(Cover Page) Next Generation Internet "White Paper" Author: Lyndon G. Pierson, Title: Distinguished Member of Technical Staff Affilliation: Sandia National Laboratories Postal Address: Lyndon G. Pierson Information Processes Center - Advanced Network Integration Department Sandia National Laboratories Dept. 4616 MS 0806 Albuquerque NM 87185-5800 email: lgpiers@sandia.gov Voice: (505) 845-8212 Fax: (505) 844-2067 =========== This paper discusses three "missing pieces" of technology thought to be required for success of a secure, high speed "next generation internet". 1) Super High Speed Communication Protocol Processing in the Optical Domain. 2) Super High Speed Encryption Processing of ATM Cell Streams, interoperable with low cost low speed encryptors. 3) The broad range of technical problems associated with the deployment and management of a constellation of Low Earth Orbit ATM Switching Satellites to provide high speed, ubiquitous, wireless voice/video/data service to/from any point on earth (and even space-borne applications). 1) Optical Communication Protocol Processing Communication switching speeds must increase faster than computation switching speeds. Though it is feasible to implement thousands of parallel data paths between computing elements of a single computing system, it is currently only feasible to implement one (or few) path(s) between systems separated by even moderate distances. Therefore, even though we need faster ways to process serial communications, an appropriately scalable solution is to make "massively parallel communication" feasible over local and wide area distances. This research is to develop methods of processing communication protocols entirely in the optical domain to achieve computer communication throughput not achievable using electronics, but which are required to support high performance massively parallel applications. These techniques will then be applied to explore the feasible limits of massively parallel "wavelength division multiplexing" (WDM) on a single optical fiber. As communication rates increase (in the 1 to 100 Gb/s arena), the switching speed of GaAs electronic components is thought extensible only into the range of 40 to 60 Gb/s, requiring the processing of communication protocols via alternate means. Two technologies that show promise of supporting higher data rates are 1) super conducting josephson junction technology (SJJ), and 2) all-optical processing of data. Though advances will be made and applied in SJJ technology, long haul transmission has already become "optical", and many communication protocol functions lend themselves to "matrix operations" which have been shown to be feasible in a parallel processing fashion in the optical domain. Other research efforts have begun to apply optical means to the processing of image data and to "sequential computing" in the "Von Neuman" sense, but not to the processing of communication functions. Therefore, this research will focus on developing methods of processing communication protocols entirely in the optical domain and methods of massively parallel communication to scale/accelerate local and wide area computer-to-computer network throughput far past that achievable in the electrical domain. 2) Super High Speed Encryption Processing of ATM Cell Streams This effort is to influence and accellerate the deployment of scalable, variable bit rate ATM encryption equipment needed to satisfy DOE/ASCI requirements for communication security. The MCNC "Atilla" prototype ATM end-to-end encryptor is the fastest known, operating at 622 Mb/s. Sandia's "Scalable ATM Encryptor Prototype" is the second fastest, operating at 155 Mb/s. The NSA "Milkbush" prototype encryptor is the third fastest, operating at about 100 Mb/s. NSA's efforts have evolved into plans to make ATM end-to-end encryptors commercially available which operate at up to 0.6 Gb/s, yet no work is ongoing to deliver ATM encryptors for higher data rates (2 to 10 Gb/s) expected to soon be required to meet ASCI objectives. An effort is needed to integrate innovative methods of key management, crypto synchronization, and key agility while scaling encryption speed. Viability of these methods for encryption of ATM cell payloads at the SONET OC-192 data rate (10 Gb/s), and for operation at OC-48 rates (2.5 Gb/s) is to be shown. Informal collaborative relationships have been initiated with research efforts into related areas of interest at Motorola and at Fore Systems, Inc., which may allow certain leverage of this work. Development of this technology will: a) Enable national defense applications requiring the secure exchange of massive amounts of data between widely separated sites. b) Enable the interoperation of low cost, "low" speed encryptors for High Performance Workstations and super-high-speed encryptors for Massively Parallel Processing (MPP) computers, and c) Significantly reduce the cost of encryptors for industry standard ATM/SONET data transmission interfaces, presently costing $30k per encryptor. 3) Wireless Low Earth Orbit (LEO) ATM Switching Satellites for Variable Data Rate Transmission Although there has been a significant amount of work in developing wireless networks, most of the work has been Proprietary in nature. In addition to the lack of standards, there is no variable bit rate, cell based, mobile node wireless infrastructure to support the growing demand on wireless communications. These demands include the DOD's C4I for the Warrior (a concept that wishes to deliver ATM to the foxhole) and extended cellular communications systems to provide worldwide coverage. The approach we wish to take focuses on the use of Asynchronous Transfer Mode (ATM) in conjunction with the Packet Radio concepts to develop a wireless ATM switching fabric to support variable data rate transmission. The problem is similar to current cellular telephone difficulties with the added feature that now the base also becomes mobile. In addition to this complexity, the system must establish virtual circuits between radios, thus creating a wireless ATM fabric. The technical issues that would be addressed are: the signaling protocols for establishing the virtual channels; the shortest distance routing protocols; the keying issues around the use of spread spectrum; and problems surrounding the high bit error rates for wireless technologies as they apply to the very low bit error rates of ATM. In addition, we would begin to look at how to embed a wireless ATM switch onto existing wireless physical layer technologies such as spread spectrum, packet cellular, packet radio, etc. Allowing wireless protocols to utilize ATM is the next step in truly integrating all communication mediums. The flexible switching fabric would allow users the (transparent) ability to hop from one satellite to another as the user changes locations, or as different satellites pass overhead. This would allow a virtually seamless connection between the current wired networks and wireless systems, while providing a common standard to the wireless community for video, voice and data transmissions. Lyndon Pierson Information Processes Center - Advanced Network Integration Department Sandia National Laboratories Dept. 4616 MS 0806 Albuquerque NM 87185-5800