An Operations Cyberinfrastructure: Using Cyberinfrastructure and Operations Research to Improve Productivity in the Enterprise
Figure 2: Functions and examples of an Operations Cyberinfrastructure, distinguished by system level.
Timely action is of the essence if we are to take full advantage of the OCI’s potential for revolutionizing the conduct of complex enterprise-wide applications. Opportunities exist now for establishing compatible, non-proprietary approaches to knowledge archiving, data exchange, and — most important — interconnection of diverse OR analytical methods. As the Atkins report has suggested with respect to CI generally, missed opportunities for building the OCI now will be much harder to make up later, once poorly structured local cyberconnections, incompatible standards, and proprietary tools have come to dominate the field.
Basic research will play an essential role in the development of the OCI, in part because the OCI will connect and assist researchers in the development of new tools, but more importantly because much remains to be learned about how an OCI should be constructed for maximum impact on its large community of potential users. We see the evolution of the OCI as following the pattern set by other cyberinfrastructures that have come to be universally recognized. The initial risky steps will have to be taken by funded research, because commercial enterprises lack the expertise or are overly tied to proprietary investments in computing and productivity tools. Once OCI standards and protocols have been established and have proven their value in initial applications, the broad value of the OCI will become evident while its apparent risk fades, and commercial entities will increasingly assume the effort of further development.
An NSF initiative in this area should include significant funding for basic OR research that is tied to development of the OCI. It should also include research in the development of CI tools that are particularly relevant to the OCI and, possibly most important, research projects at the interface of OR and CI. Each research project of the latter type should be motivated by one or more significant enterprise-level applications, and should aim to develop and implement algorithms and tools that can be leveraged across the OCI. Typical research projects that could be supported under such an initiative could be, for example,
development and implementation of an algorithm for solving a particular logistics problem that arises frequently in supply-network applications, with a commitment to making the resulting computational tool available through a Web server,
construction of a modeling and analysis library to demonstrate the use of relatively new OR approaches (for example, stochastic optimization or optimization of simulations) in solving important problems in enterprise-wide applications; and
development of methodologies and cybertools for design-time supply network design.
The value of these types of projects has already been seen in, for example, the success of the NEOS optimization server.4 However, in contrast to earlier projects, an OCI initiative will have to address many more interactions among a much broader variety of tools, giving rise to correspondingly more difficult challenges in CI design.
Grantee meetings should be held regularly to ensure coordination across the OCI research portfolio. The portfolio should include wide coverage of the most significant OR and CI research areas, but should not include “loose end” projects that do not contribute to some aspect of OCI development. By its very nature, the OCI is an integrative entity, so the research program for its development should be similarly integrated.
Funding for OCI-related projects should be leveraged by funds for other NSF programs in CI, information science, engineering, and social, behavioral, and economic sciences. When appropriate, collaborators from areas outside of OR and CI who could contribute to OCI development should be included in funded research teams. Examples of such participants could include human-factors researchers (for human-in-the-loop modeling of manufacturing processes), economists (for modeling behavior of participants in an online market), design engineers (for product design applications), electrical engineers (for applications involving the power grid and electricity markets), and medical informaticists (for modeling of information-sharing alliances in health care).
In supporting OCI initiatives, NSF should consider the creation of a modestly scaled research center to act as an “umbrella” organization for the CI-OR effort. Such a center would undertake community-building activities such as maintaining an OCI advisory board, coordinating standards, and organizing periodic meetings, in addition to supporting OCI research at its host institution(s).
To ensure a sufficiently interdisciplinary funding mix, however, the majority of OCI support should consist of single-investigator and small-team grants directly from NSF programs. Smaller grants, administered directly by NSF program managers, have historically provided NSF with excellent return on investment. Strong, capable program managers, assisted by an advisory panel of CI and OR researchers to help with proposal evaluation and project monitoring, will be able to provide appropriate guidance. We are confident that the OCI project mix can be assembled and configured to seed, sustain, and leverage OCI development and to have a significant impact on productivity of the enterprises.
1 Areas cited in Section 2.6 of the Atkins Report include Atmospheric Science, Forestry, Ocean Science, Environmental Science and Engineering, Space Weather, Computer Science and Engineering, Information Science and Digital Libraries, Biology/Bioinformatics, Medicine, Physics, Astronomy, Materials Science and Engineering, and Social and Behavioral Sciences.
2 See in particular the middle paragraph on page 4 of the Executive Summary.
3 See for example the NEOS Guide, http://www.mcs.anl.gov/otc/Guide, which is however limited to optimization.
4 See http://www-neos.mcs.anl.gov/neos.
Share with your friends:
The database is protected by copyright ©essaydocs.org 2020