Spatiotemporal Data Indexing using hB π - tree
description
Transcript of Spatiotemporal Data Indexing using hB π - tree
Spatiotemporal Data Indexing using hBπ-tree
Evangelos Kanoulas, Georgios EvangelidisDepartment of Applied Informatics, University of Macedonia, Hellas
Kanoulas - Evangelidis 2
Spatial and Temporal Data
• Linear data
• Spatial data
• Temporal data
Applications, which require to support past, current and future data.
Kanoulas - Evangelidis 3
Transaction Time Databases
transaction time
records (tuples) time-invariant key time – variant attributes time interval
valid records
Kanoulas - Evangelidis 4
TSB-tree 1 - structure
Kanoulas - Evangelidis 5
TSB-tree 3 – splitting
Kanoulas - Evangelidis 6
hBπ – tree (Data nodes)
Kanoulas - Evangelidis 7
hBπ – tree (Index nodes)
Kanoulas - Evangelidis 8
Data Records – an example
TShB-tree – transaction time
Kanoulas - Evangelidis 9
TShB-tree - splitting
• Index nodes
•Using the D/fp algorithm
• Data nodes
•Time split
•Key split
Kanoulas - Evangelidis 10
D/fp – Τ/Κ (1)
Kanoulas - Evangelidis 11
D/fp – Τ/Κ (2)
Kanoulas - Evangelidis 12
D/fp – Τ/ΤΚ (1)
Kanoulas - Evangelidis 13
D/fp – Τ/ΤΚ (2)
Kanoulas - Evangelidis 14
% historical data nodes (node capacity 4K)
0 0.35
4.6
13.4
22.28
34.46 34.47 3537.4
42.72
0
5
10
15
20
25
30
35
40
45
70 75 80 85 90
% active records that indicate the type of node splitting
%
Τ/Κ Τ/ΤΚ
Kanoulas - Evangelidis 15
Index nodes efficiency (node capacity 4Κ)
69.78 69.78 69.17
64.96
59.51
46.33 46.35 46.76 46.29
43.14
40
45
50
55
60
65
70
75
70 75 80 85 90
% active records that indicate the type of node splitting
%
Τ/Κ Τ/ΤΚ
Kanoulas - Evangelidis 16
% historical data nodes
13
46
94
37
54
76
78 93
0
10
20
30
40
50
60
70
80
90
100
0 5 10 15 20 25 30 35
update - insertion rate
%
T/K T/TK
Kanoulas - Evangelidis 17
Data nodes efficiency
64.96
53.57
47.72 49.2
43.9745.7146.29
49.57
0
10
20
30
40
50
60
70
0 5 10 15 20 25 30 35
update - insertion rate
%
T/K T/TK