Database Reference
In-Depth Information
of the item aboard the ship can be reduced to two scans. In prac-
tice, the location is associated with the identifier of a fixed sensor
reader with a virtual location, such as “
ship
”. These two scans cor-
respond to the time that the item was loaded on the ship, and the
time at which the item was removed from the ship. Therefore, by
storing the times when the item was first moved to the vicinity of
the reader and the time that it was moved away from the reader, all
the relevant information is represented [29]. This allows us to rep-
resent the item in the form (
EPC,Location,TimeIn,TimeOut
).
We note that the
TimeIn
and
TimeOut
variables are similar to
the
tstart
and
tend
variables proposed in [68] for maintaining
the object location tables.
Group-based Compression:
In most real scenarios, RFID items
often move together in groups or consignments. For example, all
items which are loaded onto the ship stay together throughout the
trip. Therefore, all the individual RFID of the items can be re-
placed by a single
generalized identifier
, or GID. In practice, groups
of items may split or merge, as items are loaded at ports from dif-
ferent sources, or split into different destinations. Correspondingly,
the generalized identifiers can be arranged hierarchically, in order
to effectively represent these merges and splits.
One challenge with the use of RFID data with traditional data ware-
housing techniques, is that traditional warehousing methods do not prop-
erly consider the spatial links between different data records, which are
essential in the RFID scenario. Therefore, traditional dare warehousing
techniques may fail, when they are directly applied to RFID data. For
example, consider the situation, where the cleaned RFID representation
is of the form (
EPC,Location,TimeIn,TimeOut
:
Measure
), where
“Measure” could correspond to a value such as the quantity of the item
present at the given location. Such a representation could be used in or-
der to respond to queries such as the number of items which are present
at a given location at any given period. However, it cannot be used
to determine the number of items which moved from one location to
another in a given period, at least with traditional data warehousing
operations.
Therefore, RFID warehouses can be represented in the form of three
different tables [29]: (a) an
info
table which contains location indepen-
dent information about the items, such as its SKU, Product type etc.,
(b) a
stay
table which essentially contains all the set of facts in the form
(
EPC,Location,TimeIn,TimeOut
:
Measure
) (or in aggregate form
as GIDs instead of EPCs), and (c) a
map
table which contains the links
