SubscribeI. Introduction - For the past few years, the Department of Defense has been focused on how to implement Network Centric Warfare (NCW). This concept encourages interoperability and information sharing among joint forces in order to "allow warfighters to take full advantage of all available information and bring all available assets to bear in a rapid and flexible manner."
It assumes that a greater quality of and access to immediate information enhances both situational awareness and speed of command. In essence, NCW translates information superiority into combat power by effectively linking knowledgeable entities in the battlespace."
Network Centric Warfare applies communication-networked applications in the fastest, most universal method available.The network needs to be equally accessible by all branches of service, for issuing and executing battle command decisions and support the real-time and wireless applications envisioned for Network Centric Warfare. It also needs to provide secure communications.
This concept is a daunting task when the vision of Network Centric Warfare is contrasted with the reality of the current DoD communications infrastructure. Today’s DoD network is burdened with vast quantities of legacy equipment. The slow speed of these networks, the significant differences among vendor equipment and protocols, and the heavy reliance on point-to-point communications have the DoD system entrenched in older technology.Even the DoD concedes that implementing NCW is a “monumental task [that will] will span a quarter century or more” .
Lastly, funding to replace such equipment is usually redirected to higher priority programs and deployed troops.In essence, the DoD has an installed, legacy-laden infrastructure that struggles to meet the capabilities needed for today’s deployed troops and has many impediments to overcome to achieve information superiority.
The DoD, then, has two tasks. One is to continue to provide essential communications systems to today’s troops.The second is to overcome the challenges legacy equipment, bandwidth and budget constraints, and interoperability between services and equipment in order to build a networked tactical communications system that supports real-time applications, wireless access, and future applications.
The DoD is considering two wide-area network protocols for these tasks.One is Internet Protocol (IP) and the other is Asynchronous Transfer Mode (ATM).An evaluation follows of both, as well as a migration map to achieve the DoD’s vision of Network Centric Warfare beginning today.
II. ”Where are we now?” – Beginning the Evolution
Today, much of the deployed tactical DoD communications equipment relies on point-to-point communications with the widespread installed base of Time-Division Multiplexers (TDM).These TDM circuits terminate at the STEP site and channels are dropped onto public networks or passed to the DISN core ATM and circuit technology network.
The two significant applications on the network today are voice and IP data. Currently the IP traffic consists of predominantly message-based applications such as email, http browsers and bulk file transfers. In the tactical environment, voice is often carried over individual analog POTS circuits that use voice compression.
Today, the DoD network is deploying the new TACLANE and FASTLANE encryption technology that supports transport of Type 1 encrypted data at broadband speeds over ATM and IP networks. Most data encryption, however, is still supported with legacy bulk encryption that is transported over unstructured, synchronous circuits. Voice encryption in the strategic environment is supported with STE technology, while the analog STU technology is still heavily used in the tactical environment due to bandwidth constraints.
With the plethora of installed serial and unstructured devices in the DoD, the network link layer needs to have both the capability and the interfaces available to transport these types of traffic. Many of these devices operate at extremely low data bit rates and most new equipment does not support these rates and interfaces.It will require considerable time and dollars to replace the large base of legacy encryption equipment currently deployed.
The US Navy, Joint Battle Command, DISA, CENTCOM, US Air Force, US Marine Corps, and US Army and MITRE all agree that there are key issues that a new DoD communications infrastructure will have to address:
•Interoperability between forces
•Ease of Use for all who need information
•Legacy application support
•Bandwidth on demand
•IP transport
•Video transport
III. A Definition of Protocols
In 2002, the DoD announced a strategy to begin IP network planning. The IP protocol resides in Layer 3 of the Open Systems Interconnection (OSI) Networking Model.This layer is responsible for managing network addressing and routing considerations. IP protocol needs the support of a Layer 2 protocol, such as PPP, HDLC, Frame Relay, or ATM, to handle link data format and flow control issues when implemented in a network.
The IP Network will use IP consistently across Layer 3 of the entire network. The network designer then has a choice whether to select Layer 2 protocols that manage data applications with varying service requirements - such as those offered with ATM - or to try to manage these service requirements from the IP protocol. While the debate often sounds like an “either-or” proposition, these protocols can compliment each other in a multi-service network.
A misperception revolves around IP being easier to use than ATM. The more accurate statement is that best-effort networks are easier to plan and operate than deterministic networks.There is nothing inherent in either protocol that effects ease of use.In fact, operating a deterministic network with a connection-oriented protocol like ATM may be simpler than operating deterministic networks with a connectionless technology like TCP/IP since each virtual circuit has a bandwidth consumption limit that is policed by the network.
Recently, implementations of Internet Protocol Version 6 have become commercially available. The key difference between Version 6 and today’s widely implemented Version 4 is that the packet address field in Version 6 has been extended from 32 to 128 bits, adding addresses and 8 traffic management bits that can be used to define traffic flows.While IPV6 has interesting traffic management features, the extended address field presents complex challenges to the technical community.Coexistence of V4 and V6 nodes in the DoD network will be difficult since each node needs to retain a unique address to route traffic correctly.Microsoft claims, “The conversion from IPv4 to IPv6 will be a larger task for the industry than the preparation for Year 2000…Unlike the Year 2000 issue, the conversion to IPv6 has no specific timeline”.Furthermore, no standards exist to ensure all prioritization is handled congruently throughout the IPV6 network.
While some people point to IPV6 as a next-generation network solution, it is a “work in progress” that is evolving toward a solution.IP is still a “best-effort” protocol and cannot offer guaranteed, or even deterministic, delivery of data over a multi-hop network.From a DoD perspective, it would be beneficial to have IP support a networked constant bit rate and a standards-based approach to voice trunking to gain bandwidth efficiency.This would permit the support of legacy applications with existing satellite capacity.
IV: Moving Forward: What are the Alternatives?
TDM: Time-division multiplexers (TDMs) are pervasive in the DoD today and still fulfilling their mission.TDM transports almost any digital communication signal and its hardware-managed channels still offer the lowest overhead and latency possible.Yet cost, size, weight, and inability to support rapid change and evolution of communication applications will force its retirement. While TDM equipment is reasonably easy to set up, changes in configuration usually require replacing hardware modules.This technology runs counter to the theme of developing a light and flexible fighting force, allowing multiple applications running simultaneously.
IPV6 over a simple link layer: This technology has enabled the construction of highly cost-effective optical networks designed to carry increasing levels of best-effort IP traffic.Likewise, VOIP is being implemented on managed, point-to-point links in enterprise networks.The near-universal use of IP to support network applications in the enterprise space makes IP a dominant force in the DoD future. The popularity of IP and the emergence of real-time applications are driving the data networking community to evaluate IP Version 6 and transition strategies.
IPV6 allows users to identify a class of service in the traffic management bits in the IP header so that routers can apply buffering to manage the traffic.An important distinction from ATM is that IPV6 manages QoS on a per-hop basis and not on a per-flow basis. For simple enterprise networks and broadband networks that are intentionally under subscribed, this may be all the traffic management that is required.Users running IPV6 in a multi-hop network will observe that the packet order of their traffic will get altered or lost, spelling trouble for critical DoD communications.
A larger issue for an IP network is how the network will support traffic requirements. IP is a connectionless technology, and thus the network places no constraint or policing of the level of resource used by a connection.Large IP networks require extensive planning and engineering to maintain high service levels, since service levels require assurance that network capacity exceeds user demand.DoD TDM users are not accustomed to taking traffic measurements to assure the predictability of a network.
There are also issues when using IPV6 in a SATCOM environment. It is possible to assemble constant bit rate traffic in IP packets and send them across an IP network, but all the IP packets on the link need to be fragmented so that the CBR and voice traffic are not impaired by the latency of large packets traversing a slow link.There is no way to assure that the CBR traffic can travel more than one hop without corruption.This corruption could conceivably translate into a critical communications failure between Battlefield Command and the Warfighter.
ATM: ATM protocols were conceived as a “universal” transport, defined and evolved under the direction of standards organizations, to accept data streams with varying formats and service requirements and deliver them in accordance with pre-defined service levels. ATM has a set of standard adaptation layers to transport synchronous serial data, variable bit rate data, and packet data. This capability maps well to voice, ISDN, encrypted data, and best-effort IP traffic that flow over the network today. ATM has a robust traffic management capability that supports five classes of service ranging from constant bit rate to unspecified bit rate.
ATM has a number of features that play well with the DoD SATCOM environment. Most notable is ATM’s ability to handle unstructured, encrypted data and its ability to scale down to narrowband environments. Users using ATM in an all-IP environment complain about the small ATM cell size and the “cell tax” it produces.But in the narrowband SATCOM world, the small ATM cell size permits the network to support both real-time and best-effort traffic over bandwidth-constrained links.In addition, ATM supports voice trunking over its AAL2 variable bit rate service, permitting compressed voice packets from multiple telephone conversations to ride together in one ATM cell, reducing latency and yielding greater efficiency in filling ATM cells.
Both ATM and IP have the ability to allocate bandwidth to applications dynamically.Since bandwidth is assigned to the application and not a physical port, packet data links can generally support more applications and roll out new applications faster than legacy communication systems.Unlike IP, the statistical multiplexing scheme of ATM permits quality of service to be managed to the per flow level.This ability to manage QoS at the flow level is extremely important to a bandwidth-constrained environment like SATCOM.
Some people question the deployment of ATM equipment based on the perception that ATM is fading into being a legacy technology.The predictable ATM Quality of Service is a foundation for Service Level Agreements and voice/video services.A 2002 report by Probe Research says that sales of ATM switches will grow 41% by 2006 - a healthy future being aided by a currently unhealthy economy. The combination of dynamic bandwidth allocation, multiple traffic types, and QoS will continue to be a unique, valuable – and profitable - technology offering.
V. Conclusion – “Achieving the Vision”
As new technology is assimilated into the DoD network, bandwidth increases and applications are expanded. Today, ATM technology is being introduced into the DoD network, not only at the core, but also over satellite access circuits. A key factor guiding the DoD’s transition to ATM is its ability to converge legacy unstructured data, voice, and packet data onto a single circuit while preserving the quality of service for each application.ATM’s quality of service at the virtual circuit level is also critical to supporting the bandwidth-constrained SATCOM environment. ATM’s links have software-controlled dynamic bandwidth, supporting simpler changes for the DoD network than hardware replacement.
While ATM is an immediate and future solution for the DoD, the move to an IP core will gain momentum in the coming years.High-speed routers and optical networks will relax congestion and quality service considerations.It is still unclear how IP address translation and the compatibility of the version 4 and version 6 IP protocols will be managed.What is clear is that the economies of optical switching will greatly reshape the core architecture.
After the core offers IPV6 services, the natural evolution of IPV6 is to move out to Teleports and STEP sites.Ultimately, IPV6 will be implemented at the access layer of the DoD Wide Area Network. During the transition, when not all sites and not all applications are running the V6 protocol, the connection-based ATM technology can separate the different versions of IP protocol into different virtual circuits and route the IPV4 traffic to the appropriate router or media gateway.This ability to facilitate the IP protocol transition assures that any ATM asset purchased today will have a long service life.
For the DoD, developing an ATM access and edge layer while developing an IP optical core offers the best plan to support the Network Centric Warfare vision. ATM access is a long-lived and stable technology that supports the variety of applications the DoD requires supporting the warfighter’s mission.It supports the QoS needed in the trenches to support decisions based on real-time data – whether that be a video, a satellite image, or a secure telephone call. This architecture can support future applications and achieve the vision for Network Centric Warfare.
Lastly, the best argument for using ATM as a transition technology within the DoD is that it is available today.This allows the legacy equipment that permeates the DoD to continue to be utilized until funding is available for its replacement.In the meantime, new program funding can begin to – and, in fact, is – supporting ATM. The DoD itself envisions a long transition to Network Centric Warfare, but its beginnings lie in ATM.
Appendix A: Transition Strategies Issues, Compared to the Net-Centric Version
Appendix B: Transition Strategy Dummary, Short Term Implementation Issues
