This is a modification to correct typographical errors in the notice issued on March 14, 2003. This is a Request for Information (RFI). The Missile Defense Agency (MDA), in support of the Department of Defense (DoD), seeks practical, yet innovative methods for solving difficult, real-time resource allocation and scheduling problems in uncertain, partially-observable environments. MDA desires to interact with academia and industry to review cutting-edge techniques and explore new solution concepts potentially across domains beyond missile defense with significant promise for applicability in supporting the Ballistic Missile Defense Decision Architecture (BMD-DA). The objective of the BMD-DA is to identify, within the pressing, real-time constraints of a ballistic missile engagement, a sensor and weapon allocation policy that will be effective in defeating the threat. During such an engagement, this system must determine when, and in what mode, various sensors will be employed to estimate the position and projected movement of a given target complex, comprised of re-entry vehicles, decoys, debris, etc. as well as the classification of its individual components. This information, in turn, must be used to determine the most effective assignments for a limited number of weapons which may be used to destroy incoming targets. Due to the complexity and high number of interacting factors in this system, determining the optimal allocation of sensor and weapon resources in real-time has proved to be challenging. DoD is currently working to deploy an initial missile defense capability in 2004, with an increasingly robust system to be implemented in subsequent years. Initial testing of the BMD-DA is scheduled to begin in 2004 with initial deployment beginning in 2006. To address the challenge of developing a practical BMD-DA, MDA desires to review promising mathematical and computing techniques with potential applicability to the problem domain. Research areas of interest include, but are not limited to the following: (1) Markov Decision Processes (particularly partially-observable MDPs); (2) Bayesian networks (BN), decision networks & other computational models for probabilistic inference and decision making; (3) approximate stochastic dynamic programming; (4) expert systems and other artificial intelligence methods; (5) neural networks & other adaptive techniques; (6) planning and scheduling approaches; (7) game-theoretic approaches; (8) stochastic optimization; (9) parallel computing algorithms, and other methods for improved computational efficiency; and (10) methods to evaluate the quality of a given resource allocation policy when the global optimum is unavailable. Given the limited time available before deployment of the BMD-DA, proposed approaches should emphasize run-time efficiency and scalability to realistic problem scales. MDA is particularly interested in solutions with potential to transition to the BMD-DA over the next decade. It is strongly preferred that proposed solutions rely only on technology expected to be available over that time horizon. MDA will review responses to this RFI to determine compatibility with the goals stated above, and intends to invite authors of selected responses to a conference scheduled for the week of May 12, 2003 to review future research directions for advanced algorithms supporting Command & Control, Battle Management, & Communications (C2BMC) in national missile defense. The conference will be based on a working panel format in which authors will brief their ideas and field questions and comments from a review panel. The outcome of the conference will be the generation of a roadmap for model and algorithm development that will support the deployment of an effective missile defense decision architecture over the next decade. The government invites interested parties to submit presentation abstracts and candidate solution methods. Abstracts should be 1-2 pages in length, and provide the following information: (1) title; (2) author & contact information; (3) abstract to include the following: summary of the solution concept; discussion of why this approach appears to be promising; suggested success criteria; representative performance results, if available; and anticipated applicability to missile defense weapon & sensor allocation; (4) recent work performed relevant to the abstract; (5) organization(s) for which the work was performed; and (6) references and papers supporting the abstract. Any anticipated limits, constraints, or costs related to testing or distributing required software should be clearly stated, as well as any proprietary & non-disclosure requirements. The Government shall not be liable for or suffer any consequential damages for any proprietary information not properly identified. Proprietary information will be safeguarded in accordance with applicable Government regulations. Authors should send responses electronically (in Word or PDF format) to Ms. Joy Smith (
[email protected], telephone 703-695-9209) at MDA/CTI no later than 04/18/2003. This is a request for information (RFI) and no contract will be awarded pursuant to this announcement. This RFI does not constitute a commitment, implied or otherwise, that the government will issue a procurement. The government does not intend to otherwise pay for information received in response to the RFI.