Deep Space Network Scheduling Using Multi-Objective Optimization with Uncertainty

Abstract

*We have developed a novel technique to incorporate uncertainty modeling within an evolutionary algorithm approach to multi-objective scheduling, with the goal of identifying a Pareto frontier (tradeoff curve) that recognizes the likelihood of events that can impact the schedule outcome. Our approach is particularly applicable to the generation of multiobjective optimized robust schedules, where objectives are assigned a service level, for example that we require an objective value to be !X with Y% confidence. We have demonstrated that such an approach can, for example, minimize scheduling on less reliable resources, based solely on a resource reliability model and not on any ad hoc heuristics. We have also investigated an alternative method of optimizing for robustness, in which we add to the set of objectives a failure risk objective to minimize. We compare the advantages and disadvantages of these two approaches. Future plans for further developing this technology include its application to space-based observatory scheduling problems. I. Introduction In the context of multi-mission scheduling of expensive shared systems such as communications resources, a critical challenge is that of exploring and managing tradeoffs among missions. For NASA’s Deep Space Network (DSN), this will become more acute if the network architecture evolves to incorporate arrays of smaller antennas that can be grouped dynamically and allocated in highly flexible ways. The objective of the work reported here is to develop techniques to explicitly optimize the multiple simultaneous and competing objectives of individual mission users as well as the network system as a whole. For the former, objectives center around maximally satisfying communications needs in terms of link quality, quantity, timing, and other factors. For the latter, objectives are based on minimizing cost and maximizing network availability. Along with the problem of competing multiple objectives, the DSN array would be subject to significant additional sources of uncertainty that complicate planning and scheduling. Chief among these is the sensitivity of Ka-band antennas to atmospheric moisture levels, which implies that weather will impact advance planning in ways that make longer lead-time scheduling more difficult. Other sources of uncertainty include equipment failures and return-to-service times, and unanticipated disruptive spacecraft events. In the following section (II) we first give an overview of the Deep Space Network (DSN) and its potential evolution to an array-based architecture — the proposed Deep Space Array Network (DSAN). We describe how this evolution would present both opportunities and challenges, and how uncertainty enters to complicate scheduling in ways not present in today's network. We then describe briefly our multi-objective approach to schedule optimization (III), and the evolutionary algorithm solution technique we are using. Next we discuss two approaches to incorporating uncertainty into the solution approach (IV), one based on explicitly modeling probability of failure as an objective to optimize (IV.A), the other based on a stochastic assessment of objective values (IV.B). We include an illustrative sample problem and show how it is solved using each technique. Finally we summarize our conclusions and describe some next steps (V). II. Overview of the Deep Space Network (DSN) and Array The Deep Space Network is NASA's collection of assets for communicating with spacecraft beyond near-earth orbit. It currently comprises dozens of large antennas of diameters 26m, 34m, and 70m, distributed geographically over three complexes spaced sufficiently far apart in longitude to afford full sky coverage (Goldstone, California; Madrid, Spain; and Canberra, Australia). In addition to the antennas, the complexes contain a variety of supporting