At each node, the histograms
obtained from the children can be used to determine if the matching events
exist, thereby pruning the descent that contains only unrelated events. For geographically
constrained range queries (e.g., all events within the temperature
range [50, 60] and in region A), the querying node can pick some intermediate
nodes in the hierarchy (those that cover part of the query range and the
querying region A) and initiate traversal from there.
4.6 Distributed Multi-dimensional Indices
We mentioned earlier that DIFS allows range queries over two keys or attributes
where one of the attributes is geographical location and the other is
a range of values. This kind of query is called a two-dimensional range query.
In this section we consider an index that permits range queries over multidimensional
data. A multi-dimensional range query has some constraints on
multi-attributes. As an example of the need for such queries, consider scientists
analyzing the growth of marine micro-organisms. They might be interested in
events that occurred within the temperature range from 50?—¦C to 60?—¦C and the
light range from 10 to 20. The query may be expressed as ???List all events that
have temperatures between 50?—¦C and 60?—¦C, and light levels between 10 and
20??™. In this example, the scientists are interested in the e?®ects of geographical
region, temperature and light on the growth of marine micro-organisms.
Pages:
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494