How to Develop Predictable On-the- Move Communications Faster and at Lower Cost Using Real-Time Network Emulation

Transcription

1 White Paper How to Develop Predictable On-the- Move Communications Faster and at Lower Cost Using Real-Time Network Emulation Introduction Communication Demands of Network-Centric Operations Challenges to Implementing Network-Centric Systems Historical Approaches to Net-Centric Systems Development The Solution Real-Time Wireless Network Emulation What To Look for in an Emulation Solution Applications of Real-Time Emulation Technology Real-Time Emulation with EXata the Parallel Processing Advantage S C A L A B L E NETWORK TECHNOLOGIES

2 Introduction Network-centric systems are widely believed to be the platform from which the U.S. and other Western nations will conduct their military operations in the future. The U.S. General Accounting Office reported in April 2008 that the Army s Future Combat System (FCS), the centerpiece program comprising fourteen integrated weapon systems and an advanced information network, is about halfway through its development phase 1. By definition, network-centric systems must rely heavily on a communications infrastructure that delivers on the move communication of voice, video, and data messages. Communication networks are also a critical force multiplier, supporting a broad range of situations and decisions such as logistics, battlefield awareness, and time-critical call-for-fire commands. Each situation and message type carries different quality of service requirements, yet all must be carried over a common infrastructure capable of connecting thousands to tens of thousands of users a very difficult challenge. Indeed, the GAO report states that It is not yet clear if or when the information network that is at the heart of the FCS concept can be developed, built, and demonstrated 2. On-the-move network performance is affected by several critical factors including traffic load, degree of mobility, terrain, and environment. Each of these factors can cause dramatic changes in link capacities, end-to-end latency, and message completion rates. Even as the technology for building scalable on-the-move networks matures, we still must accumulate the experiential base that will govern how to design and build these systems, how people are likely to use them, and why forces should be able to depend on them in mission- and life-critical situations. The approach to date has been to build and deploy prototype systems and then gain the experience of overcoming performance barriers through in-the-field testing. New technology is now available that brings the same kind of virtual design and engineering power that developers of other complex systems such as aircraft, automobiles, and buildings already enjoy, to net-centric communications development, training, and operations. This white paper outlines the key challenges to developing predictable on-the-move communications, reviews historical approaches to on-the-move network design, and presents a new approach real-time wireless network emulation as the solution to gaining the experience of meeting service requirements and overcoming technical obstacles faster and at much lower cost than designing in a trial-and-error vacuum. 1 U.S. GAO, Defense Acquisitions: 2009 Review of Future Combat System is Critical to Program s Direction, 4/10/ Ibid. White Paper: How to Develop Predictable On-the-Move Communications, Page 1

3 Communication Demands of Network-Centric Operations The aim of network-centric operations is to translate information advantage into warfighting advantage through robust networking of well informed, geographically dispersed forces 3. Three fundamental principles guide this military doctrine: A robustly-networked force improves information sharing Information sharing enhances the quality of information and shared situational awareness Shared situational awareness enables collaboration and self-synchronization, and enhances sustainability and speed of command 4 Successful application of these principles is expected to dramatically increase mission effectiveness. But the first principle robust networking is the prerequisite to realizing network-centric operations full potential. Challenges to Implementing Network-Centric Systems The April 2008 GAO report identified the scalability of the network as a major risk factor to the FCS program. The current conclusion is the proposed system will be unable to network all the units into one self-forming, self-healing network 5. The problem of coordinating bandwidth usage in a battlespace is a significant challenge, when every piece of mobile equipment and human participant becomes a potential source or relay of signals. It is difficult to efficiently transfer information between networks having different levels of security classification. Although multi-level security systems provide part of the solution, human intervention and decision-making is still needed to determine what specific data can and cannot be transferred. Accurate locational awareness is limited when maneuvering in areas where Global Positioning System (GPS) coverage is weak or non-existent. These areas include the inside of buildings, caves, and built-up areas and urban canyons, the setting for many modern military operations. Much work on reliable fusion of positional data from multiple sensors remains to be done 6. Providing secure communications in network-centric operations is difficult, since successful key management for encryption is typically the most difficult aspect of cryptography, especially with mobile systems. The problem is exacerbated with the need for speedy deployment and nimble reconfiguration of military teams to respond to rapidly changing conditions in the modern battlespace. 3 D. Alberts, J. Garstka, F. Stein, Network Centric Warfare: Developing and Leveraging Information Superiority, 2nd Edition (Revised), OASD Command and Control Research Program, p Ibid, pp U.S. GAO, Defense Acquisitions: 2009 Review of Future Combat System is Critical to Program s Direction, 4/10/ Network-centric Warfare, Wikipedia.com White Paper: How to Develop Predictable On-the-Move Communications, Page 2

4 Historical Approaches to Net-Centric Systems Development Simulation Network simulation models very complex systems. Over the last 20 years, scientists have developed models of physical networks or networked systems, most of which abstract, or simplify, the system. Abstraction is practiced in network simulations because complex networks can take a prohibitively long time to generate simulation results, even on fast computing platforms. The use of abstraction makes simulation an impractical method of testing extremely complex systems. When a model is abstracted, it can mask the very phenomena that enhancements are intended to produce. Much of the challenge in network modeling resides in building high fidelity yet efficient radio, channel, and mobility models. Network models that fail to take into consideration details like radio frequency propagation, channel interference and precise node location may render experimental results meaningless. When abstraction isn t used, many simulators are overwhelmed by the computational load. Results from non-abstracted simulations may take weeks to generate for a one-hour experiment. Thus, simulation is only valuable for testing complex networks, if there is little or no abstraction and results can be generated quickly. Physical Testbeds The purposes of prototypes are: To verify that a system provides the advertised capabilities, ensure that these capabilities are accessible by users, and to allow random, possibly unexpected events to reveal deficiencies in the technology. When building a network system, the most accurate way to test the design is through testing of the actual devices. However, there are many limitations with physical testbeds, most notably the cost. An early stage military radio prototype can cost upwards of $100K. Plus, if the design is allowed to reach the prototyping stage without proper testing in the design phase, the cost to go back to the drawing board is even greater. Another problem associated with physical prototyping is hardware management. Monitoring nodes and maintaining them, as well as managing experimental controls, is difficult and time-consuming. Configuring topology, applications, mobility and experiment details is arduous when you are moving tanks, UAVs and soldiers around a full scale field test. In order to produce a scientifically rigorous study, repeated trials must be performed. Being able start and stop trials, as well as pause and modify parameters on the fly is also necessary to efficiently conduct experiments. All these are difficult to do quickly and efficiently with physical testbeds. White Paper: How to Develop Predictable On-the-Move Communications, Page 3

5 The Solution Real-Time Wireless Network Emulation In computer science, emulation is the ability to replace live physical components of a complex system with counterpart digital representations. True emulation an exact digital replication is achieved when any player in the system hardware, software or human is not able to discern any difference between a real system component and its emulated replacement. True emulation then, by definition, is real-time emulation. Real-time network emulation uses high fidelity models of protocols and applications in real-time. All aspects of the network are emulated at the highest fidelity except the physical hardware, environmental effects and mobility, which are simulated. Network emulation blends the flexibility and low cost of simulation with the highly accurate, highly detailed results of a testbed. It is valuable throughout the life-cycle of the network-centric system and can be employed in increasing level of detail and realism as the project demands. How Emulation Speeds Technology Maturity Best of breed network testing and evaluation solutions incorporate hybrid systems of simulated, physical (real) and emulated components into a single evaluation platform for scalable and realistic network experimentation. This hybrid system provides the experimenter with the best trade-off in terms of realism, scalability and repeatability, depending on the target network scenario. Type of Emulation Prototype Emulated System Total Cost Physical Devices Only 200 $50K N/A $10,000,000 each Emulation + Physical Devices, Small Scale Emulation + Physical Devices, Medium Scale Emulation + Physical Devices, Large Scale 2-3 $50K each Dual-core computing platform emulation 200 nodes in QualNet 5 $50K each 8-processor system emulating 2000 nodes in QualNet 5 $50K each 32-processor system emulating 5000 nodes in QualNet Figure 1: Cost of Prototype vs. Emulated Systems for Testing and Evaluation Figure 1 is a cost analysis for physical testbeds alone vs. combining emulation with physical testbeds. The cost estimates include hardware and software costs but do not include labor costs for setup of prototypes, which adds considerably to the Physical Device Only setup. Setting up prototypes is a time-consuming, exacting science. Following are some examples of network constructs you could test with emulation that would be costprohibitive with physical devices: urban vs. rural environments weather effects, such as the influence of snow or rainfall on network performance $200,000 $500,000 $1,000,000 White Paper: How to Develop Predictable On-the-Move Communications, Page 4

6 compromised or destroyed nodes on a large scale Design flaws can be corrected at early stages in the design process, saving the network designer considerable time and money versus if these flaws were detected later in the development cycle. We term this emulation-based analysis. Figure 2: Emulation Enables Greater Technology Maturity in System Design and Testing The V-chart in Figure 2 shows the effect of emulation on the development cycle of network-centric systems. Development progresses from left to right. In the System Design stage, or left half of the V, first broad ideas are developed, and then the detailed ideas / component level details are fleshed out. Flaws in the designs, when caught in this phase, are relatively inexpensive to rectify. Once the System Design phase is over, System Testing takes place, which is depicted as the right half of the V. During System Testing, first the components are evaluated with respect to their original requirements, followed by increasingly larger chunks of the solution, culminating in the test and evaluation of the entire system itself. Flaws caught in the System Testing phase are far more costly to correct than in the System Design phase. Before emulation, developers of network technologies could use network simulation in the early system-level design phase, and then physical testbeds in the later system testing phase. Emulation allows many more opportunities to test and evaluate designs throughout the maturity-realism continuum. In other words, emulation allows testing that s better, cheaper and at a variety of granularities. The end result is network-centric systems can are better tested and understood more deeply at much earlier stages in the development cycle. With emulation, network and component designers need build only a few prototypes for most or all of the testing and evaluation process. Network protocols running on actual hardware interface with unabstracted high fidelity protocol models, which are emulations of the protocols running on the real hardware. The prototype hardware nodes interact with emulated nodes in various scenarios in the simulation construct, thus validating protocol behavior with real hardware in the design phase. White Paper: How to Develop Predictable On-the-Move Communications, Page 5

7 High-Fidelity, Unmodified Application Testing The interactivity of real-time emulation also extends to software applications. For example, a command and control application that will be deployed over the on-themove communication infrastructure can be run on top of the digitally-represented network so that a warfighter will get the same experience of using this as they would if they were out in the field. Urban, rural, and mixed contexts can be emulated, giving warfighters and commanders those experiences in advance. Given what s at stake to meet network centric operations objectives, it is absolutely critical that we study network designs in as high a fidelity and complex an environment in virtual space as we would in physical space. Emulation allows you to represent the communication infrastructure at such high levels of fidelity that applications running on it say a mix of third-party streaming video, voice over IP, , chat, video web conferencing, video teleconferencing can be deployed unmodified on top of large emulated networks of both legacy and future communication devices. What To Look for in an Emulation Solution Real-time Operation For an emulation to perform exactly the same way as the physical system, it must run in real time, all the time. The physical sub-network must not be able to distinguish physical devices from its real-time emulations. Real Packet Transmission Packet contents inside the emulation must be the same as packet contents outside the emulation. This allows for packet capture anywhere within the network, regardless of whether a node is emulated or real. Transparency Since are the packets are real within the emulation, we also need transparency to take advantage of this fact. Transparency allows network designers to interact with the network live, including: passive monitoring, such as collection of state information, traffic insertion, such as introduction of traffic to measure achievable bandwidth or delay triggers, such as conditional commands that notify a network engineer when a particular node moves into a particular region, or execution speed control, such as pausing, slowing down, or even fast forwarding uneventful periods in the scenario. Real-time emulations using real packets that allow dynamic interaction are extremely powerful tools. They are true replacements for networks and allow a myriad of third party applications, such as Semi Automated Forces (SAF) applications, HP Open View, and Ethereal to interact with the emulation. Examples of ways that third party applications can broaden the capabilities of an EXata emulation include: White Paper: How to Develop Predictable On-the-Move Communications, Page 6

8 applications that locate the source of a bottleneck can assist in solving network design problems applications that suggest alternate network topologies and constructs. Applications of Real-Time Emulation Technology Upgrading Existing Networks Existing tactical communications networks, such as the TADIL-J in use today by the U.S. armed services and the National Security Agency (NSA) as part of the NATO Link 16 system, lack the modern capability to route specific data packets to specific addresses. TADIL-J only supports broadcasting, making it susceptible to malicious network-centric threats such as jamming and eavesdropping. In the emerging Global Information Grid (GIG), packet-based, routable wireless networks based on the e standard enable new capabilities to integrate voice, data, and video information in real time on the battlefield. Real-time emulation enables engineers to rapidly develop detailed understanding of security in every layer of the wireless protocol stack: physical, media access control, network routing, transport/transmission, and applications. Integrating Specialized Networks Today, every government agency that depends on mobile communications to fulfill its mission operates with its own private network, with unique specifications for equipment and transmission signals. The benefit of this approach has always been rock-solid reliability and interference-free field communications. The downside: these networks present a major obstacle to cross-agency, first-responder coordination, particularly during widespread emergencies like 9-11 or Hurricane Katrina that quickly overload the public telecommunications network. Government agencies at all levels are now making interoperability connecting specialized networks in a way that will enable the agencies to break down communication barriers a top priority. One of the first large-scale projects to address interoperability was the High-Speed Data Communications System Project in Pinellas County, Florida. The mission of this project is to develop a new high-speed wireless network to augment the existing public safety wireless voice and data system. Real-time emulation is being used to develop and test a national interoperable system-of-systems framework. Integrating Homeland Defenses with Adaptive Networks National defense is becoming network-centric. Mobile communications networks used by the military will need to interoperate with legacy and emerging communications technologies used by domestic public safety agencies, relief organizations, and private sector entities in order to maximize execution effectiveness in the field. Federal, state, and local governments will need to perform two communications improvement feats simultaneously: integrate new technologies into existing infrastructures to improve interoperability and balance the ability to share information with the need to protect it from external threats. White Paper: How to Develop Predictable On-the-Move Communications, Page 7

9 The next generation of field tactical communications for both warfighting and homeland defense operations will be adaptive networks software-based radios and transmission equipment that can adapt dynamically to conditions and seamlessly interoperate with disparate communications technologies. Real-time emulation provides the accuracy and scalability that will be essential to the development of networks based on adaptive technologies because of sensitive interactions between the complex software layers involved. Adding Communications Realism to Training Real-time emulation can improve warfighter and commander training in three ways. The first is in exercising users in the specific ways they can use on-the-move communications to maximum advantage in accomplishing their mission. An important element of this type of training is to let signal officers and warfighters experience system failures, such as loss of communication assets or the system not performing as expected. Exposing warfighters to those frustrations and threats in a virtual context enables development of strategies to deal with them and to mitigate their impact in advance of actual deployment. The second important application of real-time emulation is in updating TTPs tactics, techniques, and procedures for network-centric operations. By exposing the warfighters, commanders, and fighting units to a digital representation of a future on-the-move communications infrastructure, in the context of the activities in which they will engage, effective and appropriate modifications of TTPs can be made to increase warfighter effectiveness. The same virtual network can also be used to train warfighters and commanders on the modified TTPs. The third way that real-time emulation can improve training is in war gaming. Most war gaming systems assume perfect communications, meaning that a message sent by entity A to entity B, is assumed to be received instantaneously and with 100% reliability. Anyone who uses a cell phone knows that s a myth. On-the-move communications are not likely to be instantaneous or guaranteed. Real-time emulation can inject realistic communication infrastructure and environmental effects into constructive, live, or virtual systems to better prepare the warfighters for the transition to networkcentric operations. Accelerating Tactical Communications System Development First-generation emulation technology is being used by L-3 Communications to develop the MR-TCDL, a Wireless, IP-based, Mobile Ad Hoc Networking System that provides high bandwidth capacity over very long ranges. SNT s first-generation emulation technology is used to validate MR-TCDL applications and hardware and refine system designs. Prior to test flights featuring real aircraft, L-3 connects real /MR-TCDL prototypes to an emulated fleet of aircraft. The pre-test-flight validation process, which takes place early in the system design and development phases, is intended to reduce the risk of system or network failure during system testing and actual flight test acceptance phases. With real hardware and actual algorithms integrated into the emulation, protocol service verification takes place throughout various stages of product development, not just at the final step. Emulation is critical to understanding and refining the perfor- White Paper: How to Develop Predictable On-the-Move Communications, Page 8

10 Flight Testing Simulation/Emulation Network Ground Site Emulated Emulated Emulated Emulation Node Simulation Node True HW SWNode Emulated QualNet Cluster STK Visualization L-3 SIM View Figure 3: Sample emulation architecture with flight testing. The scenario includes real, emulated and simulated nodes. mance of IPCM, network protocols and other network services. Figure 3 is a schematic of the emulation architecture with flight testing. Once the test flights are underway, emulation is used to provide flight test support with real-time 3D visualization of link status and telemetry data. By providing operational-level validation of networking software and hardware, emulation helps L-3 build a better solution faster at lower cost. Real-Time Emulation with EXata the Parallel Processing Advantage Real-time emulation has two fundamental requirements. First, you must represent communication infrastructure at a very high level of fidelity, so you can t tell whether it s real or a digital replica. Second, you must be able to scale the emulation to very large numbers so that you can evaluate networks in the sizes in which you would actually deploy them. EXata from Scalable Network Technologies breaks through the computational barrier faced by current simulation and emulation tools. EXata is built on a robust parallel simulation kernel, enabling the models within to be arbitrarily complex and arbitrarily large. The kernel of EXata, and its sister product QualNet, are the only commerciallyavailable products in the simulation domain that natively run on parallel architectures. Our kernel design enables EXata to operate on every generation of parallel computer architectures from multi-core PCs to supercomputers, and also gives us the ability to scale the size and complexity of emulations without compromising on speed. Our advanced parallel technology in the emulation space is analogous to Google s search engine compared to previous search engines. Google correctly anticipated White Paper: How to Develop Predictable On-the-Move Communications, Page 9

11 the growth of World Wide Web and designed its search engine to be inherently scalable through the power of clusters and farms of computers. In the same way, Scalable Network Technologies predicted that the complexity of on-the-move communication networks would only increase, at a much greater rate than the speed of computing power. We chose to rely on the power of parallel computing to allow us to keep pace with the increasing complexity of emulating these networks without compromising on speed or fidelity. As we said in the beginning of this paper, the military does not yet know if the future on-the-move communication networks that have so much promise will perform as needed to support network-centric operations. That s the question that EXata technology is designed to answer. The best way to demonstrate the effectiveness of groundbreaking, self-healing on-the-move networks is to operate them first in virtual space. Warfighters and commanders can then watch their wireless networks perform before they depend on them, building confidence that the demands they will place on these networks will not cause them to buckle at precisely the wrong moment. They can also develop alternative methods to recover performance and modify operations to deal with the possibility that the network might fail. In order for network designers to know that the network can deliver the throughput, latency, and the jitter that it s intended to, they can look at alternative designs, not just in terms of network level metrics but also view their impact on the performance of the actual application mix that is likely to be deployed on these networks. They can conduct realistic testing, not just in well laid down, unrealistic settings, but in the context of urban environments, so that the points of field error can be identified early in the design cycles and ways to mitigate that effect can be designed in and built in to their evolving network. EXata offers another unique advantage in the training arena. EXata is designed to work in an embedded context. By embedding EXata software within the tactical equipment and infrastructure used by our fighting forces, we directly enable our forces to train as they fight. By embedding a virtual network within tactical equipment, we directly enable our forces to deploy the set of applications, the mix of C2 and other battle command applications that they d be using, on top of the virtual network and deploy it within the context of the environment that they would use to actually wage war. White Paper: How to Develop Predictable On-the-Move Communications, Page 10

12 EXata: The Fast Track to Mature Network-Centric Systems Technology Real-time emulation is the solution for rapidly maturing the robust network infrastructure technology on which mobile network operations will depend. EXata from Scalable Network Technologies is the only commercially-available emulation tool built on a fully parallel kernel, enabling scientists and engineers to test networks of up to thousands of nodes, at real-time speed in virtual space, with the highest fidelity. To take your first step to realizing the full potential of network-centric systems, please call or sales@scalable-networks.com. About Scalable Network Technologies Headquartered in Los Angeles, California, Scalable Network Technologies is the leader in parallel processing technology for network performance evaluation. The company develops and supports high fidelity evaluation software tools used for predicting the performance of computing and communications networks and network devices. SNT has created a new category of evaluation tools for today s sophisticated networks that meets the demand for real-time, real-network performance testing. Widely recognized for its flagship product, QualNet, the company s customers include major aerospace and defense contractors, the US Department of Defense, mobile network operators, as well as research agencies and universities. For more information, please visit: Scalable Network Technologies, Inc Center Drive Suite 1250 Los Angeles, CA Copyright 2008, Scalable Network Technologies, Inc. All Rights Reserved. White Paper: How to Develop Predictable On-the-Move Communications, Page 11