Distributed Radio Telescope as Massively Parallel Computing System: A Case Study

A giant distributed radio telescope is an unusual kind of massive parallel computing system. The design of such a system is a multidisciplinary effort that is hard to accomplish within a reasonable period of time, and with a high level of confidence that the resulting system will ultimately be what it has to be, now and in the course of its entire life span. Issues like deriving specifications from sometimes incomplete and conflicting requirements, exploring alternatives in behaviors and organizations, and answering what if questions in an interactive way are not easily mastered without having a methodology and a framework in which all these crucial aspects of a design trajectory can be integrated. There is still a large gap between what that methodology and this framework should be and what the current approaches are. In an attempt to shed more light on that gap, a few telescope concepts, like the Low Frequency Phased Array (LOFAR), and the Square Kilometer Array (SKA) telescopes that have been proposed to be built in the coming decades have been analyzed, and down-scaled versions of these telescopes have been taken through an iterative exploratory specification process with the objective of identifying what the methods in the methodology and the tools in the framework should be for the design of the giant telescope to be successful. The paper reports on one of these case studies which is the Thousand Element Array (THEA), a down-scaled version of the SKA radio telescope