Research Horizons
R&D as a Tool to Enhance Technical Development
John Fujita
Independent research opportunities help keep scientists and engineers engaged, motivated, and technically proficient.
The Aerospace Independent Research and Development (IR&D) program is one of many corporate investments that enable the technical staff to develop the skills and tools needed to support the evolving requirements of national security space. Within Aerospace, a culture of accountability has been embraced that challenges the staff to strive toward independent thinking and technical excellence. This accountability encourages the scientists and engineers to think about each program and to consider many different aspects of the technical issues facing the customers' missions. If there are potential problems, each employee is accountable for ensuring that Aerospace management is fully engaged. Thus, there is a need and motivation for Aerospace staff members to keep their technical edge sharp.
Aerospace serves an important function as the corporate memory and knowledge repository for the national security space community. It doesn't manufacture products in the traditional sense—its primary product is technical expertise. So, while commercial ventures may use R&D funding to create and develop new product lines to sell, Aerospace uses these resources to make scientific and engineering contributions in support of the national security space community and to continuously build its technical know-how and capabilities. Aerospace's IR&D program challenges the technical staff to think creatively and independently. It also enables them to explore scientific discoveries along with new technologies and applications.
Aerospace's IR&D program is competitive. Each year, the program reviews significantly more proposals than it can explore.
Aerospace works with its partners to ensure that its research efforts are aligned with their mission areas. Each year, Aerospace gathers a list of technology topics from major customers, which are used to measure each IR&D proposal's relevance to the national security space mission. Proposals that do not directly tie to a technical topic are always welcome, but they bear the additional burden of demonstrating their national security space relevance. The technology topics are grouped into seven broad areas: communications and navigation, electronic devices, information science, spacecraft and launch vehicles, space and launch environments, surveillance, and systems engineering and modeling. Within these groupings, the research can be fairly diverse.
Each proposal is further scrutinized to gauge its appropriateness, innovation, feasibility, potential for skills development, and capacity for expanding the corporation's technical breadth and reputation. While some projects seek to create an entirely new set of technical skills, many serve to enhance existing skills and to safeguard endangered skills—that is, the critical know-how acquired through years of experience that might get lost if it were not passed on to the next generation.
John Fujita |
At the core of Aerospace's technical investment program are the ideas generated by the staff and customers. Some ideas could be categorized as linear or evolutionary, while others are revolutionary and seek to change the order of things. Both have a place and are valued. The key is to capture and exploit as many of these ideas as practical and to keep employees continuously thinking. Some of these projects require the construction of specialized hardware to test the underlying theories and concepts, and the development and maintenance of these installations enhances both staff expertise and corporate capability.
One recent example is an IR&D project that involves a testing and modeling program designed to investigate complex cavitation dynamics in turbo-pump inducers. This project required the construction of a state-of-the-art flow-test facility. Another example is a project geared toward optimizing and automating acquisition of laser signals, which produced a test bed that simulates the separation between two satellites in a laser communications link. This effort helped enhance cross-disciplinary activity by combining the talents of both the Electromechanical Controls Department and the Digital and Integrated Circuit Electronics Department.
Some projects can seem highly specific, even esoteric. For example, one recent initiative found a creative way to predict thermal transfer in ball bearings. Another focused on developing numeric models for predicting wall erosion in Hall-effect thrusters. However, these projects play a critical role in keeping Aerospace's technical staff close to many of the technology areas that are affecting space systems.
Other projects seek to produce more immediately tangible benefits. For example, a novel process for making microelectromechanical systems (MEMS) from silicon dioxide and implanting them on or within a silicon substrate could have broad applications. Similarly, a set of nondestructive, noncontact techniques to inspect multilayered photovoltaic semiconductors for crystalline defects could lead to the first statistical model to predict solar-array failure.
Where possible, Aerospace shares the results of these research endeavors so that the lessons are available to the broader space systems and technical communities. Aerospace's Technical Investment Program produces numerous publications each year and supports many conferences and symposia. Some projects result in patents. Aerospace regularly showcases some of the intriguing research and creative ideas that have sprung from the IR&D program in Crosslink's Research Horizons.
Analysis of Security Threat Group Networks
Gangs, antigovernment groups, and terrorist organizations pose increasing concerns for the homeland security, law enforcement, and intelligence communities. Understanding how the members of an adversarial group communicate and interact could help to contain the threats they pose.
Karen Jones of Civil and Commercial Operations heads a team investigating whether data-mining techniques such as link discovery and social network analysis can be used to monitor such groups. The team, which includes Donna Nystrom and Frank Meng of the Advanced Information Systems Technology Department (AISTD), is developing a data-mining prototype based on open-source software.
The Aerospace data-mining process entails sifting through vast troves of data, applying link discovery techniques, and analyzing any patterns that result. |
While the technology could be applicable to various security threat groups, Jones's team will initially focus on prison gangs. "We have chosen prison gangs as our focus for several reasons," Jones explained. "The corrections domain presents a small, regulated, and closed-world social system, where it is feasible to record outcomes and measure overall system performance for testing Aerospace's link-discovery and social-network analysis tool," she said. Corrections systems collect a vast amount of data on each prisoner: Interactions such as phone calls and personal visits are logged, prison bank transactions are recorded, and incident and investigation reports are written up with valuable historical and personal information. Moreover, within the corrections domain, the cost of false positives is much lower than for systems operating within the general society. In addition, because a prisoner's gang affiliation is typically known, the researchers should be able to identify which interactions are within gangs and which ones are between gangs.
Link discovery identifies both obvious and nonobvious relationships between entities—in this case, inmates and some of the people they interact with. "For example," Jones explained, "an obvious relationship exists between inmate A and inmate B when both are caught together trying to smuggle drugs while distributing laundry." But suppose that prior to this incident, prison records showed that inmate A called person C (outside the prison) and that inmate D wired money to person C; then, nonobvious relationships exist between person B and persons C and D. Persons C and D could be related by financial transactions, but such information does not mean much unless there is a link to the drug-smuggling incident. This link could indicate that C and D might be part of the drug-smuggling ring to finance and transport drugs into the prison. "A money transfer might appear innocuous as a stand-alone incident," Jones said, "but within the context of the drug smuggling incident, it appears as a suspicious link between a larger group of people." With enough links, social networks begin to emerge. Some of these networks could represent security threat groups or, in this case, prison gangs that continue to orchestrate crimes, traffic drugs, and commit violent acts while in prison. "Applying social-network analysis techniques allows the Aerospace data-mining team to analyze and map flows and relationships within groups—identifying core members, ringleaders and peripheral members—as well as hostility or collusion between gangs," Jones said.
Jones and her team met with representatives of one of the largest state correctional departments and developed a memorandum of agreement. The department has a vast amount of data on the prisoners and staff in its 53 correctional institutions and maintains statistics to better understand trends, costs, and prisoner demographics. The Aerospace team hopes to identify specific patterns of behavior and, when certain patterns emerge or change, to use the tools to predict events such as violent gang incidents, drug smuggling, or escape attempts. Such data could help correctional officers secure the safety of the inmates and staff. Prison officials could separate hostile factions to preempt violence and separate those who have colluded to commit crimes or engage in disruptive behavior.
AISTD has established a data-mining center, which includes a secure server and computer operating on an isolated network. Nystrom manages the center, where the project team is now mining the last three years of such inmate-related data as visitor logs, prison bank records, incident records, investigation reports, offense and disciplinary records, and home addresses.
Jones notes that rigorous ethical standards have been applied throughout the project design: "In a country where there is little tolerance for privacy-rights violations, mining corrections data stands up as an activity that serves the legitimate institutional interests of corrections—to monitor and protect inmates."
Devising a Ground System Cost Model
Space systems cost estimation has a rich history and a collection of established cost-estimating methodologies. Yet, one area that is still not well understood is satellite ground segment cost estimation. Tim Anderson of the Systems Architecture, Engineering, and Cost department notes that while the cost-analysis community has focused much attention on estimating the cost of spacecraft acquisition, the ability to predict the cost of a new ground segment acquisition remains elusive. There are many reasons for this, but perhaps the most important is that, unlike satellite acquisitions, in which a fully developed system is delivered with all of its hardware and most of its software functionality, ground segments tend to lack such temporal standardization. "One organization might acquire an entire ground segment, including ground terminals, mission management and mission data processing, new facilities and the like, while another ground segment acquisition might include nothing more than an upgrade to an existing function, such as data dissemination improvements, or added capability to handle an additional mission," Anderson said. Thus, the database from which to develop ground segment cost-estimating relationships is highly disparate, and it is difficult or impossible to develop reusable hardware-based or function-based cost models for these systems. Consequently, ground segment cost estimates are usually based on one or two loose analogies, or worse, constructed using engineering build-up techniques in which the technical variables (e.g., software lines of code, quantities, staffing requirements) are rough estimates themselves. "The result is that acquisition managers usually get burned by unrealistic ground segment cost estimates," Anderson said.
Since modern ground segment acquisitions span the gamut from plug-in components of an existing system to an entirely new system built from scratch, an optimal cost model should contain a set of fundamental building blocks, or "costable" elements, that can be pieced together depending on the specific architecture under consideration. In addition, the cost model should be based on recent and relevant historical cost data. Various organizations—including Aerospace—have attempted to construct robust models in the past, but these were generally limited in scope and not widely applicable. Thus, there is no comprehensive ground segment cost model in widespread use in the industry.
Anderson is attempting to bridge this gap by developing such a model. His research so far has focused on defining a fundamental Work Breakdown Structure (WBS)—the list of everything that has to be paid for to bring a system to its full operational capability—and initial data collection. Anderson's WBS is different from a traditional WBS in that it is described in the form of a matrix rather than a hierarchical decomposition of cost elements.
Most published WBS's are linear, and represent a laundry list of hardware, software, staff hours, and other items. The Aerospace WBS is a matrix, which allows the user to select specific architectural and temporal elements. This approach provides more flexibility for cost estimating. |
"The WBS separates typical ground segment functions from temporal acquisition functions, enabling one to cherry-pick the elements to be costed for any given acquisition," Anderson said. The structure was conceived and developed by Anderson with the assistance of a group of six systems engineering and operations research masters' students from George Mason University. The students, all with military or industry backgrounds, volunteered to develop the structure as their senior project. They spent a semester working with Anderson and Frank Donivan of Civil and Commercial Operations to develop a draft WBS. Following delivery of the draft WBS, the structure was further refined through extensive consultation with Aerospace program offices representing the Space and Missile Systems Center, the NRO, and NOAA.
The elements on Anderson's list, which can be tailored to any foreseeable ground segment acquisition, have been chosen such that their costs can be theoretically predicted on the basis of independent variables such as functionality, complexity, acquisition volatility, and the like. "The idea is to be able to select from the set of WBS elements in the matrix only those elements that are relevant to the acquisition in question," Anderson said. "This selection then forms the basis of the elements whose costs are to be estimated." With the completion of the WBS design, cost data for relevant historical ground segments are now being researched to develop realistic cost models for each element. Anderson hopes that this cost model will enable cost analysts to produce high-quality rough-order-of-magnitude cost estimates for use on programs in the earliest acquisition phases, when systems are not yet sufficiently defined to produce an engineering buildup cost estimate.
Successful completion and application of this model will position Aerospace as a leader in the ability to estimate the cost of virtually any satellite ground segment acquisition. "This is a natural role for our company," says Anderson. "Like the Aerospace Small Satellite Cost Model, the Ground System Cost Model will be the model to turn to when our customers need a realistic general-purpose ground segment cost-estimating methodology."
Human Systems Integration Risk and Cost Models
Human systems integration (HSI) combines human factors analysis, safety assurance, and training to ensure that systems accommodate the capabilities and limitations of their users. Space systems acquisitions have inherent HSI challenges, particularly in regard to the increasing information demands on operators, the need for operators to perform new and different jobs, the pressure to reduce staff size, and the rising demand for rapid and accurate responses to dynamic mission environments.
Integrating users in complex systems represents 40–60 percent of life-cycle costs, according to a 2004 report from the U.S. Air Force Scientific Advisory Board. Successful implementation of HSI can result in major reduction in costs in areas such as the number of personnel required by the system, time and resources for training, error and accident rates, error recovery time, and speed and proficiency with which personnel operate, maintain, repair, and deploy the system. Aerospace-supported programs that did not adequately address user integration have encountered a number of problems during turnover to operations, such as the need for significant system redesign, delays in operational acceptance, the need for more personnel with greater skills, and the need for additional operator training. One program in particular may require more than $60 million in additional funds to resolve outstanding user integration issues and sustain the increased number and skill level of personnel needed to operate and maintain the system, according to Aerospace estimates.
|
Clearly, program offices need tools to assess the potential impact of decisions regarding user integration at various phases of acquisition, and Aerospace needs a method for systematically assessing the risks and impacts of deferred, descoped, or eliminated HSI work.
Suzanne Dawes of the Ground Systems Support Office heads a team that has been trying to identify or develop such a method. The project has been challenging, partly because of the historical lack of focus on HSI as a distinct cost factor in previous programs. For example, Dawes's team began with an extensive literature review to determine what had already been done in the area of risk impact and cost modeling. The review focused on HSI in the acquisition process, searching in particular for information regarding cost-benefit analyses for HSI investments. The review revealed that two commonly used cost models at Aerospace—COCOMO/COSYSMO and SEER-SEM—do not support easy separation of HSI costs.
Dawes's team assembled a focus group to elicit information regarding the management and cost containment of HSI efforts during system acquisition. The goal was to gauge the interest in addressing these concerns and to determine which areas to address first. The group included Aerospace personnel from the engineering division and program offices as well as Air Force program office and user representatives.
The results from the focus group fell into two distinct categories: early HSI acquisition questions, and concerns about the risks and impact of deleted and deferred HSI work. In particular, decisions about staffing profiles and automation during system development proved most problematic. "Of significant concern was the increased cost associated with adding personnel (either military or contractor), changing personnel from military to contractor, and maintaining a larger-than-expected cadre of contractor personnel for years after operational turnover to support operations," said Dawes. Also significant was the concern for the cost trade-off between the level of automation designed into the system and the proposed staffing mix. "Both these key concerns require that risk assessment and cost estimation be addressed simultaneously and account for changes in risk-control options throughout the acquisition lifecycle," Dawes said.
Working with Carolyn Latta of the Cost and Requirements Department, the team (which also included Bettina Babbitt, Stephanie Heers, and Barbara Jex Courter) selected a formal probabilistic risk assessment approach that supports the estimation of the cost and benefit of each specific risk-control option, including prevention and mitigation. This approach takes as the basis of cost-risk the actual risks that a program faces, determines the probability distribution of each risk, and uses this information to determine the uncertainty in the cost. This approach links cost-risk to risk management.
The team is now using this approach to develop and test models using scenarios that reflect the major areas of concern expressed by the focus group. "For example," said Dawes, "if the cost and risk of making a system highly automated far outweighs the benefit of reducing the skill level and number of personnel, then the program office may make that decision earlier in the development life cycle. This provides the program office with quantitative data to support their decision and enables the government team adequate time to plan for changes in the personnel mix." On the other hand, results for a program may indicate that spending resources on developing a more automated system will significantly reduce personnel costs over the long term. "To the extent that programs have this information earlier in the acquisition life cycle," Dawes said, "costs may be reduced even further." Once these models are developed, additional focus groups will be convened to assess their effectiveness in addressing HSI cost drivers across the acquisition life cycle.
To Spring 2007 Table of Contents
