Antenna Network Optimization using Artificial Intelligence Techniques

Scheduling Air Force''''s Space Communications resources is a large, complex, distributed problem.  We have developed an innovative planning and scheduling system architecture which facilitates the development of more intelligent, optimizing scheduling algorithms.  It has been applied to NASA Ground Network (space communications) scheduling and several domains including those involving similarly complex constraints, such as the problem of optimizing ballistic missile engagements which includes complex line of sight (LOS) and range calculations between satellite-based sensors and three-dimensional ballistic missile trajectories.  The innovation is to apply that expertise, the planning and scheduling architecture, and a new domain-independent algorithm, to the problem of scheduling and optimizing the antenna network. The ultimate goal of this proposed effort is to improve the satellite communications resource scheduling capability.  The goals of the Phase I research are to understand the current and future antenna network resource and beam  scheduling domain, investigate integration requirements, elaborate the heuristics, algorithms and techniques for improved scheduling, analyze them as to their feasibility, define the metrics for scheduling performance, further prove the feasibility through prototype development, and develop the Phase II system design. BENEFIT: The most direct target for the results of this effort is the Air Force Satellite Control Network (AFSCN).  By both demonstrating significantly improved performance during Phase II and ensuring that the ultimate results can be integrated into the operational scheduling system, the Air Force will have a strong incentive to operationalize the Phase II development of the proposed Scheduler.  We are already marketing Aurora, our general intelligent planning and scheduling framework, and customizing it for a variety of domains.  The additional capabilities developed for communication scheduling can be synergistically marketed to the same clients.  For example, the bottleneck avoidance algorithm proposed to be fully developed here would be applicable to many Aurora customers.  We anticipate that this effort will result in additional scheduling algorithms that we will be able to incorporate into our existing scheduling products, thus increasing the benefits they provide and their value.