![](bg2.png)
Internet is augmented with the possibility of physi-
cal objects referencing services on the Internet, and
the ability for electronic services to track physical
objects, and access information stored on them, we
move towards a network that is no longer confined to
the electronic domain but deals instead with objects
and concepts that are present both in the electronic
and physical worlds.
The emergence of the IoT requires a network that
enables objects to reference electronic services on the
Internet, as well as technologies that enable entities
in the electronic domain to gain access to information
stored on individual product items. At the low end of
spectrum, ”things”, or physical objects participating
in the IoT just have some kind of identification, which
is machine readable in an automated fashion, such as
RFID or bar code tags. These objects are ”on-line”
only momentarily, perhaps just by an one-way link
that enables a tracker application to receive updates
on the new position of the object. Examples include
pallets with RFID tags in a logistics network. At the
high end of the spectrum, the objects are intelligent
themselves, they can gather statistical data about their
usage and have means of communicating that infor-
mation.
The IoT raises questions how the right to privacy
can be ensured. Privacy should be taken into account
already at the design phase of the product, by apply-
ing the principles of Notice, Consent, Anonymity, Lo-
cality, and Adequate security presented in (Langhein-
rich, 2001). The user should be made aware that data
is being recorded, and recording should not happen
without the user’s express approval. Collected data
should be anonymized, communicated using secure
means, and the user should have control on the data
also after it has been stored for analysis.
To enable the product items to have individual
data and events associated with them, they must be
identifiable in some manner. This means that a global
registry of some sort must be created, with every
product item having its own Globally Unique Product
Identifier (GUPI) (Fr¨amling et al., 2007b). The identi-
fication is used to tie the physical object to some kind
of proxy service, or product agent (K¨arkkinen et al.,
2003) running on a host on the Internet. The proxy
acts as a representant providing information or ser-
vices on the behalf of the object.
There are several approaches to the identity-to-
proxy mapping. The address (as an Uniform Re-
source Locator, URL) of the proxy is either directly
or indirectly obtainable from the object, the so-called
direct or indirect sensing described in (Kindberget al.,
2002). In the direct sensing case, the object itself con-
tains the complete URL of the proxy service. In the
indirect sensing case the identity of the product is an
Uniform Resource Name (URN), which is used as a
search key to find the URLs of the proxies from a cen-
tral repository of identity-to-URL mappings, much in
the same way that the Domain Name System (DNS)
of the Internet works for converting Fully Qualified
Domain Names to Internet Protocol addresses.
The ID@URI naming scheme used in the Dialog
collaborativelogistics system (K¨arkkinen et al., 2003)
developed at the Helsinki University of Technology
represents the direct sensing approach. In the Dia-
log system, every item has an identity of the form
identity@URI
, where the URI part is the address
(URL) where the proxy service can be found, and the
identity part is used to identify that particular item to
the proxy. This scheme has the advantage of the in-
herent simplicity of the direct sensing approach, i.e.
no look-up service is required. On the other hand, if
the URL of the proxy service was to change at some
point, it would require a re-programming of every
item with the new address, a huge task. The ID@URI
scheme works well in a logistics network environ-
ment, where the existence of a particular item on the
network is temporally limited, but in an Internet of
Things environment, where the product is present on
the network for its entire lifetime, the static nature of
the URI could be problematic.
The EPCglobal network developed by the Auto-
ID consortium is using the indirect sensing approach
to map product identities to proxy services. In
this case, the GUPI is the Electronic Product Code
(EPC), which defines a way to encode existing prod-
uct numbering schemes for use in RFID tags and
inter-organizational data exchange. The EPCglobal
implementation of the identity-to-proxy mapping is
named the Object Naming Service (ONS), and is an
extension to the DNS system (Armenio et al., 2007).
This approach has the inherent advantage that the ad-
dress of the proxy can change freely. On the other
hand, this scheme requires an organization that con-
trols the allocation of id’s, because every id must be
unique. This could make the scheme less attractive to
manufacturers, because they are then dependent on an
external organization for obtaining the unique IDs.
The World Wide Article Information (WWAI)
system proposed by Trackway is a peer-to-peer net-
work for propagating requests related to product in-
formation. This approach represents the indirect sens-
ing approach, with the responsibility of maintaining
the mappings from GUPIs to product agent URLs dis-
tributed among the nodes of the network. The infor-
mation providers, manufacturers, form a network of
trusted nodes, which are identified by company num-
bers defined in existing product numbering schemes.
GATHERING PRODUCT DATA FROM SMART PRODUCTS
253