Author: Francisco Corella

Comparing the Privacy Features of Eighteen Authentication Technologies

This blog post motivates and elaborates on the paper Privacy Postures of Authentication Technologies, which we presented at the recent ID360 conference.

There is a great variety of user authentication technologies, and some of them are very different from each other. Consider, for example, one-time passwords, OAuth, Idemix, and ICAM’s Backend Attribute Exchange: any two of them have little in common.

Different authentication technologies have been developed by different communities, which have created their own vocabularies to describe them. Furthermore, some of the technologies are extremely complex: U-Prove and Idemix are based on mathematical theories that may be impenetrable to non-specialists; and OpenID Connect, which is an extension of OAuth, adds seven specifications to a large number of OAuth specifications. As a result, it is difficult to compare authentication technologies to each other.

This is unfortunate because decision makers in corporations and governments need to decide what technologies or combinations of technologies should replace passwords, which have been rendered even more inadequate by the shift from traditional personal computers to smart phones and tablets. Decision makers need to evaluate and compare the security, usability, deployability, interoperability and, last but not least, privacy, provided by the very large number of very different authentication technologies that are competing in the marketplace of technology innovations.

But all these technologies are trying to do the same thing: authenticate the user. So it should be possible to develop a common conceptual framework that makes it possible to describe them in functional terms without getting lost in the details, to compare their features, and to evaluate their adequacy to different use cases.

The paper that we presented at the recent ID360 conference can be viewed as a step in that direction. It focuses on privacy, an aspect of authentication technology which I think is in need of particular attention. It surveys eighteen technologies, including: four flavors of passwords and one-time passwords; the old Microsoft Passport (of historical interest); the browser SSO profile of SAML; Shibboleth; OpenID; the ICAM profile of OpenID; OAuth; OpenID Connect; uncertified key pairs; public key certificates; structured certificates; Idemix pseudonyms; Idemix anonymous credentials; U-Prove tokens; and ICAM’s Backend Attribute Exchange.

The paper classifies the technologies along four different dimensions or facets, and builds a matrix indicating which of the technologies provide seven privacy features: unobservability by an identity or attribute provider; free choice of identity or attribute provider; anonymity; selective disclosure; issue-show unlinkability; multishow unlinkability by different parties; and multishow unlinkability by the same party. I will not try to recap the details here; instead I will elaborate on observations made in the paper regarding privacy enhancements that have been used to improve the privacy postures of some closed-loop authentication technologies.

Privacy Enhancements for Closed-Loop Authentication

One of the classification facets that the paper considers for authentication technologies is the distinction between closed-loop and open-loop authentication, which I discussed in an earlier post. Closed-loop authentication means that the credential authority that issues or registers a credential is later responsible for verifying possession of the credential at authentication time. Closed-loop authentication may involve two parties, or may use a third-party as a credential authority, which is usually referred to as an identity provider. Examples of third-party closed-loop authentication technologies include the browser SSO profile of SAML, Shibboleth, OpenID, OAuth, and OpenID Connect.

I’ve pointed out before that third-party closed-loop authentication lacks unobservability by the identity provider. Most third-party closed-loop authentication technologies also lack anonymity and multishow unlinkability. However, some of them implement privacy enhancements that provide anonymity and a form of multishow unlinkability. There are two such enhancements, suitable for two different use cases.

The first enhancement consists of omitting the user identifier that the identity provider usually conveys to the relying party. The credential authority is then an attribute provider rather than an identity provider: it conveys attributes that do not necessarily identify the user. This enhancement provides anonymity, and multishow unlinkability assuming no collusion between the attribute provider and the relying parties. It is useful when the purpose of authentication is to verify that the user is entitled to access a service without necessarily having an account with the service provider. This functionality is provided by Shibboleth, which can be used, e.g., to allow a student enrolled in one educational institution to access the library services of another institution without having an account at that other institution.

The core OpenID 2.0 specification specifies how an identity provider conveys an identifier to a relying party. Extensions of the protocol such as the Simple Registration Extension specify methods by which the identity provider can convey user attributes in addition to the user identifier; and the core specification hints that the identifier could be omitted when extensions are used. It would be interesting to know whether any OpenID server or client implementations allow the identifier to be omitted. Any comments?

The second enhancement consists of requiring the identity provider to convey different identifiers for the same user to different relying parties. The identity provider can meet the requirement without allocating large amounts of storage by computing a user identifier specific to a relying party as a cryptographic hash of a generic user identifier and an identifier of the relying party such as a URL. This privacy enhancement is required by the ICAM profile of OpenID. It achieves user anonymity and multishow unlinkability by different parties assuming no collusion between the identity provider and the relying parties; but not multishow unlinkability by the same party. It is useful for returning user authentication.

Two Methods of Cryptographic Single Sign-On on Mobile Devices

This is the sixth and last post of a series discussing the paper A Comprehensive Approach to Cryptographic and Biometric Authentication from a Mobile Perspective.

To conclude this series I am going to discuss briefly two methods of single sign-on (SSO) described in the paper, one based on data protection, the other on shared login sessions.

SSO Based on Data Protection

Section 5 of the paper explains how the multifactor closed-loop authentication method described in the third and fourth posts of the series provides an effective mechanism for protecting data stored in a mobile device against an adversary who captures the device. The data is encrypted under a data encryption key that is entrusted to a key storage service. To retrieve the key, the user provides a PIN and/or a biometric that are used to regenerate an uncertified key pair, which is used to authenticate to the storage service.

An adversary who captures the device needs the PIN and/or the biometric sample to regenerate the key pair, and cannot mount an offline attack to guess the PIN or to guess a biometric key derived from the biometric sample; so the adversary cannot authenticate to the key storage service, and cannot retrieve the key. For additional security the data encryption key can be cryptographically split in several portions entrusted to different storages services. Furthermore a protokey can be entrusted to those services instead of the data encryption key, the key being then derived from the protokey and the same non-stored secrets that are used to regenerate the authentication key pair as described in Section 5.4.

This data protection mechanism can be used to protect any kind of data. In particular, it can be used to protect credentials used for open-loop authentication or one-factor closed-loop authentication to any number of mobile applications or, more precisely, to the back-ends of those applications, which may be have browser-based or native front-ends. As discussed in Section 5.5, this amounts to single sign-on to those applications because, after the user enters a PIN and/or provides a biometric sample, the data encryption key retrieved from the storage service(s) can be kept in memory for a certain amount of time, making it possible to authenticate to the applications without further user intervention.

SSO Based On Shared Login Sessions

Whereas SSO based on data protection can be used for any collection of applications, SSO based on shared login sessions, described in Section 7.5, is best suited for authenticating to enterprise applications from a mobile device. A dedicated PBB in the mobile device and a VBB in the enterprise cloud are used to that purpose. The PBB contains a single protocredential shared by all the enterprise applications, which is used to regenerate an uncertified key pair, in conjunction with a PIN and/or a biometric sample supplied by the user. The VBB has access to an enterprise database that contains device and user records and where the VBB stores shared session records, as illustrated in Figure 8.

It is not difficult to share login sessions among a group of web-based applications owned by an enterprise, using a mechanism readily available on the web. Once the user has logged in to one of the web-based applications in the group, that application can set in the browser a session cookie whose scope (defined explicitly or implicitly by the domain and path attributes of the cookie) comprises the applications in the group and no others. The browser will send the cookie along with every HTTP request targeting an application in the scope of the cookie, thus authenticating the request without user intervention.

But we want to share login sessions among a group of enterprise applications comprising applications with native front-ends in addition to web-based applications. To that purpose we use the mobile authentication architecture that I discussed in the previous post, modifying it as follows.

Recall that an authentication event in the architecture consists of a cryptographic authentication of the PBB to the VBB, followed by a secondary non-cryptographic authentication using a one-time authentication token, which plays the role of a bearer token, as illustrated in Figure 6 for the case of an application with a native front-end, and in Figure 7 for the case of a web-based application. The authentication token is only used once because of the risk of a Referer leak in the case of a web-based application. However there is no such risk in the case of an application with a native front-end.

To implement shared login sessions we replace the one-time authentication token with a pair of session tokens, a one-time session token and a reusable session token. After successful cryptographic authentication of the PBB to the VBB, the VBB creates a pair of session tokens and a shared session record containing the two tokens, and sends the two tokens to the PBB, which stores them.

A native front-end obtains a reusable session token from the PBB and uses it repeatedly to authenticate to its back-end until the back-end rejects it because the session referenced by the token has expired because an expiration time in the shared session record has been reached or some other reason. Then the native front-end sends the reusable token to the PBB asking for a replacement. If the PBB has a different reusable token, it sends it to the native front-end. If not, it prompts the user for a PIN and/or a biometric sample, regenerates the uncertified key pair, authenticates cryptographically to the VBB, obtains from the VBB a pair of session tokens pertaining to a new session, and sends the new reusable token to the native front-end.

A web-based application obtains a one-time session token from the PBB and uses it to locate a shared session record and retrieve a reusable session token, which it sets in the browser as the value of a session cookie. After the PBB sends the one-time token to the application, it erases the one-time token from its storage; and after the application uses the one-time token to retrieve the reusable token, it erases the one-time token from the shared session record. The session cookie is used to authenticate HTTP requests sent by the browser to web-based applications in the group, until one of the applications finds that the session referenced by the reusable token contained in the cookie has expired. Then that application sends the reusable token to the PBB and asks for a one-time token. If the PBB has a one-time token paired with a reusable token different from the one sent by the application, it sends the one-time token to the application. Otherwise it authenticates cryptographically to the VBB as in the case of a native front-end, obtaining a pair of fresh tokens and sending the new one-time token to the application.

Pros and Cons of the Two Methods

The method based on data protection is more flexible than the method based on shared sessions. It can be used to implement SSO for any set of applications, whether or not those applications are related to each other. By contrast, the method based on shared sessions can only be used to implement SSO for a group of related applications: the set of web-based applications in the group must be circumscribable by the scope of a cookie; and, as explained in Section 8.2.2, native front-ends of applications in the group must be signed with the same code-signing key pair in Android, or must have the same Team ID in iOS, so that the PBB can refuse requests from applications not in the group.

On the other, the method based on shared login sessions has performance and security advantages, as explained in Section 7.5.3. In the method based on data protection, SSO is accomplished by making cryptographic authentication transparent to the user, whereas in the method based on shared login sessions cryptographic authentication is avoided altogether; hence the performance advantage. In the method based on data protection, the data encryption key must be present in the device while the user interacts with the applications, whereas in the method based on shared login sessions the uncertified key pair is only needed when a new session is created, and can be erased after it is used; hence the security advantage.

Using Cryptographic Authentication without a Cryptographic API on iOS and Android Devices

This is the fifth of a series of posts discussing the paper A Comprehensive Approach to Cryptographic and Biometric Authentication from a Mobile Perspective.

Everybody agrees that passwords provide very poor security for user authentication, being vulnerable to capture by phishing attacks or database breaches, or by being reused at malicious sites. Authentication using public key cryptography does not have any of these vulnerabilities, and yet, after being available for several decades, it is only used in limited contexts. As computing shifts from traditional PCs to mobile devices, everybody agrees that passwords are terribly inconvenient on touchscreen keyboards, in addition to being insecure; and yet I don’t see a rush to adopting cryptographic authentication methods on mobile devices.

What obstacles stand in the way of widespread adoption of cryptographic authentication?

One obstacle is no doubt the complexity of cryptography. Implementing cryptographic functionality is difficult even when cryptographic libraries are available. Using a cryptographic API is no trivial matter, as documented by Martin Georgiev et al. in a recent paper (reference [39] in the paper).

Another obstacle is poor support by web browsers for the deployment and use of cryptographic credentials. In particular, there are no easy-to-use standards generally supported by browser vendors for issuing cryptographic credentials to a browser and requesting the presentation by the browser of particular credentials or credentials asserting particular attributes.

In Section 7 the paper proposes an architecture for cryptographic authentication on mobile devices that addresses these two obstacles. It does that by encapsulating cryptographic authentication of a mobile device to an application back-end inside a Prover Black Box (PBB) located in the device and a Verifier Black Box (VBB) located in the cloud, as shown in figures 6 (page 48) and 7 (page 54).

The PBB may contain one or more protocredentials for multifactor closed-loop authentication, or credentials for single factor closed-loop or open-loop authentication; and it takes care of proving possession of credentials to the VBB. After a cryptographic authentication event in which the PBB proves possession of one or more credentials, the VBB creates an authentication object that records the event and contains authentication data such as the hash of a public key or attributes asserted by a public key certificate, a U-Prove token, or an Idemix anonymous credential. The authentication object is retrievable by a one-time authentication token, which the VBB passes to the PBB and the PBB passes to the application back-end via a native front-end or via the web browser. The authentication token plays the role of a bearer token in a secondary non-cryptographic authentication of the native front-end or web browser to the back-end, and allows the application back-end to retrieve the authentication data.

In Figure 6 the native front-end of a mobile application receives the authentication token from the PBB and uses it to authenticate to the back-end of the same application, which presents it to the VBB to retrieve the authentication data.

In Figure 7, the PBB sends the token via the browser to the back-end of a web-based application, thus authenticating the browser to the back-end, which again uses the token to retrieve the authentication data from the VBB. (As a matter of terminology, we view a web-based application as having a back-end and a front-end, the back-end being its cloud portion, while the front-end consists of web pages and client-side code running in the browser.)

This architecture circumvents the two obstacles identified above to the adoption of cryptographic authentication.

The browser obstacle is avoided in Figure 6 because no browser is involved, and in Figure 7 because the browser is not involved in storing or presenting credentials, and no modification of standard browser functionality is required.

The obstacle presented by the complexity of cryptography is avoided by the encapsulation of cryptographic functionality in the PBB and the VBB and by making the PBB and the VBB accessible through non-cryptographic APIs in a manner familiar to native and web-based application developers.

In Figure 6, arrows (1) and (4) represent messages sent via the operating system of the mobile device using inter-application communication mechanisms available in iOS and Android; each message is a URL having a custom scheme, with message parameters embedded as usual in the query portion of the URL. Arrow (6) represents an HTTP POST request, and arrow (7) the corresponding response. Arrow (5) is internal to the application and can be implemented as part of a standard web API through which the native front-end accesses its back-end.

In Figure 7, arrow (1) represents an HTTP response that redirects the browser to a custom scheme that targets the PBB, with parameters included in the query portion of the URL; when the browser receives the response, it forwards it to the PBB as a message, using the inter-application communication mechanism provided by the operating system. Arrow (4) represents a message sent by the PBB using the same mechanism, with scheme https; the operating system delivers it to the browser, which forwards it as an HTTP GET request to the application back-end. Arrow (5) represents an HTTP POST request, and arrow (6) the corresponding response.

The architecture is very flexible. It covers a wide variety of use cases, some of which are sketched out in Section 7.1.

A PBB-VBB pair may be used for returning-user authentication to one particular application. In that case the PBB contains a single credential (for one-factor authentication) or protocredential (for multifactor authentication).

Alternatively, a general purpose PBB may be made available to any mobile application that has a native front-end on the device or is accessed from the device through a browser, each application having its own VBB. In that case the PBB may contain any number of credentials or protocredentials used for closed-loop authentication, as well as credentials used for open-loop authentication.

An application may ask a general purpose PBB to prove possession of an uncertified key pair to the application’s VBB for returning-user authentication, or to the VBB of an identity/attribute provider or a social network for third-party closed-loop authentication or social login. The VBB of an identity/attribute provider delivers the user’s identity or attributes to the application back-end as authentication data upon presentation of the authentication token. The VBB of a social network may instead deliver an access token that provides limited access to the user’s account, thus allowing the application to obtain the user’s identity and attributes from the user’s profile, to issue social updates on behalf of the user, and more generally to provide an alternative user interface to the social network.

An application may also ask a general purpose PBB to demonstrate that the user has certain attributes by presenting public key certificates, U-Prove tokens or Idemix anonymous credentials to the application’s VBB in open-loop authentication.

For enterprise use, a PBB-VBB pair may be shared by a group of enterprise applications, including web-based applications and applications with native front-ends, with single sign-on based on shared login sessions. I will discuss this functionality in the next post.

A security analysis of the architecture is provided in Section 8. Among other security considerations, it discusses protection against leaks through so-called Referer headers, protection against misuse of an authentication token by its recipient to impersonate the user, a countermeasure against a form of Login CSRF, identification of the application that requests presentation of one or more credentials kept by a general purpose browser, and countermeasures against a malicious application masquerading as a different application or as the system browser.

Strong Authentication with a Low-Entropy Biometric Key

This is the fourth of a series of posts discussing the paper A Comprehensive Approach to Cryptographic and Biometric Authentication from a Mobile Perspective.

Biometrics are a strong form of authentication when there is assurance of liveness, i.e. assurance that the biometric sample submitted for authentication belongs to the individual seeking authentication. Assurance of liveness may be relatively easy to achieve when a biometric sample is submitted to a reader in the presence of human operator, if the reader and the operator are trusted by the party to which the user is authenticating; but it is practically impossible to achieve for remote authentication with a reader controlled by the authenticating user. When there is no assurance of liveness, security must rely on the relative secrecy of biometric features, which is never absolute, and may be non-existent. Fingerprints, in particular, cannot be considered a secret, since you leave fingerprints on most surfaces you touch. Using a fingerprint as a login password would mean leaving sticky notes with your password everywhere you go.

In addition to these security caveats, biometric authentication raises acute privacy concerns. Online transactions authenticated with biometric features would be linkable not only to other online transactions, but also to offline activities of the user. And both online and offline transactions would become linked to the user’s identity if a biometric sample or template pertaining to the user became public knowledge or were acquired by an adversary.

Yet, in Section 3, the paper proposes a method of using biometrics for user authentication on a mobile device to an application back-end. The method addresses the above security and privacy concerns as follows:

  1. First, biometrics is not used by itself, but rather as one factor in multifactor authentication, another required factor being possession of a protocredential stored in the user’s device, and another optional factor being knowledge of a passcode such as a PIN.
  2. Second, the paper suggests using an iris scan, which provides more secrecy than fingerprints. (The scan could be taken by a camera on the user’s mobile device. The paper cites the work of Hao, Anderson, and Daugman at the University of Cambridge, which achieved good results with iris scans using a near-infrared camera. I have just been told that phone cameras filter our near-infrared light, so a special camera may be needed. The Wikipedia article on iris recognition discusses the use of near-infrared vs. visible light for iris scanning.)
  3. Third, no biometric-related data is sent by the user’s device to the application back-end, neither at authentication time nor at enrollment time. The biometric sample is used to regenerate a key pair on the device, and the key pair is used to authenticate the device to the back-end.
  4. Fourth, neither a biometric sample nor a biometric template are stored in the user’s device. Instead, the paper proposes to use one of several methods described in the literature, cited in Section 3.2, for consistently producing a biometric key from auxiliary data and genuine but varying biometric samples. Only the auxiliary data is stored in the device, and it is deemed unfeasible to recover any biometric information from the auxiliary data.

The resulting security and privacy posture is discussed in Section 4.4 of the paper.

As shown in Figure 3 (in page 22 of the paper), we combine the biometric key generation process with the key pair regeneration process of our protocredential-based authentication method. The biometric sample (the iris image in the figure) is a non-stored secret (the only one in this case), and the auxiliary data is kept in the protocredential as a non-stored-secret related parameter. The auxiliary data and the biometric sample are combined to produce the biometric key. A randomized hash of the biometric key is computed using a salt which is also kept in the protocredential, as a second non-stored-secret related parameter. The randomized hash of the biometric key is used to regenerate the key pair, in conjunction with the key-pair related parameters. The key pair regeneration process produces a DSA, ECDSA or RSA key pair as described in sections 2.6.2, 2.6.3 and 2.6.4 respectively. The public key is sent to the application back-end, and the private key is used to demonstrate possession of the credential by signing a challenge. Figure 4 (in page 23 of the paper) adds a PIN as a second non-stored secret for three-factor authentication; in that case the auxiliary data is kept encrypted in the protocredential, and decrypted by x-oring the ciphertext with a randomized hash of the PIN.

The combination of biometric key generation with our protocredential-based authentication method represents a significant improvement on biometric authentication methodology. There is an intrinsic trade-off between the consistency of a biometric key across genuine biometric samples and the entropy of the key, because the need to accommodate large enough variations among genuine biometric samples reduces the entropy of the key. In the above mentioned paper by Hao et al., the authors are apologetic about the fact that their biometric key has only 44 bits of entropy when the auxiliary data is known. But this is not a problem in our authentication framework, for two reasons:

  1. The auxiliary data is not public. An adversary must capture the user’s device to obtain it.
  2. An adversary who captures the user’s device and obtains the auxiliary data cannot mount an offline guessing attack against the biometric key. All biometric keys produce well-formed DSA or ECDSA key pairs, and most biometric keys produce well-formed RSA key pairs. To determine if a guessed biometric key is valid, the adversary must therefore use it to generate a key pair, and use the key pair to authenticate online against the application back-end, which will limit the number of guesses to a small number. Forty-four bits of entropy is plenty if the adversary can only make, say, 10 guesses.

Therefore our authentication method makes it possible to use low-entropy biometric keys without compromising security. This may enable the use of biometric modalities or techniques that otherwise would not provide sufficient security.

Nevertheless we do not advocate the routine use of biometrics for authentication. As pointed out in Section 10, while malware running on the user’s device after an adversary has captured it cannot obtain biometric data, malware running on the device while the user is using it could obtain a biometric sample by prompting the user for the sample. A biometric authentication factor should only be used when exceptional security requirements demand it and exceptional security precautions are in place to protect the confidentiality of the user’s biometric features.

Defense in Depth of Cryptographic Credentials on a Mobile Device

This is the third of a series of posts discussing the paper A Comprehensive Approach to Cryptographic and Biometric Authentication from a Mobile Perspective.

Credentials based on public key cryptography provide much stronger security than ordinary passwords or one-time passwords. But a mobile device can be lost or stolen. How can a credential kept in a mobile device be protected if the user’s device is captured by an adversary? Two methods are traditionally used:

  1. Private key encryption. The private key is encoded as specified by PKCS #8, together with cryptographic parameters that typically include the public key or a public key certificate, and the resulting encoded string is encrypted under a symmetric data-encryption key derived from a passcode. This method is used, for example to protect SSH credentials used to manage cloud-hosted virtual servers. But as explained in Section 4.3.1 of the paper this method requires a high-entropy password, which is exceedingly difficult to type on the touchscreen keyboard of a smart phone.
  2. Tamper resistance. This is relied upon, for example to protect credentials in smart cards such as PIV or CAC cards. But few mobile devices have tamper resistant modules.

On an iPhone or an iPad one could think of relying on the data protection method introduced in iOS 4, which encrypts data in a locked device under a key derived from the passcode that the user enters to unlock the device and a key embedded in a hardware encryption chip. But, as explained in section 5.1 of the paper, that method has not proved to be effective.

Instead, Section 2 of the paper proposes a method for using an uncertified key pair for multifactor closed-loop authentication that makes it possible to protect the key pair without relying on any special hardware. The method is generally applicable, but is particularly useful for authentication on a mobile device. The idea is to store in the device cryptographic parameters obtained during initial credential generation, at least one of them being a secret, and later, at authentication time, to regenerate the credential from the stored cryptographic parameters and non-stored secrets supplied by the user such as a PIN and/or a biometric sample. (The non-stored secrets could be supplied by a physical uncloneable function, a PUF, in the case of an autonomous device; but the paper is not concerned with autonomous devices.) We refer to the stored parameters as a protocredential. Possession of the protocredential counts as one authentication factor, while the non-stored secrets play the role of additional authentication factors.

The paper distinguishes between parameters related to the key pair and parameters related to the non-stored secrets. In the case where a PIN is the only non-stored secret, illustrated in Figure 2, there is one non-stored-secret related parameter, a salt used to compute a randomized hash of the PIN. (Two-factor authentication with a biometric sample and three-factor authentication with a PIN and a biometric sample are discussed in Figures 3 and 4. I will discuss biometric authentication in the next blog post.) The key-pair related parameters depend on the public key cryptosystem being used. In the case of DSA and ECDSA, the key-pair related parameters are the domain parameters specified in the Digital Signature Standard. In the case of RSA, there is one key-pair related parameter, the least common multiple &#x03BB of p-1 and q-1, where p and q are the prime factors of the modulus. The key pair regeneration procedures for DSA, ECDSA and RSA are described in sections 2.6.2, 2.6.3 and 2.6.4 respectively.

In a mobile device, once the key pair has been regenerated, it is used by the device to authenticate to a mobile application with which the device has been previously registered. The application may have a native front-end or use a web browser as its front-end. The application back-end has a database that contains a record for the device, identified by a device record handle (a database primary key). To authenticate, the device sends the device record handle and the public key to the application back-end and demonstrates knowledge of the private key by signing a challenge. The back-end verifies the signature, uses the device record handle to locate the device record, computes a cryptographic hash of the public key, and verifies that the hash coincides with a hash stored in the device record. (A mobile authentication architecture that allows application developers to implement the authentication process without using a cryptographic API is described in Section 7; I will discuss it in another post later in this series.)

An adversary who captures the device and is able to read the protocredential needs the non-stored secrets to be able to regenerate the credential and authenticate. The adversary can try to guess the non-stored secrets. If a PIN is the only non-stored secret and the user chooses a 4-digit PIN, the adversary only has to try 10,000 PINs. If the adversary can test each PIN offline, it is trivial to go through all 10,000 PINs. But all PINs (in the case of DSA and ECDSA) or most PINs (in the case of RSA) produce well-formed key pairs. If the adversary does not know the public key (nor a hash of the public key), the only way to test a PIN is to try to use the key pair that it produces to authenticate online against the application back-end; and the back-end can limit the number of guesses to a very small number, such as 3 or 5 or 10. A 4-digit PIN can then be deemed to provide sufficient security, just as 4-digit PIN is usually considered secure enough for withdrawing cash from an ATM, which also limits the number of PIN guesses.

To ensure that the adversary does not know the public key, the public key should be treated as a shared secret between the device and the application back-end. Treating a public key as a secret is an unconventional and paradoxical use of public key cryptography. Section 4.1 explains that a shared symmetric secret could be used instead of a key pair but would result in a weaker security posture.

To prevent a man-in-the-middle attack, the device connects to the back-end using TLS (or some other kind of secure connection). Furthermore, the challenge signed by the device to demonstrate knowledge of the private key includes the TLS server certificate of the application back-end. Section 2.1 explains how this prevents a man-in-the-middle attack even if the adversary is able to spoof the TLS server certificate of the back-end.

All this results in a very strong defense-in-depth security posture. As discussed in Section 4.2 and summarized in Table 1, authentication is secure even against an adversary who is able to:

  1. Capture the user’s device and read the protocredential from the device storage; or
  2. Breach the security of the database back-end and obtain the hash of the public key. The adversary cannot mount an offline attack against a PIN used as single non-stored secret because the adversary does not have the protocredential, which contains at least one secret parameter. Compare this to the effect of a breach of database security when the database contains hashes of passwords, all of which become vulnerable to offline dictionary attacks; or
  3. Breach network security and read the traffic from the device to the back-end (e.g. after the TLS connection has been terminated at a reverse proxy, in a misconfigured infrastructure-as-a-service cloud). Again, the adversary cannot mount an offline attack against the PIN; or
  4. Spoof the TLS server certificate of the application back-end, as discussed above.

Also, in use cases demanding exceptionally high security, by using a high-entropy set of non-stored secrets, it is possible to achieve security even against an adversary who breaches database or communication security and then captures the device and obtains the protocredential.

We have seen how to protect an uncertified key pair used for closed-loop authentication. How about other types of credentials? Section 5 shows how the multifactor closed-loop authentication method discussed above can be used to provide effective protection for data stored in a mobile device, and in particular to provide protection for any kind of credentials, including credentials used for open-loop authentication, such as such as public key certificates, U-Prove tokens or Idemix anonymous credentials.

In the next post I will discuss the use of a biometric sample as a non-stored secret, and explain how it can achieve strong security without putting at risk the confidentiality of the user’s biometric features.

Closed-Loop vs. Open-Loop Authentication

This is the second of a series of posts discussing the paper A Comprehensive Approach to Cryptographic and Biometric Authentication from a Mobile Perspective.

In this post I want to take the time to explain and emphasize the distinction made in the paper between closed-loop authentication and open-loop authentication. This may seem an unimportant matter of vocabulary, but the distinction is essential for two reasons: first, because it helps understand the privacy posture of authentication technologies; second, because it leads to what we think is the best choice of cryptographic authentication technologies for mobile devices.

The concepts of closed-loop and open-loop authentication are defined in the introduction, and examples are given. In open-loop authentication, a party such as a certificate authority or, more generally a credential authority, issues a cryptographic credential to the user’s device, and then is “out of the loop” when the device presents the credential to a relying party. Credentials used in open-loop authentication are typically public key certificates, but could also be U-Prove tokens or Idemix anonymous credentials. In closed-loop authentication, on the other hand, the credential authority is involved in the authentication process, taking care of verifying possession of the credential by the device. In third-party closed-loop authentication, the credential authority is an identity or attribute provider, which communicates user attributes to a relying party after verifying that the device possesses the credential. In two-party authentication, there is only one party besides the user’s device, so two-party authentication can only be closed-loop authentication.

The distinction between closed-loop and open-loop authentication makes it possible to make two observations.

The first observation is that closed-loop authentication can rely on an uncertified key pair, i.e. a key pair that is not bound to any attributes by a certificate. (As a matter of vocabulary, we say that an uncertified key pair is registered by the device with the credential authority, rather than issued by the credential authority to the device, because the credential authority plays no role in generating the key pair; the paper refers to the credential authority as “the party that issues or registers the credential”.) An uncertified key pair can be used because the credential authority can store user attributes in its internal storage and retrieve them at authentication time. Therefore the attributes need not be included in the credential.

The second observation is that, in third-party closed-loop authentication, the credential authority, i.e. the identity or attribute provider, is informed of the authentication transaction and, typically, is told what relying party the user is authenticating to. This impinges on the user’s privacy, especially if the user has no choice of identity or attribute provider and does not trust the provider. This is not just a theoretical consideration. The identity providers most commonly used today have track records of privacy violations, and users are wary of being spied upon.

Some time ago, before being concerned with mobile authentication, we wrote a white paper proposing to eliminate this privacy drawback by using the browser to hide the identity of the relying party. However, this would require substantial modifications of core browser functionality. More recently, in an ICAM blog post, Anil John has proposed hiding the identity of the relying party behind a proxy. But that complicates authentication and serves only to shift the trust issue from the identity provider to the proxy.

Open-loop authentication, on the other hand, does not suffer from this privacy drawback.

These observations led us to the following choice of technologies for cryptographic authentication on mobile devices:

  • For the sake of simplicity, an uncertified key pair should be used for two-party authentication.
  • For the sake of privacy, open-loop authentication should be used when attributes are asserted by a third party, except in special cases. Credentials used in open-loop authentication could be public key certificates, U-Prove tokens, or Idemix anonymous credentials, depending on the privacy requirements, as explained in section 6.1.

There are two special cases where it makes sense to use third-party closed-loop authentication. One is social login, where an application is granted limited access to the user’s account at a social network such as Facebook or Twitter and authenticates the user as side-effect, by obtaining user attributes from the user’s profile. In social login, the social network is necessarily involved in the authentication transaction. The other is third-party login using as identity provider a personal data repository service that emphasizes privacy and is freely chosen and trusted by the user. A company participating in the Personal Data Ecosystem Consortium (PDEC), for example, could play the role of identity provider.

However, this choice of technologies posed the problem of how to protect the credentials used in open-loop authentication against an adversary who captures the user’s mobile device, because the key pair regeneration method, which I mentioned in the previous post and will discuss in more detail in the next post, does not work for open-loop authentication.

We were happy to find a simple solution to that problem. As described in Section 5, key pair regeneration can be used to implement effective data protection against an adversary who captures the device, by encrypting the data under a data-encryption key, entrusting the key to a key storage service (or splitting it cryptographically across multiple services), and authenticating to the service(s) with a regenerated key pair to retrieve the key. A credential used in open-loop authentication can be protected as data in this way, thus benefiting indirectly from the security provided by the key pair regeneration technique.

In the next post I will finally get into the technical details of the paper.

New Research on Mobile Authentication

This is the first of a series of posts discussing the paper A Comprehensive Approach to Cryptographic and Biometric Authentication from a Mobile Perspective

In the next few posts I will be reporting on research that we have been doing over the last six months related to cryptographic and biometric authentication, focused on mobile devices. I have held off from writing while we were doing the research but now I have a lot to say, so stay tuned.

By the way, in the last six months we have also moved from San Diego to San Jose. I used to work in Silicon Valley, so it’s nice to be back here and renew old friendships. If you are interested in cryptographic and/or biometric authentication and you are based in Silicon Valley or passing by, let me know; I would be happy to meet for coffee and chat.

The starting point of the this latest research was the work we presented at the NIST Cryptographic Key Management workshop last September (Key Management Challenges of Derived Credentials and Techniques for Addressing Them) and at the Internet Identity Workshop last October (New Authentication Method for Mobile Devices), and wrote up in the paper Strong and Convenient Multi-Factor Authentication on Mobile Devices.

In that early work we devised a mobile authentication architecture where the user authenticates with an uncertified key pair, and a method for regenerating an RSA key pair from a PIN and/or a biometric key. The architecture facilitates implementation by encapsulating the complexities of cryptography and biometrics in a Prover Black Box located in the device and Verifier Black Box located in the cloud, while the key pair regeneration method protects the credential against an adversary who captures the user’s mobile device, by preventing an offline attack against the PIN and/or the biometric key. The architecture was primarily intended for mobile devices but could be adapted for use in traditional PCs by means of browser extensions.

The early work left three questions open:

  1. Can the key pair regeneration method be adapted to cryptosystems other than RSA? This question is practically important because RSA can be used for encryption, and is therefore subject to export controls. The export restrictions have been relaxed a lot since the nineties, but they are so complex that consultation with a lawyer may be required to figure out whether and to what extent they are applicable to a particular product.
  2. Can the mobile authentication architecture accomodate credentials other than uncertified key pairs, including public key certificates and privacy-enhancing credentials such as U-Prove tokens and Idemix anonymous credentials? Uncertified key pairs are ideal for returning-user authentication, but they cannot be used to provide evidence that the user is entitled to attributes asserted by authoritative third parties.
  3. Does the architecture support single sign-on (SSO)? SSO is an essential usability feature when multiple frequently used applications require multifactor authentication.

I am happy to report that we have found good answers to all three questions. First, we have found efficient regeneration methods for DSA and ECDSA key pairs; since DSA and ECDSA can only be used for digital signature, they are not subject to export restrictions. Second, we have found a way of extending the architecture to accomodate a variety of credentials, including public key certificates and privacy-enhancing credentials, without giving up on the strong security properties of the original architecture. Third, we found have found two different ways of providing SSO, one of them well suited for web-wide consumer SSO, the other for enterprise SSO; and both applicable to a mix of web-based apps and apps with native front-ends.

An unanticipated result of the research was the discovery of a defense against an adversary who has succeeded in spoofing a TLS server certificate. Spoofing a certificate is difficult, but not unheard of. The defense, which relies on a form of mutual cryptographic authentication, prevents a man-in-the-middle attack and helps the user detect that a server controlled by the adversary is masquerading as a legitimate server using the spoofed certificate.

We have written all this up in a technical whitepaper,

The paper is quite long, because we thought it was important to describe everything in one place, showing how it all fits together. It would be difficult to discuss the entire paper at once, but in the next few posts I will go one by one over some of the topics in the paper; hopefully that will make it easier to discuss each topic. Watch for the next post in a few days.

Consistent Results from Inconsistent Data

This post is a continuation of my report on the NIST Cryptographic Key Management Workshop. In the previous post I said I had more to say about Rene Struik’s presentation [1] because there are multiple connections between his talk and our own presentation [2].

Dr. Struik proposed to use a physical uncloneable function (PUF) to generate a cryptographic key when the device is turned on. A PUF is a function implemented by a physical device that is difficult to predict and difficult to duplicate in a different device. One way of implementing a PUF is by applying stimuli to a device such as a chip, and reading responses that are not determined by the design of the chip, but depend instead on small random variations of the fabrication process within the tolerances of the process specification. For key generation, one stimulus is applied to the device, viz. power, and one response is produced, an uncloneable key.

One problem with this key generation method is that the response of the chip will be slightly different each time the device is turned on. Struik proposed an elegant method of removing these differences in order to obtain the same key consistently, and using these same differences as an entropy source for true random number generation.

I saw an interesting connection between Struick’s uncloneable key and the hardware key used by Apple for data protection in iOS, which I discussed during my talk.

Both keys are intended to be difficult to extract from the device where they are used. The hardware key is built into the silicon of a hardware encryption chip, whose interface does not allow the key to be exported from the chip; but no claims are made as to the chip being tamper resistant, so a well-equipped attacker may be able to read the key by probing the silicon. By contrast, the uncloneable key is not present in the device when the device is turned off and thus cannot be extracted by probing the silicon; but custom code running on the device could obtain the key when the device is turned on.

The two methods could be combined. For example, a device could contain an encryption chip that would generate an uncloneable key when power is applied, and the interface of the chip could prevent the key from being exported. However, custom code running on the device could still make use of the key, just like forensic tools are able to use the hardware key in the iPhone in brute-force attacks to crack the user’s passcode [3] [4].

Another connection between the two presentations is that they both proposed regenerating keys instead of storing them. In Struik’s presentation a symmetric key was regenerated from the physical properties of the device. In our presentation an RSA key pair was regenerated from user input (a PIN and/or a biometric sample). Our regeneration method is not applicable if there is no user, while Struik’s is; on the other hand, when used in a user-controlled mobile device, our method has the advantage that the key pair caonnot be obtained by custom code running on a device that has been lost or stolen.

But the most interesting connection, for me, was between Struik’s method for consistently producing the same uncloneable key from a succession of slightly different physical responses to power-on, and the method of Hao, Anderson and Daugman [5] for producing a consistent biometric key from a succession of genuine but slightly different iris image samples. (In our presentation, we proposed to use the method of Hao et al. in two- and/or three-factor authentication.) Both methods are based on the same ingenious adaptation of the error correction paradigm for producing consistent results from inconsistent data (which was pioneered by Juels and Wattenberg [6] for implementing fuzzy commitments as explained below).

Error correction techniques were originally intended, and are most often used, for correcting errors caused by the transmission of data over a noisy channel. Redundancy added to the data at the source makes it possible to correct a small number of errors at the destination. (A unit of data to which redundancy has been added is called a codeword.)

The small variations that occur between successive biometric samples or between successive samples of an uncloneable key are similar to errors produced by channel noise, but there is no correct source data to which redundancy could be added to eliminate errors. If fact, no errors are involved because, although one particular sample may serve as a reference, no sample is more correct than any other sample.

To remove data variations, error correction is used as follows:

  1. A random codeword c is generated by adding redundancy to a random string. The random codeword is of same length as a data sample, but is otherwise unrelated to any data sample.
  2. The random codeword c is bitwise x-ored with a reference sample r to produce helper data h:
    h = c ⊕ r.
  3. When a new sample s is obtained, it is bitwise x-ored with the helper data to produce
    (c ⊕ r) ⊕ s.
    Since x-or is associative, this is the same as
    c ⊕ (r ⊕ s)
    and since r and s are similar, i.e. differ only in a few bits, the bit-string
    d = r ⊕ s
    consists mostly of zeros. Bitwise x-oring c with d flips a few bits in c, thus having the same effect that could be produced by transmitting c over a noisy channel.
  4. Error correction is applied to recover the random codeword c from c ⊕ d. If desired, c can then be bitwise x-ored with h = c ⊕ r to obtain r.

What this accomplishes is that the same result is consistently produced from the variable sample s, as long as s in not too different from r. Either c or r can be used as the consistent result. Struik uses r, but I believe he could equivalently use c. Hao et al. use c, as their biometric key. In biometric applications, it is important to use c rather than r to avoid revealing the user’s biometric sample r, which could be a privacy violation. The codeword c contains no biometric information.

Juels and Wattenberg [6] used the same technique for implementing fuzzy commitments; as far as I can tell, they were the original inventors of the technique. In a cryptographic commitment scheme, a sender commits to a value by combining it with a witness to produce a blob, and sends the blob to a receiver. The sender can later reveal the value by providing the witness, which the receiver uses to compute the blob and verify that it coincides with the one received earlier. The receiver cannot derive the committed value from the blob without the witness, and only one value, the original committed value, can be successfully revealed by the sender. Juels and Wattenberg used the above technique to devise a fuzzy commitment scheme that tolerates small variations in the witness. With the above notations (different from their own), the committed value is the codeword c, the witness is r, and the blob is an ordered pair consisting of a conventional cryptographic hash of c and the helper data h. When the sender provides s to decommit the value, the receiver computes c as above and verifies that the hash of the computed c coincides with the hash in the blob received from the sender.

In their paper [6], Juels and Wattenberg made the connection with biometric authentication, but not with physical uncloneable functions.

References

[1] Rene Struik. Secure Key Storage and True Random Number Generation – An Overview. NIST Cryptographic Key Management Workshop, September 2012.
 
[2] Francisco Corella and Karen Lewison. Key Management Challenges of Derived Credentials and Techniques for Addressing Them. NIST Cryptographic Key Management Workshop, September 2012.
 
[3] Andrey Belenko. Overcoming iOS Data Protection to Re-enable iPhone Forensics. Black Hat USA, July-August 2011.
 
[4] Jean-Baptiste Bedrune and Jean Sigwald. iPhone data protection in depth. HITB Security Conference, May 2011.
 
[5] Feng Hao, Ross Anderson and John Daugman. Combining Cryptography with Biometrics Effectively. IEEE Transactions on Computers, volume 5, number 9, pages 1081 – 1088, 2006. Available as a Cambridge Computer Laboratory technical report.
 
[6] Ari Juels and Martin Wattenberg. A Fuzzy Commitment Scheme. ACM Conference on Computer and Communications Security, 1999.
 

Report on the NIST Cryptographic Key Management Workshop

This is a belated report on the Cryptographic Key Management Workshop that was held by NIST on September 10-11. Karen Lewison and I went to Washington DC for the workshop, where we presented a talk on techniques for addressing the key management challenges of derived credentials.

Cryptographic key management may seem to be a dry topic, but the workshop was quite interesting, especially the second day, which looked at the future. It was attended by about 50 cryptographers, and was webcast. It began with a fascinating keynote address by Whitfield Diffie on the history of cryptographic key management. His presentation is online, but slides cannot do justice to the wealth of stories and anecdotes that he narrated.

A Framework for Designing Cryptographic Key Management Systems

The main purpose of the workshop was to discuss the current drafts of NIST Special Publication 800-130, and NIST Special Publication 800-152 and solicit comments on them. (Instructions for sending comments on draft NIST publications can be found at http://csrc.nist.gov/publications/PubsDrafts.html.) SP 800-130 is a comprehensive framework of topics that should be considered by anybody who has to specify a Cryptographic Key Management System (CKMS); since key management is an essential aspect of cryptography, the framework should be invaluable to anybody designing a system that incorporates cryptographic functionality. SP 800-152 profiles the framework for cryptographic key management systems that will be used in US Federal agencies, but goes beyond the systems themselves to cover their procurement, installation, management, and operation.

The two publications were discussed during the first day of the workshop. I cannot possibly go over the very detailed discussions that took place, so I will limit myself to repeating one comment I made regarding Section 4.7 of SP 800-130, “Anonymity, Unlinkability and Unobservability”, and expanding upon it.

Anonymity, unlinkability and unobservability are privacy features that may not be directly relevant to the authentication of Federal employees in the course of their work, but they are very relevant to the authentication of both consumers on the Web at large, and citizens who access Federal information systems. Traditional authentication by username and password provides these three privacy features; but passwords have well-known security and usability drawbacks, one of them being the difficulty of remembering many different passwords. One way of reducing the number of passwords to be remembered is to rely on a third-party identity provider (IdP), so that one password (presented to the IdP) can be used to authenticate to any number of relying parties. The Federal Government allows citizens to access government web sites through redirection to several Approved Identity Providers.

But third party login has privacy drawbacks. In usual implementations, anonymity is lost because the relying party learns the user’s identity at the IdP, unlinkability is lost by the use of that identity at multiple relying parties, and unobservability is lost because the IdP is informed of the user’s logins. Profiles of third-party login protocols approved for citizen login to government sites mitigate some of these drawbacks by asking the identity provider to provide different identities for the same user to different relying parties. This mitigates the loss of anonymity, and the loss of unlinkability to a certain extent. (Relying parties by themselves cannot track the user, but they can track the user in collusion with the IdP.) But the loss of unobservability is not mitigated, because the IdP is still informed of the user’s activities.

I believe that the Government should work to develop and promote authentication methods that eliminate passwords while preserving anonymity, unlinkability and unobservability. Cryptographic authentication with a key pair, using different key pairs for different relying parties, can be a basis for such methods.

A Look at the Future

The second day of the workshop featured presentations on capabilities of future cryptographic key management systems, ranging from innovative to futuristic. (Both days’ presentations can be found in the workshop web page.)

Tim Polk, manager of the Cryptographic Technology Group at NIST, motivated the talks that followed by going over challenges identified during the development of the CKMS framework, related to interoperability across security domains, algorithmic agility, constrained devices, privacy, and scalability. He also stressed the need to develop CKMSs that are resilient to quantum computing attacks before it is too late.

Dennis Branstad of NIST discussed security policies, stating as a goal their automated specification, negotiation and enforcement.

Anna Lysyanskaya of Brown University discussed her work on anonymous credentials. She mentioned a new technique for revocation of anonymous credentials that was presented at Crypto 2012 by Libert, Peters and Yung, and said she thought it deserved the best paper award. I believe a full version of the conference paper can be found at http://eprint.iacr.org/2012/442. I haven’t read the paper yet. Revocation of privacy-enhancing credentials is practically difficult; I have discussed the topic in several earlier posts.

Paul Lambert of Marvell Semiconductors discussed authentication and privilege management for devices connected by wireless area networks. I was glad to hear him propose the use of a raw key pair as a credential. I later proposed the same thing in the talk on derived credentials.

Lily Chen of NIST discussed the difficult key management problem of handing over a secure link as a smart phone travels from one network to another, when the networks use technologies that may be as different as UMTS and WiFi.

Sarbari Gupta of Electrosoft discussed key management in a cloud environment. She argued that the Federal Risk and Authorization Management Program (FedRAMP) does not have sufficient requirements for secure key management, and advocated the establishment of a Federal Profile for Cloud Key Management.

Elaine Barker of NIST went over the intricacies and subtleties of random bit generation, and solicited comments on Draft Special Publication 800-90B (entropy sources) and Draft Special Publication 800-90C (RBG Constructions, DRBGs and NRBGs). Comments are due December 3rd.

Rene Struik discussed a method of secure key storage and true random number generation using physical unclonable functions (PUFs). The idea is to use accidental properties of a device to generate a unique key when the device is turned on. (So I would say that his technique is closer to key generation than key storage.) Error correction is used to remove minor differences in subsequent key generations. As an additional benefit, those differences are used for random number generation. This very interesting work is related in multiple ways to our own work on mobile authentication and derived credentials; I plan to discuss it in more detail in the next blog post.

Mary Theofanos of NIST went over two case studies of usability of key management procedures: a PKI deployment, and a PIV pilot. My personal getaways: the designer of a key management system must know the users and their mental models of security; must provide multiple authentication methods, e.g. by retaining username-password as a backup for a cryptographic credential; and must not require frequent PIN changes.

The usability talk was followed by a panel that presented three use cases of cross-domain interactions. Bob Griffin of RSA discussed key management in the cloud. Saikat Saha of SafeNet discussed virtualized hardware security modules. John Leiseboer of Quintessencelabs discussed quantum key distribution; this was the first presentation I’ve attended related to quantum cryptography, and it motivated me to find out more about this futuristic topic.

Derived Credentials

Finally, I gave a presentation on mobile authentication and derived credentials, co-authored with Karen Lewison. Even though this was the last presentation at the end of a long day of talks, I was gratified that, as far as I know, nobody snuck out early to the Dogfish Head brewery across the street from the NIST campus 🙂 . Derived credentials is a NIST concept referring to credentials that, in the future, will be installed in a mobile device after the user of the device authenticates with a PIV card. Our presentation went over three techniques for implementing derived credentials that we proposed earlier in a blog post and a white paper, viz. public key cryptography without certificates, key pair regeneration as an alternative to tamper resistance, and encapsulation of cryptographic and biometric processing in a “prover black box” and a “verifier black box” to insulate app developers from the complexities of cryptography and biometrics.

But we also went beyond derived credentials, in response to a request made by Elaine Barker on behalf of Dennis Branstad before the workshop. We discussed extensions of our techniques, for authentication across security domains, for social login without passwords, and for data protection at rest without tamper resistance. Since then we have put online a whitepaper on the data protection work. We have not yet written whitepapers on authentication across security domain or social login without passwords.

Wrap-up

Tim Polk wrapped up the workshop by encouraging everybody to send comments. Although there is an official comment period for each draft publication, NIST welcomes comments at any time.

Like the workshop on privacy-enhancing technology I attended last year, this workshop was both enjoyable and very useful. I’m glad to be on the email distribution list, and I’m looking forward to the next cryptography workshop at NIST.

Effective Data Protection for Mobile Devices

With iOS 4 Apple introduced a new method for protecting the data stored in an iPhone, an iPad or an iPod when the device is lost or stolen and the user has locked the device with a passcode (usually a 4-digit PIN) [1]. Different categories of data are encrypted under different keys, which are arranged in a key hierarchy. Some of the keys in the hierarchy are derived from the user’s passcode. However, every key that is derived from the passcode is also derived from a hardware key, which is hardwired into the silicon of a “hardware encryption chip” and cannot be extracted from the device by a casual attacker. This means that an attacker who steals a mobile device, opens it, and extracts data encrypted under the passcode cannot mount an offline attack against the passcode. That is, the attacker cannot try passcodes at high speed to find the one that decrypts the data, because trying passcodes requires the hardware key.

But the protection provided by the hardware encryption key, at least in iOS 4 and 5, has been defeated [2] [3] [4]. An attacker who steals an iPhone does not need to open it. The attacker can exploit an OS vulnerability or a firmware vulnerability to run custom code on the device. The custom code can read the encrypted data and, although it cannot extract the hardware key, it can use it to mount an “offline” attack on the device itself. The practical effect of the hardware key is only to force the attacker to run the attack on the CPU of the device, which is relatively slow. But it only takes 40 minutes to try all 4-digit PINs [2].

Would the hardware encryption key provide effective protection if Apple succeeded in eliminating all vulnerabilities that make it possible to run custom code? (It does not seem to have eliminated them in iOS 6, used in the recently released iPhone 5 [5].) An attack would be more difficult, but probably still possible. The attacker would have to open the phone to read the contents of the flash memory. The attacker might be able to not just read but also modify the contents of the flash memory, and thus run custom code, which could then use the hardware key to mount an offline attack against the passcode. The attacker might even be able to probe the silicon of the hardware encryption chip and extract the hardware key. Nobody has claimed to have done any of these fancy attacks, but that’s because exploiting vulnerabilities is much easier.

In the white paper

we have proposed a data protection method for mobile devices that does not require a hardware key, or any tamper resistance, and is effective even if the attacker is able to run custom code on the device. The data is encrypted under a (symmetric) data encryption key that is entrusted to an online server provided by the mobile network operator, or by the mobile device manufacturer, or by the provider of the mobile operating system, or by an independent data protection service provider trusted by the user. (The key can also be split using Shamir’s secret sharing technique into n “portions” entrusted to n different servers, so that k portions are needed to reconstruct the key. For example, with n = 5 and k = 3, portions of the key can be distributed to 5 different servers, and any 3 of those portions can be used to reconstruct the key.)

The key, or the portions of a split key, are retrieved when the user authenticates to unlock the phone, with a PIN, an iris image taken by a camera carried by the device, or both. User authentication regenerates an RSA key pair which the device uses to authenticate to the server(s). When a PIN is used, it is not vulnerable to an offline attack by an attacker who tampers with the phone; and when an iris image is used, no biometric sample or template is stored anywhere.

References

[1] Apple. iOS: Understanding data protection. October 28, 2011.
 
[2] Vladimir Katalov. ElcomSoft Breaks iPhone Encryption, Offers Forensic Access to FileSystem Dumps. March 23, 2011.
 
[3] Andrey Belenko. Overcoming iOS Data Protection to Re-enable iPhone Forensics. 2011.
 
[4] Jean-Baptiste Bedrune and Jean Sigwald. iPhone data protection in depth.
 
[5] Emil Protalinski, The Next Web. The Apple iPhone 5 has reportedly been jailbroken. September 22, 2012.