For the military, decisive force, power projection, overseas presence, and strategic agility will be the strategic concepts to meet the challenges of the future. As first explained in Joint Vision 2010, today's military capabilities must transition to dominant maneuver, precision engagement, focused logistics, and full-dimensional protection. The evolution of these elements over the next two decades will be strongly influenced, first and foremost, by the continued development and proliferation of information technologies. Information superiority is the key enabler.
Information superiority--knowing more than enough about an adversary who knows much less than enough--is the key enabler for the practitioners of US diplomatic and economic policy, as well. Geospatial information is nearly always the key to an international engagement, whether on the grand strategic level or at the "tactical" level of flesh and blood and mud. From international borders to artillery aim points, from the flow of goods and services to the mobility of a tank, geospatial information paves the way and points out the opportunities.
Moreover, with the advent of commercially available, high-resolution (less than 1-meter) satellite imagery, the United States has lost the exclusivity it once had. These images will be available, as never before, to any potential adversary. While it may be regrettable, it is not possible (nor even desirable, on other grounds) to turn back the clock. The US answer must be to use its still considerable advantage faster and better. To state the obvious, imagery TPED, in all its dimensions, is the key to "faster and better." Our use of imagery and imagery-derived intelligence must put us "inside the adversary's decision cycle." The importance of TPED for information dominance cannot be overstated.
Everyone agrees that imagery TPED is critical for information dominance; not everyone agrees on just what TPED is!
Literally, "TPED" is an intelligence insider's acronym that stands for "tasking, processing, exploitation and dissemination" and is usually juxtaposed to a specific intelligence collection discipline--e.g., imagery, SIGINT, etc.--or to a specific intelligence collection asset. Thus, we speak of "tasking" an imagery reconnaissance satellite, "processing" its raw collection, "exploiting" its processed collection take, and "disseminating" the resultant information products. Such a recitation, however, may lead one to conclude that TPED is a neat, serial process. It is not.28
Nor is TPED a system. There is no single set of engineering specifications, nor will there be. There is no single systems architecture, in the strictest sense. By some lights TPED is a "system of systems" but even that construct is misleading. TPED does embrace a concept of operations from which one may infer certain architectural concepts and, looking to the future, one can substitute newer architectural concepts and modify--hopefully improve--TPED.
Some have suggested that we view TPED as the (real-time) supply-chain management for the Imagery and Geospatial Community (IGC).
Alternatively, think of TPED as shorthand for the ensemble of (people,) systems, and processes that add value to an intelligence collection system. This construct is especially useful insofar as it leads us to question whether a collection system by itself--no matter how technically elegant--is of value commensurate with its cost. The construct also allows us to consider separate elements of TPED functionality and ask, too, whether the value each adds justifies its respective cost.
Tasking is the value-adding process by which we try to ensure that the right image gets taken, at the right time. If collection capacity is a scarce resource, then tasking includes the optimization of that scarcity. Today--and, arguably for the indefinite future--technical insight into specific collection systems is necessary to accomplish good tasking. Consequently, a corps of trained intermediaries--who mediate between the information needs of intelligence consumers (as well as all-source analysts) and the tasking of collection systems--are, and will remain, a necessary fixture in the TPED process. Despite the intermediation, we must maintain a thread to those whose needs initiated the tasking and provide feedback--ideally with a predictive component--to the end-users as to the status of a request.
Processing is the automated, rote application of algorithms that transform raw collection take into a product better suited for exploitation by a diverse set of analysts and for a diverse set of purposes. There is a continuum between collection, processing, and exploitation. The collector can have embedded and/or "on-board" processing. Or processing can be at a "down-link" site. In any case, there usually are heavy computing demands and consequent economies of scale in processing, as well as a requirement for intimate technical knowledge of the collector. For these reasons, processing is more closely tied to collection than to exploitation, both in systems design and organizational responsibility.
Because the processing "system" has as its input a well-defined collection system specification, and because it controls explicitly its output specifications, it is arguably the easiest function of TPED to architect. Said differently, it largely is isolated from the vagaries of human interaction--"free will" being the archenemy of system architecture. There is a valid interest in migrating "upstream" into the processing segment those exploitation tasks that can be routinized and automated. This complicates only slightly the processing system architecture.
We might think of processing as the link in the chain that transforms "data" into "information" accessible to human analysts.
"Exploitation" is the most abstract of the concepts and, perhaps for that reason, the easiest of the TPED functions to define. Exploitation comprises all those value-adding activities that transform imagery into intelligence or, more generally, the link in the chain that transforms "information" into "knowledge."
Because there are still an infinite number and variety of exploitation algorithms yet to be discovered, one is challenged to devise a meaningful exploitation architecture.
Generally, dissemination is thought of, simply, as getting the right information to the right place, at the right time. It is sometimes useful to decompose dissemination into two parts: the physical process of getting it there, "distribution;" and the logical process of deciding "what goes where." Of the two, the distribution historically appears to be the more expensive and difficult, and the most boring. The logical process of dissemination is by far the more intellectually challenging.
Literally, USIGS stands for the United States Imagery and Geospatial Information System: the extensive network of systems used by the Department of Defense (DoD) and the Intelligence Community that share and exploit imagery, imagery intelligence, and geospatial information. These systems provide capabilities involved with the integrated management, collection, production, exploitation, dissemination and archive, and infrastructure of this information. Organizations that have some level of interface with USIGS, but are not part of DoD and the Intelligence Community, are considered participants in USIGS if they adhere to the technical and system standards.29
USIGS includes organizations, doctrine, standards, procedures, libraries, and hardware/software that collectively provide fused imagery, imagery intelligence, and geospatial information.
The Commission appreciates the Director's reformulation of NIMA as custodian of USIGS. Sometimes misunderstood, this reformulation is emblematic of a healthy change in focus, away from systems, away from products, away from processes, and toward information services.30
For this report, however, we persist in using "TPED" in deference to the sensibilities of the reader. In most cases, a simple substitution of "USIGS" for "TPED" or vice versa works. Thus, TPED acquisition is equated to USIGS modernization, for the most part--i.e., except for purposes of budgetary and programmatic continuity, perhaps.
TPED is truly a global enterprise that includes multiple suppliers (collectors), operating in different environments, and requiring significant supporting infrastructure. NIMA has (at times) described TPED as a system of systems that will provide the tasking, processing, exploitation, and information dissemination service for all imagery. This includes imagery collected by (theater) airborne assets and by national technical means (NTM) as well as those services provided by Commercial Imagery entities. Commercial services can range from raw images to value-added products and fully exploited information.
Programmatically, TPED more or less includes all the people, hardware, software, communications and "O&M" for the entire Imagery and Geospatial Community (IGC) from the "national" level down to the theater JTF/component level.
The approach taken by NIMA is to fully modernize USIGS/TPED rather than incrementally upgrade individual components as necessary to be compatible with the NTM collectors of the FIA era. This comprehensive approach, which demands significant investment, is the only way to transition quickly to the information-centric architecture, which the Commission endorses.
Costs are proportional to a number of factors; among the big swingers are size of the IGC, size of the images, number of images. Note that if an image improves in resolution, say from 1 meter to º meter, the storage required, the bandwidth required, and the processing power required all go up by a factor of four if the area covered remains constant. But, of course, the area covered might drive each cost up by another factor of four. If the number of images per day increases by several score, these costs, again, rise proportionately. As the uses of imagery and geospatial information become more widespread, the community of users can double. And of course, multiplying all these numbers together, as we must, results in an answer that is large, impressively large, daunting to some. Such is the price of information dominance.
One of the challenges to NIMA is how to manage the significant increase in collection capability that will result from (EIS and then) FIA, and from increasing availability and capability of commercial imagers. Ensuring that tasking is assigned to the right collector is particularly challenging as airborne assets are under theater control, and commercial imagery is subject to the various terms and conditions negotiated with the respective vendors. Ensuring timely exploitation in the face of higher volumes and fewer analysts is challenging, as well. Not to mention ensuring timely distribution over communications channels managed by another agency and procured from various commercial sources.
NIMA is not yet well-positioned to acquire TPED (i.e., to modernize USIGS). As a new organization, it did not inherit from its forebears the systems engineering and acquisition personnel and institutional knowledge. This is reflected in lack of a stationary baseline architecture. As we discuss below, growing this competency is particularly difficult in this economy where the civilian sector easily outbids traditional government organizations for the needed talent; it will require extraordinary measures.
Despite administration neglect, Congress may provide NIMA with the necessary infusion of resources to start innovative TPED architecture work. To take full advantage, NIMA will have to consider innovative TPED "suppliers" beyond traditional aerospace contractors.
NIMA's TPED system is increasingly akin to an information system built for commercial customers by commercial contractors using commercial methods and commercial standards and employing technology to which DoD adds little. True, NIMA's TPED system is not quite identical to anything else (but no sufficiently complex system is without some unique features). It will be huge and girdle the globe, but there are other systems of comparable size (e.g., oil company seismographic records), data complexity (automaker-supplier CAD networks, inventory systems, commercial GIS products, market data warehouses), and reach (many large banks and credit card companies).
Because of the enormous potential for commercial technology, the Commission feels that NIMA should be more an acquiring organization, less a developing organization except in very specific areas such as imagery science. Nor should NIMA take on the role of system integrator. The Commission has not seen evidence that NIMA currently has the expertise or experience to prepare a comprehensive plan to acquire and integrate a system of systems such as TPED. This lack of expertise is exacerbated by the fact that NIMA must migrate a large number of legacy systems while maintaining operations.
As we reemphasize below, the Commission believes that a Technical Advisory Board of outside experts could serve the Director of NIMA well.
As stated previously, the Commission does not believe NIMA is making maximum effective use of commercial hardware and software. It appears to be depending heavily upon its current processes and products and persists in developing government standards that diverge from emerging commercial standards.
While it is recognized that use of GOTS may appear to be the most cost-effective short-term solution, a coherent strategy is needed which balances the use of COTS, GOTS, and customized hardware/software, recognizes the advantages and disadvantages of COTS and GOTS, and plans for the long term. The long-term view is of particular importance because TPED, and USIGS, must be able to infuse new capabilities and technologies.
In addition, it is becoming evident that future capabilities in TPED will be very dependent upon COTS. The Commission recognizes that use of COTS presents new challenges to the government to be a smart buyer and user. NIMA has not shown that it has the necessary expertise and experience to effectively integrate many COTS products into a large system of systems such as TPED.
The Commission stresses that an important step on the road to realizing fully the benefits of commercial technology will be the use of commercial, rather than government standards.31 Without standards that interface with the commercial world, it will be very difficult to accommodate future products and NIMA will be maintaining yet another obsolete system.
The rationale for COTS products is obvious: they exist, they work, and they evolve quickly as the marketplace expands. Because development and maintenance costs are amortized over many users, COTS products are usually less expensive to acquire. Buying a COTS product worth hundreds of dollars allows the USG to cash in on sometimes millions of dollars of corporate development. Buying into a solution that someone has already devised means less need for reinvention. Being able to "try before you buy" means less likelihood of error. With a large user base, COTS is more likely to be supported by third-party applications, tools, services, and training. And widely used COTS products mean that NIMA and its users can interoperate more easily with each other, with other developers, and with other geospatial data providers.
Not all COTS products are equal. Ideally, if a COTS product is to be considered it must be able to succeed in--that is, ship in volume to--the commercial marketplace. Even better, it should have evidenced some staying power already, and had the kinks worked out (e.g., version 3.0 or later).
A recent study performed by Aerospace Corporation32 indicates that the government has yet to develop an effective acquisition model for commercial technology-especially software. Much has been written about the benefits of COTS technology, however, the government, according to the study, has yet to let go of the outdated acquisition and development cycle models that require customization and duplication. NIMA must discipline itself to avoid following a commercial path for only part of the way, then reverting to blind satisfaction of requirements without performing cost and benefit trade-offs.
Will commercial products provide everything NIMA wants? A good architecture ought to make it easy to know whether a given requirement can be so satisfied. As a guess, commercial database and GIS tools are likely to satisfy a very high percentage of NIMA's requirements out of the box. The percentage of analytic tools (e.g., for modeling and simulation) that are commercially available is likely to be far less. When NIMA has a requirement unsatisfied within COTS, it has three choices besides reinventing the wheel: pay commercial contractors to support certain features in these versions, wait for subsequent versions, or make do without. Paying for additional features should be a seldom-exercised option lest COTS acquire the meaning: customized off-the-shelf (often, additional features have to be rewritten every time a new version of the base software is issued).33
The IEC program--a sad story, but with a potentially happy ending--illustrates the value of COTS products. The Commission has met with imagery analysts who expressed dissatisfaction with IEC--their complaint is that the IEC's effective, smooth "roam rate" is half that of the system it replaces.
The Commission is perplexed that NIMA would approve, fund, and execute a project to replace IDEX II with a design that, from the start, did not meet one of the most critical requirements for imagery analysis. In addition, the Commission has concerns over the large integration efforts to cobble together various software packages, especially where many of these applications are already available as integrated solutions. Addressing those two issues will likely cause both deployment delays in and cost growth of the IEC program. And the Commission is dismayed that cost of, and or delay in, fielding IEC terminals may impel NIMA to consider purchasing additional mechanical light tables. However, the Commission is buoyed by a recent NIMA initiative investigating a low-cost imagery workstation that meets most specifications, including a faster roam rate, and promises to be significantly cheaper, besides. Other agencies are also aware of this situation and are concerned enough to have started their own in-house programs--clearly a step in the wrong direction and a disappointing development.
Of additional concern is the shift in the commercial world away from UNIX and toward Windows for the very functionality of interest to NIMA. To benefit fully from the COTS cycle NIMA must heed tomorrow's trends, which for client workstation is toward Windows-based solutions and away from UNIX. The cost of high-end Windows workstations is half that of UNIX workstations and the power of graphics engines, fueled by the PC gaming market, is doubling every nine months while the price is being halved. WINTEL34 hardware and software manufacturers are continuously improving bandwidth and memory access to further enhance performance. So, while capable UNIX designs are currently available, inherent design limitations, less capable graphics cards, and less frequent design improvements, put the current IEC design at a distinct disadvantage, which will only increase with time.
The Commission also learned that many of the "electronic light table" applications that are critical for imagery and geospatial analyses are now being designed for the WINTEL. In fact, UNIX applications are likely to be offered only if requested and not as an "out of the box" solution.
The current IDEX replacement program is an example where NIMA has taken its first steps to employ some disruptive techniques in its system acquisition model. The IDEX replacement has actually followed two tracks-the first, a more traditional large-scale system integration program in which NIMA has used one of the usual government contractors as a designer, developer, and integrator of the IDEX replacement system, called IEC. IEC was to be a commercially based system. Following the normal large-scale development process, IEC has an expensive design, development, and maintenance cycle, and does not meet the existing IDEX capability. NIMA allowed the contractor to decide that CORBA would be the basis for all interfaces between all devices and processes-data would be passed and handled via CORBA-based ORBs. While the use of object-oriented programming to allow heterogeneous data types and processes to intercommunicate is laudable, adopting an emerging standard that is not commercially viable is not. The commercial world has looked at CORBA and has not adopted it as a basis for commercial systems development. CORBA compliance requires the use and development of additional software to act as the "glue" between the heterogeneous data types and processes. This "glueware" will be one-of-a-kind software, generated by the contractor, tied to a specific vendor's ORB, which must be maintained in perpertuo, thereby defeating the original intent of utilizing CORBA. This "glueware" is necessary if and only if the system requires tight integration to overcome a perceived ineptitude of the user. This tight integration is necessary to keep the user from making mistakes. NIMA's users are not inept-as evidenced by their ability to innovate the marriage between IA and GIS tools-and they should be afforded the flexibility to design by discovery.
In parallel to this effort, NIMA sponsored an in-house team to examine whether a purely COTS solution to the IDEX replacement could be found. A WINTEL-based system using COTS that are built to the WINTEL application programming interfaces (APIs) was built and tested. It performed as well as or better than both the original IDEX and the current IEC. (This is an example of a disruptive business model and is to the credit of NIMA, assuming it is implemented.) The COTS-based WINTEL solution should not be viewed as a COTS panacea; rather, it should be viewed as being a successful attempt at leveraging the existing base of commercially viable products to solve NIMA's IDEX replacement problem. Now that NIMA has a solution that is in step with the forces driving the commercial market, it will be able to take advantage of the advances that are being made in graphical technology in support of home entertainment. This will also allow NIMA to take advantage of the Web technology that will make it possible for NIMA to leverage its customer base for innovations that will give it the information edge.
Now that NIMA has taken the first step in disrupting its normal acquisition cycle, it must follow this innovative development with an equally innovative deployment plan. Using grand designs to replace other grand designs is unsound in light of current disruptive business models. NIMA should be applauded for using existing commercial standards and hardware and software in its in-house IEC replacement system; however, the deployment of this system will require NIMA to overcome its usual bureaucratic inertia that has plagued its other efforts in both TPED and USIGS.
This implementation should not be just an integration of the WINTEL architecture into the existing IEC as another software set that requires a coating of glue; rather, it should be a replacement for the existing IEC, the deployment of which should be stopped. An independent review board reporting directly to the current D/NIMA should be convened to analyze the existing WINTEL IDEX replacement system. This board-composed of non-NIMA systems analysts-should report to the current D/NIMA on the viability of the WINTEL architecture as a cost-effective replacement for IDEX.
To NIMA's credit it tasked a team to monitor IEC developments and pursue a simpler, less costly IDEX replacement. This netted a lower-cost imagery workstation, based on Windows 2000 (W2K) that meets almost all of the specifications identified for the IDEX II workstations, including a much faster roam rate than either IDEX or IEC. Initially certain capabilities35 were not available but as a testament to commercial ingenuity, these have been addressed and resolved. NIMA plans to evaluate this capability by deploying 30 workstations in a joint production cell. Assuming success, NIMA will face a dilemma: it can continue deploying IEC and offer the W2K option or fully compete the two designs, "winner take all". The Commission favors the latter approach.36
Use of commercial alternatives places great emphasis on getting the requirements right at the outset and managing the process smartly. The Commission notes that IEC is merely one segment37 of the IDEX II Replacement Project (IRP), which is managed via an Integrated Product Team (IPT) whose roles and responsibilities do not appear to be explicit. There does not appear to be a consistent understanding of either how the IPT is organized or the level of commitment expected from the various segments and/or users. This is not a recipe for success, irrespective of the use or misuse of commercial technology.
Several challenges exist in determining to what extent a commercial approach to TPED would work. A well-defined architecture will prove to be the key to well-placed confidence in commercial alternatives. A check list for success in utilizing commercial alternatives would include: demonstrating the scalability of the COTS systems under consideration; architectural "elegance," which reduces systems complexity, dependent in turn on identification of good architects; an inclusive, user-informed, prototyping strategy; and a well-vetted plan for smooth transition from legacy systems to new architecture.
This question is especially important in the database area. NIMA's online database will have a vector and raster component. The vector component is likely to have a high transaction rate but the total size can be easily measured in terabytes. The imagery component is much larger and while its ultimate size is both speculative and highly classified, a planning figure of several petabytes will do. Except for chunks associated with specific features, however, it is likely to have a relatively low hit rate (perhaps no more than 100,000 requests per day). Will COTS solutions to smaller data problems fail to scale? Or, will explicit systems integration be necessary--leaving no good choice but for NIMA to hand its architecture over to a traditional (read "aerospace") systems integration house?
Although NIMA's database is large, in many respects NIMA's problem is simpler than those of other database managers. Smaller databases such as those of banks, credit card bureaus, and server farms have higher transaction rates, more complex transactions, and more input points. A raster-image database may be huge in overall size, but manageable in terms of the number of items; and the transaction rate is low, most client transactions are straightforward (e.g., file calls), and the number of initial data feeds is limited by the number of (expensive) collection systems. No greater than the number of imaging satellites (with airborne collectors the number may approach a hundred). A vector database may have higher transactions rates and more input points but the total data set size is comparatively smaller.
It will be essential to model painstakingly the expected demands on NIMA's database to determine exactly what scalability problems will exist--storage, file complexity, number of nodes, service requests, or the support of specific applications.
Because it forces developers to produce an integrated system periodically rather than at the end, spiral development encourages light and loose versus heavy and tight systems integration. While the latter may promise to be more efficient ultimately, the former is easier to acquire and maintain; in any event, Moore's law usually rescues the less efficient design.
Reducing unnecessary systems integration also makes the overall effort accessible to more contractors, permits the total task to be managed in terms of smaller and faster deliverables, and ultimately, permits unexpected capabilities and requirements to be accommodated more easily.
The integrating mechanisms of NIMA's information architecture are a common communications stratum (e.g., TCP/IP), a common data model, and a common geodesic model (i.e., WGS 84). Systems integration is to be understood as a light appliqué, not the main event, and certainly not the primary criterion for selecting architects and contractors. And whatever systems integration experience is sought should be demonstrated against at least some significant GIS problems.
Still, one cannot ignore completely the systems integration process that ensures that everything that works apart also works together.
Should NIMA mount an in-house systems engineering and architectural effort? Can it attract enough talented outsiders through the Intergovernmental Placement Act (IPA) or other programs? Even if NIMA plans to outsource its architecture, the Commission believes that absent some intimate organic capability, NIMA cannot be a sufficiently wise buyer. Absent such expertise, it cannot readily evaluate its own requirements, the architecture that meets its requirements, and the systems that instantiate the architecture. Ineluctably, NIMA must put in place a set of (formal) procedures to validate the architecture.
An architectural goal is to end up with one "TPED" that includes imagery and geospatial data and processes. An architecture that is data-centric seems more satisfying to the Commission than one designed around (legacy) products and/or processes.
Embracing data-centric and Web-centric designs and moving to a new data model could be somewhat perilous. Test beds can play a useful role in validating and instantiating new architectures. Two approaches are possible. One is to run NIMA's architecture and data model off an extant test-bed architecture such as the one being operated by the Open GIS Consortium (OGC). The other is to sponsor a full-up Advanced Concept Technology Demonstration (ACTD). NIMA may want to do both: use OGC (or a like entity) to perform a rapid check on its geospatial model, and use the ACTD to explore the ramifications of a multi-INT database.
Not all of the database's ultimate features need be in place immediately. Some have to be part of the prototype but others can be installed later. Continuous improvement means tomorrow's capabilities are better than today's in some respects, and never worse. Mistakes should be caught while small and young. Feature expansion will await positive feedback. Most important of all, today's satisfied users will not become tomorrow's dissatisfied ones.
During the transition, users should be able to see familiar products--whether originally hardcopy or soft-copy--and it should be easy for someone to "find the button to push" that can recall the same map from the database as before. The period in which old and new coexist is a trying time, but wholesale conversion of NIMA's legacy database at the outset is probably unwarranted; initially, at least, applications should translate legacy data into usable terms (while writing new data according to the data model).
Some data will prove to be worth less than conversion costs because of age, error, or inaccuracies; other data will be found redundant. The rest have to be moved both across media and from the legacy data structures to the newly developed ones. Great care will be needed for those applications (algorithms) that can only work with legacy data structures--here conversion will be less automatic and more expensive.
In many (more) cases, old algorithms, having lost their customers, will simply be dropped. But the rest have to be painstakingly converted.
What should govern when information is to be converted: when it is needed or when it is received? Working on demand leads to crash programs and delays the availability of information (it is usually too late to inspect details up close once a crisis erupts). Working on receipt risks spending money where it is not needed.38 No easy answers.
Long-term goals can be approached through short steps. Fielding capabilities as they mature rather than at the project's end permits mistakes to be surfaced early and research has shown that early detection of mistakes reduces life-cycle costs. The development of unexpectedly popular features can be accelerated. If something does not work out, one knows early and can adjust requirements (and expectations) accordingly. However, emphasizing periodic improvements places a premium on backward compatibility and changes the training and configuration management regimes. No free lunch, here.
The Commission does not have high confidence in NIMA's current ability to accomplish its TPED system acquisition successfully. The current TPED acquisition effort lacks a clear baseline, which should tie clearly to overall strategy, requirements, and cost constraints. In addition to the lack of a common definition of TPED, there is similarly confusion as to the requirements that TPED must satisfy.39 The Commission learned that in a comprehensive requirements review that helped define FIA, considerable imaging requirements were allocated to commercial and airborne imagery:
In peacetime less than 50 percent of required area coverage is allocated to FIA, while commercial and airborne assets accounted for the majority of peacetime area allocations. For peacetime point coverage the reverse is true, with the bulk of peacetime point targets allocated to FIA, and a minority to airborne and commercial assets.
During a major theater conflict, about half of both area and point coverage are allocated to FIA, while commercial and airborne assets combine to meet the other half of all requirements.
FIA holds to the claim that it will meet all its allocations; however, because of negligible budgeting to date for commercial imagery, and proposed reductions in airborne investment, OPSTEMPO and PERSTEMPO--the FIA era still might not live up to its billing as eliminating collection scarcity. Further, the allocation of requirements to airborne sensors implies a concept of operations (CONOPS) that has not yet been articulated. Compounding the problem further still, the Commission could find no credible plans to integrate commercial and airborne products into FIA and/or TPED. Without agreement within the community of what is included in TPED and what requirements are to be met it is difficult to envision a successful acquisition effort.
The Commission received a number of briefings meant to describe TPED and its status. What becomes clear is that NIMA has not articulated a single definition of TPED. One is easily confused about where TPED ends and USIGS begins, or are they one and the same? Does TPED, as specified, support only the collectors that the NRO is acquiring under FIA, or does it also embrace airborne and commercial collectors? Does TPED extend to multi-INT capabilities? These, and other, ambiguities suggest those responsible for its implementation do not adequately understand TPED.
It appears that an acronym for the functions of tasking, processing, exploitation, and dissemination has somehow become the name for an entity without benefit of a common understanding of the content. TPED needs stability in definition and scope (and funding) so there is a common ground for describing and successfully implementing the capabilities needed to support the users. The Commission was treated to a multi-phase view of TPED by ASD(C3I) which clearly shows, in successive phases, the integration of commercial and airborne imagery assets, and multi-INT integration. If fleshed out, funded, and adhered to, the plan seems satisfactory to the Commission.
In addition, NIMA's current acquisition strategy requires NIMA to be its own system integrator. However, the Commission is not confident that NIMA currently has the system engineering experience, acquisition experience, appropriate business practices, and performance measures to so acquire TPED systems. The Commission sees high risk in NIMA's taking on responsibilities and risks above and beyond that of a simple acquisition agent. But, as argued earlier, NIMA must have sufficient organic capability to be a wise buyer.
As discussed in a preceding section, the Commission observes that TPED is not adequately utilizing commercial hardware and software. Again the Commission is somewhat conflicted as to whether or not NIMA should restrict itself to an acquisition role, ceding most development and systems integration activities.
The Commission observes that current TPED plans only tangentially increase the convergence of imagery and geospatial processes, and also notes that current TPED plans do not effectively integrate airborne and commercial imagery with national technical means. Nor do current TPED plans speak to the issue of multi-INT integration.
As an aside, the Commission notes that the FIA baseline does not support production of film, on which TPED must still rely unless NIMA receives additional resources to move the entire community to soft-copy.
The imagery TPED program increasingly strains at the fabric of the NIMA organization as a whole. Repairing the problems cited above, while necessarily adhering to the schedule imposed by the successive generations of imagery satellites--EIS and then FIA--makes the current program far more risky than previously supposed. While we cannot afford to fail, it is not clear that we are prepared to afford success. The stakes are high, the job is monumental, the time is short, the resources are marginal, and the skilled personnel are slim pickings.
NIMA does not have the organic capability or the experienced technical leadership to successfully acquire TPED, nor can it "get there from here," in time, using normal government practice. There is no help on the horizon because neither the NRO nor NSA has the talent to spare. If the US is to have a good chance of achieving a TPED capability to give the nation the information edge in the 21st century, special steps must be taken to ensure success.
The Commission recommends creation of an Extraordinary Program Office (EPO) armed with special authorities of the Director of Central Intelligence and the Secretary of Defense, augmented by Congress, and staffed beyond ceiling and above "cap" through an heroic partnership between industry, NIMA, and the NRO. The EPO, to be constituted within NIMA from the best national talent, shall be charged with and resourced for all preacquisition, systems engineering, and acquisition of imagery TPED--from end to end, from "national" to "tactical". The first milestone shall be completion of a comprehensive, understandable, modern-day "architecture" for imagery TPED. Other provisions of law notwithstanding, the Congress shall empower the Director of the EPO to commingle any and all funds duly authorized and appropriated for the purpose of the "TPED enterprise," as defined jointly by the Secretary of Defense and the Director of Central Intelligence.
An accelerated schedule helps avoid mission creep. The Commission estimates that the first four months should see (1) a preliminary data model constructed, (2) estimates of the time and resources required to convert legacy data into standard digital form (see below), and (3) a succinct requirements statement based on the principles above. Architect selection should proceed expeditiously with the actual work completed in three phases of six months each. The first phase should be specific enough so that the work of converting legacy data can begin. The second phase should be good enough to budget the next five years of TPED acquisition. The last phase should be the basis upon which software can be written and acquisitions begun.
TPED should not be based upon NIMA's current processes and products. Instead, as elaborated upon in the succeeding section, processes should be considered as Web-enabled transactions against a database; products can be pulled from the database or created by "servelets," "applets," and/or client software. The design should inherently foster imagery-GIS convergence.
The Commission has not seen evidence that an integrated plan exists that utilizes airborne, national, and commercial imagery in a cooperative effort to meet all imagery collection requirements. In addition to the comments above concerning requirements allocation among the various collectors, the Commission was not exposed to an integrated CONOPS utilizing imagery from all three sources--national, airborne, and commercial. Such a CONOPS requires close coordination with CINCs who currently have control over theater assets. An operational plan would also require agreement with commercial providers on issues such as amount of imagery to be provided, quality control, responsiveness to USG needs, and methods of exploitation.
Further concerns about the lack of integration among airborne, national, and commercial imagery are made evident by the fact that the TPED functions; namely, tasking, processing, exploitation, and dissemination for each of these imagery providers are essentially different. The fact that NIMA has not discussed these functions individually nor indicated how these functions would be accomplished for each imagery source in a cooperative environment is an indicator of the lack of an integrated plan.
Dissemination (including the communications for distribution) is arguably one of the more expensive portions of the imagery intelligence cycle. One of the critical elements of this service is the communication links. These links connect tasking authorities to collectors, collector data to processors, processors to exploiters, information to users, and users to tasking. These links must be secure, robust, high capacity, and both long and short haul.
It appears to the Commission that the lines of responsibility between TPED and communications systems, both terrestrial and space, have been blurred. The danger in this approach is that no one becomes responsible for the enterprise operating as a unit. The dialogue so far between NIMA, DISA, NRO, and the user community engenders no confidence that the links will be there when needed. It was not made clear to the Commission as to who has responsibility for the "last tactical mile." It does not appear that NIMA signed up for that responsibility--and it certainly is not resourced for that, nor should it be from "national" funds, by some accounts. However, the CINCs and Services conveniently profess not to know where TPED ends. This is not good.
Clearly more dialogue is needed to define the boundaries of TPED, responsibilities, and interfaces. Part of the difficulty in having this dialogue is that communications is considered both multi-user and multi-use; it is expensive given the bandwidth needed for imagery and geospatial product delivery--in fact, once imagery-quality bandwidth is provided, almost everyone else "rides for free." The Commission is uncertain whether an Intelligence Community communications architecture exists.40 The Commission is pretty certain that if it does, it does not stretch to the foxhole, wheelhouse, or cockpit. While such architecture is not necessarily a NIMA responsibility, it is necessary for TPED to be successful. Given this situation, it is difficult for the Commission to have confidence that the capacity for FIA and/or USIGS will be available when needed.
Despite the fact that material describing USIGS implies use of, and integration with, other Intelligence sources such as SIGINT and MASINT, the Commission found little evidence that integration is inherent in the TPED program.41 Solutions to portions of the imagery problem set generally require the integration and fusion from all sources with very short timelines and the Commission agrees that all-source TPED is needed. Multi-INT requires as a minimum the following elements: tasking processes based on required information rather than INT-specific observables; interoperability between TPED systems, MASINT, and SIGINT information embedded in the USIGS library; and multi-INT workstations equipped with exploitation aids.
A review of the current operational and planned space and airborne capabilities indicate efforts to support TPED functions within each discipline with little planning for integrated systems or functions across the current stovepipes. The NIMA TPED program does not fully address this problem. Moreover, there is some question if NIMA has the authority, expertise, and budget to execute the necessary programs. As a minimum, NIMA should have complete understanding of the relevant programs that its mission partners and others are pursuing and efforts made to coordinate these efforts. The Commission was not exposed to relevant TPED efforts at NSA and CMO regarding SIGINT and MASINT nor did it hear of cooperative efforts among NIMA, NRO, NSA, Central MASINT Office (CMO), or others for multi-INT TPED other than plans to develop a shared requirements database.
There are five significant FIA shortfalls defined by JCS that have major TPED implications and have not been considered in the current architecture. Without going into the specifics, which are classified, the Commission wants to plant the marker that augmenting FIA with any or all of the shortfall-capabilities must also provide for the TPED implications of the FIA improvements. In the spirit of Total Cost of Ownership (TCO), the Commission expects the bills for the upgrades to be calculated taking TPED modifications into account, and budgeted for as a piece.
The special authorities of the DCI should be used to create the "spaces" and the DCI and SECDEF should intercede personally with the private sector to get the "faces" to fill those spaces. Congress should codify the exceptional measures needed to set up and operate this Extraordinary Program Office (EPO). The Commission believes that the EPO should be created within NIMA.
It is anticipated that the EPO shall have a five-year lease on life, after which the Director of the EPO and D/NIMA will have arranged for a smooth transition of the required capabilities into NIMA proper.
Elements of an EPO;
The Commission feels that the Director of NIMA would benefit from outside technical expertise, in the form of a Technical Advisory Board with whom he might meet periodically to review key TPED acquisition (USIGS modernization) milestones and top-level design presentations. The Board would also represent a resource on which the Director and his senior acquisition and technology officers could call as required.
28 Some have suggested that the literal definition of imagery TPED is an anachronism and needlessly constrains our thinking. Alternative constructs are proposed:Gathering versus Tasking -"Tasking," it is argued, stems from a model based on scarcity, where the collector is limited. "Gathering" is a more useful term, deriving from a model based on abundance where discovery is the issue.Creation versus Processing--where a multisensor view of information is contrasted with a single-sensor view of data formation.Analysis versus Exploitation--"Exploitation," it is argued, is an overly narrow Indications-and-Warning (I&W) view of imagery; "Analysis," by contrast, is the function people perform best, seeing patterns in information.Sharing versus Dissemination--where "sharing" is a many-to-many model of information communication, while "dissemination" is a one-to-one, or one-to-many model of data movement.TPED, they argue, is derogated as needlessly implying a linear view of data. The alternative formulation--Gathering, Creation, Analysis, and Sharing (GCAS)--is billed as a cyclic view of information.
30 This is not to say that NIMA will no longer produce its hallmark products: maps and imagery intelligence products. As NIMA focuses on information services, the maps and intelligence reports are by-products--intentionally useful derivatives, but not the essence of NIMA.
31 As mentioned elsewhere, the Commission is partial to the definition, variously attributed to Scott McNealy, of Sun MicroSystems, that "standards" are products that ship in volume.
32 COTS-Based Systems: COTS Software Lessons Learned, Recommendations and Conclusions, Computer Systems Division, The Aerospace Corporation.
33 The Commission does offer one caution: increasingly, COTS products are marketed and produced globally. This means that a critical COTS product might have been produced by, or within easy reach of, a potential adversary. Information assurance should be a Key Performance Parameter of every significant acquisition.
34 Windows operating system on a box with "Intel inside."
35 For example, mensuration, display of stereo pair data, and the continuous paging of the data from the server environment.
36 Data on the IEC and W2K workstation that the Commission reviewed or discussed with various contractors show that the WINTEL workstation hardware would be significantly cheaper (costing no more than $25,000) than the UNIX-based IEC (currently priced upwards of $45,000). Clearly NIMA could field a larger number of workstation or recapitalize at a faster pace than it is planning to. The unsettled debate is in the cost of the software for the W2K workstation. The software costs for each IEC workstation is estimated at about $100,000. It is not clear what the software costs on a W2K would be since the current design has very little integration involved (see CORBA discussion on pg. 97). If no other differences exist, clearly, NIMA could save integration costs and benefit from the economies of scale resulting from using the Windows standard.
37 Each segment is a separately managed contract, but the relationship of these contracts to the integration contract is not clear. The nature of the delays the IDEX II Replacement Project (IRP) is currently experiencing suggests that the roles and responsibilities for integration were not clearly defined or understood. In addition, it appears that the IRP IPT has limited control over the total life cycle costs (TLCC). As a consumer of components managed via other contracts, the IRP is dependent upon decisions of the segment developers for TLCC impacts. Additionally, the operations phase of the total life cycle includes O&M, which is apparently the responsibility of a sister directorate (Information Services). There was no clear indication that members of this organization participate regularly in the IPT.
38 Of course, if we knew when and where the next crisis would develop, we could forgo the intelligence establishment.
39 The Commission has labored mightily to get this right. It's not easy. We think we are close, but each time the question is posed, the sands shift. It is legitimately difficult to gauge requirements: some requirements are point targets, others are for area coverage; not all point targets are equal, not all areas are equally interesting; peacetime is different from wartime. Complicate this by the fact that some require higher resolution, some require stereo, etc. Without making this a life's work, one may still conclude that there will be a disconnect if airborne and commercial do not deliver as originally anticipated.
40 By some accounts, the Defense Information Services Agency (DISA) is responsible (for DOD) end-to-end architecture; indeed, DISA's Global Information Grid (GIG) presumes to extend across the last tactical mile, although the Services have not yet been heard from on the notion. Even if DISA harmonizes with the Services, the situation is clouded by the fact that intelligence networks have traditionally been separate from DISA networks. They can run at a higher classification and, given the out-bound imagery bandwidth requirements and the in-bound SIGINT requirements, intelligence traffic would dominate by far a common use network. For these and other reasons, the Intelligence Community has been noticeably reticent in placing its future in DISA's hands.
41 However, the multi-phase view of TPED espoused by ASD/C3I clearly shows multi-INT integration as a later phase. As the C3I vision becomes better defined and funded it will alleviate Commission concern.
| Executive Summary and Key Judgments
| Introduction | NIMA
from the Beginning