Database Reference
In-Depth Information
them. Using some spatial database capability, these line segments can then
be checked for intersection with the river Rhine. If we have kept times and
positions for the observations corresponding to the start and the end of the
line segment, we may again do interpolation as for the previous query.
Assume we have a table for rivers.
Rivers(Name: string, Curve: line)
In the MOD, the query is written as follows.
select v.Id, inst(initial(v.Trip at r.Curve)) as PassingTime,
val(initial(v.Trip at r.Curve)) as PassingPos
from Vehicles as v, Rivers as r
where r.Name = "Rhine" and v.Trip passes r.Curve.
Again, query operations are explained in the following section.
Besides easier formulation of queries, a MOD system can offer more efficient
implementation techniques including indexing and query optimization as the
system is “aware” of the moving objects.
The rest of the chapter is structured as follows. Section
3.2
describes the data
model and query language for a MOD based on abstract data types. There are
two prototypical implementations of this model, SECONDO and Hermes. In Sec-
tion
3.3
we describe SECONDO. Section
3.4
discusses alternative representations
of sets of moving objects in the context of this model, including creating the
representations from raw trajectories. Section
3.5
addresses indexing of moving
objects. Section
3.6
provides a short introduction to Hermes, the other MOD
prototype, and explains some differences. The chapter ends with conclusions
(Section
3.7
) and bibliographic notes (Section
3.8
).
3.2 Data Model and Query Language
In this section we address the extensions of a DBMS data model and query
language to support representation and querying of moving objects. We have
already seen in the introduction that the basic idea is to use abstract data types.
These can be embedded in the role of attribute types into a relational or other
DBMS model, and the ADT operations can be embedded into the DBMS query
language, typically SQL.
The fundamental data type
moving point
(
mpoint
) to represent a trajectory
also has been introduced already. To obtain an expressive query language, the
model provides several further data types together with a carefully designed set
of operations. In the following section we motivate and introduce these types
and operations by examples. Later we consider the design principles that have
led to this model and we briefly sketch its implementation.
