Payload Scheduling with Dynamic (Re)Tasking
Total Page:16
File Type:pdf, Size:1020Kb
Tasking Information for Spring 2006 UCD Math Clinic
Dynamic Scheduling of Prioritized Tasks Tasking and Allocation information using commercial remote sensing satellite tasking as a real-world example
Task Data A task represents a user’s request for the satellite to perform some activity. In the context of the 2006 Math Clinic, a task represents a request to perform remote sensing of the Earth (e.g. to take a picture). A more accurate term for task is Collection Requirement (CR). In some systems, the CR is the external request and the Task is the translation of the collection requirement into a software object that the mission scheduling system can interpret. For our purposes the two terms can be used interchangeably.
The format for a task will be an ASCII text file in XML format. There are freeware libraries that support XML parsing, such as Xerces (for both Java and C++, at http://xml.apache.org) or xmltools.m (for Matlab Release R13, at http://mathworks.com/matlabcentral/ under the File Exchange link do a search for “xml parsing”). XML parsing can also be done natively within Matlab if you know JAXP.
XML is a text format that holds elements and attributes. An element is defined with angle brackets (
For our XML tasking document, it will have this basic structure,
1 Tasking Information for Spring 2006 UCD Math Clinic units are in seconds. The priority is an integer (0 – 999), and latitude and longitude are units of degrees (expressed as a float or double). The strategy is either simple or monitor. If the strategy is monitor, then the sensor must periodically revisit the same target. All the monitor collections must be within the preferred window (hard constraint) and the min/max revisit values will be greater than zero. The minRevisit attribute will specify the minimum number of collections that must take place on the target, and the maxRevisit specifies the maximum number of collections that could take place on the target. Scheduling more than the minRevisit and less or equal to the maxRevisit is a bonus (objective function incentive). Scheduling more monitor collections than the maxRevisit is a soft constraint violation (a waste of the resource). The rule of thumb for monitor tasks is that they should be evenly distributed across the preferred (surveillance) window.
Soft Constraint Violation
Preferred Interval Effective Interval Simple Task
Hard Constraint Violation
Preferred Interval Effective Interval Monitor Task
2 Tasking Information for Spring 2006 UCD Math Clinic
Allocation Data To schedule a task means that you have “allocated” that task to a resource, or that you have created an “allocation”. The allocations you produce, that represent your schedule, should have the following format: ID a string that identifies this allocation TaskID a string that references the associated task ID State a single digit (0-2) that indicates NEW/MODIFY/DELETE StartTime a date + time, down to the second EndTime a date + time, down to the second Your schedule could be held in memory, a flat file or a database. If it is held in memory for some number of runs, then there should be a facility to eventually persist the schedule to a flat file or database for post analysis.
3