Subscribe
Issue No.12 - December (2008 vol.9)
pp: 3
Michael C. Jaeger , Berlin University of Technology
Gero, Mühl , Berlin University of Technology
ABSTRACT
Satellite-based Internet access, the preferred solution for remote locations on the ocean, either offers a low-bandwidth connection or is very expensive to deploy. A backbone structure that provides ocean-wide Internet coverage could provide an alternative to satellite uplinks. With a wide-area network forming a mesh of nodes using floating and moving objects as well as coastal facilities, the network would use next-generation long-range surface radio technology to provide medium- to high-bandwidth Internet access. To achieve high-bandwidth Internet access under these circumstances, the backbone must leverage state-of-the-art sensor network technology, autonomous routing mechanisms, and self-organizing abilities.
In remote areas of the world, data transmission and access to the Internet often require expensive satellite connections or low-bandwidth radio links. Access at sea adds the difficulties of a drifting hostin platform and significant cost, and high-bandwidth Internet connections unfeasible. For example, equipping an aircraft with Boeing's Connexion service 1 has cost about US$500,000, and equipping a ship costs$60,000. 2 Additionally, such services have monthly fees—more than \$1,000 per month per ship using Connexion. Sufficient customer demand can help distribute costs, such as on a long-haul airplane, which usually carries more than 500 passengers over 24 hours. However, remote locations, cargo or small ships, and so on don't have the means to distribute high base costs over many clients.
Remote areas need a low-cost, non-satellite-based solution that provides reliable high-bandwidth Internet coverage. Our system offers such a solution, in which a mesh of individual sensor nodes over the ocean—essentially a wide-area sensor network—forms an Internet backbone.
Motivation
Our system particularly supports providing Internet access on remote islands and at sea. It could also create more reliable Internet connections in remote countries by increasing redundancy.
Remote Islands
Many remote islands either have few or no inhabitants. Such islands often have meteorological stations to capture data, so a low-bandwidth radio connection is available, but it's not possible to lay underwater cables to connect such islands (see Figure 1a).

Figure 1. Three applications of a low-cost, high-bandwidth, non-satellite-based Internet connection.

For small islands, independent, low-cost, high-bandwidth Internet access would provide redundancy and let residents download multimedia data. For uninhabited weather stations, such access would permit transmission of multimedia sensor data or video streams to support in-situ observations of remote places, such as a remote weather station streaming video of actual fog conditions.
The Open Sea
Large cruise ships can afford to invest in satellite-based Internet connections. However, plenty of smaller cargo or container ships have either standard surface radio connections or low-bandwidth satellite connections. Floating objects such as exploration buoys have very low-bandwidth connections using satellite radio technology; these objects on the open sea don't have access to a standard Internet connection (see Figure 1b).
Smaller ships could use Internet access to report their status (enriched with multimedia data) to their home bases. Crew members of commercial ships could send larger emails with photo and video data or host video conferences to stay in touch with their families. For small, low-budget expeditions on the open sea, an Internet connection with similar characteristics as on the mainland would let researchers continue Internet-based data retrieval.
Increasing Redundancy
Some countries have one or more underwater cables connecting to the coast of another country that provide an Internet connection. However, reliable service requires more than a single line serving an entire region (see Figure 1c). For example, in March 2007, Vietnam lost 11 km of an underwater cable that carried 82 percent of Vietnam's Internet traffic ( http://lirneasia.net/2007/06/vietnams-submarine-cable-lost-and-found). Local fishermen consider such cables lucrative as scrap. Cameroon experienced a worse case in November 2007, when a technical failure resulted in the country being offline for an entire week. 3 An underwater cable off the west coast of Africa (the SAT-3 line) provides some African countries with their only link to the Internet. This line's provider reportedly charges noncompetitive prices compared with other providers in industrialized nations, obviously because of its monopoly. 4
Each country with a coastline could access the Internet over the ocean using our proposed backbone. This solution would provide supplemental and redundant Internet access independent of a single cable or provider, reduce the risks of an entire country losing its Internet access, and decrease the likelihood of a monopoly.
Foundations
An alternative to underwater cables is a chain of buoys that provide a hop-based Internet connection to the mainland. Existing installations with a similar structure consist of floating sensor devices for oceanic research or meteorological purposes. Events where objects have spilled out of ship's during heavy storms have drawn public attention to this research area. For example, in a heavy storm in the north Pacific Ocean in 1990, a ship lost a container that hit the sea's surface and opened, spilling 60,000 pairs of sport shoes. 5 Similarly, in 1992, about 20,000 bathtub ducks spilled out, again in the north Pacific ocean. 6,7 In both cases, civilians picked up some of the objects, which had washed ashore. From the locations where the objects were found, scientists deduced the characteristics of the ocean currents on the surface. So far, about 1,400 shoes have been found on the west coast of North America and the Hawaiian islands.
These events have helped scientists understand the characteristics of the ocean's surface drift in the Pacific. The oceanic drift influences the temperature and humidity, so it's relevant for the world climate. Increased discussion of temperature warming and changes in the world climate has made simulation models of the ocean's characteristics increasingly important. Moreover, understanding the ocean's surface drift is also important for determining a ship's route. Drift often affects travel duration, which must be considered when planning a transoceanic route.
Scientists can't rely on floating objects accidentally spilling out of ships, so several research efforts have deployed various types of scientific floating objects or buoys in the ocean. For example, the Global Drifter Program (GDP), launched in 1998, deploys drifting buoys to track currents, capturing sea surface temperature and recording further meteorological data. As of May 2007, 1,247 surface drifting buoys had been deployed (see details on the GDP Web site at www.aoml.noaa.gov/phod/dac/gdp.html.)
Another research project called Argos aims to examine the climatic changes in different seasons and water's characteristics at different depths, including its salinity. The Argos project plans to deploy about 3,000 floats in its final stage. 8 These floats can sink and capture data at depths up to 2,000 meters.
Besides the floats, the project mounts additional sensor devices on volunteering commercial ships, called voluntary observing ships (VOSs), to capture data along main merchant routes. Shipping companies from different countries let researchers place sensor hardware onto their ships or install trawling buoys to gain a better understanding about the ocean's drift and currents. The companies also benefit; this information lets them plan their ships' routes more precisely.
The ocean is now full of sensors. With thousands of buoys deployed, we have a system of drifting objects that works with ship-mounted devices. The Argos project encourages contributions from its stakeholders; our proposal for an ocean-wide Internet aims to reuse this infrastructure along with coastline facilities.
Current Communication System
Satellites generally transmit sensor data from ocean-based research projects. In the Argos project, the floats and devices mounted on VOSs communicate via the Global Telecommunications System, or GTS, a project of the World Meteorological Organization ( www.wmo.ch/pages/prog/www/TEM/gts.html), a United Nations agency. The GTS provides a packet-oriented communication system implementing the X.25 protocol that builds on a satellite-based communication network. The GTS typically provides bandwidth below 9,600 baud because of the distances it has to overcome—far below the standards of state-of-the-art landline or cell-phone data communication systems.
Strengths of the GTS
The GTS has two main purposes: data transmission and data sharing. The GTS permits weather and climate data transmission from probes, sensors, buoys, and VOSs all over the world, and the WMO aims to share this data with meteorologists worldwide. So, in addition to the transmission infrastructure, the GTS provides meteorologists with database servers and data archives.
Shortcomings
The GTS provides a reliable and proven communication system that represents the backbone of today's worldwide meteorological data transmission. However, this network is specifically designed for small message sizes and requires a rather sophisticated satellite communication system. Although useful for data transmission and sharing, the GTS can't handle many other applications, such as multimedia, in situ observations, or personal messages from ocean crews. Additionally, providing Internet access at sea could enable new sensor applications and let scientists integrate and exchange sensor devices more flexibly. For example, a deployed float might have a communication link to a satellite, but because of limited bandwidth, it can't share this link with other devices. In this situation, sparsely distributed nodes with Internet access must establish their connections to the rest of the world individually—relay network or a mesh-based backbone structure isn't feasible.
Next-Generation Communication
The next generation of a global communication system for remote devices such as floats, buoys, or VOS-mounted devices should take advantage of the state of the art in sensor networks, autonomous computing, and self-organizing network infrastructure. Recent developments have led to self-configuring network nodes, self-stabilizing routing protocols, and self-coordinating sensor networks, which make possible an advanced communication network that can cope with the unreliable and unfavorable conditions on the high seas.
High-bandwidth transmissions require a sophisticated communication link that covers the host object's movement and orientation (as in Boeing's Connexion service). However, this technology doesn't support small devices, such as buoys. We propose using individual nodes to provide coverage to remote areas, supported by a satellite uplink. In the GDP, floats are an average of a few hundred kilometers apart. Considering that a communications backbone such as ours could also involve VOSs and floats from other projects, the average distance to overcome between two nodes might be even smaller. The following objects could become part of such a wide-area mesh network:

• VOSs. Commercial ships carry equipment for different research efforts, so they could also carry a mobile Internet node unit.

• Exploration and exploitation platforms. In some oceanic areas, the exploration and exploitation of hydrocarbons takes place on platforms installed offshore. These platforms can also host communications infrastructure.

• Sensor buoys already carry communications equipment, and they can be equipped with additional hardware to take part in our proposed network. Buoys can be drifting (for example, for observation of the oceanic drift) or stationary.

• Coastline support stations. On islands or at exposed spots on a coastline, fixed nodes can serve as part of the node network. Additionally, future offshore wind energy or wave energy farms could provide a relay point. Such farms usually have a (low bandwidth) Internet connection for maintenance purposes.

Figure 2 shows the overall setup of such a networking structure. For remote islands hundreds of kilometers away (not shown to scale), a chain of buoys could replace an underwater cable using state-of-the-art wireless wide-area network technology. A chain of VOSs along a merchant route could establish a ship-to-ship network that ends at a coastal harbor. Coastal access points could additionally provide a gateway to the Internet, as could special VOSs equipped with mobile satellite-based Internet access.

Figure 2. Buoys, voluntary observing ships, and coastal access points form a "floating backbone" for Internet coverage at sea and on remote islands.

Creating such a network requires overcoming several technological challenges. First, the distance between the nodes and their ability to drift creates a situation that we address using delay-tolerant networking (DTN). Networks without a stable end-to-end connection—for example, networks in mobile environments such as the ocean—use DTN. In this case, volatile weather and surface conditions change the foundation for reliable transmissions between nodes. Kevin Fall proposed a work on a DTN, developing the current Internet architecture, 9 and then expanding this idea with his colleagues. 10 Interested researchers have formed a research group, the Delay Tolerant Network Research Group (see www.dtnrg.org), arranged by the Internet Research Task Force.
The underlying network must cope with the location changes of the floats and ships, as well as challenging surface conditions, and be at least as fault-tolerant as the current GTS. These characteristics require research in different areas of autonomous systems.
Self-Configuring Routing
Because VOSs and floating objects will change positions, the network routing must be able to dynamically adapt to traveling nodes to update routing decisions. In such a scenario, nodes show a drift resembles that of personal cellular phone use. However, the drift of nodes causes volatile network links requiring a continuously updating routing algorithm. Silvia Nittel and her colleagues published a work about data management in a network of drifting nodes over the ocean. 11 Nittel's team aims to establish a similar setup for observing the ocean in a dedicated area, such as the Gulf of Maine.
Self-Organizing Network Structure
In general, we can expect that buoys won't distribute equally over the ocean. They might form clusters and leave sparsely covered areas, which isn't beneficial to maintaining wide coverage. Transmitting buoys similar to those the Argos project uses could autonomously correct their positions to maintain a well-distributed node structure over the ocean. So, mechanisms must be developed that allow for self-organization in the correction of node positions.
Self-Managing QoS-Oriented Communication
Because unfavorable weather conditions or the swell of the waves can interrupt communication links, the network must handle volatile connection characteristics that influence the QoS in a self-organized way. In the case of multimedia or video data, this requires soft real-time characteristics. So, our proposed network must apply algorithms and techniques that facilitate the redundant distribution of data depending on a particular transmission's QoS requirements.
High-Bandwidth, Low-Energy Communication
As existing architectural proposals have already pointed out, a new communication system must offer a clear advantage in terms of available bandwidth over the GTS. However, in addition to meeting bandwidth requirements, our proposed network must also overcome the long distances between nodes—and, in the case of buoys, with low energy consumption. In general, wave-based power sources already exist for buoys. 12 For low-range, radio-based transmission systems, technologies such as WiMAX provide a solution; WiMAX originally has a range of a few miles but could be extended for our proposed application.
However, we need more research on transmission behavior on the open sea under different weather conditions and involving the drift of the floating objects. We'll also need mechanisms to establish and remove connections that aren't common to wireless local area network base stations.
Conclusion
Our proposed Internet backbone involves wide-ranging research and engineering issues. Apart from classic fields in computer science, such as network routing, creating the backbone embraces emerging fields such as autonomic computing and sensor networks, as well as transmission protocol engineering and aspects of oceanography.
Engineering and deploying such a system doesn't necessarily require a large undertaking. Rather, we anticipate that the scientific community will establish a common standard for the proposed mesh that allows independent design and development of various node types for the different deployment options. Some standards and protocols already exist. However, when it comes to negotiation and routing, some areas require proposals, evaluation, and testing before being standardized.
After conducting a conceptual analysis of our proposed solution and setting the theoretical foundations, researchers must evaluate the system within a realistic testbed under real-world conditions at a low-budget scale. A preliminary version of the backbone could be installed in a contained area such as the Baltic Sea.

#### References

Gero Mühl is a postdoctoral researcher at the Berlin University of Technology. His research interests include distributed systems, middleware, event-based systems, self-organization, and ubiquitous computing. He has a PhD in computer science from the Darmstadt University of Technology. He is a member of the ACM and the German Computer Science Society (GI). Contact him at g_muehl@acm.org.
Michael C. Jaeger formerly worked at the Berlin University of Technology and currently works in the architecture group of the software and engineering department (CT SE 2) at Siemens Corporate Research and Technology in Munich. His research interests include modeling architecture and autonomy in distributed systems. He has a PhD in computer science from the Berlin University of Technology. Contact him at mcj@cs.tu-berlin.de.