Problem formulation for optimal array modeling and planning

To support NASA's vision to increase the Deep Space Network (DSN) communications capability by at least 100 times the current capability of the 70m antennas, the option of large arrays of thousands of 12m antennas is being studied at the Jet Propulsion Laboratory (JPL), and an operational prototype is planned for the 2010 time frame. The flexibility of dynamically subdividing a large antenna array into smaller array clusters of various sizes to support different concurrent missions and the ability to add or remove antenna elements from an array cluster without interrupting the signal tracking definitely enable new network operation concepts. Yet there are unique challenges to the modeling and planning of the large array. Current DSN antenna planning and scheduling is done based on network support requests from individual missions that perform their own communication link analysis and in-view period analysis. The largely manual 'horse-trading' among missions is done, by and large, based on antenna tracking-time metrics and does not take into account antenna network performance and reliability. Recognizing the iterative nature of negotiations for resources, missions tend to grossly overestimate their required coverage time, thus reducing the overall network efficiency. The above manual approach will break down in the operation of a large array of thousands of antennas. To efficiently utilize the large array, the modeling and planning process needs to 1) be highly automated, 2) take into account link capability and antenna element reliability, and 3) support long-term, short-term, and instantaneous planning. In this paper we describe an optimal modeling and planning framework for the future large array of DSN antennas. This framework takes into account the array link performance models, reliability models, constraint models, and objective functions, and determines the optimal sub-array clusters configuration that will support the maximum number of concurrent missions based on mission link properties, antenna element reliabilities, mission requests, and array operation constraints. Array cluster size can vary dynamically during the support of a mission. Larger numbers of small antennas may be needed during the acquisition/calibration phase. Some of these antennas may be released for other uses during the tracking phase. Thus, resources will be efficiently allocated to achieve full utilization.