A value-based indexing technique for time sequences
Ling Lin · 1997
A time sequence is a discrete sequence of values, e.g. temperature measurements, varying over time. Conventional indexes support efficient querying of explicit (stored) values of time sequences. In real life time sequences are often viewed as continuous and it is very important to support efficient querying of implicit values (interpolated values). For inverse queries (i.e., computing the times when the values satisfy some conditions) concerning implicit values, the entire time sequence has to be scanned, which is unacceptable when the volume of the time sequence becomes large. This thesis presents a dynamic indexing technique, termed the IP-index (Interpolation -index) for large time sequences. This index supports efficient processing of inverse queries concerning implicit values. It can be implemented on top of regular ordered indexing techniques such as B-trees. Performance measurements show that this index dramatically improves the query processing time of inverse queries. For peri...