Talking underwater, disruption-safe

EarthzineUnderwater Communication

Figure 4: Data muling, an application of DTNs in UANs

How disruption-tolerant-networks help us communicate in the harsh environments underwater

February 7, 2020

by Arnav Dhamija,
Student, University of Pennsylvania

Underwater Communication techniques

Communicating underwater is a challenging task. It faces a different set of challenges from those encountered in terrestrial networks used to communicate on land. Radio waves used in terrestrial communication propagate poorly underwater due to attenuation in the highly conductive environment of seawater [1]. Acoustic waves suffer lower attenuation (degradation in strength) and propagate better in these environments. It is hence the most commonly used medium for underwater wireless communication.

Some underwater network applications also use optical waves for communication. These have higher bandwidth, are more energy efficient, and offer better bitrates and lower latency than acoustic links. However, they have shorter range as they are more adversely affected by absorption underwater than acoustic waves, and are also susceptible to scattering. Optical links can also be disrupted due to turbidity in the water [2]. Thus, optical links may not always work in conditions where acoustic links can be used.  Wired networks, known as underwater tethers use specially reinforced cables providing power and networking to underwater nodes. However, these tethers can be expensive, heavy and too limited in range for mobile nodes such as Autonomous Underwater Vehicles (AUVs). Hence, acoustics is the most commonly used medium for underwater communication, especially over longer ranges.

Story Index
  • Underwater communication techniques
  • Challenges
  • Disruption Tolerant Networks
  • Applications
Related Stories

Oceans conference at Marseille was covered on Earthzine

On the Trail of Sea Life with an Autonomous Underwater Vehicle

The hardware developed for acoustic communication also differs significantly from the kind used in radio communication. Examples of acoustic modems include those from Teledyne Benthos, Evalogics, Subnero, Linkquest, Aquatec and Desert Star systems, to name a few [3]. Modems used in Underwater Acoustic Networks (UANs) typically have a wet-end and a dry-end. The wet-end is an analog system which consists of a waterproof transducer and its accompanying circuits. The dry-end contains the signal processing hardware and single board computers to run applications and drivers. In most cases, the acoustic modems are half-duplex, meaning that the modems can send and receive signals, but not both at the same time.

Challenges in Underwater Acoustic Communication

Figure 1: Example of Underwater Acoustic Modems - Subnero’s (left) and Teledyne Benthos’ (right). Credits: Subnero, Teledyne Marine

Figure 1: Example of Underwater Acoustic Modems - Subnero’s (left) and Teledyne Benthos’ (right). Credits: Subnero, Teledyne Marine

Though acoustic waves can travel further underwater, communication using acoustic waves faces challenges due to sources of noise such as wind, shipping and marine biological activity, and interference from other acoustic waves transmitted at the same time. Other physical factors affecting transmission include Doppler shifts when the sender and/or receiver is not stationary, and surface effects in shallow water bodies [4]. A combination of these factors may contribute to high error rates in the communication depending on the channel. Altogether, this means underwater networks can be frequently impacted by disruptions.

Besides this, data transmissions using acoustic waves can have lower bitrates and considerable delays due to the low propagation speed of sound in water, which is several orders of magnitude lower than that of radio waves. Some modems may only employ half-duplex communication, which further increases the time needed for two modems to transmit data. Moreover, the power consumption of acoustic modems to operate their transmitters is high, and this limits the number of times a communication system tries persistently to send messages before dropping them.

Figure 2: Autonomous underwater vehicles may use an acoustic modem for communication. Credit: Bluefin Robotics Corporation

Figure 2: Autonomous underwater vehicles may use an acoustic modem for communication. Credit: Bluefin Robotics Corporation

These issues with disruptions and delays are not typically considered in network protocols used for the Internet, where the expected packet loss and propagation time is much lower. To deal with the challenging scenarios of frequent disruptions and delays, we need a different set of protocols and architectures. One such architecture is the disruption tolerant network (DTN) architecture.

Disruption Tolerant Networks

DTNs are a set of rules and protocols for the functioning of the nodes of the network in talking to each other, that imparts more robustness to their operation in situations with high delays/disruptions. DTNs are specially designed to be resilient when a part of the network fails or gets disconnected due to unavoidable factors such as change in the network topology or unfavorable environmental conditions.

DTNs have enjoyed success in applications such as interplanetary communication networks in space, especially in cases where the variable alignment of planets may make it impossible to directly send a message to a receiver on Earth. Due to the great distances and limited windows of contact involved in space exploration, transmitting a message in space communication may have a delay in the order of several hours. The bitrates are also low. For example, the Mars-to-Earth network link of the Curiosity rover uses a Mars orbiter spacecraft to relay messages to the Deep Space Network antennas spread across Earth [5]. The Mars orbiter is only within the communication range of Curiosity for eight minutes per Martian day and the orbiter itself can only communicate with Earth for a few hours per day. This DTN operates over nearly 200 million kilometers - even at the speed of light, messages from the Curiosity rover can take up to 20 minutes to reach Earth!

In most communication networks, messages are transmitted via ‘packets’ – think of them as individual letters carrying bits of information. Relaying of information may take place by forwarding the message packets via several nodes in the network. The nodes at the extreme ends of the network may not be able to talk to each other, but they may be able to ‘pass the message’ by finding a suitable route through the nodes. In cases where high delays or disruptions may occur, DTNs establish a route to the receiver using multiple communication nodes which forward the message when the next node in the route is available.

Figure 3: In this example of a space DTN, the rover on Mars stores the message until it is possible to send it to Earth via the Mars orbiting satellite. Credit: NASA

Figure 3: In this example of a space DTN, the rover on Mars stores the message until it is possible to send it to Earth via the Mars orbiting satellite. Credit: NASA

Similarly, in these space networks, a route from the sender to the receiver is established using multiple communication nodes such as space probes and satellites. Each intermediate node on the route saves the message intended for the recipient and forwards it when the next node in the route is within its communication range [6]. This is known as the Store-Carry-And-Forward approach to message delivery, and it underpins the working of all DTN protocols.

Every implementation of a DTN stores an expiry time (also known as a Time-To-Live) for each message. If the message cannot be sent to the next node in the route by this time, it is deleted from the node’s storage. Implementations of DTNs vary widely and the type of DTN used depends on the application. For example, some DTNs might use pre-defined routes whereas others may use “opportunistic” routes, which are dynamically created whenever another node is within communication range. As DTNs are used when the network is inherently unreliable, they cannot guarantee that a message will be successfully delivered. To improve the odds of successful delivery, some DTN protocols send out multiple copies of a message to different nodes [7].

UANs share some features in common with interplanetary communication networks discussed above. For one, the expected bitrate in both types of networks is low. Secondly, the delay encountered for the waves to reach from transmitter to receiver are extremely large – in the case of UANs, due to the limited sound speed, and in the case of interplanetary networks, due to the large distances involved. Lastly, both UANs and space networks may encounter frequent disruptions due to environmental conditions. Moreover, the range of acoustic waves can be much lower than the distance between the sender and the receiver, requiring multiple communication nodes to forward the message. As DTNs have had great success in space networks, they are good candidates for UANs as well.

However, the DTN protocols used for space networks are not directly adaptable to UANs. Unlike space networks which have predictable lapses of connectivity due to orbital alignments, UANs can face disconnections due to other sources of unpredictability such as variations in the underwater sound communication channel, or time-varying sources of underwater noise. DTN protocols must account for these differences in operation to be useful in UANs. Research on developing DTN protocols for UANs can be found in the literature, with the first of these papers published in 2011 [9]. A detailed tutorial and survey on DTNs are available in [8] and [10], and the reader may refer to these for more detailed technical information.


Figure 4: Data muling, an application of DTNs in UANs

Figure 4: Data muling, an application of DTNs in UANs

DTNs can also offer new topologies in UANs. For example, consider the case of sensor nodes which are sometimes deployed to collect measurements from parts of the ocean, such as in the case of the Nexos sensors. Conventionally, the sensor would have to store the data until a diver can retrieve the sensor, which is a labor-intensive procedure. Instead, we could solve this problem by using an AUV running a DTN protocol as an in-between messenger (also called a ‘data mule’). The AUV can be sent to patrol the area between the research vessel and the sensor nodes as shown in Figure 4. The sensor running a DTN protocol will send the data to the patrolling AUV as soon as it comes in communication range during one of its patrols. In this example, the DTN protocol is a suitable choice for a network protocol as it requires the communication node to store the message on its internal storage until it is successfully delivered or expires at a time set by the operator. This is similar to how data can be transferred between space probes which have only a small window of contact when orbiting differently planets.

DTNs can also be used in swarm robotics applications, where each robot in the swarm collects some data about its local environment. To obtain a more accurate model of the environment, robots in the swarm can combine data by sharing their sensor readings with each other. However, the network formed in a swarm may be ad-hoc without predefined routes for each pair of senders and receivers. An opportunistic DTN protocol to determine the route for a message for a swarm of aquatic surface robots has been suggested in [11].

A possible extension to a DTN protocol can be a feature to switch between acoustic and optical communication. The protocol could use acoustic communication by default and switch over to optical communication when the node comes close enough. As such, there is no “one size fits all” approach which can fulfill all the different applications where UANs can be applied due to the need to support different kinds of communicating media and frequency ranges (such as high-frequency for a short-range high-bandwidth use case, and low-frequency for a long-range low-bandwidth use cases).

With its wide range of possible implementations, DTNs have been shown to be essential in making communication possible in space networks. As underwater networks often face similar challenges as in space, DTNs are one of the tools we may have to employ to make these communication networks more reliable and help us deal with disruptions while we 'talk' underwater.


[1] E. Jimenez, G. Quintana, P. Mena, P. Dorta, I. Perez-Alvarez, S. Zazo, M. Perez and E. Quevedo, "Investigation on radio wave propagation in shallow seawater: Simulations and measurements," 3rd Underwater Communications and Networking Conference, Ucomms 2016, pp. 1-4, 2016.

[2] N. Saeed, A. Celik and T. Y. Al-naffouri, "Underwater Optical Wireless Communications , Networking , and Localization : A Survey," pp. 1-40.

[3] S. Sendra, J. Lloret, J. M. Jimenez and L. Parra, "Underwater Acoustic Modems," IEEE Sensors Journal, vol. 16, pp. 4063-4071, 2016.

[4] M. Chitre, I. Topor, R. Bhatnagar and V. Pallayil, "Variability in link performance of an underwater acoustic network," 2013.

[5] NASA, "Communications with Earth," [Online]. Available: [Accessed 4 January 2020].

[6] NASA, "Disruption Tolerant Networking," [Online]. Available: [Accessed 4 January 2020].

[7] F. Warthman and Others, "Delay-and Disruption-Tolerant Networks (DTNs):A Tutorial," A Tutorial. V.. 0, Interplanetary Internet Special Interest Group, pp. 5-9, 2012.

[8] "Delay tolerance in underwater wireless communications: A routing perspective," Mobile Information Systems, vol. 2016, 2016.

[9] D. Merani, A. Berni, J. Potter and R. Martins, "An Underwater Convergence Layer for Disruption tolerant networking," 2011 Baltic Congress on Future Internet and Communications, BCFIC Riga 2011, pp. 103-108, 2011.

[10] H.-H. Cho, C.-Y. Chen and T. K. Shih, "Survey on Underwater delay/disruption tolerant wireless sensor network routing," IET Wireless Sensor Systems, vol. 4, no. 3, pp. 112-121, 2014.

[11] D. Sousa, M. Luis, S. Sargento and A. Pereira, "An Aquatic Mobile Sensing USV Swarm with a Link Quality-Based Delay Tolerant Network," Sensors, vol. 18, no. 10, pp. 1-31, 2018.

About the Author
Arnav Dhamija is currently pursuing a master’s degree in Robotics from the University of Pennsylvania. He interned at the Acoustic Research Lab, NUS for his undergraduate thesis in early 2019. His blog can be found at