Copyright © 2000 HITACHI Ltd.
Visual marks play critical roles in the physical world, but their use in the cyber world is limited because there they are easy to forge, tamper with and copy onto unauthorized contents. This note therefore describes a new type of visual marks that are secure because digital signatures are embedded in them and that can be used with a wide variety of cyber-world systems. It also shows their application to seal systems for WWW site authentication.
This document is a submission to the World Wide Web Consortium from Hitachi Ltd. (see Submission Request, W3C Staff Comment). For a full list of all acknowledged Submissions, please see Acknowledged Submissions to W3C.
This document is a NOTE made available by W3C for discussion only. This indicates no endorsement of its content, nor that W3C has had any editorial control in its preparation, nor that W3C has, is, or will be allocating any resources to the issues addressed by the NOTE.
A list of current W3C technical documents can be found at the Technical Reports page.
The number of Internet users is continuously increasing, and network-based social activities such as EC (electronic commerce) are becoming popular in many fields. On the other hand, however, social problems related to the networks are also increasing. It is therefore necessary to clarify the requirements that must be met by networks that serve as social infrastructures and to establish techniques for meeting these requirements. One of the most important requirements is that the networks be trustworthy, and various authentication techniques for meeting this requirement by using digital signatures have been developed. They are, however, not sufficiently clear. That is, they are not easily understood by nonprofessional people. In SSL (Secure Sockets Layer) and S-HTTP (Secure Hyper Text Transfer Protocol), for example, users can not easily understand what authentication means. Networks that are trusted therefore are networks providing not only security but also clarity.
Visual marks, such as traffic signals, play critical roles in the physical world. They are familiar to nonprofessional people and can convey information clearly. Thus one promising approach to meeting the clarity requirement is to incorporate visual marks into the cyber world. Such marks are already used in the following ways:
Thus visual marks are widely used in cyber world. These marks have serious problems, however, due to the properties of cyber worlds. Specifically, they are easy to forge, tamper with, and copy onto other Web sites. In summary, conventional visual marks are clear but are insufficiently secure.
A Visual Mark for the Cyber World(VM) is a new type of visual mark suitable for use in the cyber world. It is intended to make networks trustworthy by making visual marks secure.
VMs are made of drawings and are image data, such as bitmap graphics or JPEG files. They are placed on data that represent objects in the cyber world, and they carry information about that data.
FIGUREs are drawings from which VMs are made, and DATA is data on which VMs are pasted.
Figure 1 shows examples of FIGUREs. And as shown in Fig. 2, a VM is a simply FIGURE into which a digital signature has been embedded by digital watermarking . This signature is a signature for both the FIGURE and the DATA on which the VM is pasted. Additional application-specific information may also be embedded into the FIGURE.
Fig. 1. Examples of FIGUREs.
Fig. 2. VMs issuing system.
White and gray area respectively represent data and processes.
A VM is issued (Fig. 2) by having the issuer sign for the DATA and the FIGURE and then embedding the signature in the FIGURE. And as shown in Fig. 3, a VM is verified by first cutting it out of the DATA and then extracting and verifying the signature. If the verification is successful the system guarantees the following:
Fig. 3. VMs verifying system.
VMs are easily understood because the watermarking does not degrade the clarity of the FIGUREs from which they are made.
It is obvious from Figs. 2 and 3 that the security of VMs is equivalent to that of their signatures.
Equivalent clarity and security could be provided by using a simple combination of visual marks and digital signatures, but VMs are more portable than the simple combination because the clarity measures (visual figures) and the security measures (digital signatures) are consolidated in a single object.
This chapter illustrates a VM application to a WWW authentication scenario in which there are four players:
WWW site owner (Owner).
WWW site user who accesses the Owner's WWW site (User).
Person who issues a guarantee mark for the Owner's WWW site (Guarantor).
Certificate Authority who authenticates the Owner's public key (CA).
When the owner asks the guarantor to issue a guarantee mark for the owner's WWW site, the guarantor issues it and sends it to the owner, who places it on the pages of the WWW site. The guarantee mark may be a rating of the site, a certification of its suitability for use in schools, or any other information relevant to the site. A user accessing the site can get information about it simply by looking at the guarantee mark and can verify the guarantee mark when necessary (e.g., when sending a credit card number to the site).
We use VMs for guarantee marks, and Fig. 4 shows that the system for issuing these guarantee VMs is simply the basic VM issuing system extended to include a signature for the IP address (URL) of the WWW site to be guaranteed. This signature is needed to prevent WWW site disguise: the copying of both the WWW page data and its guarantee mark to the WWW site of an attacker who pretends to be the legal owner of its content.
Fig. 4. System issuing an VMs for WWW authentication.
This protocol description uses the following terminology.
SKx: secret key of player x.
PKx: public key of player x.
Enc(DATA, K): result of encrypting DATA with key K.
IP-ADDRESS: IP address (URL) of the WWW site to be marked.
W-DATA: DATA defining pages of the WWW site. This may be XML source code.
X | Y: concatenation of data X and data Y.
The protocols for issuing and authenticating VMs are illustrated in Fig. 5, and there are four steps in the protocol for issuing them.
Fig. 5. Protocols for VMs application.
The protocol for authenticating VMs has three steps.
These application protocols have been implemented in C language, and Fig. 6 shows VMs made from the FIGUREs in Fig. 1. A roughly 2-K Bytes digital signature is embedded in each VM.
Fig. 6. VMs generated from the FIGUREs in Fig. 1.
As can be seen by comparing Figures 1 and 6, the clarity of the FIGUREs is not degraded by watermarking.
Three types of attack are possible.
As mentioned in Section2, the security against the first and second types of attacks is equivalent to the security of the embedded signature. The security against third type is also the same as that of the signature because VMs contain the signature for the IP address of the correct WWW site.
This issue will be discussed in next chapter by comparing VMs with alternative approaches.
This chapter compares VMs with the following three alternative approaches to WWW authentication.
It is clear from the comparison results summarize in Table 1 that simple marks are clear and portable but of course not secure. And because a digital signature expresses nothing by itself, it tells a WWW user nothing about the WWW page unless the user goes through the verification process. Simple signatures, although secure and reasonably portable, thus do not meet a user's needs because their meaning is not clear.
|1||Simple Marks||No Good||Good||Good|
|2||Simple Signature||Good||No Good||Not so Bad|
|3||Simple combination of marks and Signature||Good||Good||No Good|
|4||VMs||Good||Good||Not so Bad|
A WWW owner using a combination of marks and signatures needs to attach a mark and the corresponding signature to the WWW page in such a way that there is a link between them (e.g., clicking the mark causes the signature to be verified). This attachment needs to be standardized so that WWW users can use a common program to verify signatures. Such standardization requires extensive and continuous effort because the languages for describing WWW pages are continuously evolving along with the WWW managing systems. VMs also require standardization, but this standardization should be much easier because all that needs to be standardized is the way of attaching VMs. That is, there is no need for standardization of the way of linking marks and signatures.
This paper described VMs; trusted visual marks for the authentication of Web sites. The paper has shown that VMs are as easily understood as visual marks, as secure as digital signatures, and are more portable than a simple combination of visual marks and digital signatures. VMs can be used with other Web related authentication systems such as Signed-XML and can then improve their clarity.