Concise Binary Object Representation (CBOR) is data format designed for small code size and small message size. There is a need for the ability to have the basic security services defined for this data format. This document specifies how to do signatures, message authentication codes and encryption using this data format.

The source for this draft is being maintained in GitHub. Suggested changes should be submitted as pull requests at <https://github.com/cose-wg/cose-spec>. Instructions are on that page as well. Editorial changes can be managed in GitHub, but any substantial issues need to be discussed on the COSE mailing list.

This Internet-Draft is submitted in full conformance with the provisions of BCP 78 and BCP 79.

Internet-Drafts are working documents of the Internet Engineering Task Force (IETF). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet-Drafts is at http://datatracker.ietf.org/drafts/current/.

Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress."

This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (http://trustee.ietf.org/license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Simplified BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Simplified BSD License.

1. Introduction

There has been an increased focus on the small, constrained devices that make up the Internet of Things (IOT). One of the standards that has come of of this process is the Concise Binary Object Representation (CBOR). CBOR extended the data model of the JavaScript Object Notation (JSON) by allowing for binary data among other changes. CBOR is being adopted by several of the IETF working groups dealing with the IOT world as their encoding of data structures. CBOR was designed specifically to be both small in terms of messages transport and implementation size as well having a schema free decoder. A need exists to provide basic message security services for IOT and using CBOR as the message encoding format makes sense.

The JOSE working group produced a set of documents [RFC7515][RFC7516][RFC7517][RFC7518] that defined how to perform encryption, signatures and message authentication (MAC) operations for JavaScript Object Notation (JSON) documents and then to encode the results using the JSON format [RFC7159]. This document does the same work for use with the Concise Binary Object Representation (CBOR) [RFC7049] document format. While there is a strong attempt to keep the flavor of the original JOSE documents, two considerations are taken into account:

1.1. Design changes from JOSE

1.2. Requirements Terminology

The key words “MUST”, “MUST NOT”, “REQUIRED”, “SHALL”, “SHALL NOT”, “SHOULD”, “SHOULD NOT”, “RECOMMENDED”, “NOT RECOMMENDED”, “MAY”, and “OPTIONAL” in this document are to be interpreted as described in [RFC2119].

1.3. CBOR Grammar

There currently is no standard CBOR grammar available for use by specifications. While we describe the CBOR structures in prose, they are agumented in the text by the use of the CBOR Data Definition Language (CDDL) [I-D.greevenbosch-appsawg-cbor-cddl]. The use of CDDL is intended to be explanitory. In the event of a conflict between the text and the CDDL grammar, the text is authorative. (Problems may be introduced at a later point because the CDDL grammar is not yet fixed.)

CDDL productions that together define the grammar are interspersed in the document like this:

The collected CDDL can be extracted from the XML version of this document via the following XPath expression below. (Depending on the XPath evaluator one is using, it may be necessary to deal with > as an entity.)

1.4. CBOR Related Terminology

In JSON, maps are called objects and only have one kind of map key: a string. In COSE, we use both strings and integers (both negative and non-negative integers) as map keys, as well as data items to identify specific choices. The integers (both positive and negative) are used for compactness of encoding and easy comparison. (Generally, in this document the value zero is going to be reserved and not used.) Since the work "key" is mainly used in its other meaning, as a cryptographic key, we use the term "label" for this usage of either an integer or a string to identify map keys and choice data items.

1.5. Document Terminology

Key management is used as a term to describe how a key at level n is obtained from level n+1 in encrypted and MACed messages. The term is also used to discuss key life cycle management, this document does not discuss key life cycle operations.

1.6. Mandatory to Implement Algorithms

It is expected that COSE is going to be used in a wide variety of applications and on a wide variety of devices. Many of the constrained devices are going to be setup to used a small fixed set of algorithms, and this set of algorithms may not match those available on a device. We therefore have deferred to the application protocols the decision of what to specify for mandatory algorithms.

Since the set of algorithms in an environment of constrained devices may depend on what the set of devices are and how long they have been in operation, we want to highlight that application protocols will need to specify some type of discovery method of algorithm capabilities. The discovery method may be as simple as requiring preconfiguration of the set of algorithms to providing a discovery method built into the protocol. S/MIME provided a number of different ways to approach the problem:

2. The COSE_MSG structure

The COSE_MSG structure is a top level CBOR object that corresponds to the DataContent type in the Cryptographic Message Syntax (CMS) [RFC5652]. This structure allows for a top level message to be sent that could be any of the different security services. The security service is identified within the message.

The COSE_Tagged_MSG CBOR type takes the COSE_MSG and prepends a CBOR tag of TBD1 to the encoding of COSE_MSG. By having both a tagged and untagged version of the COSE_MSG structure, it becomes easy to either use COSE_MSG as a top level object or embedded in another object. The tagged version allows for a method of placing the COSE_MSG structure into a choice, using a consistent tag value to determine that this is a COSE object.

The existence of the COSE_MSG and COSE_Tagged_MSG CBOR data types are not intended to prevent protocols from using the individual security primitives directly. Where only a single service is required, that structure can be used directly.

Each of the top-level security objects use a CBOR map as the base structure. Items in the map at the top level are identified by a label. The type of the value associated with the label is determined by the definition of the label.

The set of labels present in a security object is not restricted to those defined in this document. However, it is not recommended that additional fields be added to a structure unless this is going to be done in a closed environment. When new fields need to be added, it is recommended that a new message type be created so that processing of the field can be ensured. Using an older structure with a new field means that any security properties of the new field will not be enforced. Before a new field is added at the outer level, strong consideration needs to be given to defining a new header field and placing it into the protected headers. Applications should make a determination if non-standardized fields are going to be permitted. It is suggested that libraries allow for an option to fail parsing if non-standardized fields exist, this is especially true if they do not allow for access to the fields in other ways.

Implementations MUST be prepared to find an integer under this label that does not correspond to the values 1 to 3. If this is found then the client MUST stop attempting to parse the structure and fail. The value of 0 is reserved and not to be used. If the value of 0 is found, then clients MUST fail processing the structure. Implementations need to recognize that the set of values might be extended at a later date, but they should not provide a security service based on guesses of what is there.

NOTE: Is there any reason to allow for a marker of a COSE_Key structure and allow it to be a COSE_MSG? Doing so does allow for a security risk, but may simplify the code. [CREF6]JLS: OPEN ISSUE

The top level of each of the COSE message structures are encoded as maps. We use an integer to distinguish between the different security message types. By searching for the integer under the label identified by msg_type (which is in turn an integer), one can determine which security message is being used and thus what syntax is for the rest of the elements in the map.

COSE Map Labels
name	number	comments
msg_type	1	Occurs only in top level messages
protected	2	Occurs in all structures
unprotected	3	Occurs in all structures
payload	4	Contains the content of the structure
signatures	5	For COSE_Sign - array of signatures
signature	6	For COSE_signature only
ciphertext	4	TODO: Should we reuse the same as payload or not?
recipients	9	For COSE_encrypt and COSE_mac
tag	10	For COSE_mac only

The CDDL grammar that provides the label values is:

; message_labels
msg_type=1
protected=2
unprotected=3
payload=4
signatures=5
signature=6
ciphertext=4
recipients=9
tag=10

3. Header Parameters

The structure of COSE has been designed to have two buckets of information that are not considered to be part of the payload itself, but are used for holding information about algorithms, keys, or evaluation hints for the processing of the layer. These two buckets are available for use in all of the structures in this document except for keys. While these buckets can be present, they may not all be usable in all instances. For example, while the protected bucket is defined as part of recipient structures, most of the algorithms that are used for recipients do not provide the necessary functionality to provide the needed protection and thus the bucket should not be used.

Both buckets are implemented as CBOR maps. The map key is a 'label' (Section 1.4). The value portion is dependent on the definition for the label. Both maps use the same set of label/value pairs. The integer range for labels has been divided into several sections with a standard range, a private range, and a range that is dependent on the algorithm selected. The defined labels can be found in the 'COSE Header Parameters' IANA registry (Section 15.3).

Two buckets are provided for each layer: [CREF7]JLS: A completest version of this grammar would list the options available in the protected and unprotected headers. Do we want to head that direction?

protected: contains parameters about the current layer that are to be cryptographically protected. This bucket MUST NOT be used if it is not going to be included in a cryptographic computation. This bucket is encoded in the message as a binary object. This value is obtained by CBOR encoding the protected map and wrapping it in a bstr object. This wrapping allows for the encoding of the protected map to be transported with a greater chance that it will not be altered in transit. (Badly behaved intermediates could decode and re-encode, but this will result in a failure to verify unless the re-encoded byte string is identical to the decoded byte string.) This finesses the problem of all parties needing to be able to do a common connical encoding.
unprotected: contains parameters about the current layer that are not cryptographically protected.

Only parameters that deal with the current layer are to be placed at that layer. As an example of this, the parameter 'content type' describes the content of the message being carried in the message. As such this parameter is placed only the the content layer and is not placed in the key managment or signature layers. In principle, one should be able to process any given layer without reference to any other layer. (The only data that should need to cross layers is the cryptographic key.)

The presence of both buckets is optional, however the requirement that the 'alg' parameter be present at each level effectively imposes a requirement that one of the buckets will always be present. The parameters that go into the buckets come from the IANA "COSE Header Parameters" (Section 15.3). Some common parameters are defined in the next section, but a number of parameters are defined throughout this document.

The CDDL fragment that describes the two buckets is:

header_map = {+ label => any }

Headers = (
    ? protected => bstr,                ; Contains a header_map
    ? unprotected => header_map
)

3.1. COSE Headers

This document defines a set of common header parameters. A summary of the parameters defined in this section can be found in Table 2. This table should be consulted to determine the value of label used as well as the type of the value.

The set of header parameters defined in this section are:

alg

This parameter is used to indicate the algorithm used for the security processing. This parameter MUST be present at each level of a signed, encrypted or authenticated message. The value is taken from the 'COSE Algorithm Registry' (see Section 15.4).

crit

This parameter is used to ensure that applications will take appropriate action based on the values found. The parameter is used to indicate which protected header labels an application that is processing a message is required to understand. The value is an array of COSE Header Labels. When present, this parameter MUST be placed in the protected header bucket.

Integer labels in the range of 0 to 10 SHOULD be omitted.
Integer labels in the range -1 to -255 can be omitted as they are algorithm dependent. If an application can correctly process an algorithm, it can be assumed that it will correctly process all of the parameters associated with that algorithm. (The algorithm range is -1 to -65536, it is assumed that the higher end will deal with more optional algorithm specific items.)

The header parameter values indicated by 'crit' can be processed by either the security library code or by an application using a security library, the only requirement is that the parameter is processed.

content type

This parameter is used to indicate the content type of the data in the payload or ciphertext fields. [CREF8]JLS: After looking at this, I am wondering if the type for this should be: [int int]/[int tstr] so that we can keep the major/minor difference of media-types. This does cost a couple of bytes in the message. Integers are from the XXXXX[CREF9]JLS: Need to figure out how we are going to go about creating this registry -or are we going to modify the current mime-content table? IANA registry table. Strings are from the IANA 'mime-content types' registry. Applications SHOULD provide this parameter if the content structure is potentially ambiguous.

kid

This parameter one of the ways that can be used to find the key to be used. The value of this parameter is matched against the 'kid' field in a COSE_Key structure. Applications MUST NOT assume that 'kid' values are unique. There may be more than one key with the same 'kid' value, it may be required that all of the keys need to be checked to find the correct one. The internal structure of 'kid' values is not defined and generally cannot be relied on by applications. Key identifier values are hints about which key to use, they are not directly a security critical field, for this reason they can normally be placed in the unprotected headers bucket.

nonce

This parameter holds either a nonce or Initialization Vector value. The value can be used either as a counter value for a protocol or as an IV for an algorithm. TODO: Talk about zero extending the value in some cases. [CREF10]JLS: Open to do.

Common Header Parameters
name	label	value type	value registry	description
alg	1	int / tstr	COSE Algorithm Registry	Integers are taken from table --POINT TO REGISTRY--
crit	2	[+ label]	COSE Header Label Registry	integer values are from -- POINT TO REGISTRY --
content type	3	tstr / int	media-types registry	Value is either a media-type or an integer from the media-type registry
jku	*	tstr		URL to COSE key object
jwk	*	COSE_Key		contains a COSE key not a JWK key
kid	4	bstr		key identifier
nonce	5	bstr		Nonce or Initialization Vector (IV)
x5c	*	bstr*		X.509 Certificate Chain
x5t	*	bstr		SHA-1 thumbprint of key
x5t#S256	*	bstr		SHA-256 thumbprint of key
x5u	*	tstr		URL for X.509 certificate
zip	*	int / tstr		Integers are taken from the table --POINT TO REGISTRY--

OPEN ISSUES:

Which of the following items do we want to have standardized in this document: jku, jwk, x5c, x5t, x5t#S256, x5u, zip
I am currently torn on the question "Should epk and iv/nonce be algorithm specific or generic headers?" They are really specific to an algorithm and can potentially be defined in different ways for different algorithms. As an example, it would make sense to defined nonce for CCM and GCM modes that can have the leading zero bytes stripped, while for other algorithms this might be undesirable.
We might want to define some additional items. What are they? A possible example would be a sequence number as this might be common. On the other hand, this is the type of things that is frequently used as the nonce in some places and thus should not be used in the same way. Other items might be challenge/response fields for freshness as these are likely to be common.

4. Signing Structure

The signature structure allows for one or more signatures to be applied to a message payload. There are provisions for parameters about the content and parameters about the signature to be carried along with the signature itself. These parameters may be authenticated by the signature, or just present. Examples of parameters about the content would be the type of content, when the content was created, and who created the content. Examples of parameters about the signature would be the algorithm and key used to create the signature, when the signature was created, and counter-signatures.

When more than one signature is present, the successful validation of one signature associated with a given signer is usually treated as a successful signature by that signer. However, there are some application environments where other rules are needed. An application that employs a rule other than one valid signature for each signer must specify those rules. Also, where simple matching of the signer identifier is not sufficient to determine whether the signatures were generated by the same signer, the application specification must describe how to determine which signatures were generated by the same signer. Support of different communities of recipients is the primary reason that signers choose to include more than one signature. For example, the COSE_Sign structure might include signatures generated with the RSA signature algorithm and with the Elliptic Curve Digital Signature Algorithm (ECDSA) signature algorithm. This allows recipients to verify the signature associated with one algorithm or the other. (The original source of this text is [RFC5652].) More detailed information on multiple signature evaluation can be found in [RFC5752].

The CDDL grammar for a signature message is:

COSE_Sign = {
    msg_type => msg_type_signed,
    Headers,
    ? payload => bstr,
    signatures => [+ COSE_signature]
}

The fields is the structure have the following semantics:

msg_type: identifies this as providing the signed security service. The value MUST be msg_type_signed (1).
protected: is described in Section 3.
unprotected: is described in Section 3.
payload: contains the serialized content to be signed. If the payload is not present in the message, the application is required to supply the payload separately. The payload is wrapped in a bstr to ensure that it is transported without changes. If the payload is transported separately, it is the responsibility of the application to ensure that it will be transported without changes.
signatures: is an array of signature items. Each of these items uses the COSE_signature structure for its representation.

We use the values in Table 1 as the labels in the COSE_Sign map. While other labels can be present in the map, it is not generally a recommended practice. The other labels can be either of integer or string type, use of other types SHOULD be treated as an error.

The CDDL grammar structure for a signature is:

COSE_signature =  {
    Headers,      
    signature => bstr
}

The fields in the structure have the following semantics:

protected: is described in Section 3.
unprotected: is described in Section 3.
signature: contains the computed signature value.

4.1. Externally Supplied Data

One of the features that we supply in the COSE document is the ability for applications to provide additional data to be authenticated as part of the security, but that is not carried as part of the COSE object. The primary reason for supporting this can be seen by looking at the CoAP message struture [RFC7252] where the facility exists for options to be carried before the payload. An example of data that can be placed in this location would be transaction ids and nonces to check for replay protection. If the data is in the options section, then it is available for routers to help in performing the replay detection and prevention. However, it may also be desired to protect these values so that they cannot be modified in transit. This is the purpose of the externally supplied data field.

This document describes the process for using a byte array of externally supplied authenticated data, however the method of constructing the byte array is a function of the application. Applications which use this feature need to define how the externally supplied authenticated data is to be constructed. Such a construction needs to take into account the following issues:

If multiple items are included, care needs to be taken that data cannot bleed between the items. This is usually addressed by making fields fixed width and/or encoding the length of the field. Using options from CoAP as an example, these fields use a TLV structure so they can be concatenated without any problems.
If multiple items are included, a defined order for the items needs to be defined. Using options from CoAP as an example, an application could state that the fields are to be ordered by the option number.

4.2. Signing and Verification Process

The COSE structure used to create the byte stream to be signed uses the following CDDL grammar structure:

Sig_structure = [
    body_protected: bstr,
    sign_protected: bstr,
    external_aad: bstr,
    payload: bstr
]

How to compute a signature:

Create a Sig_structure object and populate it with the appropriate fields. For body_protected and sign_protected, if the fields are not present in their corresponding maps, an bstr of length zero is used.
If the application has supplied external additional authenticated data to be included in the computation, then it is placed in the 'external_aad' field. If no data was supplied, then a zero length binary value is used.
Create the value ToBeSigned by encoding the Sig_structure to a byte string.
Call the signature creation algorithm passing in K (the key to sign with), alg (the algorithm to sign with) and ToBeSigned (the value to sign).
Place the resulting signature value in the 'signature' field of the map.

How to verify a signature:

Create a Sig_structure object and populate it with the appropriate fields. For body_protected and sign_protected, if the fields are not present in their corresponding maps, an bstr of length zero is used.
If the application has supplied external additional authenticated data to be included in the computation, then it is placed in the 'external_aad' field. If no data was supplied, then a zero length binary value is used.
Create the value ToBeSigned by encoding the Sig_structure to a byte string.
Call the signature verification algorithm passing in K (the key to verify with), alg (the algorithm to sign with), ToBeSigned (the value to sign), and sig (the signature to be verified).

In addition to performing the signature verification, one must also perform the appropriate checks to ensure that the key is correctly paired with the signing identity and that the appropriate authorization is done.

5. Encryption object

In this section we describe the structure and methods to be used when doing an encryption in COSE. In COSE, we use the same techniques and structures for encrypting both the plain text and the keys used to protect the text. This is different from the approach used by both [RFC5652] and [RFC7516] where different structures are used for the plain text and for the different key management techniques.

One of the byproducts of using the same technique for encrypting and encoding both the content and the keys using the various key management techniques, is a requirement that all of the key management techniques use an Authenticated Encryption (AE) algorithm. (For the purpose of this document we use a slightly loose definition of AE algorithms.) When encrypting the plain text, it is normal to use an Authenticated Encryption with Additional Data (AEAD) algorithm. For key management, either AE or AEAD algorithms can be used. See Appendix A for more details about the different types of algorithms. [CREF11]Ilari: I don't follow/understand this text

The CDDL grammar structure for encryption is:

COSE_encrypt = {
    msg_type=>msg_type_encrypted,
    COSE_encrypt_fields
}

COSE_encrypt_fields = (
    Headers,
    ? ciphertext => bstr,
    ? recipients => [+{COSE_encrypt_fields}]
)

Description of the fields:

msg_type: identifies this as providing the encrypted security service. The value MUST be msg_type_encrypted (2).
protected: is described in Section 3.
unprotected: is described in Section 3.
ciphertext: contains the encrypted plain text. If the ciphertext is to be transported independently of the control information about the encryption process (i.e. detached content) then the field is omitted.
recipients: contains the recipient information. It is required that at least one recipient MUST be present for the content encryption layer.

5.1. Key Management Methods

A typical encrypted message consists of an encrypted content and an encrypted CEK for one or more recipients. The content-encryption key is encrypted for each recipient. The details of this encryption depends on the key management technique used, but the six generally techniques are:

none:: The CEK is the same as as the identified previously distributed symmetric key.
symmetric key-encryption keys:: The CEK is encrypted using a previously distributed symmetric key-encryption key.
key agreement:: the recipient's public key and a sender's private key are used to generate a pairwise symmetric key, then the CEK is either the derived key or encrypted by the derived key.
key transport:: the CEK is encrypted in the recipient's public key
passwords:: the CEK is encrypted in a key-encryption key that is derived from a password or other shared secret value.

Section 12 provides details on a number of different key management algorithms and discusses which parameters need to be present for each of the key management techniques.

5.2. Encryption Algorithm for AEAD algorithms

The encryption algorithm for AEAD algorithms is fairly simple. In order to get a consistent encoding of the data to be authenticated, the Enc_structure is used to have canonical form of the AAD.

Enc_structure = [
    protected: bstr,
    external_aad: bstr
]

Copy the protected header field from the message to be sent.
If the application has supplied external additional authenticated data to be included in the computation, then it is placed in the 'external_aad' field. If no data was supplied, then a zero length binary value is used. (See Section 4.1 for application guidance on constructing this field.)
Encode the Enc_structure using a CBOR Canonical encoding Section 14 to get the AAD value.
Determine the encryption key. This step is dependent on the key management method being used: For:
No Recipients:
The key to be used is determined by the algorithm and key at the current level.
Direct and Direct Key Agreement:
The key is determined by the key and algorithm in the recipient structure. The encryption algorithm and size of the key to be used are inputs into the KDF used for the recipient. (For direct, the KDF can be thought of as the identity operation.)
Other:
The key is randomly generated.
Call the encryption algorithm with K (the encryption key to use), P (the plain text) and AAD (the additional authenticated data). Place the returned cipher text into the 'ciphertext' field of the structure.
For recipients of the message, recursively perform the encryption algorithm for that recipient using the encryption key as the plain text.

5.3. Encryption algorithm for AE algorithms

Verify that the 'protected' field is absent.
Verify that there was no external additional authenticated data supplied for this operation.
Determine the encryption key. This step is dependent on the key management method being used: For:
No Recipients:
The key to be used is determined by the algorithm and key at the current level.
Direct and Direct Key Agreement:
The key is determined by the key and algorithm in the recipient structure. The encryption algorithm and size of the key to be used are inputs into the KDF used for the recipient. (For direct, the KDF can be thought of as the identity operation.)
Other:
The key is randomly generated.
Call the encryption algorithm with K (the encryption key to use) and the P (the plain text). Place the returned cipher text into the 'ciphertext' field of the structure.
For recipients of the message, recursively perform the encryption algorithm for that recipient using the encryption key as the plain text.

6. MAC objects

In this section we describe the structure and methods to be used when doing MAC authentication in COSE. This document allows for the use of all of the same methods of key management as are allowed for encryption.

When using MAC operations, there are two modes in which it can be used. The first is just a check that the content has not been changed since the MAC was computed. Any of the key management methods can be used for this purpose. The second mode is to both check that the content has not been changed since the MAC was computed, and to use key management to verify who sent it. The key management modes that support this are ones that either use a pre-shared secret, or do static-static key agreement. In both of these cases the entity MACing the message can be validated by a key binding. (The binding of identity assumes that there are only two parties involved and you did not send the message yourself.)

COSE_mac = {
   msg_type=>msg_type_mac, 
   Headers,
   ? payload => bstr,
   tag => bstr,
   recipients => [+{COSE_encrypt_fields}]
}

Field descriptions:

msg_type: identifies this as providing the encrypted security service. The value MUST be msg_type_mac (3).
protected: is described in Section 3.
unprotected: is described in Section 3.
payload: contains the serialized content to be MACed. If the payload is not present in the message, the application is required to supply the payload separately. The payload is wrapped in a bstr to ensure that it is transported without changes, if the payload is transported separately it is the responsibility of the application to ensure that it will be transported without changes.
tag: contains the MAC value.
recipients: contains the recipient information. See the description under COSE_Encryption for more info.

MAC_structure = [
     protected: bstr,
     external_aad: bstr,
     payload: bstr
]

How to compute a MAC:

Create a MAC_structure and copy the protected and payload fields from the COSE_mac structure.
If the application has supplied external authenticated data, encode it as a binary value and place in the MAC_structure. If there is no external authenticated data, then use a zero length 'bstr'. (See Section 4.1 for application guidance on constructing this field.)
Encode the MAC_structure using a canonical CBOR encoder. The resulting bytes is the value to compute the MAC on.
Compute the MAC and place the result in the 'tag' field of the COSE_mac structure.
Encrypt and encode the MAC key for each recipient of the message.

7. Key Structure

A COSE Key structure is built on a CBOR map object. The set of common parameters that can appear in a COSE Key can be found in the IANA registry 'COSE Key Common Parameter Registry' (Section 15.6). Additional parameters defined for different key types can be found in the IANA registry 'COSE Key Type Parameters' (Section 15.7).

A COSE Key Set uses a CBOR array object as it's underlying type. The values of the array elements are COSE Keys. A Key Set MUST have at least one element in the array. [CREF12]JLS: Is there a reason to assign a CBOR tag to identify keys and/or key sets?

The CDDL grammar describing a COSE_Key and COSE_KeySet is: [CREF13]JLS: We can really simplify the grammar for COSE_Key to be just the kty (the one required field) and the generic item. The reason to do this is that it makes things simpler. The reason not to do this says that we really need to add a lot more items so that a grammar check can be done that is more tightly enforced.

COSE_Key = {
    key_kty => tstr / int,
    ? key_ops => [+ (tstr / int) ],
    ? key_alg => tstr / int,
    ? key_kid => bstr,
    * label => values
}

COSE_KeySet = [+COSE_Key]

The element “kty” is a required element in a COSE_Key map.

7.1. COSE Key Common Parameters

This document defines a set of common parameters for a COSE Key object. Table 3 provides a summary of the parameters defined in this section. There are also a set of parameters that are defined for a specific key type. Key type specific parameters can be found in Section 13.

Key Map Labels
name	label	CBOR type	registry	description
kty	1	tstr / int	COSE General Values	Identification of the key type
key_ops	4	[* (tstr/int)]		Restrict set of permissible operations
alg	3	tstr / int	COSE Algorithm Values	Key usage restriction to this algorithm
kid	2	bstr		Key Identification value - match to kid in message
x5u	*	tstr
x5c	*	bstr*
x5t	*	bstr
x5t#S256	*	bstr
use	*	tstr		deprecated - don't use

kty:: This parameter is used to identify the family of keys for this structure, and thus the set of key type specific parameters to be found. The set of values can be found in Table 21.
alg:: This parameter is used to restrict the algorithms that are to be used with this key. If this parameter is present in the key structure, the application MUST verify that this algorithm matches the algorithm for which the key is being used. If the algorthms do not match, then this key object MUST NOT be used to perform the cryptographic operation. Note that the same key can be in a different key structure with a different or no algorithm specified, however this is considered to be a poor security practice.
kid:: This parameter is used to give an identifier for a key. The identifier is not structured and can be anything from a user provided string to a value computed on the public portion of the key. This field is intended for matching against a 'kid' parameter in a message in order to filter down the set of keys that need to be checked.
key_ops:: This parameter is defined to restrict the set of operations that a key is to be used for. The value of the field is an array of values from Table 4.

Only the 'kty' field MUST be present in a key object. All other parameters can be omitted if their behavior is not needed.

Key Operation Values
name	value	description
sign	1	The key is used to create signatures. Requires private key fields.
verify	2	The key is used for verification of signatures.
encrypt	3	The key is used for key transport encryption.
decrypt	4	The key is used for key transport decryption. Requires private key fields.
wrap key	5	The key is used for key wrapping.
unwrap key	6	The key is used for key unwrapping. Requires private key fields.
key agree	7	The key is used for key agreement.

The following provides a CDDL fragment which duplicates the assignment labels from Table 3 and Table 4.

;key_labels
key_kty=1
key_kid=2
key_alg=3
key_ops=4

;key_ops values
key_ops_sign=1
key_ops_verify=2
key_ops_encrypt=3
key_ops_decrypt=4
key_ops_wrap=5
key_ops_unwrap=6
key_ops_agree=7

8. Signature Algorithms

There are two basic signature algorithm structures that can be used. The first is the common signature with appendix. In this structure, the message content is processed and a signature is produced, the signature is called the appendix. This is the message structure used by our common algorithms such as ECDSA and RSASSA-PSS. (In fact the SSA in RSASSA-PSS stands for Signature Scheme with Appendix.) The basic structure becomes:

        
signature = Sign(message content, key)

valid = Verification(message content, key, signature)

The second is a signature with message recovery. (An example of such an algorithm is [PVSig].) In this structure, the message content is processed, but part of is included in the signature. Moving bytes of the message content into the signature allows for an effectively smaller signature, the signature size is still potentially large, but the message content is shrunk. This has implications for systems implementing these algoritms and for applications that use them. The first is that the message content is not fully available until after a signature has been validated. Until that point the part of the message contained inside of the signature is unrecoverable. The second is that the security analysis of the strength of the signature is very much based on the structure of the message content. Messages which are highly predictable require additional randomness to be supplied as part of the signature process, in the worst case it because the same as doing a singature with appendix. Thirdly, in the event that multple signatures are applied to a message, all of the signature algorithms are going to be required to consume the same number of bytes of message content.

        
signature, message sent = Sign(message content, key)

valid, message content = Verification(message sent, key, signature)

At this time, only signatures with appendixes are defined for use with COSE, however considerable interest has been expressed in using a signature with message recovery algorithm due to the effective size reduction that is possible. Implementations will need to keep this in mind for later possible integration.

8.1. ECDSA

ECDSA [DSS] defines a signature algorithm using ECC.

The ECDSA signature algorithm is parameterized with a hash function (h. In the event that the length of the hash function output is greater than group of the key, the left most bytes of the hash output are used.

The algorithms defined in this document can be found in Table 5.

ECDSA Algorithm Values
name	value	hash	description
ES256	-7	SHA-256	ECDSA w/ SHA-256
ES384	-8	SHA-384	ECDSA w/ SHA-384
ES512	-9	SHA-512	ECDSA w/ SHA-512

In order to promote interoperability, it is suggested that SHA-256 be used only with keys of length 256, SHA-384 be used only with keys of length 384 and SHA-512 be used only with keys of length 521. This is aligned with the recommendation in Section 4 of [RFC5480].

The signature algorithm results in a pair of integers (R, S). These integers will be of the same order as length of the key used for the signature process. The signature is encoded by converting the integers into byte strings of the same length as the key size. The length is rounded up to the nearest byte and is left padded with zero bits to get to the correct length. The two integers are then concatenated together to form a byte string that is the resulting signature.

Using the function defined in [RFC3447] the signature is:
Signature = I2OSP(R, n) | I2OSP(S, n)
where n = ceiling(key_length / 8)

8.1.1. Security Considerations

The security strength of the signature is no greater than the minimum of the security strength associated with the bit length of the key and the security strength of the hash function. When a hash function is used that has greater security than is provided by the length of the key, the signature algorithm uses the leftmost key length bits of the hash function output.

System which have poor random number generation can leak their keys by signing two messages with the same value of 'k'. [RFC6979] provides a method to deal with this problem by making 'k' be deterministic based on the message content rather than randomly generated. Applications which specify ECDSA should evaluate the ability to get good random number generation and recommend this when it is not possible. Note: Use of this technique even when good random number generation exists may still be a good idea.

There are two substitution that can theoretically be mounted against the ECDSA signature algorithm.

Changing the curve used to validate the signature: If one changes the curve used to validate the signature, then potentially one could have a two messages with the same signature each computed under a different curve. The only requirement on the new curve is that it's order be the same as the old one and it be acceptable to the client. An example would be to change from using the curve P-256 to using Curve25519. (Both are 256 bit curves.) We current do not have any way to deal with this version of the attack except to restrict the overall set of curves that can be used.
Change the hash function used to validate the signature: If one has either two different hash functions of the same length, or one can truncate a hash function down, then one could potentially find collisions between the hash functions rather than within a single hash function. (For example, truncating SHA-512 to 256 bits might collide with a SHA-256 bit hash value.) This attack can be mitigated by including the signature algorithm identifier in the data to be signed.

8.2. RSASSA-PSS

The RSASSA-PSS signature algorithm is defined in [RFC3447].

The RSASSA-PSS signature algorithm is parametized with a hash function (h), a mask generation function (mgf) and a salt length (sLen). For this specification, the mask generation function is fixed to be MGF1 as defined in [RFC3447]. It has been recommended that the same hash function be used for hashing the data as well as in the mask generation function, for this specification we following this recommendation. The salt length is the same length as the hash function output.

The algorithms defined in this document can be found in Table 6.

RSASSA-PSS Algorithm Values
name	value	hash	salt length	description
PS256	-26	SHA-256	32	RSASSA-PSS w/ SHA-256
PS384	-27	SHA-384	48	RSASSA-PSS w/ SHA-384
PS512	-28	SHA-512	64	RSASSA-PSS w/ SHA-512

8.2.1. Security Considerations

In addition to needing to worry about keys that are too small to provide the required security, there are issues with keys that are too large. Denial of service attacks have been mounted with overly large keys. This has the potential to consume resources with potentially bad keys. There are two reasonable ways to address this attack. First, a key should not be used for a cryptographic operation until it has been matched back to an authorized user. This approach means that no cryptography would be done except for authorized users. Second, applications can impose maximum as well as minimum length requirements on keys. This limits the resources consumed even if the matching is not performed until the cryptography has been done.

There is a theoretical hash substitution attack that can be mounted against RSASSA-PSS. However, the requirement that the same hash function be used consistently for all operations is an effective mitigation against it. Unlike ECDSA, hash functions are not truncated so that the full hash value is always signed. The internal padding structure of RSASSA-PSS means that one needs to have multiple collisions between the two hash functions in order to be successful in producing a forgery based on changing the hash function. This is highly unlikely.

9. Message Authentication (MAC) Algorithms

Message Authentication Codes (MACs) provide data authentication and integrity protection. They provide either no or very limited data origination. (One cannot, for example, be used to prove the identity of the sender to a third party.)

MAC algorithms can be based on either a block cipher algorithm (i.e. AES-MAC) or a hash algorithm (i.e. HMAC). This document defines a MAC algorithm for each of these two constructions.

9.1. Hash-based Message Authentication Codes (HMAC)

The Hash-base Message Authentication Code algorithm (HMAC) [RFC2104][RFC4231] was designed to deal with length extension attacks. The algorithm was also designed to allow for new hash algorithms to be directly plugged in without changes to the hash function. The HMAC design process has been vindicated as, while the security of hash algorithms such as MD5 has decreased over time, the security of HMAC combined with MD5 has not yet been shown to be compromised [RFC6151].

The HMAC algorithm is parameterized by an inner and outer padding, a hash function (h) and an authentication tag value length. For this specification, the inner and outer padding are fixed to the values set in [RFC2104]. The length of the authentication tag corresponds to the difficulty of producing a forgery. For use in constrained environments, we define a set of HMAC algorithms that are truncated. There are currently no known issues when truncating, however the security strength of the message tag is correspondingly reduced in strength. When truncating, the left most tag length bits are kept and transmitted.

The algorithm defined in this document can be found in Table 7.

HMAC Algorithm Values
name	value	Hash	Length	description
HMAC 256/64	*	SHA-256	64	HMAC w/ SHA-256 truncated to 64 bits
HMAC 256/256	4	SHA-256	256	HMAC w/ SHA-256
HMAC 384/384	5	SHA-384	384	HMAC w/ SHA-384
HMAC 512/512	6	SHA-512	512	HMAC w/ SHA-512

For some key management methods, the length of the key is known or determinable from the key management method. For example, if RSA-OAEP is used then the key will be output at the correct length. However, if any of the key derivation methods are used, then the size of the key to be obtained is an input parameter to the key derivation step. For all HMAC methods defined in this document, the key size for a key derivation methods MUST be the same size as the hash function used. It is RECOMMENDED that the key size be the same size as the hash function for all other key management methods.

9.1.1. Security Considerations

HMAC has proved to be resistant even when used with weakening hash algorithms. The current best method appears to be a brute force attack on the key. This means that key size is going to be directly related to the security of an HMAC operation.

9.2. AES Message Authentication Code (AES-CBC-MAC)

AES-CBC-MAC is defined in [MAC].

AES-CBC-MAC is parameterized by the key length, the authentication tag length and the IV used. For all of these algorithms, the IV is fixed to all zeros. We provide an array of algorithms for various key lengths and tag lengths. The algorithms defined in this document are found in Table 8.

AES-MAC Algorithm Values
name	value	key length	tag length	description
AES-MAC 128/64	*	128	64	AES-MAC 128 bit key, 64-bit tag
AES-MAC 256/64	*	256	64	AES-MAC 256 bit key, 64-bit tag
AES-MAC 128/128	*	128	128	AES-MAC 128 bit key, 128-bit tag
AES-MAC 256/128	*	256	128	AES-MAC 256 bit key, 128-bit tag

9.2.1. Security Considerations

A number of attacks exist against CBC-MAC that need to be considered.

A single key must only be used for messages of a fixed and known length. If this is not the case, an attacker will be able to generated a message with a valid tag given two message, tag pairs. This can be addressed by using different keys for different length messages. (CMAC mode also addresses this issue.)
If the same key is used for both encryption and authentication operations, using CBC modes an attacker can produce messages with a valid authentication code.
If the IV can be modified, then messages can be forged. This is addressed by fixing the IV to all zeros.

10. Content Encryption Algorithms

10.1. AES GCM

The GCM mode is is a generic authenticated encryption block cipher mode defined in [AES-GCM]. The GCM mode is combined with the AES block encryption algorithm to define a an AEAD cipher.

The GCM mode is parameterized with by the size of the authentication tag. The size of the authentication tag is limited to a small set of values. For this document however, the size of the authentication tag is fixed at 128-bits.

The set of algorithms defined in this document are in Table 9.

Algorithm Value for AES-GCM
name	value	description
A128GCM	1	AES-GCM mode w/ 128-bit key
A192GCM	2	AES-GCM mode w/ 192-bit key
A256GCM	3	AES-GCM mode w/ 256-bit key

10.1.1. Security Considerations

When using AES-CCM the following restrictions MUST be enforced:

The key and nonce pair MUST be unique for every message encrypted.
The total amount of data encrypted MUST NOT exceed 2^39 − 256 bits . An explicit check is required only in environments where it is expected that it might be exceeded.

10.2. AES CCM

Counter with CBC-MAC (CCM) is a generic authentication encryption block cipher mode defined in [RFC3610]. The CCM mode is combined with the AES block encryption algorithm to define a commonly used content encryption algorithm used in constrainted devices.

The CCM mode has two parameter choices. The first choice is M, the size of the authentication field. The choice of the value for M involves a trade-off between message expansion and the probably that an attacker can undetecably modify a message. The second choice is L, the size of the length field. This value requires a trade-off between the maximum message size and the size of the Nonce.

It is unfortunate that the specification for CCM specified L and M as a count of bytes rather than a count of bits. This leads to possible misunderstandings where AES-CCM-8 is frequently used to refer to a version of CCM mode where the size of the authentication is 64-bits and not 8-bits. These values have traditionally been specified as bit counts rather than byte counts. This document will follow the tradition of using bit counts so that it is easier to compare the different algorithms presented in this document.

We define a matrix of algorithms in this document over the values of L and M. Constrained devices are usually operating in situations where they use short messages and want to avoid doing key management operations. This favors smaller values of M and larger values of L. Less constrained devices do will want to be able to user larger messages and are more willing to generate new keys for every operation. This favors larger values of M and smaller values of L. (The use of a large nonce means that random generation of both the key and the nonce will decrease the chances of repeating the pair on two different messages.)

The following values are used for L:

16-bits (2): limits messages to 2^16 bytes (64Kbyte) in length. This sufficently long for messages in the constrainted world. The nonce length is 13 bytes allowing for 2^(13*8) possible values of the nonce without repeating.
64-bits (8): limits messages to 2^64 byes in length. The nonce length is 7 bytes allowing for 2^56 possible values of the nonce without repeating.

The following values are used for M:

64-bits (8): produces a 64-bit authentication tag. This implies that there is a 1 in 2^64 chance that an modified message will authenticate.
128-bits (16): produces a 128-bit authentication tag. This implies that there is a 1 in 2^128 chance that an modified message will authenticate.

Algorithm Values for AES-CCM
name	value	L	M	k	description
AES-CCM-16-64-128	10	16	64	128	AES-CCM mode 128-bit key, 64-bit tag, 13-byte nonce
AES-CCM-16-64-256	11	16	64	256	AES-CCM mode 256-bit key, 64-bit tag, 13-byte nonce
AES-CCM-64-64-128	30	64	64	128	AES-CCM mode 128-bit key, 64-bit tag, 7-byte nonce
AES-CCM-64-64-256	31	64	64	256	AES-CCM mode 256-bit key, 64-bit tag, 7-byte nonce
AES-CCM-16-128-128	12	16	128	128	AES-CCM mode 128-bit key, 128-bit tag, 13-byte nonce
AES-CCM-16-128-256	13	16	128	256	AES-CCM mode 256-bit key, 128-bit tag, 13-byte nonce
AES-CCM-64-128-128	32	64	128	128	AES-CCM mode 128-bit key, 128-bit tag, 7-byte nonce
AES-CCM-64-128-256	33	64	128	256	AES-CCM mode 256-bit key, 128-bit tag, 7-byte nonce

10.2.1. Security Considerations

When using AES-CCM the following restrictions MUST be enforced:

The key and nonce pair MUST be unique for every message encrypted.
The total number of times the AES block cipher is used MUST NOT exceed 2^61 operations. This limitation is the sum of times the block cipher is used in computing the MAC value and in performing stream encryption operations. An explicit check is required only in environments where it is expected that it might be exceeded.

[RFC3610] additionally calls out one other consideration of note. It is possible to do a pre-computation attack against the algorithm in cases where the portions encryption content is highly predictable. This reduces the security of the key size by half. Ways to deal with this attack include adding a random portion to the nonce value and/or increasing the key size used. Using a portion of the nonce for a random value will decrease the number of messages that a single key can be used for. Increasing the key size may require more resources in the constrained device. See sections 5 and 10 of [RFC3610] for more information.

10.3. ChaCha20 and Poly1305

ChaCha20 and Poly1305 combined together is a new AEAD mode that is defined in [RFC7539]. This is a new mode defined to be a cipher which is not AES and thus would not suffer from any future weaknesses found in AES. These cryptographic functions are designed to be fast in software only implementations.

The ChaCha20/Poly1305 AEAD construction defined in [RFC7539] has no parameterization. It takes a 256-bit key and an a 96-bit nonce as well as the plain text and additional data as inputs and produces the cipher text as an option. We define one algorithm identifier for this algorithm in Table 11.

Algorithm Value for AES-GCM
name	value	description
ChaCha20/Poly1305	11	ChaCha20/Poly1305 w/ 256-bit key

10.3.1. Security Considerations

The pair of key, nonce MUST be unique for every invocation of the algorithm. Nonce counters are considered to be an acceptable way of ensuring that they are unique.

11. Key Derivation Functions (KDF)

11.1. HMAC-based Extract-and-Expand Key Derivation Function (HKDF)

The HKDF key derivation algorithm is defined in [RFC5869].

The HKDF algorithm is defined to take a number of inputs These inputs are:

secret - a shared value that is secret. Secrets may be either previously shared or derived from operations like a DH key agreement.
salt - an optional public value that is used to change the generation process. If specified, the salt is carried using the 'salt' algorithm parameter. While [RFC5869] suggests that the length of the salt be the same as the length of the underlying hash value, any amount of salt will improve the security as different key values will be generated. A parameter to carry the salt is defined in Table 13. This parameter is protected by being included in the key computation and does not need to be separately authenticated.
length - the number of bytes of output that need to be generated.
context information - Information that describes the context in which the resulting value will be used. Making this information specific to the context that the material is going to be used ensures that the resulting material will always be unique. The context structure used is encoded into the algorithm identifier.
hash function - The underlying hash function to be used in the HKDF algorithm. The hash function is encoded into the HKDF algorithm selection.

HKDF is defined to use HMAC as the underlying PRF. However, it is possible to use other functions in the same construct to provide a different KDF function that may be more appropriate in the constrained world. Specifically, one can use AES-CBC-MAC as the PRF for the expand step, but not for the extract step. When using a good random shared secret of the correct length, the extract step can be skipped. The extract cannot be skipped if the secret is not uniformly random, for example if it is the result of a ECDH key agreement step.

The algorithms defined in this document are found in Table 12

HKDF algorithms
name	hash	Skip extract	context
HKDF SHA-256	SHA-256	no	XXX
HKDF SHA-512	SHA-512	no	XXX
HKDF AES-MAC-128	AES-CBC-128	yes	HKDF using AES-MAC as the PRF w/ 128-bit key
HKDF AES-MAC-256	AES-CBC-128	yes	HKDF using AES-MAC as the PRF w/ 256-bit key

HKDF Algorithm Parameters
name	label	type	description
salt	-20	bstr	Random salt

11.2. Context Information Structure

The context information structure is used to ensure that the derived keying material is "bound" to the context of the transaction. The context information structure used here is based on that defined in [SP800-56A]. By using CBOR for the encoding of the context information structure, we automatically get the same type of type and length separation of fields that is obtained by the use of ASN.1. This means that there is no need to encode the lengths for the base elements as it is done by the CBOR encoding.

The context information structure refers to PartyU and PartyV as the two parties which are doing the key derivation. Unless the application protocol defines differently, we assign PartyU to the entity that is creating the message and PartyV to the entity that is receiving the message. This is because we are assuming a set of stand alone store and forward messaging processes. In [SP800-56A], PartyU is the initiator and PartyV is the responder. The specification is written with the idea of on-line protocols rather than store and forward protocols as the main consumer.

Application protocols are free to define the roles differently. For example, they could assign the PartyU role to the entity that initiates the connection and allow directly sending multiple messages over the line without changing the role information.

Using the PartyU and PartyV fields is the easiest way to get different keys in each direction. The use of a transaction identifier, either in one of the supplemental fields or as the salt if one is using HKDF, ensures that a unique key is generated for each set of transactions. Combining nonce fields with the transaction identifier provides a method so that a different key is used for each message in each direction.

We encode the context specific information using a CBOR array type. For the fields that we define an algorithm parameter, the details of the parameters can be found in Table 14. The fields in the array are:

AlgorithmID

This field indicates the algorithm for which the key material will be used. This field is required to be present and is a copy of the algorithm identifier in the message. The field exists in the context information so that if the same environment is used for different algorithms, then completely different keys will be generated each of those algorithms. (This practice means if algorithm A is broken and thus can is easier to find, the key derived for algorithm B will not be the same as the key for algorithm B.)

PartyUInfo

This field holds information about party U. The ParytUInfo structure is divided into three pieces:

identity: This contains the identity information for party U. The identities can be assigned in one of two manners. Firstly, a protocol can assign identities based on roles. For example, the roles of "client" and "server" may be assigned to different entities in the protocol. Each entity would then use the correct label for the data they they send or receive. The second way is for a protocol to assign identities is to use a name based on a naming system (i.e. DNS, X.509 names).
We define an algorithm parameter 'PartyU identity' that can be used to carry identity information in the message. However, identity information is often known as part of the protocol and can thus be inferred rather than made explicit. If identity information is carried in the message, applications SHOULD have a way of validating the supplied identity information. The identity information does not need to be specified and can be left as absent.
The identity value supplied will be integrity checked as part of the key derivation process. If the identity string is wrong, then the wrong key will be created.
nonce: This contains a one time nonce value. The nonce can either be implicit from the protocol or carried as a value in the unprotected headers.
We define an algorithm parameter 'PartyU nonce' that can be used to carry this value in the message However, the nonce value could be determined by the application and the value determined from elsewhere.
This item is optional and can be absent.
other: This contains other information that is defined by the protocol.
This item is optional and can be absent.

PartyVInfo

M00TODO: Copy down from PartyUInfo when that text is ready.

SuppPubInfo

This field contains public information that is mutually known to both parties.

keyDataLength: This is set to the number of bits of the desired output value. (This practice means if algorithm A can use two different key lengths, the key derived for longer key size will not contain the key for shorter key size as a prefix.)
other: The field other is for free form data defined by the application. An example is that an application could defined two different strings to be placed here to generate different keys for a data stream vs a control stream. This field is optional and will only be present if the application defines a structure for this information. Applications that define this SHOULD use CBOR to encode the data so that types and lengths are correctly include.

SuppPrivInfo

This field contains private information that is mutually known information. An example of this information would be a pre-existing shared secret. The field is optional and will only be present if the application defines a structure for this information. Applications that define this SHOULD use CBOR to encode the data so that types and lengths are correctly include.


COSE_KDF_Context = [
    AlgorithmID : int / tstr,
    PartyUInfo : [
        ? nonce : bstr / int,
        ? identity : bstr,
        ? other : bstr
    ],
    PartyVInfo : [
        ? nonce : bstr,
        ? identity : bstr / tstr,
        ? other : bstr
    ],
    SuppPubInfo : [
        keyDataLength : uint,
        ? other : bstr
    ],
    ? SuppPrivInfo : bstr
]

Context Algorithm Parameters
name	label	type	description
PartyU identity	-21	bstr	Party U identity Information
PartyU nonce	-22	bstr / int	Party U provided nonce
PartyU other	-23	bstr	Party U other provided information
PartyV identity	-24	bstr	Party V identity Information
PartyV nonce	-25	bstr / int	Party V provided nonce
PartyV other	-26	bstr	Party V other provided information

12. Key Management Algorithms

There are a number of different key management methods that can be used in the COSE encryption system. In this section we will discuss each of the key management methods, what fields need to be specified, and which algorithms are defined in this document to deal with each of them.

The names of the key management methods used here are the same as are defined in [RFC7517]. Other specifications use different terms for the key management methods or do not support some of the key management methods.

At the moment we do not have any key management methods that allow for the use of protected headers. This may be changed in the future if, for example, the AES-GCM Key wrap method defined in [RFC7518] were extended to allow for authenticated data. In that event, the use of the 'protected' field, which is current forbidden below, would be permitted.

12.1. Direct Encryption

In direct encryption mode, a shared secret between the sender and the recipient is used as the key. When direct encryption mode is used, it MUST be the only mode used on the message. It is a massive security leak to have both direct encryption and a different key management mode on the same message.

For JOSE, direct encryption key management is the only key management method allowed for doing MACed messages. In COSE, all of the key management methods can be used for MACed messages.

The COSE_encrypt structure for the recipient is organized as follows:

The 'protected', 'ciphertext' and 'recipients' fields MUST be absent.
At a minimum, the 'unprotected' field MUST contain the 'alg' parameter and SHOULD contain a parameter identifying the shared secret.

12.1.1. Direct Key

This key management technique is the simplest method, the supplied key is directly used as the key for the next layer down in the message. There are no algorithm parameters defined for this key management methods. The algorithm identifier assignment can be found in Table 15.

Direct Key
name	value	description
direct	-6	Direct use of CEK

12.1.1.1. Security Considerations

The direct key management technique has several potential problems that need to be considered:

These keys need to have some method to be regularly updated over time. All of the content encryption algorithms specified in this document have limits on how many times a key can be used without significant loss of security.
These keys need to be dedicated to a single algorithm. There have been a number of attacks developed over time when a single key is used for multiple different algorithms. One example of this is the use of a single key both for CBC encryption mode and CBC-MAC authentication mode.
Breaking one message means all messages are broken. If an adversary succeeds in determining the key for a single message, then the key for all messages is also determined.

12.1.2. Direct Key with KDF

This key managment takes a common shared secret between the two parties and applies the HKDF function (Section 11.1) using the context structure defined in Section 11.2 to transform the shared secret into the necessary key. Either the 'salt' parameter of HKDF or the partyU 'nonce' parameter of the context structure MUST be present. This parameter can be generated either randomly or deterministically, the requirement is that it be a unique value for the key pair in question.

If the salt/nonce value is generated randomly, then it is suggested that the length of the random value be the same length as the hash function underlying HKDF, i.e 256-bits. While there is no way to guarantee that it will be unique, there is a high probability that it will be unique. If the salt/nonce value is generated deterministically, it can be guaranteed to be unique and thus there is no length requirement.

Since with this technique a new key can be generated for every message, the restrictions on IVs can frequently be relaxed. For the content encryption algorithms used in this document IVs must be unique for a specific key. If the key is altered then the IV can be re-used. Alternatively, an application can be the IV be generated from the same context as the key is by changing the algorithm identifier to the string "IV-GENERATION".

The set of algorithms defined in this document can be found in Table 16.

Direct Key
name	value	KDF	description
direct+KDF-SHA-256	*	HKDF SHA-256	Shared secret w/ KDF
direct+KDF-SHA-512	*	HKDF SHA-512	Shared secret w/ KDF
direct+KDF-AES-128	*	HKDF AES-MAC-128	Shared secret w/ AES-MAC 128-bit key
direct+KDF-AES-256	*	HKDF AES-MAC-256	Shared secret w/ AES-MAC 256-bit key

12.1.2.1. Security Considerations

The shared secret need to have some method to be regularly updated over time. The shared secret is forming the basis of trust, although not used directly it should still be subject to scheduled rotation.

12.2. Key Wrapping

In key wrapping mode, the CEK is randomly generated and that key is then encrypted by a shared secret between the sender and the recipient. All of the currently defined key wrapping algorithms for JOSE (and thus for COSE) are AE algorithms. Key wrapping mode is considered to be superior to direct encryption if the system has any capability for doing random key generation. This is because the shared key is used to wrap random data rather than data has some degree of organization and may in fact be repeating the same content.

The COSE_encrypt structure for the recipient is organized as follows:

The 'protected' field MUST be absent if the key wrap algorithm is an AE algorithm.
The 'recipients' field is normally absent, but can be used. Applications MUST deal with a recipients field present, not being able to decrypt that recipient is an acceptable way of dealing with it. Failing to process the message is not an acceptable way of dealing with it.
The plain text to be encrypted is the key from next layer down (usually the content layer).
At a minimum, the 'unprotected' field MUST contain the 'alg' parameter and SHOULD contain a parameter identifying the shared secret.

12.2.1. AES Key Wrapping

The AES Key Wrapping algorithm is defined in [RFC3394]. This algorithm uses an AES key to wrap a value that is a multiple of 64-bits, as such it can be used to wrap a key for any of the content encryption algorithms defined in this document. [CREF14]JLS: Do we also want to document the use of RFC 5649 as well? It allows for other sizes of keys that might be used for HMAC - i.e. a 200 bit key. The algorithm exists, but I do not personally know of any standard uses of it. The algorithm requires a single fixed parameter, the initial value. This is fixed to the value specified in Section 2.2.3.1 of [RFC3394]. There are no public parameters that vary on a per invocation basis.

AES Key Wrap Algorithm Values
name	value	key size	description
A128KW	-3	128	AES Key Wrap w/ 128-bit key
A192KW	-4	192	AES Key Wrap w/ 192-bit key
A256KW	-5	256	AES Key Wrap w/ 256-bit key

12.2.1.1. Security Considerations for AES-KW

12.3. Key Encryption

Key Encryption mode is also called key transport mode in some standards. Key Encryption mode differs from Key Wrap mode in that it uses an asymmetric encryption algorithm rather than a symmetric encryption algorithm to protect the key. This document defines one Key Encryption mode algorithm.

When using a key encryption algorithm, the COSE_encrypt structure for the recipient is organized as follows:

The 'protected' field MUST be absent.
The plain text to be encrypted is the key from next layer down (usually the content layer).
At a minimum, the 'unprotected' field MUST contain the 'alg' parameter and SHOULD contain a parameter identifying the asymmetric key.

12.3.1. RSAES-OAEP

RSAES-OAEP is an asymmetric key encryption algorithm. The defintion of RSAEA-OAEP can be find in Section 7.1 of [RFC3447]. The algorithm is parameterized using a masking generation function (mgf), a hash function (h) and encoding parameters (P). For the algorithm identifiers defined in this section: Table 18 summarizes the rest of the values.

mgf is always set to MFG1 from [RFC3447] and uses the same hash function as h.
P is always set to the empty octet string.

RSAES-OAEP Algorithm Values
name	value	hash	description
RSAES-OAEP w/SHA-256	-25	SHA-256	RSAES OAEP w/ SHA-256
RSAES-OAEP w/SHA-512	-26	SHA-512	RSAES OAEP w/ SHA-512

12.3.1.1. Security Considerations for RSAES-OAEP

A key size of 2048 bits or larger MUST be used with these algorithms. This key size corresponds roughly to the same strength as provided by a 128-bit symmetric encryption algorithm.

It is highly recommended that checks on the key length be done before starting a decryption operation. One potential denial of service operation is to provide encrypted objects using either abnormally long or oddly sized RSA modulus values. Implementations SHOULD be able to encrypt and decrypt with modulus between 2048 and 16K bits in length.[CREF15]JLS: Is this range we want to specify? Applications can impose additional restrictions on the length of the modulus.

12.4. Direct Key Agreement

When using the 'Direct Key Agreement' key managment method, the two parties use a key agreement method to create a shared secret. A KDF is then applied to the shared secret to derive a key to be used in protecting the data. This key is normally used as a CEK or MAC key, but could be used for other purposes if more than two layers are in use (see Appendix B).

The most commonly used key agreement algorithm used is Diffie-Hellman, but other variants exist. Since COSE is designed for a store and forward environment rather than an on-line environment, many of the DH variants cannot be used as the receiver of the message cannot provide any key material. One side-effect of this is that perfect forward security is not achievable, a static key will always be used for the receiver of the COSE message.

Two variants of DH that are easily supported are:

- Ephemeral-Static DH: where the sender of the message creates a one time DH key and uses a static key for the recipient. The use of the ephemeral sender key means that no additional random input is needed as this is randomly generated for each message.
Static-Static DH: where a static key is used for both the sender and the recipient. The use of static keys allows for recipient to get a weak version of data origination for the message. When static-static key agreement is used, then some piece of unique data is require to ensure that a different key is created for each message

In this specification, both variants are specified. This has been done to provide the weak data origination option for use with MAC operations.

When direct key agreement mode is used, it MUST be the only key management mode used on the message and there MUST be only one recipient. This method creates the key directly and that makes it difficult to mix with additional recipients. If multiple recipients are needed, then the version with key wrap needs to be used.

The COSE_encrypt structure for the recipient is organized as follows:

The 'protected' field MUST be absent.
At a minimum, the 'unprotected' field MUST contain the 'alg' parameter and SHOULD contain a parameter identifying the recipient's asymmetric key.
The 'unprotected' field MUST contain the 'epk' parameter.

12.4.1. ECDH

The basic mathematics for Elliptic Curve Diffie-Hellman can be found in [RFC6090]. Two new curves have been defined in [I-D.irtf-cfrg-curves].

ECDH is parameterized by the following:

Curve Type/Curve: The curve selected controls not only the size of the shared secret, but the mathematics for computing the shared secret. The curve selected also controls how a point in the curve is represented and what happens for the identity points on the curve. In this specification we allow for a number of different curves to be used. The curves are defined in Table 22.
Since the only the math is changed by changing the curve, the curve is not fixed for any of the algorithm identifiers we define, instead it is defined by the points used.
Ephemeral-static or static-static: The key agreement process may be done using either a static or an ephemeral key at the senders side. When using ephemeral keys, the sender MUST generate a new ephemeral key for every key agreement operation. The ephemeral key is placed in in the 'ephemeral key' parameter and MUST be present for all algorithm identifiers which use ephemeral keys. When using static keys, the sender MUST either generate a new random value placed in either in the KDF parameters or the context structure. For the KDF functions used, this means either in the 'salt' parameter for HKDF (Table 13) or in in the 'PartyU nonce' parameter for the context struture (Table 14) MUST be present. (Both may be present if desired.) The value in the parameter MUST be unique for the key pair being used. It is acceptable to use a global counter which is incremented for every static-static operation and use the resulting value. When using static keys, the static key needs to be identified to the recipient. The static key can be identified either by providing the key ('static key') or by providing a key identifier for the static key ('static key id'). Both of these parameters are defined in Table 20
Key derivation algorithm: The result of an ECDH key agreement process does not provide a uniformly random secret, as such it needs to be run through a KDF in order to produce a usable key. Processing the secret through a KDF also allows for the introduction of both context material, how the key is going to be used, and one time material in the even to of a static-static key agreement.
Key Wrap algorithm: The key wrap algorithm can be 'none' if the result of the KDF is going to be used as the key directly. This option, along with static-static, should be used if knowledge about the sender is desired. If 'none' is used then the content layer encryption algorithm size is value fed to the context structure. Support is also provided for any of the key wrap algorithms defined in section Section 12.2.1. If one of these options is used, the input key size to the key wrap algorithm is the value fed into the context structure as the key size.

The set of algorithms direct ECDH defined in this document are found in Table 19.

ECDH Algorithm Values
name	value	KDF	Ephemeral-Static	Key Wrap	description
ECDH-ES + HKDF-256	50	HKDF - SHA-256	yes	none	ECDH ES w/ HKDF - generate key directly
ECDH-ES + HKDF-512	51	HKDF - SHA-256	yes	none	ECDH ES w/ HKDF - generate key directly
ECDH-SS + HKDF-256	52	HKDF - SHA-256	no	none	ECDH ES w/ HKDF - generate key directly
ECDH-SS + HKDF-512	53	HKDF - SHA-256	no	none	ECDH ES w/ HKDF - generate key directly
ECDH-ES+A128KW	54	HKDF - SHA-256	yes	A128KW	ECDH ES w/ Concat KDF and AES Key wrap w/ 128 bit key
ECDH-ES+A192KW	55	HKDF - SHA-256	yes	A192KW	ECDH ES w/ Concat KDF and AES Key wrap w/ 192 bit key
ECDH-ES+A256KW	56	HKDF - SHA-256	yes	A256KW	ECDH ES w/ Concat KDF and AES Key wrap w/ 256 bit key
ECDH-SS+A128KW	57	HKDF - SHA-256	no	A128KW	ECDH SS w/ Concat KDF and AES Key wrap w/ 128 bit key
ECDH-SS+A192KW	58	HKDF - SHA-256	no	A192KW	ECDH SS w/ Concat KDF and AES Key wrap w/ 192 bit key
ECDH-SS+A256KW	59	HKDF - SHA-256	no	A256KW	ECDH SS w/ Concat KDF and AES Key wrap w/ 256 bit key

ECDH Algorithm Parameters
name	label	type	algorithm	description
ephemeral key	-1	COSE_Key	ECDH-ES	Ephemeral Public key for the sender
static key	-2	COSE_Key	ECDH-ES	Static Public key for the sender
static key id	-3	bstr	ECDH-SS	Static Public key identifier for the sender

12.5. Key Agreement with KDF

Key Agreement with Key Wrapping uses a randomly generated CEK. The CEK is then encrypted using a Key Wrapping algorithm and a key derived from the shared secret computed by the key agreement algorithm.

The COSE_encrypt structure for the recipient is organized as follows:

The 'protected' field MUST be absent if the key wrap algorithm is an AE algorithm. [CREF16]JLS: It would be possible to include the protected field in the KDF rather than the key wrap algorithm if we wanted to. This would provide the same level of security, it would not be possible to get the same key if they are different.
The plain text to be encrypted is the key from next layer down (usually the content layer).
At a minimum, the 'unprotected' field MUST contain the 'alg' parameter, a parameter identifying the recipient asymmetric key, and a parameter with the sender's asymmetric public key.

12.5.1. ECDH

These algorithms are defined in Table 19.

12.6. Password

[CREF17]JLS: Do we want/need to support this? JOSE did it mainly to support the encryption of private keys.

12.6.1. PBES2

name	value	description
PBES2-HS256+A128KW	*	PBES2 w/ HMAC SHA-256 and AES Key wrap w/ 128 bit key
PBES2-HS384+A192KW	*	PBES2 w/ HMAC SHA-384 and AES Key wrap w/ 192 bit key
PBES2-HS512+A256KW	*	PBES2 w/ HMAC SHA-512 and AES Key wrap w/ 256 bit key

13. Keys

The COSE_Key object defines a way to hold a single key object, it is still required that the members of individual key types be defined. This section of the document is where we define an initial set of members for specific key types.

For each of the key types, we define both public and private members. The public members are what is transmitted to others for their usage. We define private members mainly for the purpose of archival of keys by individuals. However, there are some circumstances where private keys may be distributed by various entities in a protocol. Examples include: Entities which have poor random number generation. Centralized key creation for multi-cast type operations. Protocols where a shared secret is used as a bearer token for authorization purposes.

Keys are identified by the 'kty' member of the COSE_Key object. In this document we define four values for the member.

Key Type Values
name	value	description
EC1	1	Elliptic Curve Keys w/ X Coordinate only
EC2	2	Elliptic Curve Keys w/ X,Y Coordinate pair
RSA	3	RSA Keys
Symmetric	4	Symmetric Keys

13.1. Elliptic Curve Keys

Two different key structures are being defined for Elliptic Curve keys. One version uses both an x and a y coordinate, potentially with point compression. This is the traditional EC point representation that is used in [RFC5480]. The other version uses only the x coordinate as the y coordinate is either to be recomputed or not needed for the key agreement operation. An example of this is Curve25519 [I-D.irtf-cfrg-curves].

EC Curves
name	key type	value	description
P-256	EC2	1	NIST P-256 also known as secp256r1
P-384	EC2	2	NIST P-384 also known as secp384r1
P-521	EC2	3	NIST P-521 also known as secp521r1
Curve25519	EC1	1	Provide reference
Goldilocks	EC1	2	Provide reference

13.1.1. Single Coordinate Curves

NOTE: This section represents at risk work depending on the ability to get good references for Curve25519 and Goldilocks.

New versions of ECC have been targeted at variants where only a single value of the EC Point need to be transmitted. This work is currently going on in the IRTF CFRG group.

For EC keys with both coordinates, the 'kty' member is set to 1 (EC1). The members that are defined for this key type are:

crv: contains an identifier of the curve to be used with the key. [CREF18]JLS: Do we create a registry for curves? Is is the same registry for both EC1 and EC2? The curves defined in this document for this key type can be found in Table 22. Other curves may be registered in the future and private curves can be used as well.
x: contains the x coordinate for the EC point. The octet string represents a little-endian encoding of x.
d: contains the private key.

For public keys, it is REQUIRED that 'crv' and 'x' be present in the structure. For private keys, it is REQUIRED that 'crv' and 'd' be present in the structure. It is RECOMMENDED that 'x' also be present, but it can be recomputed from the required elements and omitting it saves on space.

EC Key Parameters
name	key type	value	type	description
crv	1	-1	int / tstr	EC Curve identifier - Taken from the COSE General Registry
x	1	-2	bstr	X Coordinate
d	1	-4	bstr	Private key

13.1.2. Double Coordinate Curves

The traditional way of sending EC curves has been to send either both the x and y coordinates, or the x coordinate and a sign bit for the y coordinate. The latter encoding has not been recommend in the IETF due to potential IPR issues with Certicom. However, for operations in constrained environments, the ability to shrink a message by not sending the y coordinate is potentially useful.

For EC keys with both coordinates, the 'kty' member is set to 2 (EC2). The members that are defined for this key type are:

crv: contains an identifier of the curve to be used with the key. The curves defined in this document for this key type can be found in Table 22. Other curves may be registered in the future and private curves can be used as well.
x: contains the x coordinate for the EC point. The integer is converted to an octet string as defined in [SEC1]. Zero octets MUST NOT be removed from the front of the octet string. [CREF19]JLS: Should we use the integer encoding for x, y and d instead of bstr?
y: contains either the sign bit or the value of y coordinate for the EC point. For the value, the integer is converted to an octet string as defined in [SEC1]. Zero octets MUST NOT be removed from the front of the octet string. For the sign bit, the value is true if the value of y is positive.
d: contains the private key.

For public keys, it is REQUIRED that 'crv', 'x' and 'y' be present in the structure. For private keys, it is REQUIRED that 'crv' and 'd' be present in the structure. It is RECOMMENDED that 'x' and 'y' also be present, but they can be recomputed from the required elements and omitting them saves on space.

EC Key Parameters
name	key type	value	type	description
crv	2	-1	int / tstr	EC Curve identifier - Taken from the COSE General Registry
x	2	-2	bstr	X Coordinate
y	2	-3	bstr / bool	Y Coordinate
d	2	-4	bstr	Private key

13.2. RSA Keys

This document defines a key structure for both the public and private halves of RSA keys. Together, an RSA public key and an RSA private key form an RSA key pair. [CREF20]JLS: Looking at the CBOR specification, the bstr that we are looking in our table below should most likely be specified as big numbers rather than as binary strings. This means that we would use the tag 6.2 instead. From my reading of the specification, there is no difference in the encoded size of the resulting output. The specification of bignum does explicitly allow for integers encoded with leading zeros.

The document also provides support for the so-called "multi-prime" RSA where the modulus may have more than two prime factors. The benefit of multi-prime RSA is lower computational cost for the decryption and signature primitives. For a discussion on how multi-prime affects the security of RSA crypto-systems, the reader is referred to [MultiPrimeRSA].

This document follows the naming convention of [RFC3447] for the naming of the fields of an RSA public or private key. The table Table 25 provides a summary of the label values and the types associated with each of those labels. The requirements for fields for RSA keys are as follows:

For all keys, 'kty' MUST be present and MUST have a value of 3.
For public keys, the fields 'n' and 'e' MUST be present. All other fields defined in Table 25 MUST be absent.
For private keys with two primes, the fields 'other', 'r_i', 'd_i' and 't_i' MUST be absent, all other fields MUST be present.
For private keys with more than two primes, all fields MUST be present. For the third to nth primes, each of the primes is represented as a map containing the fields 'r_i', 'd_i' and 't_i'. The field 'other' is an array of those maps.

RSA Key Parameters
name	key type	value	type	description
n	3	-1	bstr	Modulus Parameter
e	3	-2	int	Exponent Parameter
d	3	-3	bstr	Private Exponent Parameter
p	3	-4	bstr	First Prime Factor
q	3	-5	bstr	Second Prime Factor
dP	3	-6	bstr	First Factor CRT Exponent
dQ	3	-7	bstr	Second Factor CRT Exponent
qInv	3	-8	bstr	First CRT Coefficient
other	3	-9	array	Other Primes Info
r_i	3	-10	bstr	i-th factor, Prime Factor
d_i	3	-11	bstr	i-th factor, Factor CRT Exponent
t_i	3	-12	bstr	i-th factor, Factor CRT Coefficient

13.3. Symmetric Keys

Occasionally it is required that a symmetric key be transported between entities. This key structure allows for that to happen.

For symmetric keys, the 'kty' member is set to 3 (Symmetric). The member that is defined for this key type is:

k: contains the value of the key.

This key structure contains only private key information, care must be taken that it is never transmitted accidentally. For public keys, there are no required fields. For private keys, it is REQUIRED that 'k' be present in the structure.

Symmetric Key Parameters
name	key type	value	type	description
k	4	-1	bstr	Key Value

14. CBOR Encoder Restrictions

There as been an attempt to limit the number of places where the document needs to impose restrictions on how the CBOR Encoder needs to work. We have managed to narrow it down to the following restrictions:

The restriction applies to the encoding the Sig_structure, the Enc_structure, and the MAC_structure.
The rules for Canonical CBOR (Section 3.9 of RFC 7049) MUST be used in these locations. The main rule that needs to be enforced is that all lengths in these structures MUST be encoded such that they are encoded using definite lengths and the minimum length encoding is used.
All parsers used SHOULD fail on both parsing and generation if the same label is used twice as a key for the same map.

15. IANA Considerations

15.1. CBOR Tag assignment

It is requested that IANA assign a new tag from the “Concise Binary Object Representation (CBOR) Tags” registry. It is requested that the tag be assigned in the 0 to 23 value range.

Tag Value: TBD1

Data Item: COSE_Msg

Semantics: COSE security message.

15.2. COSE Object Labels Registry

It is requested that IANA create a new registry entitled “COSE Object Labels Registry”. [CREF21]JLS: Finish the registration process.

This table is initially populated by the table in Table 1.

15.3. COSE Header Parameter Registry

It is requested that IANA create a new registry entitled “COSE Header Parameters”.

The columns of the registry are:

name: The name is present to make it easier to refer to and discuss the registration entry. The value is not used in the protocol. Names are to be unique in the table.
label: This is the value used for the label. The label can be either an integer or a string. Registration in the table is based on the value of the label requested. Integer values between 1 and 255 and strings of length 1 are designated as Standards Track Document required. Integer values from 256 to 65535 and strings of length 2 are designated as Specification Required. Integer values of greater than 65535 and strings of length greater than 2 are designated as first come first server. Integer values in the range -1 to -65536 are delegated to the “COSE Header Algorithm Label” registry. Integer values beyond -65536 are marked as private use.
value: This contains the CBOR type for the value portion of the label.
value registry: This contains a pointer to the registry used to contain values where the set is limited.
description: This contains a brief description of the header field.
specification: This contains a pointer to the specification defining the header field (where public).

The initial contents of the registry can be found in Table 2. The specification column for all rows in that table should be this document.

Additionally, the label of 0 is to be marked as 'Reserved'.

15.4. COSE Header Algorithm Label Table

It is requested that IANA create a new registry entitled “COSE Header Algorithm Labels”.

The columns of the registry are:

name: The name is present to make it easier to refer to and discuss the registration entry. The value is not used in the protocol.
algorithm: The algorithm(s) that this registry entry is used for. This value is taken from the “COSE Algorithm Value” registry. Multiple algorithms can be specified in this entry. For the table, the algorithm, label pair MUST be unique.
label: This is the value used for the label. The label is an integer in the range of -1 to -65536.
value: This contains the CBOR type for the value portion of the label.
value registry: This contains a pointer to the registry used to contain values where the set is limited.
description: This contains a brief description of the header field.
specification: This contains a pointer to the specification defining the header field (where public).

The initial contents of the registry can be found in: Table 13, Table 14, Table 20, and Appendix D. The specification column for all rows in that table should be this document.

15.5. COSE Algorithm Registry

It is requested that IANA create a new registry entitled “COSE Algorithm Registry”.

The columns of the registry are:
value: The value to be used to identify this algorithm. Algorithm values MUST be unique. The value can be a positive integer, a negative integer or a string. Integer values between 0 and 255 and strings of length 1 are designated as Standards Track Document required. Integer values from 256 to 65535 and strings of length 2 are designated as Specification Required. Integer values of greater than 65535 and strings of length greater than 2 are designated as first come first server. Integer values in the range -1 to -65536 are delegated to the “COSE Header Algorithm Label” registry. Integer values beyond -65536 are marked as private use.
description: A short description of the algorithm.
specification: A document where the algorithm is defined (if publicly available).

The initial contents of the registry can be found in the following: Table 10, Table 9, Table 5, Table 7, Table 15, Table 17, Table 18. The specification column for all rows in that table should be this document.

15.6. COSE Key Common Parameter Registry

It is requested that IANA create a new registry entitled “COSE Key Common Parameter" Registry.

The columns of the registry are:

name: This is a descriptive name that enables easier reference to the item. It is not used in the encoding.
label: The value to be used to identify this algorithm. Key map labels MUST be unique. The label can be a positive integer, a negative integer or a string. Integer values between 0 and 255 and strings of length 1 are designated as Standards Track Document required. Integer values from 256 to 65535 and strings of length 2 are designated as Specification Required. Integer values of greater than 65535 and strings of length greater than 2 are designated as first come first server. Integer values in the range -1 to -65536 are used for key parameters specific to a single algorithm delegated to the “COSE Key Parameter Label” registry. Integer values beyond -65536 are marked as private use.
CBOR Type: This field contains the CBOR type for the field
registry: This field denotes the registry that values come from, if one exists.
description: This field contains a brief description for the field
specification: This contains a pointer to the public specification for the field if one exists

This registry will be initially populated by the values in Section 7.1. The specification column for all of these entries will be this document.

15.7. COSE Key Type Parameter Registry

It is requested that IANA create a new registry “COSE Key Type Parameters”.

The columns of the table are:

key type: This field contains a descriptive string of a key type. This should be a value that is in the COSE General Values table and is placed in the 'kty' field of a COSE Key structure.
name: This is a descriptive name that enables easier reference to the item. It is not used in the encoding.
label: The label is to be unique for every value of key type. The range of values is from -256 to -1. Labels are expected to be reused for different keys.
CBOR type: This field contains the CBOR type for the field
description: This field contains a brief description for the field
specification: This contains a pointer to the public specification for the field if one exists

This registry will be initially populated by the values in Table 23, Table 24, Table 25, and Table 26. The specification column for all of these entries will be this document.

15.8. Media Type Registration

15.8.1. COSE Security Message

This section registers the "application/cose" and "application/cose+cbor" media types in the "Media Types" registry. [CREF22]JLS: Should we register both or just the cose+cbor one? These media types are used to indicate that the content is a COSE_MSG.

Type name: application
Subtype name: cose
Required parameters: N/A
Optional parameters: N/A
Encoding considerations: binary
Security considerations: See the Security Considerations section of RFC TBD.
Interoperability considerations: N/A
Published specification: RFC TBD
Applications that use this media type: To be identified
Fragment identifier considerations: N/A
Additional information:
- Magic number(s): N/A
- File extension(s): cbor
- Macintosh file type code(s): N/A
Person & email address to contact for further information: iesg@ietf.org
Intended usage: COMMON
Restrictions on usage: N/A
Author: Jim Schaad, ietf@augustcellars.com
Change Controller: IESG
Provisional registration? No

Type name: application
Subtype name: cose+cbor
Required parameters: N/A
Optional parameters: N/A
Encoding considerations: binary
Security considerations: See the Security Considerations section of RFC TBD.
Interoperability considerations: N/A
Published specification: RFC TBD
Applications that use this media type: To be identified
Fragment identifier considerations: N/A
Additional information:
- Magic number(s): N/A
- File extension(s): cbor
- Macintosh file type code(s): N/A
Person & email address to contact for further information: iesg@ietf.org
Intended usage: COMMON
Restrictions on usage: N/A
Author: Jim Schaad, ietf@augustcellars.com
Change Controller: IESG
Provisional registration? No

15.8.2. COSE Key media type

This section registers the "application/cose+json" and "application/cose-set+json" media types in the "Media Types" registry. These media types are used to indicate, respectively, that content is a COSE_Key or COSE_KeySet object.

Type name: application
Subtype name: cose-key+cbor
Required parameters: N/A
Optional parameters: N/A
Encoding considerations: binary
Security considerations: See the Security Considerations section of RFC TBD.
Interoperability considerations: N/A
Published specification: RFC TBD
Applications that use this media type: To be identified
Fragment identifier considerations: N/A
Additional information:
- Magic number(s): N/A
- File extension(s): cbor
- Macintosh file type code(s): N/A
Person & email address to contact for further information: iesg@ietf.org
Intended usage: COMMON
Restrictions on usage: N/A
Author: Jim Schaad, ietf@augustcellars.com
Change Controller: IESG
Provisional registration? No

Type name: application
Subtype name: cose-key-set+cbor
Required parameters: N/A
Optional parameters: N/A
Encoding considerations: binary
Security considerations: See the Security Considerations section of RFC TBD.
Interoperability considerations: N/A
Published specification: RFC TBD
Applications that use this media type: To be identified
Fragment identifier considerations: N/A
Additional information:
- Magic number(s): N/A
- File extension(s): cbor
- Macintosh file type code(s): N/A
Person & email address to contact for further information: iesg@ietf.org
Intended usage: COMMON
Restrictions on usage: N/A
Author: Jim Schaad, ietf@augustcellars.com
Change Controller: IESG
Provisional registration? No

16. Security Considerations

There are security considerations:

Protect private keys
MAC messages with more than one recipient means one cannot figure out who sent the message
Use of direct key with other recipient structures hands the key to other recipients.
Use of direct ECDH direct encryption is easy for people to leak information on if there are other recipients in the message.
Considerations about protected vs unprotected header fields.

17. References

17.1. Normative References

[RFC2119]	Bradner, S., "Key words for use in RFCs to Indicate Requirement Levels", BCP 14, RFC 2119, March 1997.
[RFC7049]	Bormann, C. and P. Hoffman, "Concise Binary Object Representation (CBOR)", RFC 7049, October 2013.

17.2. Informative References

[AES-GCM]	Dworkin, M., "NIST Special Publication 800-38D: Recommendation for Block Cipher Modes of Operation: Galois/Counter Mode (GCM) and GMAC.", Nov 2007.
[DSS]	U.S. National Institute of Standards and Technology, "Digital Signature Standard (DSS)", July 2013.
[I-D.greevenbosch-appsawg-cbor-cddl]	Vigano, C., Birkholz, H. and R. Sun, "CBOR data definition language: a notational convention to express CBOR data structures.", Internet-Draft draft-greevenbosch-appsawg-cbor-cddl-05, March 2015.
[I-D.irtf-cfrg-curves]	Langley, A. and R. Salz, "Elliptic Curves for Security", Internet-Draft draft-irtf-cfrg-curves-02, March 2015.
[I-D.mcgrew-aead-aes-cbc-hmac-sha2]	McGrew, D., Foley, J. and K. Paterson, "Authenticated Encryption with AES-CBC and HMAC-SHA", Internet-Draft draft-mcgrew-aead-aes-cbc-hmac-sha2-05, July 2014.
[MAC]	NiST, N., "FIPS PUB 113: Computer Data Authentication", May 1985.
[MultiPrimeRSA]	Hinek, M. and D. Cheriton, "On the Security of Multi-prime RSA", June 2006.
[PVSig]	Brown, D. and D. Johnson, "Formal Security Proofs for a Signature Scheme with Partial Message Recover", February 2000.
[RFC2104]	Krawczyk, H., Bellare, M. and R. Canetti, "HMAC: Keyed-Hashing for Message Authentication", RFC 2104, February 1997.
[RFC2633]	Ramsdell, B., "S/MIME Version 3 Message Specification", RFC 2633, June 1999.
[RFC3394]	Schaad, J. and R. Housley, "Advanced Encryption Standard (AES) Key Wrap Algorithm", RFC 3394, September 2002.
[RFC3447]	Jonsson, J. and B. Kaliski, "Public-Key Cryptography Standards (PKCS) #1: RSA Cryptography Specifications Version 2.1", RFC 3447, February 2003.
[RFC3610]	Whiting, D., Housley, R. and N. Ferguson, "Counter with CBC-MAC (CCM)", RFC 3610, September 2003.
[RFC4231]	Nystrom, M., "Identifiers and Test Vectors for HMAC-SHA-224, HMAC-SHA-256, HMAC-SHA-384, and HMAC-SHA-512", RFC 4231, December 2005.
[RFC4262]	Santesson, S., "X.509 Certificate Extension for Secure/Multipurpose Internet Mail Extensions (S/MIME) Capabilities", RFC 4262, December 2005.
[RFC5480]	Turner, S., Brown, D., Yiu, K., Housley, R. and T. Polk, "Elliptic Curve Cryptography Subject Public Key Information", RFC 5480, March 2009.
[RFC5652]	Housley, R., "Cryptographic Message Syntax (CMS)", STD 70, RFC 5652, September 2009.
[RFC5751]	Ramsdell, B. and S. Turner, "Secure/Multipurpose Internet Mail Extensions (S/MIME) Version 3.2 Message Specification", RFC 5751, January 2010.
[RFC5752]	Turner, S. and J. Schaad, "Multiple Signatures in Cryptographic Message Syntax (CMS)", RFC 5752, January 2010.
[RFC5869]	Krawczyk, H. and P. Eronen, "HMAC-based Extract-and-Expand Key Derivation Function (HKDF)", RFC 5869, May 2010.
[RFC5990]	Randall, J., Kaliski, B., Brainard, J. and S. Turner, "Use of the RSA-KEM Key Transport Algorithm in the Cryptographic Message Syntax (CMS)", RFC 5990, September 2010.
[RFC6090]	McGrew, D., Igoe, K. and M. Salter, "Fundamental Elliptic Curve Cryptography Algorithms", RFC 6090, February 2011.
[RFC6151]	Turner, S. and L. Chen, "Updated Security Considerations for the MD5 Message-Digest and the HMAC-MD5 Algorithms", RFC 6151, March 2011.
[RFC6979]	Pornin, T., "Deterministic Usage of the Digital Signature Algorithm (DSA) and Elliptic Curve Digital Signature Algorithm (ECDSA)", RFC 6979, DOI 10.17487/RFC6979, August 2013.
[RFC7159]	Bray, T., "The JavaScript Object Notation (JSON) Data Interchange Format", RFC 7159, March 2014.
[RFC7252]	Shelby, Z., Hartke, K. and C. Bormann, "The Constrained Application Protocol (CoAP)", RFC 7252, DOI 10.17487/RFC7252, June 2014.
[RFC7515]	Jones, M., Bradley, J. and N. Sakimura, "JSON Web Signature (JWS)", RFC 7515, May 2015.
[RFC7516]	Jones, M. and J. Hildebrand, "JSON Web Encryption (JWE)", RFC 7516, May 2015.
[RFC7517]	Jones, M., "JSON Web Key (JWK)", RFC 7517, May 2015.
[RFC7518]	Jones, M., "JSON Web Algorithms (JWA)", RFC 7518, May 2015.
[RFC7539]	Nir, Y. and A. Langley, "ChaCha20 and Poly1305 for IETF Protocols", RFC 7539, DOI 10.17487/RFC7539, May 2015.
[SEC1]	Standards for Efficient Cryptography Group, "SEC 1: Elliptic Curve Cryptography", May 2009.
[SP800-56A]	Barker, E., Chen, L., Roginsky, A. and M. Smid, "NIST Special Publication 800-56A: Recommendation for Pair-Wise Key Establishment Schemes Using Discrete Logarithm Cryptography", May 2013.

Appendix A. AEAD and AE algorithms

The set of encryption algorithms that can be used with this specification is restricted to authenticated encryption (AE) and authenticated encryption with additional data (AEAD) algorithms. This means that there is a strong check that the data decrypted by the recipient is the same as what was encrypted by the sender. Encryption modes such as counter have no check on this at all. The CBC encryption mode had a weak check that the data is correct, given a random key and random data, the CBC padding check will pass one out of 256 times. There have been several times that a normal encryption mode has been combined with an integrity check to provide a content encryption mode that does provide the necessary authentication. AES-GCM [AES-GCM], AES-CCM [RFC3610], AES-CBC-HMAC [I-D.mcgrew-aead-aes-cbc-hmac-sha2] are examples of these composite modes.

PKCS v1.5 RSA key transport does not qualify as an AE algorithm. There are only three bytes in the encoding that can be checked as having decrypted correctly, the rest of the content can only be probabilistically checked as having decrypted correctly. For this reason, PKCS v1.5 RSA key transport MUST NOT be used with this specification. RSA-OAEP was designed to have the necessary checks that that content correctly decrypted and does qualify as an AE algorithm.

When dealing with authenticated encryption algorithms, there is always some type of value that needs to be checked to see if the authentication level has passed. This authentication value may be:

A separately generated tag computed by both the encrypter and decrypter and then compared by the decryptor. This tag value may be either placed at the end of the cipher text (the decision we made) or kept separately (the decision made by the JOSE working group). This is the approach followed by AES-GCM [AES-GCM] and AES-CCM [RFC3610].
A fixed value that is part of the encoded plain text. This is the approach followed by the AES key wrap algorithm [RFC3394].
A computed value is included as part of the encoded plain text. The computed value is then checked by the decryptor using the same computation path. This is the approach followed by RSAES-OAEP [RFC3447].

Appendix B. Three Levels of Recipient Information

All of the currently defined Key Management methods only use two levels of the COSE_Encrypt structure. The first level is the message content and the second level is the content key encryption. However, if one uses a key management technique such as RSA-KEM (see Appendix A of RSA-KEM [RFC5990], then it make sense to have three levels of the COSE_Encrypt structure.

These levels would be:

Level 0: The content encryption level. This level contains the payload of the message.
Level 1: The encryption of the CEK by a KEK.
Level 2: The encryption of a long random secret using an RSA key and a key derivation function to convert that secret into the KEK.

This is an example of what a triple layer message would look like. The message has the following layers:

Level 0: Has a content encrypted with AES-GCM using a 128-bit key.
Level 1: Uses the AES Key wrap algorithm with a 128-bit key.
Level 3: Uses ECDH Ephemeral-Static direct to generate the level 1 key.

In effect this example is a decomposed version of using the ECDH-ES+A128KW algorithm.

Size of binary file is 220 bytes

{
  1: 2,
  2: h'a10101',
  3: {
    5: h'02d1f7e6f26c43d4868d87ce'
  },
  4: h'64f84d913ba60a76070a9a48f26e97e863e285295a44320878caceb076
3a334806857c67',
  9: [
    {
      3: {
        1: -3
      },
      4: h'5a15dbf5b282ecb31a6074ee3815c252405dd7583e078188',
      9: [
        {
          3: {
            1: 50,
            4: h'6d65726961646f632e6272616e64796275636b406275636b
6c616e642e6578616d706c65',
            -1: {
              1: 2,
              -1: 1,
              -2: h'b2add44368ea6d641f9ca9af308b4079aeb519f11e9b8
a55a600b21233e86e68',
              -3: h'1a2cf118b9ee6895c8f415b686d4ca1cef362d4a7630a
31ef6019c0c56d33de0'
            }
          }
        }
      ]
    }
  ]
}

Appendix C. Examples

The examples can be found at https://github.com/cose-wg/Examples. I am currently still in the process of getting the examples up there along with some control information for people to be able to check and reproduce the examples.

Examples may have some features that are in questions but not yet incorporated in the document.

To make it easier to read, the examples are presented using the CBOR's diagnostic notation rather than a binary dump. [CREF23]JLS: Do we want to keep this as diagnostic notation or should we switch to having "binary" examples instead? Using the Ruby based CBOR diagnostic tools at ????, the diagnostic notation can be converted into binary files using the following command line: (install command is?...)

          
         diag2cbor < inputfile > outputfile

The examples can be extracted from the XML version of this docuent via an XPath expression as all of the artwork is tagged with the attribute type='CBORdiag'.

C.1. Examples of MAC messages

C.1.1. Shared Secret Direct MAC

This example users the following:

MAC: AES-CMAC, 256-bit key, trucated to 64 bits
Key management: direct shared secret
File name: Mac-04

Size of binary file is 74 bytes

{
  1: 3,
  2: h'a1016f4145532d434d41432d3235362f3634',
  4: h'546869732069732074686520636f6e74656e742e',
  10: h'd9afa663dd740848',
  9: [
    {
      3: {
        1: -6,
        4: h'6f75722d736563726574'
      }
    }
  ]
}

C.1.2. ECDH Direct MAC