The following paper was originally published in the
`Proceedings of the First USENIX Workshop on Electronic Commerce
`New York, New York, July 1995.
`
`DigiBox: A Self-Protecting Container
`for Information Commerce
`
`Olin Sibert, David Bernstein, and David Van Wie
`Electronic Publishing Resources, Inc.
`Sunnyvale, California
`
`For more information about USENIX Association contact:
`1. Phone:
`510 528-8649
`2. FAX:
`510 548-5738
`3. Email:
`office@usenix.org
`4. WWW URL: http://www.usenix.org
`
`001
`
`

`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`The DigiBox:
`A Self-Protecting Container for Information Commerce
`
`Olin Sibert
`David Bernstein
`David Van Wie
`
`Electronic Publishing Resources, Inc.
`460 Oakmead Parkway
`Sunnyvale, California
`1 408 774 6100
`info@epr.com
`
`Abstract
`
`Information Commerce is a business activity carried out among several parties in which information car-
`ries value and is treated as a product. The information may be content, it may be returned usage and mar-
`keting data, and it may be representative of financial transactions.
`
`In each of these cases the information is valuable and must be kept secure and private. Traditional
`approaches secure the transmission of that information from one point to another; there are no persistent
`protections. Protection of all of these components of information commerce for all parties in a transaction
`value chain is necessary for a robust electronic infrastructure.
`
`A prerequisite to such an environment is a cryptographically protected container for packaging
`information and controls that enforce information rights. This paper describes such a container, called the
`™
`. EPR has submitted initial specifications for the DigiBox container to the ANSI IISP Electronic
`DigiBox
`Publishing Task Force (EPUB) within the User/Content Provider Standards Working Group (WG4).
`
`1 Introduction
`
`As services and products in modern commerce
`increasingly take electronic form, traditional com-
`merce is evolving into electronic commerce. This
`includes both creation and enforcement of various
`agreements between parties in an electronic com-
`mercial relationship. It also includes enforcing the
`rights of these parties with respect to the secure
`management of electronic content or services
`usage, billing, payment, and related activities.
`
`To save money, to be competitive, and to be effi-
`cient [1,2], members of modern society will shortly
`be using new information technology tools that
`
`truly support electronic commerce. These tools
`provide for the flow of products and services
`through creators’, providers’, and users’ hands.
`They enable the creation, negotiation, and enforce-
`ment of electronic agreements, including the evo-
`lution of controls that manage both the use and
`consequences of use of electronic content or ser-
`vices. In addition, these tools support “evolving”
`agreements that progressively reflect the require-
`ments of further participants in a commercial
`model.
`
`Participants in electronic commerce [3,4] will need
`rules and mechanisms such that:
`
`This document was created with FrameMaker 4.0.4
`
`002
`
`

`
`requires a substantial manufacturing investment.
`Figure 1 illustrates a simplified traditional informa-
`tion economy: physical goods flow from a pub-
`lisher (manufacturer) to a customer, in response to
`orders and followed by payments. The author’s
`relationship with the publisher may be more light-
`weight, but the author is nonetheless dependent on
`the publisher to report sales and make royalty pay-
`ments in accordance with the author’s contract. In
`addition, a financial institution provides payment
`processing and clearing services for all parties.
`
`Publisher
`
`Financial
`Institution
`
`Payment
`Product
`O rd er
`
`Usage Information
`Payment,
`Content
`
`Author
`
`Customer
`
`Figure 1. Traditional information economy.
`
`Because of the flexibility afforded by electronic
`mechanisms, information commerce is evolving
`from indirect, advertiser-supported, mass-audi-
`ence media to a new, niche-audience-oriented busi-
`ness model. In this system, members of the
`electronic community, with or without the eco-
`nomic support of advertising, pay providers
`directly for what they want to receive. Business-to-
`business purchasing is steadily evolving into a
`direct electronic ordering model.
`
`Figure 2 illustrates the flexibility possible in new
`electronic
`information
`commerce models.
`Although there is still a role for publishers, this
`role no longer involves physical goods. Rather, the
`publisher is responsible for packaging and aggre-
`gating information goods and control information,
`
`1.
`
`Information providers can be assured that their
`content is used only in authorized ways;
`
`2. Privacy rights of users of content are pre-
`served; and
`
`3. Diverse business models related to content can
`be electronically implemented.
`
`The Internet and other information commerce
`infrastructures will require a management compo-
`nent that enforces such rules, ensuring a safe,
`coherent, fair, and productive community. This
`management component will be critical to the elec-
`tronic highway’s acceptance. Without rules to pro-
`tect the rights of content providers and other
`electronic community members, the electronic
`highway will comprise nothing more than a collec-
`tion of limited, disconnected applications.
`
`Analysts have concluded that content will consti-
`tute the largest revenue-generating component of
`the information superhighway [5]. It is also clear
`that unfettered access to content requires that con-
`tent providers be able to maintain control over lit-
`erary or copyrighted assets. Many analysts
`conclude that this will be one of the key bottle-
`necks in the implementation and deployment of
`New Media.
`
`2 Information Commerce and Digital
`Value Chains
`
`Information commerce is often considered a
`wholly new concept, made possible only through
`the use of networks and computers. In fact, a
`robust information economy has existed for centu-
`ries, involving trafficking in physical
`representa-
` of information such as books, newspapers,
`tions
`and so on. Because such commerce involves physi-
`cal goods, there is a non-negligible floor to the cost
`of handling information goods. The new aspects of
`the electronic information economy are that the
`information itself is the entire product and that the
`product can be distributed at negligible marginal
`cost.
`
`The traditional information economy in physical
`goods is publisher-centric, because creation of
`information goods—particularly low-cost goods—
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`003
`
`

`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`then making them available to customers. Similar
`to a manufacturing/distribution/retail chain for
`physical goods, the electronic model permits infor-
`mation retailers, and even end customers, to re-
`package and redistribute different aggregations of
`information while ensuring that the appropriate
`control rules are maintained. A clearinghouse
`ensures that usage information and payments are
`provided directly to authors and publishers; the
`payments themselves are made through traditional
`financial institutions. Because control rules are
`associated with information, a variety of payment
`and other business models can be associated with
` versus
`the same content (e.g.,
`purchase
` pay-per-
`).
`use
`
`Author
`Creates
`
`Content
`
`Financial
`Institution
`
`Aggregated Payment,
`Usage Information
`
`Publisher
`Distributes
`
`Content
`
`Payment
`Authorizations
`
`Clearinghouse
`
`Customer
`Pays per Use
`
`Extracted
`
`Content
`
`Customer
`Purchases
`
`Payments,
`Usage Reports
`
`Content
`
`Retailer
`Redistributes
`
`Content
`
`Customer
`Purchases
`
`Figure 2. Electronic information economy.
`
`The conversion from traditional commercial distri-
`bution channels requires key foundation technolo-
`gies and results in a fundamental shift in existing
`infrastructures. This channel transformation will
`create a new electronic digital distribution industry.
`Digital distribution employing the DigiBox con-
`tainer architecture and its associated support envi-
`™
`, can play a critical role in
`ronment, InterTrust
`this transformation of the communication, media,
`and information technology markets.
`
`2.1 Protecting All the Information in
`Information Commerce
`
`The very properties that make “the net” attractive
`as a distribution medium—ease of manipulating
`information in electronic form—also appear to
`make these protections intractable. Addressing this
`dichotomy requires a paradigm shift in computer
`architecture to introduce the concept of a “secure
`processing” environment in which protected infor-
`mation can be manipulated without being subject
`to external tampering or disclosure. A prerequisite
`to such an environment is a cryptographically pro-
`tected “container” for seamlessly packaging infor-
`mation and controls that enforce information use
`rights.
`
`The DigiBox described by this paper is such a con-
`tainer.
`
`The need for various information commerce com-
`puters and appliances to interoperate requires that
`this container format and its access methods be
`standardized. EPR has submitted initial specifica-
`tions for the DigiBox container to the American
`National Standards Institute (ANSI) Information
`Infrastructure Standards Panel (IISP) through the
`Electronic Publishing Task Force (EPUB) in the
`User/Content Provider Standards Working Group
`(WG4).
`
`The primary goal of information protection is to
`permit proprietors of digital information (i.e., the
`artists, writers, distributors, packagers, market
`researchers, etc.) to have the same type and degree
`of control present in the “paper world.” Because
`digital information is intangible and easily dupli-
`cated, those rights are difficult to enforce with con-
`ventional
`information processing
`technology.
`Many types of rights (compensation, distribution,
`modification, etc.) are associated with the various
`elements of information commerce, and these
`information property rights take many forms. At a
`high level, there is the legal definition of “copy-
`right,” codified in U.S. law [6–9] and the Berne
`Convention. This gives copyright holders a legal
`right to control how copyrighted information is
`handled. In addition, various high-level rights are
`conferred by contractual arrangements between
`primary rightsholders and other parties.
`
`004
`
`

`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`For example, the protections needed for content
`elements incorporate the licensing provisions for
`the intellectual property rights of the content right-
`sholders. In a broader sense, these rights include
`control over several activities: the right to be com-
`pensated for use of the property; the right to con-
`trol how content is distributed; the right to prevent
`modification of content by a distributor; “fair use”
`rights; the rights to the usage data, privacy rights of
`individuals, and so on.
`
`(e.g., “pay $5.00 for the collection, then pay the
`creator,” “report use of each item”). Users receive
`content and operate on it, generating billing reports
`and usage reports that are delivered to a clearing-
`house and paid or summarized back for the origi-
`nating parties. This structure is very rich and is
`capable of supporting many business models.
`There are multiple flows of information in many
`different directions amongst the parties involved in
`the transactions.
`
`In the realm of physical goods, these rights are
`enforced by a combination of legal and technical
`means. However, the technical means can be (and
`are) unsophisticated because the technology for
`violating rights is relatively expensive and time-
`consuming—in comparison to equivalent activities
`with respect to digital information. Photocopying a
`book or copying a video cassette is inherently more
`labor intensive and costly than copying a file. So,
`while defeating technical means of enforcement is
`(relatively) expensive, it can be done—and often
`the legal means to deter this are inadequate.
`
`2.2
`
`Information Commerce—Not Just
`
`
`
`Payment
`
`Rights protection is also a fundamental aspect of
`commerce. Commerce is not just a way for two
`parties to pay each other for something. Rather, it
`is an extraordinarily rich web of relationships
`among parties that concerns payment, negotiation,
`control, advertising, reporting, auditing, and a vari-
`ety of other activities. These activities are impor-
`tant aspects of the transaction relationships. Often
`the information carried in these reports, audits, and
`the like is highly valuable and highly confidential,
`perhaps even more valuable than the content that is
`the subject of the information commerce at hand.
`These activities too are performed and controlled
`in the “paper world” by legal and technical means,
`but there are no widely used models for their elec-
`tronic equivalents.
`
`Figure 3 shows some of the operations that could
`occur in true electronic commerce, using the Inter-
`net World-Wide Web [10] mechanisms as an exam-
`ple. Creators originate content and apply rules
`(e.g., “pay author $1.00/use”) for its use. Distribu-
`tors repackage content, applying additional rules
`
`Another example is that of an advertiser (acting as
`distributor, or with a distributor). The advertiser
`might have a rule that offers a discount, or no
`charge at all, but only if the user views the adver-
`tisement and agrees to have that fact reported to the
`advertiser.
`
`It is relatively simple to devise schemes for parties
`to pay each other electronically (for example, Digi-
`Cash [11], NetBill [12], Open Market [13], SNPP
`[14], NetCheque [15], First Virtual [16], etc.). Pay-
`ment, however, constitutes only one—and perhaps
`the simplest one—of the means in which parties in
`commerce interact. All the other information com-
`merce components must be accomplished with the
`same needs for security, privacy, and integrity. In
`fact, these aspects of electronic commerce, includ-
`ing rights protection, are strongly intertwined in
`the digital economy, because much digital com-
`merce concerns information and innovative busi-
`ness models for information commerce.
`
`3 Existing Approaches to Information
`Commerce
`
`Information proprietors employ a variety of tech-
`nological protection approaches
`today. These
`approaches are generally “point solutions,” in that
`they protect a specific type of property in a specific
`context and enforce only specifically defined
`rights—typically only the right to compensation
`for use. Because the technologies are limited, the
`market is fragmented, and there are no general pro-
`tection solutions.
`
`005
`
`

`
`
`
`
`
`Registrar/
`Repository Managers
`
`Content
`Servers
`
`Transaction
`Server
`
`Private Transaction
`Networks
`
`Repository
`Administration
`
`WWW
`Server
`
`Clearinghouses
`
`Clearinghouse
`Interface
`
`Internet
`
`Content
`
`Business
`Rules
`
`DigiBox
`Packaging
`Application
`
`Authors
`
`Figure 3. Multi-party Internet information commerce.
`
`DigiBox-
`Aware
`Browser
`
`Users
`
`3.1 No Protection
`
`Much digital property is distributed without any
`technological enforcement for property rights, on
`the assumption that legal means suffice. This
`approach works well enough for many low-value
`properties, but it has the disadvantage of raising the
`price to legitimate users who must pay for both
`
`their own and illegitimate use. In many cases, how-
`ever, this cost is negligible, and no protection is an
`economically sound choice. Even for content that
`is free, however, a creator may wish to impose
`some rules for reporting or some access control. Of
`course, privacy rights of users will be a concern to
`many.
`
`006
`
`

`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`3.2 License Managers
`
`For some valuable software properties, license
`managers are used. Because a software property is
`dynamic (executable), it is feasible to restrict it so
`that it functions properly only through interaction
`with a license manager process. In general, there is
`no protection of usage data in these schemes. In
`some cases this technique has been applied to con-
`tent protection, but only with limited success [17,
`18].
`
`3.3 Cryptographic Unlock
`
`Some static properties (fonts, for example; also
`some installable software) are protected by a sim-
`ple “unlock” scheme: a purchaser makes a pur-
`chase, for example by telephone with a credit card,
`and receives a cryptographic key in return. This
`key can then be used to “unlock” one property
`from some widely distributed medium (e.g., CD-
`ROM or network download). This mechanism is
`relatively inflexible, and its inherently manual
`nature makes it expensive.
`
`3.4 Billing Schemes
`
`Various billing schemes (as mentioned above) per-
`mit purchase of information following what is
`essentially an electronic check or electronic credit
`draft model. These methods are suitable for con-
`ventional transactions, but not for the enormous
`volumes of (individually) very low-value transac-
`tions that would be generated using a complex dig-
`ital property.
`
`3.5
`
`Secured Delivery
`
`Various secured delivery systems (e.g., SSL [19],
`SHTTP [20]) share the same problems as crypto-
`graphic unlock, but in a network context. They are
`only point-to-point solutions, with the information
`(content, usage data, etc.) at each site being left
`unprotected once the delivery has occurred. Fur-
`thermore, they are inherently online systems: it is
`not practical to decouple the delivery of informa-
`tion from payment for its use.
`
`4 Information Protection Architecture:
`InterTrust and DigiBox
`
`EPR has produced the InterTrust Virtual Distribu-
`tion Architecture to solve unmet, critical needs of
`electronic commerce. Almost any
`imaginable
`information transaction can be supported by Inter-
`Trust. A few examples include distribution of con-
`tent (e.g., text, video, audio) over networks,
`selective release of data from a database, con-
`trolled release of sensitive information, and so on.
`InterTrust can also support the secure communica-
`tion of private information such as EDI and elec-
`tronic financial transactions, as well as delivery of
`the “back channel” marketing and usage data
`resulting from transactions.
`
`DigiBox is a foundation technology within Inter-
`Trust. It provides a secure container to package
`information so that the information cannot be used
`except as provided by the rules and controls associ-
`ated with the content. InterTrust rules and controls
`specify what types of content usage are permitted,
`as well as the consequences of usage such as
`reporting and payment.
`
`Within InterTrust, DigiBox containers can enforce
`a “distributed electronic contract” for value-chain
`activities functioning within an electronic distribu-
`tion environment. This unique approach underlies
`EPR’s information metering and digital rights pro-
`tection technology. Electronic commerce infra-
`structure participants can use
`InterTrust
`to
`substantially enhance their network, security, or
`payment method solutions
`.
`
`The DigiBox is a container for both digital prop-
`erty (content) and controls. It is used in conjunc-
`tion with a locally secured rights protection
`application (discussed further below) to make con-
`tent available as governed by arbitrarily flexible
`controls.
`
`The DigiBox container mechanism is implemented
`in a set of platform-independent class libraries that
`provide access to objects in the container and
`extensions to OpenDoc and OLE object technolo-
`gies. DigiBox allows rights management compo-
`nents to be integrated with content in highly
`flexible and configurable control structures. Digi-
`
`007
`
`

`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`Box rights management components can be inte-
`grated with content in a single deliverable, or some
`or all of the components can be delivered indepen-
`dently. DigiBox rights management components
`enable true superdistribution [21] and can support
`virtually any network topology and any number of
`participants, including distributors, redistributors,
`information retailers, corporate content users, and
`consumers.
`
`4.1 Content
`
`The digital information in a DigiBox (one or more
`“properties”) is information in any form. It may be
`mapped to a specific compound object format (e.g.,
`OpenDoc, OLE, PDF), or may be application spe-
`cific.
`
`Further, it may be delivered in stream or other
`communication-oriented forms, not just in a file-
`like container.
`
`4.2 Controls
`
`Controls specify rules and consequences for opera-
`tions on content. Controls are also delivered in a
`DigiBox, and the controls for a property may be
`delivered either with the property or independently.
`Controls are tied to properties by cryptographic
`means.
`
`Because controls can be delivered with properties
`in a container, the DigiBox supports superdistribu-
`tion.
`
`4.3 Commerce
`
`Commerce takes place governed by controls. This
`may involve metering, billing for use, reporting of
`usage, and so on. These operations take place
`locally in a secure environment, and they generate
`audit trails and reports that must be reported peri-
`odically to clearinghouses.
`
`5 DigiBox Implementation
`
`The DigiBox is a structure that can hold, in a pro-
`tected manner, information commerce elements of
`all kinds: content, usage information, representa-
`
`tion of financial transactions (e.g., electronic cash),
`and other digital elements of information com-
`merce.
`
`5.1 Container Logical Structure
`
`Figure 4 shows the logical structure of properties
`and control sets in two containers. Container C
`1
`holds two properties, P
` and P
`, and one control set,
`2
`1
`CS
`, that applies to property P
`; container C
` con-
`1
`1
`2
`tains two control sets and no properties. As shown
`in the example, each of these elements has a title
`attribute to provide a human-readable description
`of the element and, for control sets, an attribute
`indicating to what other elements the control set
`applies.
`
`A control set specifies rules and consequences,
`such as pricing, reporting, and so on, for the prop-
`erties to which it applies. A user holding just this
`container could use (e.g., view, print) content from
`—though only as specified by CS
`. Because
`P
`1
`1
`there is no control set applying to P
` in that con-
`2
`tainer, P
` would not be usable in any way.
`2
`
`A user holding both containers could use property
`P
`, as specified by CS
`, and in addition has the
`2
`2
`choice of whether to designate CS
` or CS
` when
`1
`3
`using P
`. CS
`, which describes itself as “discount,”
`1
`3
`is likely to be the user’s preferred choice.
`
`The DigiBox includes several elements: organiza-
`tional structures, properties, controls, and support-
`ing data items. Almost all the information in a
`DigiBox is encrypted, as described below, and
`access to the encrypted form is provided through a
`storage manager as appropriate, depending on how
`the DigiBox is delivered (e.g., as a file or as a data
`stream).
`
`5.2 Container Physical Structure
`
`Figure 5 is a schematic picture illustrating the
`physical structure of a DigiBox container. (Some
`elements have been omitted for clarity.) It begins
`
`container header structure containing
`with a
`descriptive and organizational information about
`the container. Part of the container header is
`encrypted (both for secrecy and for integrity pro-
`tection); the rest is public organizational informa-
`
`008
`
`

`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`Property
`P1
`
`Title = giraffe
`
`Control Set
`CS1
`
`
`
`Applies to = P1
`Title = regular
`
`Control Set
`CS3
`
`Applies to = P1
`Title = discount
`
`Control Set
`CS2
`
`Applies to = P2
`Title = discount
`
`Container C2
`
`Property
`P2
`
`Title = elephant
`
`Container C1
`
`Figure 4. Container logical structure.
`
`is followed by additional
`tion. The header
`container-wide structures such as the
`transport key
`and the
`container table of contents
`block (TKB)
`, some of which are encrypted and others
`(TOC)
`not.
`
`These organizational elements are followed by the
`structures defining the container’s content (e.g.,
` and
`
`control sets). As shown in the fig-
`properties
`
`property header,
`ure, a property is represented by a
`and data blocks composing the
`property attributes,
`property. As shown, the header is encrypted and
`
`the attributes are not; the data blocks may be
`wholly or partly encrypted, or not at all, depending
`on security requirements.
`
`The figure shows an example property consisting
`of a multimedia property formed from a pair of
`synchronized data streams for audio and video. In
`this example, each video block is mostly unen-
`crypted so that access can be rapid while still main-
`taining reasonable security—encrypting even 10
`percent of an MPEG stream renders it effectively
`useless for illicit copying. On the other hand, the
`audio is entirely encrypted, and each audio block
`
`009
`
`

`
`
`
`
`
`
`
`
`
`
`
`Video
`Block 1
`
`Video
`Block 2
`
`Audio
`Block 1
`
`Audio
`Block 2
`
`Property P1 Data
`
`Container
`TOC
`
`Property
`P1Header
`
`Property
`P1
`Attributes
`
`Container
`Header
`
`Container
`TKB
`
`Shading indicates encryption:
`
`Unencrypted
`
`Encrypted by Key 1
`
`Encrypted by Key 2
`. . .
`
`Figure 5. Container physical format.
`
`uses four distinct keys, because the content propri-
`etor requires much stronger security for audio than
`for video.
`
`frames of a movie, segments of a musical piece,
`and so on, because the mapping is performed by a
`control process specified by the control structure,
`not simply via a table-driven data structure.
`
`A property is represented as one or more property
`sections, each of which is independently associated
`with control information, and which may also be
`stored and accessed independently. A property, for
`example, might be a collection of clip-art images,
`and each image might be a property “chunk,” with
`its own control specifying how that image’s creator
`is compensated.
`
`Controls can map to property chunks at arbitrary
`granularity and can enforce arbitrary organiza-
`tional structures within the property (such as a file
`hierarchy). Controls can apply to individual bytes,
`
`5.3 Cryptographic Techniques
`
`in a DigiBox are
`The high-level elements
`
`transport key that is normally
`encrypted with a
`derived (by exclusive OR) from two parts: one that
`is delivered in the DigiBox itself, encrypted with a
`public key algorithm, and the other that is stored in
`protected storage locally. The locally stored part is
`shared among all the local nodes capable of pro-
`cessing that DigiBox, but the part in the DigiBox is
`unique. This separation provides protection against
`accidental or malicious disclosure of either part.
`
`010
`
`

`
`
`
`
`
`
`
`In Protected
`Local Storage
`
`ID = 6
`
`ID = 7
`
`ID = 8
`
`ID = 30
`
`TKEK
`Storage
`TKEK6
`TKEK7
`TKEK8
`TKEK30
`TKEK31
`TKEK32
`TKEK33
`
`ID = 31
`
`ID = 32
`
`ID = 33
`
`Partial TK
`Storage
`
`ID = 73
`
`ID = 81
`
`ID = 90
`
`ID = 142
`
`ID = 176
`
`ID = 177
`
`Partial TK73
`Partial TK81
`Partial TK90
`Partial TK142
`Partial TK176
`Partial TK177
`
`In Container
`
`Transport Key
`Block (TKB)
`
`ID = 1
`
`ID = 5
`ID = 31
`ID = 36
`ID = 40
`ID = 61
`
`Partial TK
`
`Container
`Header
`
`Public
`Header
`Information
`
`Encrypted
`Header
`Information
`
`Decrypt
`
`Decrypt
`
`ID = 142 Partial TK Value
`
`XOR
`
`Transport Key
`
`Decrypted Header Information
`
`Figure 6. Container transport security.
`
`Figure 6 illustrates how the transport key (TK) is
`(TKB) contains
`derived. The transport key block
`
`one or more slots, each of which contains a partial
`
`transport key encrypted under a different transport
`key encrypting key (TKEK). Each TKB slot identi-
`fies the TKEK used, and a matching TKEK is
`
`011
`
`

`
`
`
`
`
`
`
`
`
`
`
`selected from local protected storage. Decrypting
`the slot yields a partial TK, which is combined
`with its corresponding partial TK again from pro-
`tected local storage to yield the actual TK for
`decrypting the container header.
`
`The data for the property itself is encrypted with
`other keys (“content keys”) that are themselves
`delivered in encrypted high-level structures; this
`approach permits the keys for a property to be
`delivered entirely separately from the property or
`its controls. Multiple keys, in a wide variety of
`key-mapping schemes, are used to encrypt the
`data, limiting the loss that would occur from dis-
`closure of any one key.
`
`All DigiBox control structures are both encrypted
`and verified for integrity with a cryptographic hash
`function. Several cryptographic algorithms are
`supported for these control structures (principally
`for export control reasons), and arbitrary algo-
`rithms are supported for encryption of the data.
`
`5.4
`
`Security Characteristics
`
`The DigiBox cryptographic structures are designed
`to be secure even in the face of loss of individual
`key components, and to minimize the damage in
`case a key or processing environment is compro-
`mised. The system is designed to provide commer-
`cially acceptable risks and losses for a variety of
`business models.
`
`The basic algorithms are strong: Triple DES [22]
`and RSA [23] are preferred. This security is, of
`course, only as strong as the tamper-resistance of
`the local processing environment. The preferred
`implementation of DigiBox processing relies on a
`“secure processing unit” (SPU) that contains a
`CPU, memory, program storage, and key storage in
`a single
`tamper-resistant hardware package.
`Although these are not widely available today, the
`variety of applications they might support makes it
`likely that such SPUs will become widely inte-
`grated into common computing platforms. When
`running in an SPU, the DigiBox processing and
`control mechanisms are sufficiently well protected
`to support most commerce applications.
`
`In the absence of an SPU, other approaches are
`useful for many business models. In fact, a soft-
`ware-only implementation is sufficient for many
`applications, because much content is of relatively
`low value and is used in a context (business to
`business) where a modest level of fraud is both less
`likely and more tolerable. As long as the software
`is moderately difficult to defeat and tools to defeat
`it have no legitimate purpose, business models can
`be supported where some risk of loss is acceptable.
`In the world of electronic commerce, just as for tra-
`ditional commerce, security is not absolute: it is
`just a factor to balance against the cost of loss and
`fraud.
`
`6 Conclusions
`
`The DigiBox is one component of a general-pur-
`pose electronic commerce solution that rests on
`three basic principles: rights protection, interopera-
`bility, and strong security.
`
`Electronic commerce, and information commerce
`in particular, needs a robust information protection
`mechanism, including rights protection and con-
`trols, not just payment systems. As the electronic
`world evolves, however, and moves forward from
`simply emulating
`traditional
`transactions
`into
`entirely new business models, rights protection and
`control will become the predominant issues.
`
`Protection of intellectual property rights in infor-
`mation requires strong cryptography as well as a
`flexible infrastructure for controlling use of the
`information. A standard protected container for
`information is necessary to support interoperabil-
`ity—most existing schemes tightly bind the creator
`of protected information and the software that pro-
`cesses it. A standard container can rationalize
`information commerce and reduce costs for all par-
`ticipants.
`
`In the long term, general-purpose secure electronic
`commerce will need pervasive deployment of
`tamper-resistant hardware devices
`to perform
`secure processing of protected content. However,
`as these solutions are developed, many business
`models can be accommodated with weaker or less
`complete solutions because the risk and expected
`losses are commercially acceptable.
`
`012
`
`

`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`Business-to-business purchasing is steadily evolv-
`ing into a direct electronic ordering model. Future
`communications and media markets will become
`increasingly
`segmented
`and
`specialized
`in
`response to customer preferences and needs and
`involve increasing, and more sophisticated, direct
`interaction between consumers and providers.
`These markets and their value chains (with or with-
`out intermediary distributors) will require secure
`metering and control tools that enable a user to
`efficiently and economically tailor resources to his
`or her own desires.
`
`During the next decade, digital delivery of tradi-
`tional electronic products, such as information
`databases and software, will be joined by a rapidly
`growing array of both New Media and electroni-
`cally distributed traditional content. The conver-
`sion
`from
`traditional models
`requires key
`foundation technologies and will result in a funda-
`mental shift in current infrastructure. This transfor-
`mation will create a new distribution industry.
`Digital distribution employing a universal content
`and commerce container can play a critical role in
`this broad economic transformation.
`
`[5] R. Benjamin and R Wigand, “Electronic Mar-
`kets and Virtual Value Chains on the Informa-
`tion Superhighway,” Sloan Management
`Review, Vol. 36 No. 2 (1995).
`
`[6] U.S. Constitution, Article 1, Section 8, Clause 8
`(1787).
`
`[7] U.S. Copyright Act of 1978
`
`[8] 17 U.S.C. s107
`
`[9] 17 U.S.C s102(a)
`
`[10] T. Berners-Lee, R Caillian, and J.-F. Groff,
`“The World Wide Web,” Computer Networks
`