Multiobjective optimization of a radio telescope array with site constraints

Abstract

The next generation of radio telescope interferometric arrays requires careful design of the array configuration to optimize the performance and minimize the cost of the overall system while adhering to site constraints. A framework, based on genetic algorithms, for rapid exploration and optimization of the objective space pertaining to multiple objectives has been developed. A large space of possible designs has been evaluated for 27-, 60-, 100-, and 160-station arrays. The 27-station optimizations can be compared to the well-known VLA case, and the larger array designs apply to arrays currently under design such as LOFAR, ATA, and the SKA. In the initial implementation of the framework, designs were evaluated with respect to two metrics, array imaging performance and the length of cable necessary to connect the stations which adhere to LOFAR design objectives and parameters. Imaging performance is measured by the degree to which the sampling of the uv plane is uniform. The length of cable used to connect stations was determined to be a key cost parameter in the construction of the array. Site constraints are a key issue in any ground-based system and have been included to ascertain the effects on the objective space. Results will show non-dominated configurations for the metrics chosen as one metric is emphasized over another, as well as a comparison of how site constraints affect the overall design and convergence of the genetic algorithm simulations. The framework is general, and may be applied to other design goals and issues, such as particular schemes for sampling the uv plane, array robustness, and phased deployment of arrays.