NASA’s Deep Space Network (DSN) provides communication and other services for planetary exploration for both NASA and international users. The DSN operates antennas at three complexes in California, Spain, and Australia, with the longitudinal distribution of the complexes enabling full sky coverage and generally providing some overlap in spacecraft visibility. Beginning in 2018, the DSN will be transitioning to a remote operations paradigm where local dayshift operators at each complex will be preparing and staffing the links (or contacts) for all antennas in the DSN. In addition, the number of simultaneous links an operator will be required to support will increase from two to three. Without tools to manage the increased link complexity, there is a risk that operators will be overloaded.
A new approach was developed to incorporate link complexity into scheduling, including modeling link complexity and the generation of automated operator link assignments. Minimizing the combined complexity of the multiple links assigned to a single operator is emerging as a key tool in achieving the DSN’s overall automation objectives.
Not all activities are equally demanding, and when link control operators (LCOs) are managing multiple activities at once, it is easy to see that inadvertent overloading of the operations staff is a potential risk. As a result, the new approach models the complexity of individual activities, and avoids overloading individual LCOs with too much work at one time. There are two major parts to this effort: one that occurs during schedule generation, and one that occurs during shift planning.
During schedule generation (weeks to months ahead of execution), the technique predicts the occurrence of spikes in loading, and provides feedback to users so they can make adjustments early in the process before the schedule is firm. In addition, busier periods of link complexity could serve as early warning that additional or overtime staffing may be required to cover a particular time frame or critical event.
During shift planning (hours to days ahead of execution), it determines a good assignment of work to operators that does not exceed threshold values for number of links or overall link complexity, and, as much as possible, evenly distributes the work across the available operations staff.
The solution consists of a model of the operators and a schedule of links as inputs. The operator model contains two timelines: a link count, which is an integer resource measuring the number of links assigned to the operator; and complexity value, which is a floating-point resource measuring the total complexity of the links assigned to the operator. Each of the timelines has an associated limit, such that exceeding the limit for any duration is considered a conflict.
Four types of complexity functions are included:
- No complexity – A complexity value of zero was used for all links.
- Constant complexity – A constant value was applied for each section of the link.
- Linearly decreasing complexity – The complexity linearly decreases to 0 over the duration of each link phase (with 5-minute quantization).
- Exponentially decreasing complexity – Complexity is modeled as proportional to exp(–t/60), where t is time in minutes since phase start (also with 5-m quantization).
A Web interface was developed to visualize the schedule and timeline results. Users can choose schedules from the DSN to invoke the automated scheduling algorithm. A single-step version of the algorithm can be invoked where only one link is scheduled at a time. It can be used for debugging as well as demonstrating to users the order that links are evaluated and the operator assignment. There is also the capability to manually edit the schedule and move a link from one operator to another. Timeline values and metrics are automatically updated on a manual change, giving users the opportunity to evaluate their change.
This work was done by Daniel Q. Tran and Mark D. Johnston of Caltech for NASA’s Jet Propulsion Laboratory. NPO-49879
This Brief includes a Technical Support Package (TSP).
Link Complexity Scheduling Algorithm
(reference NPO49879) is currently available for download from the TSP library.
Don't have an account? Sign up here.