White Paper Workshop on Research Directions for the Next Generation Internet Vienna, VA May 13-14, 1997 Kul Bhasin, Acting Chief Satellite Networks and Architectures Branch NASA Lewis Research Center 21000 Brookpark Road Cleveland, OH 44135 bhasin@lerc.nasa.gov (216) 433-3676 (216) 433-8705 (Fax) Eric A. Bobinsky Terasphere 343 W. Bagley Road, Suite 405 Berea, OH 44017 eric_bobinsky@terasphere.com (216) 243-2992 (216) 243-2934 (Fax) 1. Satellites in Today's Internet Communication satellites are part of today's Internet. Since satellites are used by telecommunications companies in their network backbones, some percentage of Internet traffic transits satellites. Although most of this traffic is carried internationally between North America, South America, Europe and Asia, some of it flows within national and regional boundaries over national and regional satellite networks. Satellite service providers like Orion, Intelsat and Eutelsat offer broadband data services to international common carriers such as AT&T, MCI and Sprint as well as to individual companies that use satellite connections in their corporate intranets. New satellite Internet services include direct-to-user and ISP-to-POP connections. The best known direct-to-user service is Hughes' DirectPC, which provides wideband ISP-to-PC access via a small dish antenna. The downstream (ISP-to-PC) data rate is as high as 400 kbps and can support a small LAN. Several operators are offering economically attractive packages for connection of ISPs to Internet POPs. A number of Asia-Pacific and South American ISPs are currently using this service. As an emerging service platform, the NGI is in a position to take full advantage of the revolution in satellite technology that is now underway "from the ground floor". 2. The role of satellites in the NGI - Meeting the Initiative Development of the NGI is paralleling the deployment of the new generation of high-performance communication satellites. This new generation has technical characteristics very much in line with NGI requirements, including high data rates, inherently ubiquitous access, support of enhanced Internet capabilities and global environmental monitoring, and support of national security goals. 2.1 High data rates New satellite systems are being designed to support data rates to 155 Mbps (even 622 Mbps) with the error performance of present-day fiber optic networks. Worldwide, over a dozen of these systems are either on the drawing board or in the assembly stage. The first launches of these high- performance systems are scheduled for the end of 1997. Several of these systems are based on spacecraft operating in the geostationary orbit (GSO), 22,300 miles above the Earth's equator, while others will use low-Earth orbits (LEO) from a few hundred to a few thousand miles in altitude. LEO systems promise not only excellent error performance and high data rates, but signal propagation delays comparable to those of terrestrial networks. 2.2 Ubiquitous access Since coverage is not limited to parts of the world which are readily served by terrestrial networks, satellites can bring the NGI to users who might otherwise be denied it due to geographic or economic factors. Though it might be prohibitively expensive to bring broadband NGI access to a remote school or hospital by laying a dedicated fiber-optic cable without federal subsidies, deployment of a satellite NGI terminal may be well within the financial reach of even the poorest, remote communities. 2.3 Enhanced capabilities 2.3.1 Broadcast/Multicast By virtue of simultaneously reaching large geographic areas, satellite systems are natural platforms for broadcast and multicast applications, whatever the underlying network fabric. 2.3.2 Remote research and computing Today's Internet supports remote scientific fieldwork, but researchers are usually limited to using narrowband facilities to access email, databases and other applications resident on home-based computer networks. As the sophistication of portable research equipment increases, so too does the need for high-performance access to computational facilities that might be located an ocean away. The new generation of satellites provides the mechanism for seamlessly connecting field workers to the enormous information and computing resources of the NGI. 2.4 Environmental monitoring Global change monitoring requires interactive communications with monitoring stations located in extremely remote areas. Communication satellites are used today to link remote monitoring stations to base locations. These are typically narrow-band telemetry and command links, which will eventually give way to the broadband links needed to support more sophisticated architectures, including real-time, multispectral imaging, and other multimodal sensors. Not only will satellites provide these links, but by integrating them into the NGI, global monitoring data can be distributed directly to the research and education community. 2.5 National security A key goal of the ARPAnet program was a network that could survive complete destruction of parts of its physical structure. From that need came packet switching and the 'connectionless' network paradigm. As today's Internet (and presumably the NGI) migrates toward a 'connection-oriented' model to provide quality of service guarantees, the network becomes less survivable. Moreover, though the Internet backbone is designed to be self- healing in terms of accidental ("statistical") outages, it has not been specifically designed to withstand the new threat of info-terrororism. Several government entities, including the President's Commission on Critical Infrastructure Protection, have mandated that US national networks should be designed to thwart both accidental outages and calculated attacks on the physical and transport levels. This can be achieved by a switched network if it provides redundant routing and media diversity as an integral design feature. Basing the NGI on any single platform increases its vulnerability to determined attack. On the other hand, basing its design (at all layers) on a redundant wireline, wireless and satellite backbone decreases its vulnerability. Because the new generation of communication satellites will provide performance nearly identical to fiber optic networks, the only cost of providing such diversity is accommodation of media having only slightly different performance characteristics. 3. Directions for Research To fully utilize the capabilities of next-generation satellites in the NGI, a number of research issues need to be addressed. Chief among these is the general goal of "seamless integration" and interoperability with terrestrial networks. Because service quality is an end-to-end characteristic, NGI traffic traversing a hybrid path consisting of two or more diverse media types should not behave differently from that traversing a single-medium path. In other words, the physical characteristics of the media over which NGI traffic flows should be transparent to the user. Since the NGI will build on an existing infrastructure, research focusing on current-generation Internet performance over satellites can be naturally extended to include architectures and protocols proposed for the NGI. NASA and a satellite industry consortium (for example) currently support active interoperability research in the areas of ATM, SONET/SDH and TCP/IP (although most ATM and SONET research does not target Internet applications per se). Most of this existing body of work, particularly that focussing on data rates above 45 Mbps, can form a solid base for an active NGI-satellite research program. Research is also needed focusing on the unique service capabilities of satellites, including broadcast and multicast NGI applications. Certain applications may be better-suited to satellites than to terrestrial networks, and it is clearly in the best interest of the NGI to fully understand and exploit them. Finally, an entirely new class of space-based networks is being built around constellations of low Earth orbiting satellites. While many of these systems support voice and low-rate data applications, others are being designed expressly to support broadband data services with extremely good error performance and without the delays inherent in geostationary satellites. The potential of these systems to bring about truly ubiquitous access on a global scale is enormous, and the NGI is perhaps the first "killer app" poised to take advantage of these new platforms. Because both they and NGI are "under construction", an aggressive and near-term research program is needed to insure that their convergence occurs in an optimal and mutually beneficial way. 4. Recommendations for the NGI Initiative 4.1 Even though the NGI might constitute a totally new network, it is more probable that much of its traffic will travel over the existing, national and global physical plants, which consist of a hodge-podge of different media maintained and tariffed by a plethora of carriers and service providers. Insofar as possible the NGI should be designed to be media-independent and robust enough to perform well over this hybrid network. 4.2 As a national and global utility, the NGI should be designed to take full advantage of the unique capabilities of different transmission technologies, including wireline, wireless and satellite. 4.3 Because the NGI will be a key component of the NII and a strategic national asset, it should be designed to take maximum advantage of routing diversity, media diversity and other methods that harden and protect the physical network from intentional damage. 4.4 The broadband satellite and wireless industries taken together represent a much smaller interest than the terrestrial telecom/datacomm community, but since they represent a critical part of the NII and GII, particular attention must be paid to ensuring that adequate funds are provided for R&D efforts targeting their full utilization in the NGI.