RFC 9175

Internet Engineering Task Force (IETF)                         C. Amsüss
Request for Comments: 9175
Updates: 7252                                          J. Preuß Mattsson
Category: Standards Track                                    G. Selander
ISSN: 2070-1721                                              Ericsson AB
                                                           February 2022

Constrained Application Protocol (CoAP): Echo, Request-Tag, and Token


   This document specifies enhancements to the Constrained Application
   Protocol (CoAP) that mitigate security issues in particular use
   cases.  The Echo option enables a CoAP server to verify the freshness
   of a request or to force a client to demonstrate reachability at its
   claimed network address.  The Request-Tag option allows the CoAP
   server to match block-wise message fragments belonging to the same
   request.  This document updates RFC 7252 with respect to the
   following: processing requirements for client Tokens, forbidding non-
   secure reuse of Tokens to ensure response-to-request binding when
   CoAP is used with a security protocol, and amplification mitigation
   (where the use of the Echo option is now recommended).

Status of This Memo

   This is an Internet Standards Track document.

   This document is a product of the Internet Engineering Task Force
   (IETF).  It represents the consensus of the IETF community.  It has
   received public review and has been approved for publication by the
   Internet Engineering Steering Group (IESG).  Further information on
   Internet Standards is available in Section 2 of RFC 7841.

   Information about the current status of this document, any errata,
   and how to provide feedback on it may be obtained at

Copyright Notice

   Copyright (c) 2022 IETF Trust and the persons identified as the
   document authors.  All rights reserved.

   This document is subject to BCP 78 and the IETF Trust's Legal
   Provisions Relating to IETF Documents
   (https://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 Revised BSD License text as described in Section 4.e of the
   Trust Legal Provisions and are provided without warranty as described
   in the Revised BSD License.

Table of Contents

   1.  Introduction
     1.1.  Terminology
   2.  Request Freshness and the Echo Option
     2.1.  Request Freshness
     2.2.  The Echo Option
       2.2.1.  Echo Option Format
     2.3.  Echo Processing
     2.4.  Applications of the Echo Option
     2.5.  Characterization of Echo Applications
       2.5.1.  Time-Based versus Event-Based Freshness
       2.5.2.  Authority over Used Information
       2.5.3.  Protection by a Security Protocol
     2.6.  Updated Amplification Mitigation Requirements for Servers
   3.  Protecting Message Bodies Using Request Tags
     3.1.  Fragmented Message Body Integrity
     3.2.  The Request-Tag Option
       3.2.1.  Request-Tag Option Format
     3.3.  Request-Tag Processing by Servers
     3.4.  Setting the Request-Tag
     3.5.  Applications of the Request-Tag Option
       3.5.1.  Body Integrity Based on Payload Integrity
       3.5.2.  Multiple Concurrent Block-Wise Operations
       3.5.3.  Simplified Block-Wise Handling for Constrained Proxies
     3.6.  Rationale for the Option Properties
     3.7.  Rationale for Introducing the Option
     3.8.  Block2 and ETag Processing
   4.  Token Processing for Secure Request-Response Binding
     4.1.  Request-Response Binding
     4.2.  Updated Token Processing Requirements for Clients
   5.  Security Considerations
     5.1.  Token Reuse
   6.  Privacy Considerations
   7.  IANA Considerations
   8.  References
     8.1.  Normative References
     8.2.  Informative References
   Appendix A.  Methods for Generating Echo Option Values
   Appendix B.  Request-Tag Message Size Impact

   Authors' Addresses

1.  Introduction

   The initial suite of specifications for the Constrained Application
   Protocol (CoAP) ([RFC7252], [RFC7641], and [RFC7959]) was designed
   with the assumption that security could be provided on a separate
   layer, in particular, by using DTLS [RFC6347].  However, for some use
   cases, additional functionality or extra processing is needed to
   support secure CoAP operations.  This document specifies security
   enhancements to CoAP.

   This document specifies two CoAP options, the Echo option and the
   Request-Tag option.  The Echo option enables a CoAP server to verify
   the freshness of a request, which can be used to synchronize state,
   or to force a client to demonstrate reachability at its claimed
   network address.  The Request-Tag option allows the CoAP server to
   match message fragments belonging to the same request, fragmented
   using the CoAP block-wise transfer mechanism, which mitigates attacks
   and enables concurrent block-wise operations.  These options in
   themselves do not replace the need for a security protocol; they
   specify the format and processing of data that, when integrity
   protected using, e.g., DTLS [RFC6347], TLS [RFC8446], or Object
   Security for Constrained RESTful Environments (OSCORE) [RFC8613],
   provide the additional security features.

   This document updates [RFC7252] with a recommendation that servers
   use the Echo option to mitigate amplification attacks.

   The document also updates the Token processing requirements for
   clients specified in [RFC7252].  The updated processing forbids non-
   secure reuse of Tokens to ensure binding of responses to requests
   when CoAP is used with security, thus mitigating error cases and
   attacks where the client may erroneously associate the wrong response
   to a request.

   Each of the following sections provides a more-detailed introduction
   to the topic at hand in its first subsection.

1.1.  Terminology

   The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
   "OPTIONAL" in this document are to be interpreted as described in
   BCP 14 [RFC2119] [RFC8174] when, and only when, they appear in all
   capitals, as shown here.

   Like [RFC7252], this document relies on the Representational State
   Transfer [REST] architecture of the Web.

   Unless otherwise specified, the terms "client" and "server" refer to
   "CoAP client" and "CoAP server", respectively, as defined in

   A message's "freshness" is a measure of when a message was sent on a
   timescale of the recipient.  A server that receives a request can
   either verify that the request is fresh or determine that it cannot
   be verified that the request is fresh.  What is considered a fresh
   message is application dependent; exemplary uses are "no more than 42
   seconds ago" or "after this server's last reboot".

   The terms "payload" and "body" of a message are used as in [RFC7959].
   The complete interchange of a request and a response body is called a
   (REST) "operation".  An operation fragmented using [RFC7959] is
   called a "block-wise operation".  A block-wise operation that is
   fragmenting the request body is called a "block-wise request
   operation".  A block-wise operation that is fragmenting the response
   body is called a "block-wise response operation".

   Two request messages are said to be "matchable" if they occur between
   the same endpoint pair, have the same code, and have the same set of
   options, with the exception that elective NoCacheKey options and
   options involved in block-wise transfer (Block1, Block2, and Request-
   Tag) need not be the same.  Two blockwise request operations are said
   to be matchable if their request messages are matchable.

   Two matchable block-wise request operations are said to be
   "concurrent" if a block of the second request is exchanged even
   though the client still intends to exchange further blocks in the
   first operation.  (Concurrent block-wise request operations from a
   single endpoint are impossible with the options of [RFC7959] -- see
   the last paragraphs of Sections 2.4 and 2.5 -- because the second
   operation's block overwrites any state of the first exchange.)

   The Echo and Request-Tag options are defined in this document.

2.  Request Freshness and the Echo Option

2.1.  Request Freshness

   A CoAP server receiving a request is, in general, not able to verify
   when the request was sent by the CoAP client.  This remains true even
   if the request was protected with a security protocol, such as DTLS.
   This makes CoAP requests vulnerable to certain delay attacks that are
   particularly perilous in the case of actuators [COAP-ATTACKS].  Some
   attacks can be mitigated by establishing fresh session keys, e.g.,
   performing a DTLS handshake for each request, but, in general, this
   is not a solution suitable for constrained environments, for example,
   due to increased message overhead and latency.  Additionally, if
   there are proxies, fresh DTLS session keys between the server and the
   proxy do not say anything about when the client made the request.  In
   a general hop-by-hop setting, freshness may need to be verified in
   each hop.

   A straightforward mitigation of potential delayed requests is that
   the CoAP server rejects a request the first time it appears and asks
   the CoAP client to prove that it intended to make the request at this
   point in time.

2.2.  The Echo Option

   This document defines the Echo option, a lightweight challenge-
   response mechanism for CoAP that enables a CoAP server to verify the
   freshness of a request.  A fresh request is one whose age has not yet
   exceeded the freshness requirements set by the server.  The freshness
   requirements are application specific and may vary based on resource,
   method, and parameters outside of CoAP, such as policies.  The Echo
   option value is a challenge from the server to the client included in
   a CoAP response and echoed back to the server in one or more CoAP

   This mechanism is not only important in the case of actuators, or
   other use cases where the CoAP operations require freshness of
   requests, but also in general for synchronizing state between a CoAP
   client and server, cryptographically verifying the aliveness of the
   client or forcing a client to demonstrate reachability at its claimed
   network address.  The same functionality can be provided by echoing
   freshness indicators in CoAP payloads, but this only works for
   methods and response codes defined to have a payload.  The Echo
   option provides a convention to transfer freshness indicators that
   works for all methods and response codes.

2.2.1.  Echo Option Format

   The Echo option is elective, safe to forward, not part of the cache-
   key, and not repeatable (see Table 1, which extends Table 4 of

   | No. | C | U | N | R | Name | Format | Length | Default |
   | 252 |   |   | x |   | Echo | opaque | 1-40   | (none)  |

                  Table 1: Echo Option Summary

   C=Critical, U=Unsafe, N=NoCacheKey, R=Repeatable

   The Echo option value is generated by a server, and its content and
   structure are implementation specific.  Different methods for
   generating Echo option values are outlined in Appendix A.  Clients
   and intermediaries MUST treat an Echo option value as opaque and make
   no assumptions about its content or structure.

   When receiving an Echo option in a request, the server MUST be able
   to verify that the Echo option value (a) was generated by the server
   or some other party that the server trusts and (b) fulfills the
   freshness requirements of the application.  Depending on the
   freshness requirements, the server may verify exactly when the Echo
   option value was generated (time-based freshness) or verify that the
   Echo option was generated after a specific event (event-based
   freshness).  As the request is bound to the Echo option value, the
   server can determine that the request is not older than the Echo
   option value.

   When the Echo option is used with OSCORE [RFC8613], it MAY be an
   Inner or Outer option, and the Inner and Outer values are
   independent.  OSCORE servers MUST only produce Inner Echo options
   unless they are merely testing for reachability of the client (the
   same as proxies may do).  The Inner option is encrypted and integrity
   protected between the endpoints, whereas the Outer option is not
   protected by OSCORE.  As always with OSCORE, Outer options are
   visible to (and may be acted on by) all proxies and are visible on
   all links where no additional encryption (like TLS between client and
   proxy) is used.

2.3.  Echo Processing

   The Echo option MAY be included in any request or response (see
   Section 2.4 for different applications).

   The application decides under what conditions a CoAP request to a
   resource is required to be fresh.  These conditions can, for example,
   include what resource is requested, the request method and other data
   in the request, and conditions in the environment, such as the state
   of the server or the time of the day.

   If a certain request is required to be fresh, the request does not
   contain a fresh Echo option value, and the server cannot verify the
   freshness of the request in some other way, the server MUST NOT
   process the request further and SHOULD send a 4.01 (Unauthorized)
   response with an Echo option.  The server MAY include the same Echo
   option value in several different response messages and to different
   clients.  Examples of this could be time-based freshness (when
   several responses are sent closely after each other) or event-based
   freshness (with no event taking place between the responses).

   The server may use request freshness provided by the Echo option to
   verify the aliveness of a client or to synchronize state.  The server
   may also include the Echo option in a response to force a client to
   demonstrate reachability at its claimed network address.  Note that
   the Echo option does not bind a request to any particular previous
   response but provides an indication that the client had access to the
   previous response at the time when it created the request.

   Upon receiving a 4.01 (Unauthorized) response with the Echo option,
   the client SHOULD resend the original request with the addition of an
   Echo option with the received Echo option value.  The client MAY send
   a different request compared to the original request.  Upon receiving
   any other response with the Echo option, the client SHOULD echo the
   Echo option value in the next request to the server.  The client MAY
   include the same Echo option value in several different requests to
   the server or discard it at any time (especially to avoid tracking;
   see Section 6).

   A client MUST only send Echo option values to endpoints it received
   them from (where, as defined in Section 1.2 of [RFC7252], the
   security association is part of the endpoint).  In OSCORE processing,
   that means sending Echo option values from Outer options (or from
   non-OSCORE responses) back in Outer options and sending those from
   Inner options in Inner options in the same security context.

   Upon receiving a request with the Echo option, the server determines
   if the request is required to be fresh.  If not, the Echo option MAY
   be ignored.  If the request is required to be fresh and the server
   cannot verify the freshness of the request in some other way, the
   server MUST use the Echo option to verify that the request is fresh.
   If the server cannot verify that the request is fresh, the request is
   not processed further, and an error message MAY be sent.  The error
   message SHOULD include a new Echo option.

   One way for the server to verify freshness is to bind the Echo option
   value to a specific point in time and verify that the request is not
   older than a certain threshold T.  The server can verify this by
   checking that (t1 - t0) < T, where t1 is the request receive time and
   t0 is the time when the Echo option value was generated.  An example
   message flow over DTLS is shown Figure 1.

    Client   Server
       |       |
       +------>|        Code: 0.03 (PUT)
       |  PUT  |       Token: 0x41
       |       |    Uri-Path: lock
       |       |     Payload: 0 (Unlock)
       |       |
       |<------+        Code: 4.01 (Unauthorized)
       |  4.01 |       Token: 0x41
       |       |        Echo: 0x00000009437468756c687521 (t0 = 9, +MAC)
       |       |
       | ...   | The round trips take 1 second, time is now t1 = 10.
       |       |
       +------>|        Code: 0.03 (PUT)
       |  PUT  |       Token: 0x42
       |       |    Uri-Path: lock
       |       |        Echo: 0x00000009437468756c687521 (t0 = 9, +MAC)
       |       |     Payload: 0 (Unlock)
       |       |
       |       | Verify MAC, compare t1 - t0 = 1 < T => permitted.
       |       |
       |<------+        Code: 2.04 (Changed)
       |  2.04 |       Token: 0x42
       |       |

     Figure 1: Example Message Flow for Time-Based Freshness Using the
         'Integrity-Protected Timestamp' Construction of Appendix A

   Another way for the server to verify freshness is to maintain a cache
   of values associated to events.  The size of the cache is defined by
   the application.  In the following, we assume the cache size is 1, in
   which case, freshness is defined as "no new event has taken place".
   At each event, a new value is written into the cache.  The cache
   values MUST be different or chosen in a way so the probability for
   collisions is negligible.  The server verifies freshness by checking
   that e0 equals e1, where e0 is the cached value when the Echo option
   value was generated, and e1 is the cached value at the reception of
   the request.  An example message flow over DTLS is shown in Figure 2.

     Client   Server
        |       |
        +------>|        Code: 0.03 (PUT)
        |  PUT  |       Token: 0x41
        |       |    Uri-Path: lock
        |       |     Payload: 0 (Unlock)
        |       |
        |<------+        Code: 4.01 (Unauthorized)
        |  4.01 |       Token: 0x41
        |       |        Echo: 0x05 (e0 = 5, number of total lock
        |       |                            operations performed)
        |       |
        | ...   | No alterations happen to the lock state, e1 has the
        |       | same value e1 = 5.
        |       |
        +------>|        Code: 0.03 (PUT)
        |  PUT  |       Token: 0x42
        |       |    Uri-Path: lock
        |       |        Echo: 0x05
        |       |     Payload: 0 (Unlock)
        |       |
        |       | Compare e1 = e0 => permitted.
        |       |
        |<------+        Code: 2.04 (Changed)
        |  2.04 |       Token: 0x42
        |       |        Echo: 0x06 (e2 = 6, to allow later locking
        |       |                            without more round trips)
        |       |

       Figure 2: Example Message Flow for Event-Based Freshness Using
            the 'Persistent Counter' Construction of Appendix A

   When used to serve freshness requirements (including client aliveness
   and state synchronizing), the Echo option value MUST be integrity
   protected between the intended endpoints, e.g., using DTLS, TLS, or
   an OSCORE Inner option [RFC8613].  When used to demonstrate
   reachability at a claimed network address, the Echo option SHOULD be
   a Message Authentication Code (MAC) of the claimed address but MAY be
   unprotected.  Combining different Echo applications can necessitate
   different choices; see Appendix A, item 2 for an example.

   An Echo option MAY be sent with a successful response, i.e., even
   though the request satisfied any freshness requirements on the
   operation.  This is called a "preemptive" Echo option value and is
   useful when the server anticipates that the client will need to
   demonstrate freshness relative to the current response in the near

   A CoAP-to-CoAP proxy MAY set an Echo option on responses, both on
   forwarded ones that had no Echo option or ones generated by the proxy
   (from cache or as an error).  If it does so, it MUST remove the Echo
   option it recognizes as one generated by itself on follow-up
   requests.  When it receives an Echo option in a response, it MAY
   forward it to the client (and, not recognizing it as its own in
   future requests, relay it in the other direction as well) or process
   it on its own.  If it does so, it MUST ensure that the client's
   request was generated (or is regenerated) after the Echo option value
   used to send to the server was first seen.  (In most cases, this
   means that the proxy needs to ask the client to repeat the request
   with a new Echo option value.)

   The CoAP server side of CoAP-to-HTTP proxies MAY request freshness,
   especially if they have reason to assume that access may require it
   (e.g., because it is a PUT or POST); how this is determined is out of
   scope for this document.  The CoAP client side of HTTP-to-CoAP
   proxies MUST respond to Echo challenges itself if the proxy knows
   from the recent establishing of the connection that the HTTP request
   is fresh.  Otherwise, it MUST NOT repeat an unsafe request and SHOULD
   respond with a 503 (Service Unavailable) with a Retry-After value of
   0 seconds and terminate any underlying Keep-Alive connection.  If the
   HTTP request arrived in early data, the proxy SHOULD use a 425 (Too
   Early) response instead (see [RFC8470]).  The proxy MAY also use
   other mechanisms to establish freshness of the HTTP request that are
   not specified here.

2.4.  Applications of the Echo Option

   Unless otherwise noted, all these applications require a security
   protocol to be used and the Echo option to be protected by it.

   1.  Actuation requests often require freshness guarantees to avoid
       accidental or malicious delayed actuator actions.  In general,
       all unsafe methods (e.g., POST, PUT, and DELETE) may require
       freshness guarantees for secure operation.

       *  The same Echo option value may be used for multiple actuation
          requests to the same server, as long as the total time since
          the Echo option value was generated is below the freshness

       *  For actuator applications with low delay tolerance, to avoid
          additional round trips for multiple requests in rapid
          sequence, the server may send preemptive Echo option values in
          successful requests, irrespectively of whether or not the
          request contained an Echo option.  The client then uses the
          Echo option with the new value in the next actuation request,
          and the server compares the receive time accordingly.

   2.  A server may use the Echo option to synchronize properties (such
       as state or time) with a requesting client.  A server MUST NOT
       synchronize a property with a client that is not the authority of
       the property being synchronized.  For example, if access to a
       server resource is dependent on time, then the server MUST NOT
       synchronize time with a client requesting access unless the
       client is a time authority for the server.

       Note that the state to be synchronized is not carried inside the
       Echo option.  Any explicit state information needs to be carried
       along in the messages the Echo option value is sent in; the Echo
       mechanism only provides a partial order on the messages'

       *  If a server reboots during operation, it may need to
          synchronize state or time before continuing the interaction.
          For example, with OSCORE, it is possible to reuse a partly
          persistently stored security context by synchronizing the
          Partial IV (sequence number) using the Echo option, as
          specified in Section 7.5 of [RFC8613].

       *  A device joining a CoAP group communication [GROUP-COAP]
          protected with OSCORE [GROUP-OSCORE] may be required to
          initially synchronize its replay window state with a client by
          using the Echo option in a unicast response to a multicast
          request.  The client receiving the response with the Echo
          option includes the Echo option value in a subsequent unicast
          request to the responding server.

   3.  An attacker can perform a denial-of-service attack by putting a
       victim's address in the source address of a CoAP request and
       sending the request to a resource with a large amplification
       factor.  The amplification factor is the ratio between the size
       of the request and the total size of the response(s) to that
       request.  A server that provides a large amplification factor to
       an unauthenticated peer SHOULD mitigate amplification attacks, as
       described in Section 11.3 of [RFC7252].  One way to mitigate such
       attacks is for the server to respond to the alleged source
       address of the request with an Echo option in a short response
       message (e.g., 4.01 (Unauthorized)), thereby requesting the
       client to verify its source address.  This needs to be done only
       once per endpoint and limits the range of potential victims from
       the general Internet to endpoints that have been previously in
       contact with the server.  For this application, the Echo option
       can be used in messages that are not integrity protected, for
       example, during discovery.  (This is formally recommended in
       Section 2.6.)

       *  In the presence of a proxy, a server will not be able to
          distinguish different origin client endpoints, i.e., the
          client from which a request originates.  Following from the
          recommendation above, a proxy that provides a large
          amplification factor to unauthenticated peers SHOULD mitigate
          amplification attacks.  The proxy SHOULD use the Echo option
          to verify origin reachability, as described in Section 2.3.
          The proxy MAY forward safe requests immediately to have a
          cached result available when the client's repeated request

       *  Amplification mitigation is a trade-off between giving
          leverage to an attacker and causing overhead.  An
          amplification factor of 3 (i.e., don't send more than three
          times the number of bytes received until the peer's address is
          confirmed) is considered acceptable for unconstrained
          applications in [RFC9000], Section 8.

          When that limit is applied and no further context is
          available, a safe default is sending initial responses no
          larger than 136 bytes in CoAP serialization.  (The number is
          assuming Ethernet, IP, and UDP headers of 14, 40, and 8 bytes,
          respectively, with 4 bytes added for the CoAP header.  Triple
          that minus the non-CoAP headers gives the 136 bytes.)  Given
          the token also takes up space in the request, responding with
          132 bytes after the token is safe as well.

       *  When an Echo response is sent to mitigate amplification, it
          MUST be sent as a piggybacked or Non-confirmable response,
          never as a separate one (which would cause amplification due
          to retransmission).

   4.  A server may want to use the request freshness provided by the
       Echo option to verify the aliveness of a client.  Note that, in a
       deployment with hop-by-hop security and proxies, the server can
       only verify aliveness of the closest proxy.

2.5.  Characterization of Echo Applications

   Use cases for the Echo option can be characterized by several
   criteria that help determine the required properties of the Echo
   option value.  These criteria apply both to those listed in
   Section 2.4 and any novel applications.  They provide rationale for
   the statements in the former and guidance for the latter.

2.5.1.  Time-Based versus Event-Based Freshness

   The property a client demonstrates by sending an Echo option value is
   that the request was sent after a certain point in time or after some
   event happened on the server.

   When events are counted, they form something that can be used as a
   monotonic but very non-uniform time line.  With highly regular events
   and low-resolution time, the distinction between time-based and
   event-based freshness can be blurred: "no longer than a month ago" is
   similar to "since the last full moon".

   In an extreme form of event-based freshness, the server can place an
   event whenever an Echo option value is used.  This makes the Echo
   option value effectively single use.

   Event-based and time-based freshness can be combined in a single Echo
   option value, e.g., by encrypting a timestamp with a key that changes
   with every event to obtain semantics in the style of "usable once but
   only for 5 minutes".

2.5.2.  Authority over Used Information

   Information conveyed to the server in the request Echo option value
   has different authority depending on the application.  Understanding
   who or what is the authoritative source of that information helps the
   server implementor decide the necessary protection of the Echo option

   If all that is conveyed to the server is information that the client
   is authorized to provide arbitrarily (which is another way of saying
   that the server has to trust the client on whatever the Echo option
   is being used for), then the server can issue Echo option values that
   do not need to be protected on their own.  They still need to be
   covered by the security protocol that covers the rest of the message,
   but the Echo option value can be just short enough to be unique
   between this server and client.

   For example, the client's OSCORE Sender Sequence Number (as used in
   [RFC8613], Appendix B.1.2) is such information.

   In most other cases, there is information conveyed for which the
   server is the authority ("the request must not be older than five
   minutes" is counted on the server's clock, not the client's) or which
   even involve the network (as when performing amplification
   mitigation).  In these cases, the Echo option value itself needs to
   be protected against forgery by the client, e.g., by using a
   sufficiently large, random value or a MAC, as described in
   Appendix A, items 1 and 2.

   For some applications, the server may be able to trust the client to
   also act as the authority (e.g., when using time-based freshness
   purely to mitigate request delay attacks); these need careful case-
   by-case evaluation.

   To issue Echo option values without integrity protection of its own,
   the server needs to trust the client to never produce requests with
   attacker-controlled Echo option values.  The provisions of
   Section 2.3 (saying that an Echo option value may only be sent as
   received from the same server) allow that.  The requirement stated
   there for the client to treat the Echo option value as opaque holds
   for these applications like for all others.

   When the client is the sole authority over the synchronized property,
   the server can still use time or events to issue new Echo option
   values.  Then, the request's Echo option value not so much proves the
   indicated freshness to the server but reflects the client's intention
   to indicate reception of responses containing that value when sending
   the later ones.

   Note that a single Echo option value can be used for multiple
   purposes (e.g., to both get the sequence number information and
   perform amplification mitigation).  In this case, the stricter
   protection requirements apply.

2.5.3.  Protection by a Security Protocol

   For meaningful results, the Echo option needs to be used in
   combination with a security protocol in almost all applications.

   When the information extracted by the server is only about a part of
   the system outside of any security protocol, then the Echo option can
   also be used without a security protocol (in case of OSCORE, as an
   Outer option).

   The only known application satisfying this requirement is network
   address reachability, where unprotected Echo option values are used
   both by servers (e.g., during setup of a security context) and
   proxies (which do not necessarily have a security association with
   their clients) for amplification mitigation.

2.6.  Updated Amplification Mitigation Requirements for Servers

   This section updates the amplification mitigation requirements for
   servers in [RFC7252] to recommend the use of the Echo option to
   mitigate amplification attacks.  The requirements for clients are not
   updated.  Section 11.3 of [RFC7252] is updated by adding the
   following text:

   |  A CoAP server SHOULD mitigate potential amplification attacks by
   |  responding to unauthenticated clients with 4.01 (Unauthorized)
   |  including an Echo option, as described in item 3 in Section 2.4 of
   |  RFC 9175.

3.  Protecting Message Bodies Using Request Tags

3.1.  Fragmented Message Body Integrity

   CoAP was designed to work over unreliable transports, such as UDP,
   and includes a lightweight reliability feature to handle messages
   that are lost or arrive out of order.  In order for a security
   protocol to support CoAP operations over unreliable transports, it
   must allow out-of-order delivery of messages.

   The block-wise transfer mechanism [RFC7959] extends CoAP by defining
   the transfer of a large resource representation (CoAP message body)
   as a sequence of blocks (CoAP message payloads).  The mechanism uses
   a pair of CoAP options, Block1 and Block2, pertaining to the request
   and response payload, respectively.  The block-wise functionality
   does not support the detection of interchanged blocks between
   different message bodies to the same resource having the same block
   number.  This remains true even when CoAP is used together with a
   security protocol (such as DTLS or OSCORE) within the replay window
   [COAP-ATTACKS], which is a vulnerability of the block-wise
   functionality of CoAP [RFC7959].

   A straightforward mitigation of mixing up blocks from different
   messages is to use unique identifiers for different message bodies,
   which would provide equivalent protection to the case where the
   complete body fits into a single payload.  The ETag option [RFC7252],
   set by the CoAP server, identifies a response body fragmented using
   the Block2 option.

3.2.  The Request-Tag Option

   This document defines the Request-Tag option for identifying request
   bodies, similar to ETag, but ephemeral and set by the CoAP client.
   The Request-Tag is intended for use as a short-lived identifier for
   keeping apart distinct block-wise request operations on one resource
   from one client, addressing the issue described in Section 3.1.  It
   enables the receiving server to reliably assemble request payloads
   (blocks) to their message bodies and, if it chooses to support it, to
   reliably process simultaneous block-wise request operations on a
   single resource.  The requests must be integrity protected if they
   should protect against interchange of blocks between different
   message bodies.  The Request-Tag option is mainly used in requests
   that carry the Block1 option and in Block2 requests following these.

   In essence, it is an implementation of the "proxy-safe elective
   option" used just to "vary the cache key", as suggested in [RFC7959],
   Section 2.4.

3.2.1.  Request-Tag Option Format

   The Request-Tag option is elective, safe to forward, repeatable, and
   part of the cache key (see Table 2, which extends Table 4 of

   | No. | C | U | N | R | Name        | Format | Length | Default |
   | 292 |   |   |   | x | Request-Tag | opaque | 0-8    | (none)  |

                  Table 2: Request-Tag Option Summary

   C=Critical, U=Unsafe, N=NoCacheKey, R=Repeatable

   Request-Tag, like the Block options, is both a class E and a class U
   option in terms of OSCORE processing (see Section 4.1 of [RFC8613]).
   The Request-Tag MAY be an Inner or Outer option.  It influences the
   Inner or Outer block operations, respectively.  The Inner and Outer
   values are therefore independent of each other.  The Inner option is
   encrypted and integrity protected between the client and server, and
   it provides message body identification in case of end-to-end
   fragmentation of requests.  The Outer option is visible to proxies
   and labels message bodies in case of hop-by-hop fragmentation of

   The Request-Tag option is only used in the request messages of block-
   wise operations.

   The Request-Tag mechanism can be applied independently on the server
   and client sides of CoAP-to-CoAP proxies, as are the Block options.
   However, given it is safe to forward, a proxy is free to just forward
   it when processing an operation.  CoAP-to-HTTP proxies and HTTP-to-
   CoAP proxies can use Request-Tag on their CoAP sides; it is not
   applicable to HTTP requests.

3.3.  Request-Tag Processing by Servers

   The Request-Tag option does not require any particular processing on
   the server side outside of the processing already necessary for any
   unknown elective proxy-safe cache-key option.  The option varies the
   properties that distinguish block-wise operations (which includes all
   options except Block1, Block2, and all operations that are elective
   NoCacheKey).  Thus, the server cannot treat messages with a different
   list of Request-Tag options as belonging to the same operation.

   To keep utilizing the cache, a server (including proxies) MAY discard
   the Request-Tag option from an assembled block-wise request when
   consulting its cache, as the option relates to the operation on the
   wire and not its semantics.  For example, a FETCH request with the
   same body as an older one can be served from the cache if the older's
   Max-Age has not expired yet, even if the second operation uses a
   Request-Tag and the first did not.  (This is similar to the situation
   about ETag in that it is formally part of the cache key, but
   implementations that are aware of its meaning can cache more
   efficiently (see [RFC7252], Section 5.4.2).

   A server receiving a Request-Tag MUST treat it as opaque and make no
   assumptions about its content or structure.

   Two messages carrying the same Request-Tag is a necessary but not
   sufficient condition for being part of the same operation.  For one,
   a server may still treat them as independent messages when it sends
   2.01 (Created) and 2.04 (Changed) responses for every block.  Also, a
   client that lost interest in an old operation but wants to start over
   can overwrite the server's old state with a new initial (num=0)
   Block1 request and the same Request-Tag under some circumstances.
   Likewise, that results in the new message not being part of the old

   As it has always been, a server that can only serve a limited number
   of block-wise operations at the same time can delay the start of the
   operation by replying with 5.03 (Service Unavailable) and a Max-Age
   indicating how long it expects the existing operation to go on, or it
   can forget about the state established with the older operation and
   respond with 4.08 (Request Entity Incomplete) to later blocks on the
   first operation.

3.4.  Setting the Request-Tag

   For each separate block-wise request operation, the client can choose
   a Request-Tag value or choose not to set a Request-Tag. It needs to
   be set to the same value (or unset) in all messages belonging to the
   same operation; otherwise, they are treated as separate operations by
   the server.

   Starting a request operation matchable to a previous operation and
   even using the same Request-Tag value is called "request tag
   recycling".  The absence of a Request-Tag option is viewed as a value
   distinct from all values with a single Request-Tag option set;
   starting a request operation matchable to a previous operation where
   neither has a Request-Tag option therefore constitutes request tag
   recycling just as well (also called "recycling the absent option").

   Clients that use Request-Tag for a particular purpose (like in
   Section 3.5) MUST NOT recycle a request tag unless the first
   operation has concluded.  What constitutes a concluded operation
   depends on the purpose and is defined accordingly; see examples in
   Section 3.5.

   When Block1 and Block2 are combined in an operation, the Request-Tag
   of the Block1 phase is set in the Block2 phase as well; otherwise,
   the request would have a different set of options and would not be
   recognized any more.

   Clients are encouraged to generate compact messages.  This means
   sending messages without Request-Tag options whenever possible and
   using short values when the absent option cannot be recycled.

   Note that Request-Tag options can be present in request messages that
   carry no Block options (for example, because a proxy unaware of
   Request-Tag reassembled them).

   The Request-Tag option MUST NOT be present in response messages.

3.5.  Applications of the Request-Tag Option

3.5.1.  Body Integrity Based on Payload Integrity

   When a client fragments a request body into multiple message
   payloads, even if the individual messages are integrity protected, it
   is still possible for an attacker to maliciously replace a later
   operation's blocks with an earlier operation's blocks (see
   Section 2.5 of [COAP-ATTACKS]).  Therefore, the integrity protection
   of each block does not extend to the operation's request body.

   In order to gain that protection, use the Request-Tag mechanism as

   *  The individual exchanges MUST be integrity protected end to end
      between the client and server.

   *  The client MUST NOT recycle a request tag in a new operation
      unless the previous operation matchable to the new one has

      If any future security mechanisms allow a block-wise transfer to
      continue after an endpoint's details (like the IP address) have
      changed, then the client MUST consider messages matchable if they
      were sent to any endpoint address using the new operation's
      security context.

   *  The client MUST NOT regard a block-wise request operation as
      concluded unless all of the messages the client has sent in the
      operation would be regarded as invalid by the server if they were

      When security services are provided by OSCORE, these confirmations
      typically result either from the client receiving an OSCORE
      response message matching the request (an empty Acknowledgement
      (ACK) is insufficient) or because the message's sequence number is
      old enough to be outside the server's receive window.

      When security services are provided by DTLS, this can only be
      confirmed if there was no CoAP retransmission of the request, the
      request was responded to, and the server uses replay protection.

   Authors of other documents (e.g., applications of [RFC8613]) are
   invited to mandate this subsection's behavior for clients that
   execute block-wise interactions over secured transports.  In this
   way, the server can rely on a conforming client to set the Request-
   Tag option when required and thereby have confidence in the integrity
   of the assembled body.

   Note that this mechanism is implicitly implemented when the security
   layer guarantees ordered delivery (e.g., CoAP over TLS [RFC8323]).
   This is because, with each message, any earlier message cannot be
   replayed any more, so the client never needs to set the Request-Tag
   option unless it wants to perform concurrent operations.

   Body integrity only makes sense in applications that have stateful
   block-wise transfers.  On applications where all the state is in the
   application (e.g., because rather than POSTing a large representation
   to a collection in a stateful block-wise transfer, a collection item
   is created first, then written to once and available when written
   completely), clients need not concern themselves with body integrity
   and thus the Request-Tag.

   Body integrity is largely independent from replay protection.  When
   no replay protection is available (it is optional in DTLS), a full
   block-wise operation may be replayed, but, by adhering to the above,
   no operations will be mixed up.  The only link between body integrity
   and replay protection is that, without replay protection, recycling
   is not possible.

3.5.2.  Multiple Concurrent Block-Wise Operations

   CoAP clients, especially CoAP proxies, may initiate a block-wise
   request operation to a resource, to which a previous one is already
   in progress, which the new request should not cancel.  A CoAP proxy
   would be in such a situation when it forwards operations with the
   same cache-key options but possibly different payloads.

   For those cases, Request-Tag is the proxy-safe elective option
   suggested in the last paragraph of Section 2.4 of [RFC7959].

   When initializing a new block-wise operation, a client has to look at
   other active operations:

   *  If any of them is matchable to the new one, and the client neither
      wants to cancel the old one nor postpone the new one, it can pick
      a Request-Tag value (including the absent option) that is not in
      use by the other matchable operations for the new operation.

   *  Otherwise, it can start the new operation without setting the
      Request-Tag option on it.

3.5.3.  Simplified Block-Wise Handling for Constrained Proxies

   The Block options were defined to be unsafe to forward because a
   proxy that would forward blocks as plain messages would risk mixing
   up clients' requests.

   In some cases, for example, when forwarding block-wise request
   operations, appending a Request-Tag value unique to the client can
   satisfy the requirements on the proxy that come from the presence of
   a Block option.

   This is particularly useful to proxies that strive for stateless
   operations, as described in [RFC8974], Section 4.

   The precise classification of cases in which such a Request-Tag
   option is sufficient is not trivial, especially when both request and
   response body are fragmented, and is out of scope for this document.

3.6.  Rationale for the Option Properties

   The Request-Tag option can be elective, because to servers unaware of
   the Request-Tag option, operations with differing request tags will
   not be matchable.

   The Request-Tag option can be safe to forward but part of the cache
   key, because proxies unaware of the Request-Tag option will consider
   operations with differing request tags unmatchable but can still
   forward them.

   The Request-Tag option is repeatable because this easily allows
   several cascaded stateless proxies to each put in an origin address.
   They can perform the steps of Section 3.5.3 without the need to
   create an option value that is the concatenation of the received
   option and their own value and can simply add a new Request-Tag
   option unconditionally.

   In draft versions of this document, the Request-Tag option used to be
   critical and unsafe to forward.  That design was based on an
   erroneous understanding of which blocks could be composed according
   to [RFC7959].

3.7.  Rationale for Introducing the Option

   An alternative that was considered to the Request-Tag option for
   coping with the problem of fragmented message body integrity
   (Section 3.5.1) was to update [RFC7959] to say that blocks could only
   be assembled if their fragments' order corresponded to the sequence

   That approach would have been difficult to roll out reliably on DTLS,
   where many implementations do not expose sequence numbers, and would
   still not prevent attacks like in Section 2.5.2 of [COAP-ATTACKS].

3.8.  Block2 and ETag Processing

   The same security properties as in Section 3.5.1 can be obtained for
   block-wise response operations.  The threat model here does not
   depend on an attacker; a client can construct a wrong representation
   by assembling it from blocks from different resource states.  That
   can happen when a resource is modified during a transfer or when some
   blocks are still valid in the client's cache.

   Rules stating that response body reassembly is conditional on
   matching ETag values are already in place from Section 2.4 of

   To gain protection equivalent to that described in Section 3.5.1, a
   server MUST use the Block2 option in conjunction with the ETag option
   ([RFC7252], Section 5.10.6) and MUST NOT use the same ETag value for
   different representations of a resource.

4.  Token Processing for Secure Request-Response Binding

4.1.  Request-Response Binding

   A fundamental requirement of secure REST operations is that the
   client can bind a response to a particular request.  If this is not
   ensured, a client may erroneously associate the wrong response to a
   request.  The wrong response may be an old response for the same
   resource or a response for a completely different resource (e.g., see
   Section 2.3 of [COAP-ATTACKS]).  For example, a request for the alarm
   status "GET /status" may be associated to a prior response "on",
   instead of the correct response "off".

   In HTTP/1.1, this type of binding is always assured by the ordered
   and reliable delivery, as well as mandating that the server sends
   responses in the same order that the requests were received.  The
   same is not true for CoAP, where the server (or an attacker) can
   return responses in any order and where there can be any number of
   responses to a request (e.g., see [RFC7641]).  In CoAP, concurrent
   requests are differentiated by their Token.  Note that the CoAP
   Message ID cannot be used for this purpose since those are typically
   different for the REST request and corresponding response in case of
   "separate response" (see Section 2.2 of [RFC7252]).

   CoAP [RFC7252] does not treat the Token as a cryptographically
   important value and does not give stricter guidelines than that the
   Tokens currently "in use" SHOULD (not SHALL) be unique.  If used with
   a security protocol not providing bindings between requests and
   responses (e.g., DTLS and TLS), Token reuse may result in situations
   where a client matches a response to the wrong request.  Note that
   mismatches can also happen for other reasons than a malicious
   attacker, e.g., delayed delivery or a server sending notifications to
   an uninterested client.

   A straightforward mitigation is to mandate clients to not reuse
   Tokens until the traffic keys have been replaced.  The following
   section formalizes that.

4.2.  Updated Token Processing Requirements for Clients

   As described in Section 4.1, the client must be able to verify that a
   response corresponds to a particular request.  This section updates
   the Token processing requirements for clients in [RFC7252] to always
   assure a cryptographically secure binding of responses to requests
   for secure REST operations like "coaps".  The Token processing for
   servers is not updated.  Token processing in Section 5.3.1 of
   [RFC7252] is updated by adding the following text:

   |  When CoAP is used with a security protocol not providing bindings
   |  between requests and responses, the Tokens have cryptographic
   |  importance.  The client MUST make sure that Tokens are not used in
   |  a way so that responses risk being associated with the wrong
   |  request.
   |  One easy way to accomplish this is to implement the Token (or part
   |  of the Token) as a sequence number, starting at zero for each new
   |  or rekeyed secure connection.  This approach SHOULD be followed.

5.  Security Considerations

   The freshness assertion of the Echo option comes from the client
   reproducing the same value of the Echo option in a request as it
   received in a previous response.  If the Echo option value is a large
   random number, then there is a high probability that the request is
   generated after having seen the response.  If the Echo option value
   of the response can be guessed, e.g., if based on a small random
   number or a counter (see Appendix A), then it is possible to compose
   a request with the right Echo option value ahead of time.  Using
   guessable Echo option values is only permissible in a narrow set of
   cases described in Section 2.5.2.  Echo option values MUST be set by
   the CoAP server such that the risk associated with unintended reuse
   can be managed.

   If uniqueness of the Echo option value is based on randomness, then
   the availability of a secure pseudorandom number generator and truly
   random seeds are essential for the security of the Echo option.  If
   no true random number generator is available, a truly random seed
   must be provided from an external source.  As each pseudorandom
   number must only be used once, an implementation needs to get a new
   truly random seed after reboot or continuously store the state in
   nonvolatile memory.  See [RFC8613], Appendix B.1.1 for issues and
   approaches for writing to nonvolatile memory.

   A single active Echo option value with 64 (pseudo)random bits gives
   the same theoretical security level as a 64-bit MAC (as used in,
   e.g., AES_128_CCM_8).  If a random unique Echo option value is
   intended, the Echo option value SHOULD contain 64 (pseudo)random bits
   that are not predictable for any other party than the server.  A
   server MAY use different security levels for different use cases
   (client aliveness, request freshness, state synchronization, network
   address reachability, etc.).

   The security provided by the Echo and Request-Tag options depends on
   the security protocol used.  CoAP and HTTP proxies require (D)TLS to
   be terminated at the proxies.  The proxies are therefore able to
   manipulate, inject, delete, or reorder options or packets.  The
   security claims in such architectures only hold under the assumption
   that all intermediaries are fully trusted and have not been

   Echo option values without the protection of randomness or a MAC are
   limited to cases when the client is the trusted source of all derived
   properties (as per Section 2.5.2).  Using them needs per-application
   consideration of both the impact of a malicious client and of
   implementation errors in clients.  These Echo option values are the
   only legitimate case for Echo option values shorter than four bytes,
   which are not necessarily secret.  They MUST NOT be used unless the
   Echo option values in the request are integrity protected, as per
   Section 2.3.

   Servers SHOULD use a monotonic clock to generate timestamps and
   compute round-trip times.  Use of non-monotonic clocks is not secure,
   as the server will accept expired Echo option values if the clock is
   moved backward.  The server will also reject fresh Echo option values
   if the clock is moved forward.  Non-monotonic clocks MAY be used as
   long as they have deviations that are acceptable given the freshness
   requirements.  If the deviations from a monotonic clock are known, it
   may be possible to adjust the threshold accordingly.

   An attacker may be able to affect the server's system time in various
   ways, such as setting up a fake NTP server or broadcasting false time
   signals to radio-controlled clocks.

   For the purpose of generating timestamps for the Echo option, a
   server MAY set a timer at reboot and use the time since reboot,
   choosing the granularity such that different requests arrive at
   different times.  Servers MAY intermittently reset the timer and MAY
   generate a random offset applied to all timestamps.  When resetting
   the timer, the server MUST reject all Echo option values that were
   created before the reset.

   Servers that use the "List of Cached Random Values and Timestamps"
   method described in Appendix A may be vulnerable to resource
   exhaustion attacks.  One way to minimize the state is to use the
   "Integrity-Protected Timestamp" method described in Appendix A.

5.1.  Token Reuse

   Reusing Tokens in a way so that responses are guaranteed to not be
   associated with the wrong request is not trivial.  The server may
   process requests in any order and send multiple responses to the same
   request.  An attacker may block, delay, and reorder messages.  The
   use of a sequence number is therefore recommended when CoAP is used
   with a security protocol that does not provide bindings between
   requests and responses, such as DTLS or TLS.

   For a generic response to a Confirmable request over DTLS, binding
   can only be claimed without out-of-band knowledge if:

   *  the original request was never retransmitted and

   *  the response was piggybacked in an Acknowledgement message (as a
      Confirmable or Non-confirmable response may have been transmitted
      multiple times).

   If observation was used, the same holds for the registration, all
   reregistrations, and the cancellation.

   (In addition, for observations, any responses using that Token and a
   DTLS sequence number earlier than the cancellation Acknowledgement
   message need to be discarded.  This is typically not supported in
   DTLS implementations.)

   In some setups, Tokens can be reused without the above constraints,
   as a different component in the setup provides the associations:

   *  In CoAP over TLS, retransmissions are not handled by the CoAP
      layer and behave like a replay window size of 1.  When a client is
      sending TLS-protected requests without Observe to a single server,
      the client can reuse a Token as soon as the previous response with
      that Token has been received.

   *  Requests whose responses are cryptographically bound to the
      requests (like in OSCORE) can reuse Tokens indefinitely.

   In all other cases, a sequence number approach is RECOMMENDED, as per
   Section 4.

   Tokens that cannot be reused need to be handled appropriately.  This
   could be solved by increasing the Token as soon as the currently used
   Token cannot be reused or by keeping a list of all Tokens unsuitable
   for reuse.

   When the Token (or part of the Token) contains a sequence number, the
   encoding of the sequence number has to be chosen in a way to avoid
   any collisions.  This is especially true when the Token contains more
   information than just the sequence number, e.g., the serialized
   state, as in [RFC8974].

6.  Privacy Considerations

   Implementations SHOULD NOT put any privacy-sensitive information in
   the Echo or Request-Tag option values.  Unencrypted timestamps could
   reveal information about the server, such as location, time since
   reboot, or that the server will accept expired certificates.
   Timestamps MAY be used if the Echo option is encrypted between the
   client and the server, e.g., in the case of DTLS without proxies or
   when using OSCORE with an Inner Echo option.

   Like HTTP cookies, the Echo option could potentially be abused as a
   tracking mechanism that identifies a client across requests.  This is
   especially true for preemptive Echo option values.  Servers MUST NOT
   use the Echo option to correlate requests for other purposes than
   freshness and reachability.  Clients only send Echo option values to
   the same server from which the values were received.  Compared to
   HTTP, CoAP clients are often authenticated and non-mobile, and
   servers can therefore often correlate requests based on the security
   context, the client credentials, or the network address.  Especially
   when the Echo option increases a server's ability to correlate
   requests, clients MAY discard all preemptive Echo option values.

   Publicly visible generated identifiers, even when opaque (as all
   defined in this document are), can leak information as described in
   [NUMERIC-IDS].  To avoid the effects described there, the absent
   Request-Tag option should be recycled as much as possible.  (That is
   generally possible as long as a security mechanism is in place --
   even in the case of OSCORE outer block-wise transfers, as the OSCORE
   option's variation ensures that no matchable requests are created by
   different clients.)  When an unprotected Echo option is used to
   demonstrate reachability, the recommended mechanism of Section 2.3
   keeps the effects to a minimum.

7.  IANA Considerations

   IANA has added the following option numbers to the "CoAP Option
   Numbers" registry defined by [RFC7252]:

   | Number | Name        | Reference |
   | 252    | Echo        | RFC 9175  |
   | 292    | Request-Tag | RFC 9175  |

        Table 3: Additions to CoAP
         Option Numbers Registry

8.  References

8.1.  Normative References

   [RFC2119]  Bradner, S., "Key words for use in RFCs to Indicate
              Requirement Levels", BCP 14, RFC 2119,
              DOI 10.17487/RFC2119, March 1997,

   [RFC6347]  Rescorla, E. and N. Modadugu, "Datagram Transport Layer
              Security Version 1.2", RFC 6347, DOI 10.17487/RFC6347,
              January 2012, <https://www.rfc-editor.org/info/rfc6347>.

   [RFC7252]  Shelby, Z., Hartke, K., and C. Bormann, "The Constrained
              Application Protocol (CoAP)", RFC 7252,
              DOI 10.17487/RFC7252, June 2014,

   [RFC7959]  Bormann, C. and Z. Shelby, Ed., "Block-Wise Transfers in
              the Constrained Application Protocol (CoAP)", RFC 7959,
              DOI 10.17487/RFC7959, August 2016,

   [RFC8174]  Leiba, B., "Ambiguity of Uppercase vs Lowercase in RFC
              2119 Key Words", BCP 14, RFC 8174, DOI 10.17487/RFC8174,
              May 2017, <https://www.rfc-editor.org/info/rfc8174>.

   [RFC8470]  Thomson, M., Nottingham, M., and W. Tarreau, "Using Early
              Data in HTTP", RFC 8470, DOI 10.17487/RFC8470, September
              2018, <https://www.rfc-editor.org/info/rfc8470>.

   [RFC8613]  Selander, G., Mattsson, J., Palombini, F., and L. Seitz,
              "Object Security for Constrained RESTful Environments
              (OSCORE)", RFC 8613, DOI 10.17487/RFC8613, July 2019,

8.2.  Informative References

              Preuß Mattsson, J., Fornehed, J., Selander, G., Palombini,
              F., and C. Amsüss, "Attacks on the Constrained Application
              Protocol (CoAP)", Work in Progress, Internet-Draft, draft-
              mattsson-core-coap-attacks-01, 27 July 2021,

              Dijk, E., Wang, C., and M. Tiloca, "Group Communication
              for the Constrained Application Protocol (CoAP)", Work in
              Progress, Internet-Draft, draft-ietf-core-groupcomm-bis-
              05, 25 October 2021,

              Tiloca, M., Selander, G., Palombini, F., Preuß Mattsson,
              J., and J. Park, "Group OSCORE - Secure Group
              Communication for CoAP", Work in Progress, Internet-Draft,
              draft-ietf-core-oscore-groupcomm-13, 25 October 2021,

              Gont, F. and I. Arce, "On the Generation of Transient
              Numeric Identifiers", Work in Progress, Internet-Draft,
              draft-irtf-pearg-numeric-ids-generation-08, 31 January
              2022, <https://datatracker.ietf.org/doc/html/draft-irtf-

   [REST]     Fielding, R., "Architectural Styles and the Design of
              Network-based Software Architectures", 2000,

   [RFC7641]  Hartke, K., "Observing Resources in the Constrained
              Application Protocol (CoAP)", RFC 7641,
              DOI 10.17487/RFC7641, September 2015,

   [RFC8323]  Bormann, C., Lemay, S., Tschofenig, H., Hartke, K.,
              Silverajan, B., and B. Raymor, Ed., "CoAP (Constrained
              Application Protocol) over TCP, TLS, and WebSockets",
              RFC 8323, DOI 10.17487/RFC8323, February 2018,

   [RFC8446]  Rescorla, E., "The Transport Layer Security (TLS) Protocol
              Version 1.3", RFC 8446, DOI 10.17487/RFC8446, August 2018,

   [RFC8974]  Hartke, K. and M. Richardson, "Extended Tokens and
              Stateless Clients in the Constrained Application Protocol
              (CoAP)", RFC 8974, DOI 10.17487/RFC8974, January 2021,

   [RFC9000]  Iyengar, J., Ed. and M. Thomson, Ed., "QUIC: A UDP-Based
              Multiplexed and Secure Transport", RFC 9000,
              DOI 10.17487/RFC9000, May 2021,

Appendix A.  Methods for Generating Echo Option Values

   The content and structure of the Echo option value are implementation
   specific and determined by the server.  Two simple mechanisms for
   time-based freshness and one for event-based freshness are outlined
   in this appendix.  The "List of Cached Random Values and Timestamps"
   mechanism is RECOMMENDED in general.  The "Integrity-Protected
   Timestamp" mechanism is RECOMMENDED in case the Echo option is
   encrypted between the client and the server.

   Different mechanisms have different trade-offs between the size of
   the Echo option value, the amount of server state, the amount of
   computation, and the security properties offered.  A server MAY use
   different methods and security levels for different use cases (client
   aliveness, request freshness, state synchronization, network address
   reachability, etc.).

   1.  List of Cached Random Values and Timestamps.  The Echo option
       value is a (pseudo)random byte string called r.  The server
       caches a list containing the random byte strings and their
       initial transmission times.  Assuming 72-bit random values and
       32-bit timestamps, the size of the Echo option value is 9 bytes
       and the amount of server state is 13n bytes, where n is the
       number of active Echo option values.  The security against an
       attacker guessing Echo option values is given by s = bit length
       of r - log2(n).  The length of r and the maximum allowed n should
       be set so that the security level is harmonized with other parts
       of the deployment, e.g., s >= 64.  If the server loses time
       continuity, e.g., due to reboot, the entries in the old list MUST
       be deleted.

       Echo option value:  random value r

       Server State:  random value r, timestamp t0

       This method is suitable for both time-based and event-based
       freshness (e.g., by clearing the cache when an event occurs) and
       is independent of the client authority.

   2.  Integrity-Protected Timestamp.  The Echo option value is an
       integrity-protected timestamp.  The timestamp can have a
       different resolution and range.  A 32-bit timestamp can, e.g.,
       give a resolution of 1 second with a range of 136 years.  The
       (pseudo)random secret key is generated by the server and not
       shared with any other party.  The use of truncated HMAC-SHA-256
       is RECOMMENDED.  With a 32-bit timestamp and a 64-bit MAC, the
       size of the Echo option value is 12 bytes, and the server state
       is small and constant.  The security against an attacker guessing
       Echo option values is given by the MAC length.  If the server
       loses time continuity, e.g., due to reboot, the old key MUST be
       deleted and replaced by a new random secret key.  Note that the
       privacy considerations in Section 6 may apply to the timestamp.
       Therefore, it might be important to encrypt it.  Depending on the
       choice of encryption algorithms, this may require an
       initialization vector to be included in the Echo option value
       (see below).

       Echo option value:  timestamp t0, MAC(k, t0)

       Server State:  secret key k

       This method is suitable for both time-based and event-based
       freshness (by the server remembering the time at which the event
       took place) and independent of the client authority.

       If this method is used to additionally obtain network
       reachability of the client, the server MUST use the client's
       network address too, e.g., as in MAC(k, t0, claimed network

   3.  Persistent Counter.  This can be used in OSCORE for sequence
       number recovery, per Appendix B.1.2 of [RFC8613].  The Echo
       option value is a simple counter without integrity protection of
       its own, serialized in uint format.  The counter is incremented
       in a persistent way every time the state that needs to be
       synchronized is changed (in the case described in Appendix B.1.2
       of [RFC8613], when a reboot indicates that volatile state may
       have been lost).  An example of how such a persistent counter can
       be implemented efficiently is the OSCORE server Sender Sequence
       Number mechanism described in Appendix B.1.1 of [RFC8613].

       Echo option value:  counter

       Server State:  counter

       This method is suitable only if the client is the authority over
       the synchronized property.  Consequently, it cannot be used to
       show client aliveness.  It provides statements from the client
       similar to event-based freshness (but without a proof of

   Other mechanisms complying with the security and privacy
   considerations may be used.  The use of encrypted timestamps in the
   Echo option provides additional protection but typically requires an
   initialization vector (a.k.a. nonce) as input to the encryption
   algorithm, which adds a slight complication to the procedure as well
   as overhead.

Appendix B.  Request-Tag Message Size Impact

   In absence of concurrent operations, the Request-Tag mechanism for
   body integrity (Section 3.5.1) incurs no overhead if no messages are
   lost (more precisely, in OSCORE, if no operations are aborted due to
   repeated transmission failure and, in DTLS, if no packets are lost
   and replay protection is active) or when block-wise request
   operations happen rarely (in OSCORE, if there is always only one
   request block-wise operation in the replay window).

   In those situations, no message has any Request-Tag option set, and
   the Request-Tag value can be recycled indefinitely.

   When the absence of a Request-Tag option cannot be recycled any more
   within a security context, the messages with a present but empty
   Request-Tag option can be used (1 byte overhead), and when that is
   used up, 256 values from 1-byte options (2 bytes overhead) are

   In situations where that overhead is unacceptable (e.g., because the
   payloads are known to be at a fragmentation threshold), the absent
   Request-Tag value can be made usable again:

   *  In DTLS, a new session can be established.

   *  In OSCORE, the sequence number can be artificially increased so
      that all lost messages are outside of the replay window by the
      time the first request of the new operation gets processed, and
      all earlier operations can therefore be regarded as concluded.


   The authors want to thank Carsten Bormann, Roman Danyliw, Benjamin
   Kaduk, Murray Kucherawy, Francesca Palombini, and Jim Schaad for
   providing valuable input to the document.

Authors' Addresses

   Christian Amsüss
   Email: christian@amsuess.com

   John Preuß Mattsson
   Ericsson AB
   Email: john.mattsson@ericsson.com

   Göran Selander
   Ericsson AB
   Email: goran.selander@ericsson.com