Strategy for an Army Center for Network Science, Technology, and Experimentation (2007)

This chapter discusses the infrastructure needed for the Army to develop a world-class network science, technology, and experimentation (NSTE) capability. Many of the organizations currently engaged in NSTE will move to Aberdeen Proving Ground, but the committee recognized that much of the NSTE infrastructure would continue to be distributed. Consequently, the chapter also discusses the advantages and disadvantages of distributed and centralized approaches to organization.

All of the network areas shown in Table 2-3 (i.e., communications and information, human performance in networks, adversary understanding, and non-physical network science research areas) have infrastructure requirements that are necessary to support NSTE activities. The first two areas tend to be focused on the development of new capabilities associated with enabling information networks to support network-centric operations and on the social and human networks involved in decision making. Network-centric operations involve people, processes, and information technologies working together to enable timely and trusted access to and sharing of information, as well as collaboration among those who need it. The Global Information Grid (GIG) is the foundation for enabling essential social and human networks that lead to effective and timely decisions. Army NSTE must be capable of supporting development of these emerging complex social/human models.

In addition, there is a need to conduct experiments that can improve the understanding of the usability of these new decision-making networks in a manner that will translate into new capabilities. This improved understanding of information technologies and the decision-making process will also help the Army to develop new capabilities that will deny adversaries similar advantages.

Hence, the NSTE infrastructure being sought by the Army must also be capable of supporting experiments for the development of new technologies to disrupt the information and human networks of adversaries.

The infrastructure framework for an Army network science, technology, and experimentation center (NSTEC) must be capable of supporting:

• Development of fundamental network theory and network technologies,

• Assessment of impacts on human performance,

• Integration of new technologies and social networks into capabilities, and

• Experimentation as a means to test and confirm fundamental theories and models and/or to characterize new technologies and operational concepts while also being capable of promoting training of personnel when applicable.

Network science theory currently is focused on fundamentals of information theory, decision making, and understanding of network vulnerabilities. These theoretical foundations are based on those elements of networks that technologies can provide and support, as well as on improving the understanding of how social/human networks evolve with respect to new network technologies (e.g., getting humans connected in different waysas an example, consider how teenagers stay continuously connected (“24/7”) via wireless cell phones, both synchronous and asynchronous). The 24/7 wireless cell phone is more than just voice, including an ever-expanding richness of collaboration and social networking never seen before, as evidenced by new applications of multi-point video, gaming, music, etc.

Theories of networks and decision-making models need experimentation to test or confirm, discard, or modify the principles on which a fundamental understanding of the subject is based. These “network models” are essential to characterize how new network technologies will scale with size or degrade due to “black-outs” and malicious attacks. However, experiments on network models are very sensitive (in a very non-linear way) to the environment, and innovative approaches will be needed to design an infrastructure that enables live and virtual simulations to be performed at geographically separated laboratories.

For some time, there has been recognition that conducting complex network experiments, including social/human behavior, is very difficult in a controlled environment. That is, successfully achieving the integration of interdependent and very mobile/dynamic network technologies and social/human networks is the “Achilles’ heel” of successful network experiments. Special consideration must be given to the infrastructure testing environment (complex wireless propagation,

mobility, chaotic behavior, etc.) to allow various levels of integration (one-to-one systems, multiple-systems, and system-of-systems), and to allow testing of various technical issues, theories, and models. The importance of integration has been highlighted in earlier “digitization” efforts by the Army, and the essential role of integrating human performance in battle command exercises has been studied in multiple C2 experiments for the Future Combat Systems (FCS) (Lickteig et al., 2003). Other complexities of the system-of-systems integration environment are discussed in Krygiel (1999).

The development of new theories, models, and technologies requires a world-class environment that will foster innovation and “network thinking.” A world-class environment should have a university or campus-like feel with interconnected buildings, libraries, open areas (inside and outside) for discussions, public spaces with wireless Internet access, multi-media information displays, conference facilities, and lodging for visitors. State-of-the-art collaboration facilities are needed that incorporate best practices from industry and academiasuch as the “smart” rooms (intelligent agent technology, voice/gesture activated, etc.) present at the MIT Media Lab. There will also be a need for some dedicated facilities for secret, top-secret, and sensitive compartmented information.

Theoretical research will require extensive network modeling facilities, including a computing infrastructure with supercomputing capabilities. An improved understanding of the complex, large-scale, and dynamic nature of human and social network interactions will be advanced through the use of high-fidelity models that include physical wireless networks, i.e., complex models that will involve very intensive computations. New approaches in areas such as parallel computing, biological computers, and large-scale network computing (grid computing) will be required. In addition to the computing infrastructure, the investigation of new networking technologies (e.g., large-scale ad hoc networks) places new demands on reconfigurable network labs (with emphasis on wireless functionality) that can quickly adopt new technologies and easily be adapted to human/social networking.

The computing requirements for such research will require an infrastructure that includes a set of dedicated labs (i.e., facilities for sensitive compartmented information, command and control, advancing networking, etc.) that are designed around a controlled environment with human subjects as participants. An NSTEC would also have to establish laboratories in field locations with access to soldiers and leaders who can serve as subjects in experiments. These field-location laboratories must be dedicated sites that are highly instrumented and allow for the collection of data that can be analyzed and used in developing new models and refining a theoretical understanding of network characteristics and applications.

In addition, these dedicated labs would host new information technologies

(e.g., physical networks, command and control, networked sensors, etc.). Some labs, for example, might host a set of technologies dedicated to research on adversary networks. Due to the necessary levels of security and nature of this work, these special dedicated labs would require that isolated and secure networks be in place that could quickly improve awareness of vulnerabilities and promote the development of new technologies to deny network-centric operational capabilities to adversaries.

High-capacity (i.e., gigabyte) optical fiber and wireless (Wi-Fi and next-generation 802.xxx systems) networks are key enablers for mobile multi-point and multi-media (data, voice, video, etc.) information transmission. These networks would ensure GIG connectivity to the other Army and governmental organizations concerned with network advances, such as the Army Materiel Command, Department of the Army staff, the Army Training and Doctrine Command, the Army Intelligence and Security Command, the Defense Information Systems Agency, and the Department of Homeland Security. An optical fiber backbone would enable access to a rich set of network services (including security authentication, data management, collaboration, discovery, etc.) as well as to other NSTEC facilities and test ranges. All facilities and the computing infrastructure would need to operate through this high-capacity backbone and be interconnected with all associated sites. Given the classified nature of some work, these networks would have to be secure and approved for use at different levels of security classification (e.g., secret level, top-secret level, sensitive compartmented information, etc.).

Critical nodes in the Army networks of today and in the future are the soldiers and leaders who (1) seek and use complex information from networks; (2) must quickly make decisions based on that information and communicate and collaborate with subordinates, peers, and superiors over the network; and (3) control actions and equipment on other network nodes. In the areas of human performance, the focus of the NSTE infrastructure framework is on “intellectual capital” and research that is complementary to network science theory, integration, and experimentation. Indeed, the dynamic configuration of the networks per se should be driven by the human need for information and connection necessary to make decisions. Four actions are required to ensure that human performance in networks is optimized:

1. A substantial increase in ARI and HRED efforts is needed if research on human performance in networks and adversary understanding is to have any impact. This does not mean a change in emphasis solely to add network-related research: both ARI and HRED have important research missions without the addition of network research, and that other research

is funded at levels that are barely adequate. The committee believes that the Army definitely needs to have more personnel than the roughly 15-20 scientists who are currently conducting or scheduled to perform research in behavioral and social science areas related to networks at ARI and HRED. Substantial growth in both personnel and resources is called for, with something on the order of 20-25 percent of the Army’s fiscal and personnel resources for NSTE dedicated to human performance in networks and adversary understanding.

1. Hire or retrain in-house scientists in such relevant areas as social networks, decision making and other cognitive processes, behavioral representation and modeling, performance measurement, and neural sciences.

2. Require teaming between information, human factors, and behavioral scientists so that network interfaces are compatible with human capabilities and information requirements to promote flexible and reconfigurable networking.

3. Ensure that the impact of research in information and communication science is measured in terms of how it benefits human performance in the network, both in the laboratory and in field environments. This will require new experimental paradigms and measurement techniques and tools (as described in the section titled “Experimentation” below in this chapter).

As mentioned earlier, an integration capability that can handle the increasing complexity of interdependent technologies and a scale of experimentation that includes human/social networks is essential. This will call for a network science integration facility (NSIF) that should include several key components as follows:

• A large-open-space laboratory area to allow the integration of network technologies onto platforms (vehicles, robots, humans, etc.),

• The ability to accommodate a scale-up of as many as several hundred nodes (i.e., a number representative of the types of networks the Army plans to field; known also as the “sub-net topology”) in a virtual network with the capability to have a large fraction of the nodes in motion, and

• Full instrumentation (including for social/human networks) to allow detailed diagnostics of integration and performance issues.

The NSIF should be connected with other network test beds, such as those of the Army Future Combat Systems (FCS) program, and it should be able to exercise control over other network test beds as necessary to expand overall experimentation capabilities. The NSIF networks will require significant networking

capacity with external networks; indeed, extending the Department of Defense Information Systems Network (DISN) GIG into the NSIF environment should be a high Army priority, regardless of expense.

Integration with network systems outside the NSIF will enable the connection of various elements of research in a manner that allows interoperability and performance issues to be characterized and resolved. In addition to the DISN GIG, this integration should include the networks of other DOD services, federal agencies (intelligence, Department of Homeland Security (DHS), health, local, etc.), and selected coalition/multi-national partners.

The DOD has long embraced experimentation as a fundamental tool for building capabilities. As in the physical sciences, experimentation is essential in this case for improving fundamental understanding and expanding knowledge in the areas of information and communications, adversary understanding, and human performance in networks. In addition to the necessary infrastructure and tools, successful experimentation requires a culture for “new system thinking,” committed leaders (because experimentation is naturally disruptive of the status quo), and skilled personnel.

The basic infrastructure and tools needed for the experimentation environment are discussed in Boutelle and Grasso (1998) and NRC (2004). These include:

1. Information and physical infrastructure

   1. Networks

   2. Information repositories

   3. Architecture frameworks

   4. Test facilities and integration capability

   5. Training facilities

   6. Places and platforms

2. Tools

   1. Modeling and simulation

   2. Prototypes, surrogates, etc.

   3. Artificial environments

   4. Data capture and dissemination

Most experiments will require networking connectivity and capacity as well as computing power and data storage. In addition, experiments will be dependent on the existence of accepted operational, technical, and system architectures for the integration of assets and their application to scenarios and missions. Having data repositories with such information readily available and accessible to other

locations can ease experiment planning and support the extension of results to other cases or scenarios (NRC, 2004).

An instrumented test range requires a physical infrastructure. As discussed earlier, core facilities for such things as simulations, war games, integration and testing, and training must all be provided as part of a distributed test range. The associated facilities must be equipped with the basics of uninterrupted power, good lighting and ventilation, suitable climate control, and adequate space as well as any specialized equipment needed for specific activities. Equally critical is the timely availability of networked vehicles across distributed experimental ranges, aircraft, and the various platforms that are integral to the experiments. Major elements of an instrumented-distributed test range for Army NSTE should include:

• A facility of ~100 nodes, most or all of which can be in motion (representing vehicles, in a typical future Army sub-net);

• The CERDEC Fort Dix command, control, and communications on-the-move (C3OTM) test bed that is fully connected back to an NSTEC and other Army test ranges (e.g., central technical support facility (CTSF) at Ft. Hood, Texas, Army FCS at Ft. Bliss, Texas, etc.); and

• The ability to conduct “adversary network attacks” and appropriate diagnostics.

As the Army learned from its earlier experiments in digitization, the integration of the above test ranges will be critical to supporting future experimentation objectives (Boutelle and Grasso, 1998; Krygiel, 1999). The ranges will combine with other components to allow collaboration experiments and a wide exchange of new innovative concepts.¹

External network facilities need to include:

• Capability to connect into all elements of NSTE infrastructure, including ranges and network C2 lab facilities;

• A network C2 facility for ~50-seat fixed nodes completely under the control of an NSTEC;

• Configurable laboratory space (similar to the CTSF at Fort Hood) to allow for participation by representative command elements (i.e., C2, battle management, squad/company/battalion/brigade, joint/multi-national); and

• Ability to connect/interact with other external C2 facilities (e.g., Fort Dix, Fort Hood, universities, joint experimentation facilities, etc.).

Ultimately, experiments must support embedded and distance training and learning capabilities via a system that links platforms into a virtual environment

that will be necessary to provide an appropriate level of realism in live experimentation conducted across a large network. Connecting remotely into training systems that are integrated into platforms will help operators and experimental investigators to shift seamlessly between training and experimentation. Note that the operators of the units participating in experiments must be aware of such goals and be well trained in their usual roles as well as agile enough to handle new, and networked, surrogate systems and new concepts. The spectrum of experiments to be conducted would include human-in-the-loop experimentation, prototypes, surrogates, and stimulators. Artificial environments sometimes would be required.

As previously noted, network science experimentation requires new “tools” with emphasis on an integrated family of models ranging from customized spreadsheets and war games to high-resolution simulations. These tools should include a sophisticated system for data collection and information capture, and they should provide a capability for interpretive analysis so that learning derived through experiments can be put to use in informing decisions about future forces (NRC, 2004). Simulation and stimulation facilities, such as those currently at the CTSF at Ft. Hood and at the National Training Center (NTC) at Ft. Irwin, California, are necessary to provide capabilities regarding:

• Force-on-force and counterinsurgency;

• Intelligence via remote sensors (e.g., unattended ground sensors ); and

• Modeling of complex environments (e.g., urban/town), as well as adversary social networks (e.g., tribal, financial, political, etc.).

Table 4-1 summarizes key NSTEC infrastructure elements and the features of each element. The table highlights characteristics of the infrastructure necessary to support NSTE in priority network science areas and to accomplish the mission recommended (see Recommendations 1a and 1c in Chapter 2).

Development of the NSTEC capabilities and infrastructure could be phased over multiple years, and there are several reasons why a phased implementation approach might be the most favorable for the Army. Foremost is the erratic nature of government funding for the BRAC, which has already affected the BRAC timetable. The requisite talents and skills of the leadership team that is needed are likely to be significantly different for different phases. Moreover, the content of the NSTE R&D portfolio will change as the infrastructure for the center is established and the new organization matures. Phase 1 might include infrastructure inherited from the existing BRAC planning that could later be integrated into the NSTEC. Phase 2 might include the infrastructure needed to “stand up” distinct NSTEC facilities at a target date, and Phase 3 might encompass the infrastructure needed for future growth and modernization due to expected advances in network science knowledge and technology.

In addition to the infrastructure needed to support a center for NSTE, a full NSTE capability will also involve government, industry, and academia, and all three must be considered in determining an organizational approach for an Army NSTEC. The unique contributions of each of these groups are described below.

Chapter 3 discusses the NSTE activities that are meeting current Army needs. The primary Army organizations engaged in NSTE are ARL (ARO, HRED, and CISD) for basic research; CERDEC (C2D, S&TCD, and C3OTM test bed) for applied and advanced research and technology development; and ARI for basic, applied, and advanced research. The estimates of annual NSTE investments in these three organizations are ARL, ~$70 million; CERDEC, ~$75 million; and ARI, ~$2 million, with corresponding staffing. The committee sought more detailed information on the actual funding levels and staff related to NSTE, but the Army was unable to supply usable data.

Universities provide access to state-of-the-art research at all levels, including the global level. A steady and continuing relationship with academia can ensure responsiveness to evolving Army sponsorship requirements as well as comprehensive understanding of the sponsor’s requirements and problems. It will be

important for an Army NSTEC to engage multiple universities in providing broad and diverse thinking on relevant advances in network science.

Other contributions made by academia may include:

• Broad access to information, including proprietary data;

• Independence and objectivity;

• Quick response capability; and

• Freedom from real and/or perceived conflicts of interest.

Industry provides the path most frequently used by the government for transitioning technology to the operating forces. Relationships with industry can also provide an NSTEC with knowledge concerning best commercial practices and cost-effective approaches. In addition, industry can provide commercial surrogates and prototypes (e.g., state-of-the-art wireless systems and networks), access to large-scale system architecture, engineering and integration expertise, models for commercial networks and information systems, general logistics support in experimentation preparation and execution, and commercial tools to aid analysts in interpreting data.

Other contributions made by industry include:

• Broad access to industry information, including proprietary data;

• Broad access to specialized facilities (e.g., software integration); and

• Support for technology transfer agreements between government and industry.

An NSTEC would be established to perform S&T in network science, network technology, and network experimentation. As discussed in Chapter 3, a considerable amount of NSTE work, on the order of $150 million annually, is funded by the Army today. However, current efforts in network science research are not adequate for Army needs, and there are important network areas that are not being addressed by the Army S&T program (NRC, 2005). In order to form an NSTEC, the Army would need to consolidate its efforts in network technology and experimentation and combine them with an augmented program of network science research. The importance of networks to military performance is increasing rapidly, and NSTE deserves substantially increased attention by the S&T community.

In the long term, the NSTEC should become a joint entity serving the needs of all of the U.S. military services in a coherent, collaborative fashion. While addressing the Army’s current needs seems to be a practical first step, a longer-

term vision should be maintained by NSTEC management. This is an important issue that will impact the scale and scope of the research and basic technology efforts in NSTE.

Just what should an NSTEC look like? Should the emphasis be on an attractive central campus featuring buildings and infrastructure, or is a virtual “co-laboratory” of distributed activities more appropriate? Do the imperatives of existing resources suggest an immediate change or an incremental approach to implementation? One thing is clear—an NSTEC, to be a world-class organization and a key contributor to the Army’s future success, must attract first-class talent, and that means providing cutting-edge challenges for top-flight scientists and engineers, providing them an environment in which to strive and achieve.

An obvious but far from optimal version of an NSTEC would involve merely relocating elements of various existing Army organizations to a centralized government-operated location such as Aberdeen Proving Ground. To be effective, any such relocation should involve an attractive physical plant with state-of-the-art technical facilities. This alternative could be considerably enhanced by delegating existing special authorities available to the Army for utilization by an NSTEC in areas such as personnel management, acquisition, and other financial and business practices.

The advantages of a distributed (versus centralized) NSTEC include:

• Access to highly talented and specialized knowledge and expertise in the large field of network science that can be leveraged globally;

• Cost-effective and efficient global access to specialized facilities, such as the MIT Media Lab, Microsoft labs, National Security Agency labs, and operating force experiments (e.g., regional joint forces experiments);

• Access to specialized platforms (e.g., aircraft, ground vehicles, ships, etc.);

• Access to industry and university laboratories and expertise; and

• Distributed operations that mirror how the Army fights in the field.

The disadvantages of a distributed (versus centralized) NSTEC include:

• Increased management overhead due to many distributed moving parts;

• Increased operating expenses (external networks, management, etc.);

• Information sharing across a large distributed environment involving data repositories and content staging cost and complexity; and

• Increased security risks.

There are key elements of infrastructure that will be essential to the creation of a center for NSTE. Current and ongoing NSTE activities in the Army provide a firm basis for the infrastructure requirements that would be needed to establish an NSTEC. The key elements of the required infrastructure are listed in Table 4-1.

Options for physical realization of an NSTEC range from the extremes of a centralized facility in a single location to fully distributed facilities using networked connectivity. There are advantages to both; but regardless of which is selected, there will be important needs for contributions from academia and industry, which will require interconnections to many locations.

Conclusion 3a: The extensive infrastructure needed to support Army NSTE requirements will be developed initially from the facilities of existing organizations and will require a special planning effort to synchronize with BRAC relocations already in progress.

Conclusion 3b: The magnitude and diversity of the required infrastructure suggest a phased implementation approach to establish an Army NSTEC. A plan to develop NSTE capabilities and infrastructure could be phased over multiple years, beginning with the reorganization of existing and relocated facilities and ending with the establishment of a world-class center for network science, technology, and experimentation. An adequate plan will involve leadership with the appropriate talent and vision for all phases, especially as the content of the NSTE R&D portfolio matures.

Conclusion 3c: Based on Army needs, the NSTEC should be a hybrid operation consisting of two or three centralized facilities having interconnectivity to a variety of distributed supporting elements.

Recommendation 3: The Army should plan and fund for NSTE infrastructure resources that provide for (1) flexible configurations of network experiments and integration, both internally and externally; (2) facility designs that enhance and encourage academic and industry partnerships; and (3) an environment with world-class experimental capabilities and a campus-like atmosphere to attract truly talented people.