Towards Simplified Optimal Sector Splitting Billy Josefsson - LFV - - PowerPoint PPT Presentation
Towards Simplified Optimal Sector Splitting Billy Josefsson - LFV - - PowerPoint PPT Presentation
Towards Simplified Optimal Sector Splitting Billy Josefsson - LFV Valentin Polishchuk, Leonid Sedov - Linkping University dynamic Demand Capacity Balancing (dCDB) Traffic density - changing Sectors - static 9:00 AM Controller A - A
dynamic Demand Capacity Balancing (dCDB)
Traffic density - changing Sectors - static A B Controller A - Controller B - 9:00 AM Everyone is happy! For how long?
2
dynamic Demand Capacity Balancing (dCDB)
Traffic density - changing Sectors - static A B Controller A - Controller B - 12:00 Solution?
3
dynamic Demand Capacity Balancing (dCDB)
Traffic density - changing Sectors - static A B Controller A - Controller B - Controller C - 12:00 C Everyone is happy again
4
dynamic Demand Capacity Balancing (dCDB)
Traffic density - changing Sectors - static A B Controller A - Controller B - Controller C - 14:00 C Do we need sector C?
5
$$$
dynamic Demand Capacity Balancing (dCDB)
Traffic density - changing Sectors - static A B Controller A - Controller B - Controller C - 14:00 C Do we need sector C?
6
dynamic Demand Capacity Balancing (dCDB)
Traffic density - changing Sectors - static A B Controller A - Controller B - 14:00
7
Gluing
[Kjellin ‘14] [Yousefi and Donohue ‘04]
8
Cutting
[Gerdes et al., SID’16]
9
Binary split
Split airspace into 2 parts Need more sectors? - Do recursive split.
10
Convex boundaries
Line segment intersects convex region at most once
11
Straightline cuts
Chord joins two points on boundary
12
KPIs
Maximum imbalance - peak complexity Average imbalance - total complexity Maximum imbalance
13
KPIs
Maximum imbalance - peak complexity Average imbalance - total complexity Average imbalance Balance max or avg?
- We can both!
14
Input
- Region P
- Set S of straightline flight segments
○ Segment start and end coordinates ○ Times when aircraft enters and leaves segment
15
Сritical points
16
Сritical points
17
Сritical points
18
Сritical points
19
Сritical points
20
Сritical points
21
Сritical points
22
Ham sandwich theorem
23
http://www.etudes.ru/en/etudes/ham-sandwich-theorem/
Proof: avg = 0 split exists for each point
24
Proof: avg = 0 split exists for each point
25
1
Proof: avg = 0 split exists for each point
26
1
Proof: avg = 0 split exists for each point
27
1
Proof: avg = 0 split exists for each point
28
0.1 0.9
Proof: avg = 0 split exists for each point
29
0.6 0.4
Proof: avg = 0 split exists for each point
30
0.5 0.5
Proof: exists max & avg = 0 split
31
Proof: exists max & avg = 0 split
32
avg = 0 max > 0
B A
ML(AB) = 3 MR(AB) = 2
Proof: exists max & avg = 0 split
33
B A
avg = 0 max > 0 ML(AB) = 3 MR(AB) = 2
Proof: exists max & avg = 0 split
34
avg = 0 max > 0
B A
ML(AB) = 4 MR(AB) = 2
Proof: exists max & avg = 0 split
35
A B
avg = 0 max < 0 ML(AB) = 2 MR(AB) = 3
Proof: exists max & avg = 0 split
36
avg = 0 max > 0
B A
ML(AB) = 3 MR(AB) = 2
1: Find critical points
37
2: Lines through all pairs of critical points
38
2: Lines through all pairs of critical points
39
2: Lines through all pairs of critical points
40
3: Find pairs of intervals with 0 M-imbalance
41
3: Find pairs of intervals with 0 M-imbalance
42
3: Find pairs of intervals with 0 M-imbalance
43
3: Find pairs of intervals with 0 M-imbalance
44
3: Find pairs of intervals with 0 M-imbalance
45
4: Look for avg-balanced splits
46
4: Look for avg-balanced splits
47
Artificial example
48
M-imbalance changes
49
Rectangular boundary
50
Straightline flight segments
51
Optimal cut
52
All optimal cuts
53
Second cut
54
Second cut
55
Further research
- Multiple sectors
- Directions of flights
- Interaction between flights and boundaries
- Choose between optimal cuts (KPI?)
56
Further research
- Multiple sectors
- Directions of flights
- Interaction between flights and boundaries
- Choose between optimal cuts (KPI?)
57