Balancing Controllers' Workload by Locally Redesigning Airspace Sectors Irina Kostitsyna, Joseph Mitchell, Girishkumar Sabhnani

Embed Size (px)

Citation preview

  • Slide 1
  • Balancing Controllers' Workload by Locally Redesigning Airspace Sectors Irina Kostitsyna, Joseph Mitchell, Girishkumar Sabhnani
  • Slide 2
  • Kansas City Air Route Traffic Control Center
  • Slide 3
  • Slide 4
  • Slide 5
  • Sectorization Problem Given: Polygonal domain D Set of tracks T={ i (x,y,z,t)} Goal: Partition D into k sectors that have balanced workload conform to the traffic flow have nice shape
  • Slide 6
  • Sectorization Problem Sector Workload Avg airplane count Max airplane count Delay Number of hand-offs etc Flow Conformace Intersection angles Distances to intersection points Min dwell time Nice Sector Shape Min angle Max angle Convexity Min edge length
  • Slide 7
  • Our Method: Resectorization Start with some initial sectorization Extract dominant flows and critical points Introduce parameters MIN_SECTOR_ANGLE, MIN_DWELL_TIME, DELAY, etc Introduce a cost function on sectors Locally move sectors vertices to minimize the cost
  • Slide 8
  • TTLLT
  • Slide 9
  • GeoSect
  • Slide 10
  • Before and After
  • Slide 11
  • Human in the Loop Experiment 2x load of traffic in ZKC 3 time intervals MIP initial sectors Goal: minimize the delay with 6 or less sectors
  • Slide 12
  • Comparison to other methods
  • Slide 13
  • Theoretical Questions Partition D into min numer of sectors under some constraints on sectors geometry flow conformance workload For a given vertex v and 3 neigboring sectors find the best relocation of v
  • Slide 14
  • Sectorizing is Hard
  • Slide 15
  • Future and Current Work Dynamic rebalancing Robust sectorization Introducing new requirements weather abstract demand etc
  • Slide 16