SVG BASED SECURE UNIVERSAL MULTIMEDIA ACCESS

Ahmed Reda Kaced

GET / T

ecom Paris - LTCI UMR 5141 CNRS, Computer Science and Networks Department

37-39 Rue Dareau 75014, Paris - France

Jean-Claude Moissinac

GET / T

ecom Paris - LTCI UMR 5141 CNRS, Signal and Image Processing Department

37-39 Rue Dareau 75014, Paris - France

Keywords:

SVG, Universal Multimedia Access, content adaptation, Merkle Hash Trees.

Abstract:

In this paper, we develop and implement our Secure Universal Multimedia Access system (SUMA) for SVG

content in the following three subtasks. For content adaptation, we based on XML/RDF, CC/PP and XSLT,

for signing and authenticating SVG content we use Merkle hash tree technique and for content delivery, we

develop a mechanism for dynamic delivery of multimedia content over wired/wireless network.

We present Signature scheme and an access control system that can be used for controlling access to SVG

documents. The ﬁrst part of this paper brieﬂy describes the access control model on which the system is

based. The second part of this paper presents the design and implementation of SUMA adaptation engine.

SUMA aims to deliver an end-to-end authenticity of original SVG content exchanged in a heterogeneous

network while allowing content adaptation by intermediary proxies between the content transmitter and the

ﬁnal users. Adaptation and authentication management are done by the intermediary proxies, transparently to

connected hosts, which totally make abstraction of these processes.

1 INTRODUCTION

Delivering multimedia content to Web enabled mo-

bile devices such as phones and PDA presents a chal-

lenge; different devices have different content presen-

tation requirements. One of the most exciting devel-

oping technologies for this ﬁeld is Scalable Vector

Graphics (SVG) (W3C, 1998). SVG ﬁles are com-

pact and provide high-quality graphics on the Web, in

print, and on resource-limited handheld devices. SVG

is exciting because it offers Web developers a method

to create and animate images through an XML pro-

gramming language.

SVG is a language for describing two-dimensional

graphics and graphical applications in eXtensible

Markup Language (XML). SVG 1.1 (W3C, 2003b)

is a W3C Recommendation and forms the core of the

current SVG developments. SVG 1.2 is the speciﬁca-

tion currently being developed as is available in draft

form. The SVG Mobile Proﬁles: SVG Basic (SVGB)

and SVG Tiny (SVGT) (W3C, 2003a) are targeted

to resource-limited devices and are part of the 3GPP

platform for third generation mobile phones.

To build a Universal Multimedia Access (UMA)

system delivering SVG content in a heterogeneous

network, we have to be able to adapt SVG full to SVG

mobile proﬁles or other compliant format accepted by

the handheld terminals; this is possible using adapta-

tion techniques. There are however some drawbacks

and one of the major drawbacks at the moment is the

security problem. In the this area, while controlling

access to text-based documents has been the focus

of many research activities (Ferraiolo et al., 2003),

raster graphic information has been seldom consid-

ered, mainly because of its monolithic internal struc-

ture. However, XML-based vector images present

new and challenging feature protection problems, re-

lated to ﬁne-grained access control to their internal

structure. We have then deﬁned a novel approach to

ﬁne-grained feature protection of SVG data. The pro-

posed model allows to selectively transform SVG data

according to the user’s proﬁle, thus releasing only the

features that the user is entitled to see.

In this paper we present the design and implemen-

tation of an SVG based UMA system. In particular,

the platform is able to allow proxy adaptation opera-

tion or third party distribution of the exchanged SVG

documents.

330

Reda Kaced A. and Moissinac J. (2007).

SVG BASED SECURE UNIVERSAL MULTIMEDIA ACCESS.

In Proceedings of the Second International Conference on Signal Processing and Multimedia Applications, pages 326-331

DOI: 10.5220/0002141103260331

 SciTePress

1.1 Overview of SVG

SVG supports scripting and animation, so is ideal for

interactive, data-driven, personalized graphics. An-

imation techniques can range from a simple linear

movement to 3D double helix morphing effects. Web

developers, once they are more aware of the possibil-

ities, can ﬁnd unprecedented levels of control. The

broad support behind SVG comes from its many ad-

vantages as the following short feature list demon-

strates:

• SVG ﬁles are pure XML and can be read and mod-

iﬁed by a large range of tools (e.g. notepad)

• SVG ﬁles are smaller and more compressible than

JPEG or GIF images

• SVG images are scalable

• SVG images are zoomable without degradation

• Text in SVG is selectable and searchable

In SVG two-dimensional graphics are described

using XML. Therefore a set of basic shape elements

is deﬁned e.g. lines, rectangles, circles, ellipses, poly-

gons, paths, etc. The position, size, color, etc. of each

shape is deﬁned by attributes. Additionally there are

special tags to group shapes <g> and reference self-

deﬁned symbols <use>. SVG provides a possibility

to add alternative descriptions of the content. There-

fore each element can include a title and a description

tag (<title> and <desc>). The SVG source code

example in Figure 1 describes a server-proxy-client

architecture. Each component (that we call object in

the rest of this paper) in the path is represented in a

group (<g>) that contains the component itself, the

text, and the link paths associated. Our solution can

also use ﬁner granularity where objects are straight-

forwardly basics shape elements.

Figure 1: Example of SVG document.

1.2 Universal Multimedia Access

Universal Multimedia Access (UMA) means the ac-

cess to multimedia information by any terminal

through any network (Mohan et al., 1999). The ob-

jective of UMA systems is to make available different

presentations of the same information, suiting differ-

ent terminals, networks and user preferences (Butler

et al., 2002), (Kirda et al., 2001). This can be achieved

through customizing the content to the environment

where it shall be consumed.

This content customization (adaptation) can be

implemented in three different places: 1) at the con-

tent server, 2) at a proxy side, and 3) at the user ter-

minal.

The proxy-based adaptation is the most suitable

for the context change support (Wijnants et al., 2005),

(Endler et al., 2005), (Palit et al., 2006). In this case,

the proxy is the entity responsible of retrieving client

contexts and looking for the eventual changes that

may occur. These changes are therefore transparent

for the content server. The proxy can transform ex-

isting SVG content and thus the content server is not

directly involved in the adaptation.

Diagram in Figure 2 illustrates the UMA system

and its major elements. The developed system im-

plements the customization engine at the proxy side.

Content server needs to consider adaptation a poste-

riori by taking into account user’s proﬁle and device

capability.

Server this dimension applies to the users that is the

content source, and do not especially consume

content as end-users but perform any action re-

garding the content, in principle easing the con-

sumption task for an end-user. A server can be

a user that creates content, authenticates content,

distributes content and so on.

Proxy this dimension is introduced as an intermedi-

ary between the content provider and the client

device. It acts as transcoding proxy and/or cache

proxy to deliver the right (adapted) version of an

original content to the client.

Client this dimension applies to the user that receives

content for consumption and wishes to have the

content tailored to its user environment, the so-

called end-users.

As shown in Figure 2, our UMA mechanism con-

siders three types of actors and two kinds of access

channels between these actors. The three actors are

devices, servers, and proxies; the two access channels

are a network access between device and proxy, and

a content access between proxy and server. From the

SVG BASED SECURE UNIVERSAL MULTIMEDIA ACCESS

331

Figure 2: UMA infrastructure in SUMA.

users’ point of view, an adaptive and seamless con-

nection of the two access channels results in a univer-

sal access channel to the content. From the content

point of view, a dynamic delivery of adaptive content

results in a universal content presentation to the users.

1.3 Motivation

Let us consider for example a tele-health multimedia

application using Internet to distribute SVG record-

ings (images, videos, and sounds) of doctors and con-

ﬁdential data. Multimedia objects accessed by the ap-

plication users are content unaware and annotations

emptied. Users are equipped with different devices

(PC, Laptop, PDA, Smartphone, etc.). Exchanged

contents have to be adapted to each device proﬁle

using adaptation proxies (they can also add content

like advertisements...). For security and conﬁden-

tiality reasons, exchanged contents must be authen-

ticated. Moreover some parts of the SVG presenta-

tion appearing in these recordings is highly classiﬁed

and should not be changed by unauthorized users (in-

cluding proxies). So the problem here is how can we

allow adaptation operations by intermediaries keep-

ing the authentication and conﬁdentiality of the ex-

changed contents?

Solving such issue using current security tech-

niques and access control models is very hard as the

content is not already deﬁned. In reality, the autho-

rization manager should be able to specify authoriza-

tion rules and policies by stating the features of such

intermediaries.

2 SYSTEM ARCHITECTURE

In this section we describe SUMA (Secure Universal

Multimedia Access) our proxy-based content adapta-

tion framework. The main architecture type of SUMA

system is shown in Figure 3. In this approach, the user

takes the UMA Engine as its proxy, which will then

make the request to the content server on behalf of

this user. The content server response is then received

by the UMA Engine, which afterwards decides and

performs the adaptation (if needed), and then sends

the transformed content back to the client.

The idea is based on the concept of create content

once and used with various presentation. Instead of

preparing different formats for the different devices,

the server use standard SVG ﬁles, XML and XSL to

describe Web content and presentation, respectively.

In the proxy side, we ﬁnd an adaptation engine, im-

plemented with XSLT, and a security engine which

is used to ensure security and authentication of the

exchanged content. To ensure signature validity, the

client has an authenticity veriﬁer which tests the cor-

rectness of the received signature; it is also equipped

with an SVG reader or in the worst case a “adapted

format” reader. In the following we describe each one

of these components.

2.1 Content Authentication Engine

There are three phases available in order to build a se-

cure SVG content to send from the server to a client.

Once a client logs in the system and requests the mul-

timedia data, the server by the Access Control Policy

Generator (ACPG) veriﬁes user’s identiﬁcation and

the related permission, and generates an adaptive ver-

sion which speciﬁes what parts of the SVG can be

adapted by the intermediary proxy. After that, the ser-

vice provider signs this content using technique de-

scribed in 2.1.2 and sends it to the client passing by

the proxy.

2.1.1 ACPG

In SUMA, access control services simultaneously

support a Role-Based Access Control (RBAC) model

(Ferraiolo et al., 1995) and an access control model

based on Access Control Lists (ACL) (Barkley, 1997).

The RBAC model incorporates a user-role assign-

ment relation, a role hierarchy and a permission-role

assignment relation. Three user-roles are deﬁned

owner, adapter and consumer. The ACL model pro-

vides a straightforward way of granting or denying

access to a given resources for speciﬁed users.

The implementation for the prototype of ACPG

has been done using the JAVA Standard Edition.

Therefore, the XACML 2.0 (Anderson, 2004) is cho-

sen to be implemented in the authorization service

using sunxacml1.2 (SUN, 2003). The RBAC poli-

cies and ACL policies have been built using sun’s

XACML APIs. The generated RBAC policies and

ACL policies are shown in Figure 4. An RBAC

XACML PolicySet contain role <PolicySet> (RPS)

and permission <PolicySet> (PPS). The RPS asso-

ciates holders of a given role attribute and value with

a PPS that contains the actual permissions associated

SIGMAP 2007 - International Conference on Signal Processing and Multimedia Applications

332

Figure 3: SUMA architecture.

with the given role. This means each RPS references

a single corresponding PPS.

The SVG document is parsed by the ACPG, and

generate a secure version where every adaptive object

is tagged with the access policy associated with. This

version is transmitted to the Security Engine (SE) to

be signed.

2.1.2 Signature Scheme

To ensure the authenticity of the exchanged SVG doc-

ument, we use AMCA, a Merkle Hash Tree (MHT)

based signature technique which we have described

in (Kaced and Moissinac, 2006) adapting it to SVG

content. The method we propose allows a subject to

prove source authenticity as well as the authenticity

of the content of a document.

To accomplish this goal, the idea is to associate a

hash value with each node in the graph representation

of an XML document. More precisely, the hash value

associated with an attribute is obtained by applying a

hash function over the concatenation of the attribute

value and the attribute name. By contrast, the hash

value associated with an element is the result of the

same hash function computed over the concatenation

of the element content, the element tag name, and the

hash values associated with its children nodes, both

attributes and elements. Hash values associated with

the nodes of an SVG document are computed by the

Merkle hash function.

Concretely, the SE starts by adding new empty ob-

jects (called freeleaves) to the SVG, each one is asso-

ciated to an adaptive object. After, it uses AMCA to

sign the SVG resulted and sends it to the proxy. This

last proceeds to the adaptation operations putting the

adapted objects in the freeleaves signing them itself.

The end-user when receives the adapted version, can

recompute the signature of the sender using AMCA

techniques. Please refer to (Kaced and Moissinac,

2006) for more details of these techniques.

2.2 Content Adaptation Engine

An adaptation method D is applied on a content M ,

if the proﬁle description of the receiver R matches

the proﬁles input requirement of M , and the proﬁle

output description of M matches the client require-

ments. To avoid developing static adaptation meth-

ods for each kind of resources adaptation, its prefer-

able that the server has methods that provides many

outputs depending to the context of their application.

This is called dynamic adaptation. A dynamic adap-

tation is the one that interacts with the current client

context. Taking into account the client context by an

adaptation method can be achieved at three locations:

the client, the server, or the proxy.

In the general case, the client content adaptation

is achieved using scripts and rendering styles which

are sent inside the content and evaluated during the

rendering according to the capability of the user de-

vice. Unfortunately, this kind of adaptation has many

disadvantages and depends widely on the client pro-

cessing power.

Dynamic adaptation on the proxy represents a best

alternative to deal with the variety of client contexts.

It allows the delivery of adapted content directly. In

our system, we consider two kinds of adaptations: the

structural adaptation (transformation) and the media

adaptation. Structural transformations are based on

XSLT (Clark et al., 1999) which allows transforming

SVG document into other XML-based documents.

Media adaptations use speciﬁc plug-ins applications.

We developed then, an SVG template for SUMA

server, and several transcoding plug-ins for the adap-

tation proxy:

XML to SVG transcoder. We developed an SVG

template which is composed of three nested sets: ob-

ject templates. background image. and foreground

objects. Object templates are used to deﬁne speciﬁc

types of objects and their shapes. Foreground ob-

jects use a group of global parameters of translation

SVG BASED SECURE UNIVERSAL MULTIMEDIA ACCESS

333

Figure 4: ACPG role-based access control, (a) a user-role

assignment conﬁguration ﬁle, (b) Role <PolicySet>. for

the adapter role, (c) Permission <PolicySet> for the adapter

role.

and scaling, that can he changed while the SVGs are

cropped or scaled. Individual object has its transla-

tions and scaling parameters relative to a group origin

point.

SVG to SVG transcoder. Based on the client de-

vice proﬁle, SUMA transcodes an SVG based on the

viewbox, global shift of the foreground objects and

cropping/scaling of the background image in SVG.

Also, we embed the manipulated background image

into the SVG to reduce the work and transmission

load for proxy load.

SVG to Mobile SVG transcoder. Mobile SVG

includes two proﬁles - SVG Basic and SVG Tiny. We

develop this transcoder for converting general SVGs

into the format that is compliant to the SVG proﬁles

used by the Pocket SVG Viewer. Some transcoding is

needed such as the font style of text and strict require-

ments of DTDs in the SVG headers.

SVG to HTML transcoder. This transcoder is

based on an XSLT stylesheet. For a non-SVG render-

ing devices, general browsers that reads HTMLs can

be used to displayed a transcoded SVG ﬁles that have

been converted to HTML and an image. The hyper-

link properties of the original SVGs are utilized by the

image maps inside the HTML.

SVG to Image transcoder. Batik (Apache, 2004)

is a Java-based toolkit for applications or applets that

want to use images in the SVG format for various pur-

poses. Batik provides building blocks that developers

can assemble in various ways in their Java technol-

ogy applications or applets to generate, parse, view or

convert SVG contents. Batik can convert SVG con-

tent into other formats such as JPEG, PNG or Tiff or

other formats (transcoder API). We use Batik to gen-

erate a plug-in which converts the original SVG im-

ages to either JPEG or GIF.

3 EXPERIMENTS

SUMA performs two preliminary operations: con-

tent signature and content adaptation. We deployed

SUMA over a wireless LAN 802.11g environment.

Any browser of devices can set the URL of the adapter

server as its HTTP proxy. We have tested the sys-

tem on desktops, laptops, HP Pocket PC (Win-mobile

device), and Sony Mobile Phone Browser. Figure 5

shows an example of the SUMA content triggered by

different transcoding proﬁle. It shows the transcoded

content using the Pocket IE browser on a PocketPC

device. It also shows the intelligence information as-

sociated with the objects on the SVG. we can see

also the transcoded result on the Sony w900i Mo-

bile phones. All ﬁgures and property sheets are

transcoded to HTML and JPEG images.

An advantage of the proposed system was ob-

served. After our presentation at the debut demo, an

audience challenged whether this system can let his

Sony Clie, a Palm-OS device which was not originally

supported, access the SUMA content. Within 10 min-

utes, we added this device to the repository proﬁles,

helped the device accessing the wireless LAN, and

set the HTTP proxy. Afterwards, it could access the

customized SUMA content right away. This example

SIGMAP 2007 - International Conference on Signal Processing and Multimedia Applications

334

Figure 5: Example of the SUMA content triggered by dif-

ferent transcoding proﬁle.

showed the ﬂexibility and usability of the proposed

platform.

4 CONCLUSION

In this paper, we have highlighted the challenges for

SVG delivery access control and presented SUMA

framework, a Universal Multimedia Access system

which ensure end-to-end data authentication.

SUMA provide a proxy-based adaptation for SVG

content. It is based on two major parts, a Content Au-

thentication Engine which uses Merkle Hash Tree sig-

nature technique to allow adaptation operations keep-

ing possibility of signature veriﬁcation in the client

side, and a Content Adaptation Engine, which imple-

ment some plug-ins to convert standard SVG content

to transcoded format for different proﬁle devices.

The major advantages of SUMA is the ease of

maintenance and the end-to-end security of the ex-

changed content. However SUMA does not imple-

ment content Encryption which is a big drawback for

the moment. This is our next challenge.

REFERENCES

Anderson, A. (2004). XACML Proﬁle for Role Based

Access Control (RBAC). OASIS Access Control TC

Committee Draft, 1:13.

Apache (2004). Batik, a java-based toolkit for scalable vec-

tor graphics (svg). http://xml.apache.org/batik/.

Barkley, J. (1997). Comparing simple role based access

control models and access control lists. Proceedings

of the second ACM workshop on Role-based access

control, pages 127–132.

Butler, M., Giannetti, F., Gimson, R., and Wiley, T. (2002).

Device independence and the Web. Internet Comput-

ing, IEEE, 6(5):81–86.

Clark, J. et al. (1999). XSL Transformations (XSLT) Ver-

sion 1.0. W3C Recommendation, 16(11).

Endler, M., Rubinsztejn, H., da Rocha, R., and do Sacra-

mento Rodrigues, V. (2005). Proxy-based Adaptation

for Mobile Computing. PUC.

Ferraiolo, D., Cugini, J., and Kuhn, D. (1995). Role-Based

Access Control (RBAC): Features and Motivations.

Proceedings of 11th Annual Computer Security Ap-

plication Conference, pages 11–15.

Ferraiolo, D., Kuhn, D., and Chandramouli, R. (2003).

Role-based access controls. Artech House Boston.

Kaced, A. R. and Moissinac, J.-C. (2006). Multimedia

content authentication for proxy-side adaptation. In

Proceedings of the IEEE International Conference on

Digital Telecommunications, ICDT 06.

Kirda, E., Kerer, C., Jazayeri, M., and Kruegel, C. (2001).

Supporting multi-device enabled Web services: chal-

lenges and openproblems. Enabling Technologies:

Infrastructure for Collaborative Enterprises, 2001.

WET ICE 2001. Proceedings. Tenth IEEE Interna-

tional Workshops on, pages 49–54.

Mohan, R., Smith, J., and Li, C. (1999). Adapting Multi-

media Internet Content for Universal Access. IEEE

TRANSACTIONS ON MULTIMEDIA, 1(1).

Palit, H. N., Chi, C.-H., and Liu, L. (2006). Proxy-

based pervasive multimedia content delivery. comp-

sac, 1:255–264.

SUN, M. (2003). Sun’s XACML implementation. For

information about this implementation see web site

http://sunxacml.sourceforge.net.

W3C (1998). Scalable vector graphics (svg).

http://www.w3.org/Graphics/SVG/.

W3C (2003a). Mobile svg proﬁles: Svg tiny and svg ba-

sic. http://www.w3.org/TR/2003/REC-SVGMobile-

20030114/.

W3C (2003b). Scalable vector graphics (svg) 1.1 spec-

iﬁcation, w3c recommendation 14 january 2003.

http://www.w3.org/TR/SVG11/.

Wijnants, M., Monsieurs, P., Quax, P., and Lamotte, W.

(2005). Exploiting proxy-based transcoding to in-

crease the user quality of experience in networked ap-

plications. aaa-idea, 0:73–80.

SVG BASED SECURE UNIVERSAL MULTIMEDIA ACCESS

335