Database Reference
In-Depth Information
when a set of pallets are loaded onto a track, the set of EPC readings
for all the objects are inserted as the children of the EPC of the truck,
in the
CONTAINMENT
table. Thus, an event detector continuously mon-
itors the observation streams, and triggers actions which generate the
corresponding data.
One challenge with managing RFID data, which was noticed in [68]
was that RFID data typically have very large volume, which can lead to
accumulation of large volumes of data. This can lead to slower queries
and updates. An important observation about RFID data is that they
typically have a limited life span, starting from the time it is first tagged,
to the time when it is sold to the customer. Therefore, the database
management approach in [68] partitions the data into an
active
set of
RFID data, which corresponds to items which are frequently updated;
and an inactive set of data, which corresponds to items that are no longer
updated frequently. Since the majority of the data becomes inactive over
time, this leads to much faster queries of the active data during its life-
cycle.
3.1 E
cient Warehousing of RFID Data
A related, but somewhat different kind of RFID data management
and warehousing has been discussed in [29]. This approach is designed
towards finding the relevant paths of items in the RFID scenario. This
process is also designed towards modeling the dynamic relationships such
as
containment
, except that it does so not just for explicit containment,
but also for items which move together. Also, the mapping relationships
are modeled somewhat differently. The approach is also designed for
tracking specific
measures
associated with the RFID items, which is
typical in a data warehouse.
As in the case of [68], methods need to be designed to handle the
massive redundancy of different types. These could be because of mul-
tiple readings of the same item from the same reader at multiple times.
Consider the situation, where a typical reading from an RFID tag is of
the form (
EPC,Location,Time
). We note that the same tag may be
read many times at the same location, even though no significant event
may have occurred involving the time. As in [68], the only two readings
which are significant are the first and last moment at which the items
were read. The work in [29] uses two main kinds of compression.
Temporal Compression:
Multiple scans of the same code at the
same (virtual) location can be compressed significantly. For exam-
ple, if an item is loaded on an ship from one port to another, then
the virtual location of the item corresponds to “
ship
” and all scans
