al., 2006). The policy itself is in turn defined using
WS-Policy (S. Bajaj et al., 2006). Based on the re-
ceived policy the Subject selects an Information Card
and thus the Digital Identity that should be disclosed
in Step (2). The component that holds the Informa-
tion Cards and allows to choose which identity should
be disclosed is called Identity Selector or Card Se-
lector. The selected card, in case it is not a per-
sonal card (a card representing a self-issued digital
identity), contains the information that is necessary
to contact the responsible IdP. In Step (3), the Sub-
ject, or to be precise the Identity Selector on behalf
of the Subject, negotiates the policy with the IdP,
again using WS-MetadataExchange and WS-Policy.
The Subject requests a Security Token based on the
previously received policy using WS-Trust (OASIS
WS-Trust, 2007) in Step (4). In order to request
the token, the Subject needs to authenticate to the
IdP. In the OASIS-IMI standard, four different au-
thentication mechanisms are defined. These are user-
name and password, kerberos v5 credential, X.509
certificate credential, and a self-issued Security To-
ken (OASIS-IMI, 2009). A self-issued Security To-
ken is a Security Token issued by the subjects Card
Selector using a personal card and not by an IdP as
in case of a managed card. In Step (5), the Subject
transmits the received Security Token to the RP using
WS-Security (A. Nadalin et al., 2004). After receiv-
ing the Security Token the RP verifies the token and
grants access to the desired service.
4 AFFIRMATIVE STATEMENTS
The previous sections gave an introduction to Infor-
mation Cards, identification and authentication in e-
government and the Austrian Citizen Card. In this
section we will present our approach to use a Citizen
Card together with Information Cards for identifica-
tion and authentication in e-government services.
Our aim is to combine the strong means of au-
thentication attained by qualified electronic signatures
and identification provided by a Citizen Card with
the common user experience of Information Cards.
Therefore, it is desirable to have the user’s Citizen
Card appearing as just another card in the Card Se-
lector, such that the Citizen Card may be used in a
manner similar to the use of any other information
card.
A Citizen Card carries the same information that
is also typically available with Information Cards.
Given name, family name and date of birth are Claims
often required by RPs. In addition the qualified cer-
tificate on the Citizen Card often includes the signa-
tory’s e-mail address – another Claim that is typically
required by RPs. For RPs providing e-government
services, the most important Claim will however be
the government-issued identifier (or an derived iden-
tifier as discussed in section 2).
As already mentioned Citizen Card based identi-
fication relies on pre-issued identities. The only third
party interaction required upon identification and au-
thentication is obtaining revocation status informa-
tion for the qualified certificate. There is no identity
provider involved in the identification and authentica-
tion process.
The Information Card model introduces the con-
cept of self-issued Security Tokens for authentication
without third-party IdP interaction. In the Information
Card model, as defined in OASIS-IMI (OASIS-IMI,
2009), self-issued Security Tokens always base on
self-asserted identities. We are however going to use
self-issued Security Tokens to provide government-
issued, and therefore government-asserted and not
self-asserted, identity to the RP.
Security Tokens are signed by the issuer before
they are used for authentication at the RP. As our aim
is to use qualified electronic signatures for authentica-
tion, at a first glance it seems obvious to apply a qual-
ified electronic signature directly on the self-issued
Security Token. Considering the following issues we
have however chosen to follow a different approach.
• OASIS-IMI does not describe the usage of signa-
ture algorithms other than RSA. However, many
Citizen Cards use different signature algorithms
(e.g. ECDSA). Therefore, using such signature
algorithms would most likely result in interoper-
ability issues with existing RP implementations.
• As discussed in section 2, national implementa-
tions of the Signature Directive require that the
signatory must have the ability to view the data
before signing. However, there is no standard rep-
resentation of a Security Token in a form mean-
ingful for an end user. Transformation of the Se-
curity Token in an end user readable form before
signing would violate the OASIS-IMI specifica-
tion and would therefore again result in incom-
patibility to existing RP implementations.
For the given reasons we have decided to define a
new Affirmative Statement (ASt) Claim. The Claim
includes a statement in natural language similar to the
one shown at the end of section 2 and has to be signed
by the user with a qualified signature for the following
authentication process. The ASt Claim also includes
the government-issued identifier (either the identifier
itself or a derived identifier thereof). This generated
Claim is added to a self-issued Security Token, which
INFORMATION CARDS AND AFFIRMATIVE STATEMENTS
337