facilities cont.

Experimentations in Lake LaSalle at the University at Buffalo (UB).