130 likes | 320 Views
Spatial and Spatio-temporal Networks Reem Ali, Amr Magdy. Spatial Networks. Examples: Road, Train and River networks. Data Models. Conceptual Data Model Graphs Logical Data Model Data types: Graph, Vertex, Edge, Path, … Operations: addEdge (), getSuccessors (), ... Physical Data Model
E N D
Spatial Networks • Examples: Road, Train and River networks
Data Models • Conceptual Data Model • Graphs • Logical Data Model • Data types: Graph, Vertex, Edge, Path, … • Operations: addEdge(), getSuccessors(), ... • Physical Data Model • Memory based: Adjacency list, Adjacency Matrix • Disk based: normalized and denormalized
Data Models (cont’d) • Normalized Representation • Denormalized Representation
Query Languages for Graphs • Relational Algebra-based languages cannot compute “Transitive Closure” • Support for Graph Queries: • CONNECT (SQL92 for DAG) • RECURSIVE (SQL3) Transitive Closure
Graph Algorithms • Traversal: Breadth First Search, Depth First Search • Shortest Path: Dijkstra’s and Best First • Large graphs cannot fit in main memory => Hierarchical Strategies • Boundary Graph • Fragment Graphs
Access Methods • We need to minimize I/O cost for graph algorithms. • e.g. getSuccessors() is a dominant I/O cost for many queries • Connectivity Residue Ratio (CRR)= total no. of unsplit edges/total no. of edges • Graph Partitioning: to maximize CRR. • e.g. CCAM
Spatio-temporal Networks • Definition: a network whose status changes with time • e.g., road networks status changes from time to another due to traffic changes • e.g., air travel paths changes due to weather conditions
Applications • Road Networks • e.g., Emergency traffic planning, route finding services, Minimizing travel times, Freight Delivery Services, etc. • Air travel networks • e.g., flight route planning
Representations • Snapshot-based Graph Collection • Time-expanded Graph A A 2 A 1 C C C 3 1 2 3 B 1 B B at t=0 at t=1 at t=2 t=0 t=2 t=1 t=3 t=4 t=5 A0 A2 A1 A3 A4 A5 B0 B2 B1 B3 B4 B5 C0 C2 C1 C3 C4 C5
Representations • Time-aggregated Graph A [1,2,-] C [2,3,1] [1,-,3] B
Important Problems • Fixed start-time shortest path • All start-times shortest path • Recommending best travel start-time
All Start Times Shortest Paths • Shortest path does not change every time instant • Instead, each interval of time hasshortest path e.g., rush vs. non-rush hour • One path calculation is required per interval(big computation saving) • Start time instants when the shortest path between a source and destination may change are called critical time points