Statistical risk estimation for communication system design
Massachusetts Institute of Technology. Dept. of Aeronautics and Astronautics.
David W. Miller, Moe Z. Win, Kar-Ming Cheung, and Alvar Saenz-Otero.
MetadataShow full item record
Spacecraft are complex systems that involve many subsystems and multiple relationships among them. The design of a spacecraft is an evolutionary process that starts from requirements and evolves over time. During this process, changes can affect mass and power at component, subsystem, and system level. Each spacecraft has to respect overall constraints in terms of mass and power. The current practice in system design deals with this problem by allocating margins to individual components and to individual subsystems. However, a statistical characterization of the fluctuations in mass and power of the overall system (i.e. the spacecraft) is missing. This lack can result in a risky spacecraft design that might not fit the mission constraints and requirements, or in a conservative design that might not fully utilize the available resources. This problem is especially challenging at the initial stage of the design, when high levels of uncertainty due to lack of knowledge are unavoidable. This research proposes a statistical approach to quantify the likelihood that the design of a spacecraft would meet the mission constraints in mass and power consumption, focusing on the initial stage of the design. Due to the complexity of the problem and the different expertise required to develop a complete risk model for a spacecraft design, the scope of this research is focused on risk estimation for a specific spacecraft subsystem: the communication subsystem. The current research aims to be a "proof of concept" of a risk-based design approach, which can then be further expanded to the design of other subsystems as well as to the whole spacecraft. The approach presented in this thesis includes a baseline communication system design tool, and a statistical characterization of the design risks through a combination of historical mission data and expert opinion. Different statistical techniques are explored to ensure that the amount of information extracted from data and expert opinion is maximized. Specifically, for statistics based on data, Kernel Density Estimator is selected as the preferred technique to extract probability densities from a database of previous space missions' components. Expert elicitation is generated through a four-part model which quantifies experts' sensitivity to biases, and uses this measurement to compose properly the assessments from different experts. Finally, an optimization framework is developed to compare multiple possible design architectures, and to select the one that minimizes design objectives, like mass and power consumption, while minimizing the risk associated with the same metrics. Examples of missions are applied to validate the model. Results show that the statistical approach recognizes whether the initial estimate of the system is an overestimation or an underestimation, providing a valuable tool to measure the risk of a communication system at the initial state of the design. Specifically, statistics based on historical data and on expert elicitation allow the designer to size contingency properly, providing a reliable estimation of mass and power in the initial stage of the design. Thanks to this method, the communication system designers will be able to evaluate and compare different communication architectures in a risk trade-off prospective across the evolution of the design. Extensions to different subsystems and to additional metrics (like cost) make this model applicable to a wider range of problems.
Thesis (Ph. D.)--Massachusetts Institute of Technology, Dept. of Aeronautics and Astronautics, 2012.Cataloged from PDF version of thesis.Includes bibliographical references (p. 281-295).
DepartmentMassachusetts Institute of Technology. Dept. of Aeronautics and Astronautics.
Massachusetts Institute of Technology
Aeronautics and Astronautics.