Skip to main content

Media over QUIC Overview
draft-westerlund-moq-overview-00

Document Type Active Internet-Draft (individual)
Authors Magnus Westerlund , Zaheduzzaman Sarker
Last updated 2026-06-30
RFC stream (None)
Intended RFC status (None)
Formats
Stream Stream state (No stream defined)
Consensus boilerplate Unknown
RFC Editor Note (None)
IESG IESG state I-D Exists
Telechat date (None)
Responsible AD (None)
Send notices to (None)
draft-westerlund-moq-overview-00
moq                                                        M. Westerlund
Internet-Draft                                                  Ericsson
Intended status: Informational                                 Z. Sarker
Expires: 1 January 2027                                            Nokia
                                                            30 June 2026

                        Media over QUIC Overview
                    draft-westerlund-moq-overview-00

Abstract

   This document provides a high-level overview of the Media over QUIC
   (MoQ) protocol suite.  It describes the architecture, data model,
   transport protocol, streaming formats, and security mechanisms that
   together enable scalable low-latency media delivery over the
   Internet.  The document explains how these components relate to each
   other and how they are composed to create interoperable media
   applications.

About This Document

   This note is to be removed before publishing as an RFC.

   Status information for this document may be found at
   https://datatracker.ietf.org/doc/draft-westerlund-moq-overview/.

   Discussion of this document takes place on the Media over QUIC (MOQ)
   Working Group mailing list (mailto:moq@ietf.org), which is archived
   at https://mailarchive.ietf.org/arch/browse/moq/.  Subscribe at
   https://www.ietf.org/mailman/listinfo/moq/.

   Source for this draft and an issue tracker can be found at
   https://github.com/https://github.com/gloinul/draft-westerlund-moq-
   overview.

Status of This Memo

   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 https://datatracker.ietf.org/drafts/current/.

Westerlund & Sarker      Expires 1 January 2027                 [Page 1]
Internet-Draft                MoQ Overview                     June 2026

   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 Internet-Draft will expire on 1 January 2027.

Copyright Notice

   Copyright (c) 2026 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  . . . . . . . . . . . . . . . . . . . . . . . .   3
     1.1.  Motivation  . . . . . . . . . . . . . . . . . . . . . . .   4
     1.2.  Document Scope  . . . . . . . . . . . . . . . . . . . . .   4
     1.3.  Terminology . . . . . . . . . . . . . . . . . . . . . . .   5
   2.  Use Cases . . . . . . . . . . . . . . . . . . . . . . . . . .   6
     2.1.  Live Streaming Media Delivery . . . . . . . . . . . . . .   6
     2.2.  Real-Time Conferencing  . . . . . . . . . . . . . . . . .   7
     2.3.  Content Contribution and Primary Distribution . . . . . .   8
   3.  Architecture and Nodes  . . . . . . . . . . . . . . . . . . .   8
     3.1.  Roles . . . . . . . . . . . . . . . . . . . . . . . . . .   8
     3.2.  Relays  . . . . . . . . . . . . . . . . . . . . . . . . .   9
     3.3.  Deployment Topologies . . . . . . . . . . . . . . . . . .  10
   4.  Data Model  . . . . . . . . . . . . . . . . . . . . . . . . .  11
     4.1.  Hierarchy . . . . . . . . . . . . . . . . . . . . . . . .  11
     4.2.  Track Naming  . . . . . . . . . . . . . . . . . . . . . .  11
     4.3.  Properties  . . . . . . . . . . . . . . . . . . . . . . .  12
     4.4.  Object Immutability and Cacheability  . . . . . . . . . .  12
   5.  MOQT: The Transport Protocol  . . . . . . . . . . . . . . . .  13
     5.1.  Session Establishment . . . . . . . . . . . . . . . . . .  13
     5.2.  Publishing and Subscribing  . . . . . . . . . . . . . . .  14
     5.3.  Priorities and Delivery . . . . . . . . . . . . . . . . .  14
     5.4.  Extensibility and Session Management  . . . . . . . . . .  15
   6.  Streaming Formats . . . . . . . . . . . . . . . . . . . . . .  16
     6.1.  Role of a Streaming Format  . . . . . . . . . . . . . . .  16
     6.2.  MSF: MOQT Streaming Format  . . . . . . . . . . . . . . .  16

Westerlund & Sarker      Expires 1 January 2027                 [Page 2]
Internet-Draft                MoQ Overview                     June 2026

     6.3.  CMSF: CMAF-Compliant Streaming Format . . . . . . . . . .  17
     6.4.  Layering  . . . . . . . . . . . . . . . . . . . . . . . .  18
   7.  Security Architecture . . . . . . . . . . . . . . . . . . . .  18
     7.1.  Hop-by-Hop Security . . . . . . . . . . . . . . . . . . .  18
     7.2.  End-to-End Object Encryption  . . . . . . . . . . . . . .  19
     7.3.  Authorization . . . . . . . . . . . . . . . . . . . . . .  19
     7.4.  Privacy Considerations  . . . . . . . . . . . . . . . . .  20
   8.  Defining a MoQ Application  . . . . . . . . . . . . . . . . .  21
     8.1.  Application Design Choices  . . . . . . . . . . . . . . .  21
     8.2.  Existing Application Definitions  . . . . . . . . . . . .  22
     8.3.  Reading Guide . . . . . . . . . . . . . . . . . . . . . .  22
   9.  Security Considerations . . . . . . . . . . . . . . . . . . .  23
   10. IANA Considerations . . . . . . . . . . . . . . . . . . . . .  24
   11. Contributors  . . . . . . . . . . . . . . . . . . . . . . . .  24
   12. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . .  24
   13. Informative References  . . . . . . . . . . . . . . . . . . .  24
   Authors' Addresses  . . . . . . . . . . . . . . . . . . . . . . .  26

1.  Introduction

   Media over QUIC (MoQ) is an IETF working group that produces a suite
   of protocol specifications for scalable, low-latency delivery over
   the Internet.  The central specification is the MoQ Transport
   protocol (MOQT), a publish/subscribe protocol built on QUIC [RFC9000]
   that uses intermediate relays to achieve distribution at scale.
   Additional specifications build on MOQT to define streaming formats,
   media containers, and end-to-end security mechanisms.

   This document provides a high-level overview of these component
   protocols, how they relate to each other, and how they are composed
   to create interoperable applications.  It is intended as an entry
   point for implementers, reviewers, and protocol designers who want to
   understand the overall system before reading the individual normative
   specifications.

   Although the primary motivation of MoQ is media delivery, MOQT itself
   is content-agnostic — it transports opaque objects without knowledge
   of their payload.  Media-specific logic (codecs, containers, adaptive
   bitrate switching) is defined in separate streaming format
   specifications that layer on top of the transport.  This separation
   means the protocol can serve other applications that benefit from the
   same publish/subscribe, relay-based delivery model.

Westerlund & Sarker      Expires 1 January 2027                 [Page 3]
Internet-Draft                MoQ Overview                     June 2026

1.1.  Motivation

   Today's media delivery landscape uses separate protocols for
   different stages of the pipeline.  Content is typically ingested
   using protocols like RTMP, then repackaged at origin servers for
   distribution via HTTP-based protocols such as HLS or DASH.  This
   separation introduces repackaging overhead, limits end-to-end
   latency, and creates complexity at the boundaries between protocol
   domains.

   MoQ addresses this by providing a single protocol that can carry
   media from original publisher to end subscriber, including through
   intermediate relay nodes.  The design is driven by four goals:

   Latency:  TCP-based protocols suffer from head-of-line blocking and
      are slow to respond to congestion.  MoQ leverages QUIC's
      multiplexed streams and partial reliability to minimise queuing
      latency and react quickly to changing network conditions.

   Leveraging QUIC:  QUIC provides parallel streams with independent
      loss recovery, unreliable datagrams, stream priorities, and flow
      control.  MoQ maps its data model onto these features to achieve
      flexible trade-offs between latency and reliability.

   Convergence:  A single protocol from contribution through
      distribution eliminates the need for intermediate repackaging.
      The same MoQ objects produced by an encoder can be delivered to
      end subscribers without transformation at relay nodes.

   Relays:  Scalable media delivery requires intermediaries that can
      cache and fan out content.  MoQ treats relays as first-class
      protocol participants: object metadata is structured so that
      relays can route, prioritise, and cache content without accessing
      the media payload itself.

   The resulting protocol is a general-purpose pub/sub delivery layer.
   The media-specific aspects — how audio and video samples are
   packaged, how tracks are described in a catalog, how adaptive bitrate
   switching works — are defined by separate streaming format
   specifications that build on MOQT.  Applications outside media
   delivery can use MOQT directly with their own object formats, without
   adopting any of the media-specific layers.

1.2.  Document Scope

   This document is informational.  It does not define protocol
   behaviour or create new requirements.  Its purpose is to:

Westerlund & Sarker      Expires 1 January 2027                 [Page 4]
Internet-Draft                MoQ Overview                     June 2026

   *  Orient readers to the MoQ architecture and its components.

   *  Explain how the components relate to each other.

   *  Point readers to the relevant normative specification for each
      topic.

   The covered specifications are:

   *  MOQT (Media over QUIC Transport) [I-D.ietf-moq-transport] — the
      publish/subscribe transport protocol.

   *  MSF (MOQT Streaming Format) [I-D.ietf-moq-msf] — a streaming
      format defining media packaging, catalog, and timeline.

   *  CMSF (CMAF-compliant MSF) [I-D.ietf-moq-cmsf] — an extension of
      MSF for CMAF/ISO-BMFF packaged media.

   *  MoQ Secure Objects [I-D.ietf-moq-secure-objects] — end-to-end
      authenticated encryption for content objects.

   *  Privacy Pass Authentication [I-D.ietf-moq-privacy-pass-auth] —
      privacy-preserving token-based authorization.

   *  Common Access Tokens for MoQ (C4M) [I-D.ietf-moq-c4m] — bearer
      token authorization scheme.

   *  LOC (Low Overhead Container) [I-D.ietf-moq-loc] — a minimal media
      container format.

1.3.  Terminology

   This document uses the following terms.  Definitions attempt to be
   consistent with [I-D.ietf-moq-transport] but in case of discrepancy
   rely on the MOQT specification.

   Client:  The party initiating a Transport Session.

   Server:  The party accepting an incoming Transport Session.

   Publisher:  An endpoint that handles subscriptions by sending objects
      from a track.

   Subscriber:  An endpoint that subscribes to and receives tracks.

   Original Publisher:  The initial publisher of a given track — the
      entity that creates the media content.

Westerlund & Sarker      Expires 1 January 2027                 [Page 5]
Internet-Draft                MoQ Overview                     June 2026

   End Subscriber:  A subscriber that consumes the content and does not
      forward it to other subscribers.

   Relay:  An entity that is both a publisher and a subscriber,
      forwarding content between other endpoints.  Relays are neither
      the original publisher nor the end subscriber.

   Track:  A time-ordered sequence of groups, identified by a Full Track
      Name.  A track is the unit of subscription.

   Group:  A collection of objects within a track that forms a join
      point.  Groups are independently useful — objects within a group
      should not depend on objects in other groups.

   Object:  The smallest addressable data unit in MoQ.  An object is an
      immutable sequence of bytes identified by its track, group ID, and
      object ID.

   Subgroup:  A sequence of objects within a group that share a QUIC
      stream.  Objects in a subgroup have consistent priority and
      dependency relationships.

   Transport Session:  A QUIC connection or WebTransport session over
      which MOQT operates.

2.  Use Cases

   As noted in Section 1, MOQT is a content-agnostic publish/subscribe
   protocol.  It can carry any data that fits the track/group/object
   model — media is simply the use case that motivated its design.
   Other potential applications include IoT telemetry distribution,
   collaborative document state synchronisation, real-time game state
   replication, and financial market data feeds.

   This section describes the primary media use cases that have driven
   the protocol design.  They illustrate the requirements that shaped
   the architecture, but do not represent an exhaustive list of what
   MOQT can be used for.

2.1.  Live Streaming Media Delivery

   In live streaming, an original publisher encodes and publishes media
   content that is delivered through one or more relays to a potentially
   large number of end subscribers.

    Original                                            End
    Publisher --> Relay --> Relay --> ... --> Relay --> Subscribers
                                                        (many)

Westerlund & Sarker      Expires 1 January 2027                 [Page 6]
Internet-Draft                MoQ Overview                     June 2026

   Key characteristics:

   *  One-to-many distribution, typically through a tree of relays.

   *  Target end-to-end latency ranges from under one second to several
      seconds, depending on the application.

   *  Subscribers may join at any time and must be able to start
      playback from a recent group boundary.

   *  Adaptive bitrate (ABR) switching between alternate encodings of
      the same content allows subscribers to adapt to varying network
      conditions.

   *  The original publisher produces a catalog describing available
      tracks (resolutions, bitrates, codecs) so subscribers can make
      informed selections.

2.2.  Real-Time Conferencing

   In conferencing, multiple participants each act as both original
   publisher and end subscriber.  Each participant publishes their own
   audio and video tracks and subscribes to tracks from other
   participants.

    Participant A ---+            +--- Participant C
                     |            |
                     v            v
                 +-------+    +-------+
                 | Relay |<-->| Relay |
                 +-------+    +-------+
                     ^            ^
                     |            |
    Participant B ---+            +--- Participant D

   Key characteristics:

   *  Many-to-many communication with low participant counts (tens to
      hundreds).

   *  Target latency is interactive: below 500 milliseconds end-to-end.

   *  Each participant subscribes to a subset of available tracks (e.g.,
      active speakers only).

   *  Scalable video coding (SVC) or simulcast may be used, with the
      relay or subscriber selecting appropriate layers.

Westerlund & Sarker      Expires 1 January 2027                 [Page 7]
Internet-Draft                MoQ Overview                     June 2026

   *  End-to-end encryption is particularly important since relays
      should not have access to media content.

2.3.  Content Contribution and Primary Distribution

   In content contribution (ingest), a professional encoder at a
   production site publishes high-quality media to a relay operated by a
   distribution service (e.g., a CDN).

    Encoder/       CDN Ingress      CDN Edge
    Producer  -->  Relay       -->  Relay(s)  -->  Subscribers

   Key characteristics:

   *  The ingress relay receives the contributed content and may
      transcode it into multiple representations (resolutions,
      bitrates).

   *  After transcoding, the distribution service becomes the original
      publisher of the new representations — the transcoded tracks are
      distinct from the contributed track.

   *  Reliability is prioritised over latency for the contribution link.

   *  The same MOQT protocol is used for both the contribution link
      (encoder to ingress) and the distribution links (edge relays to
      subscribers), achieving the convergence goal.

3.  Architecture and Nodes

3.1.  Roles

   A MOQT deployment involves endpoints in the following roles:

   Publisher:  Sends objects belonging to one or more tracks.  A
      publisher responds to subscriptions and delivers matching objects
      to subscribers.

   Subscriber:  Requests tracks and receives objects.  A subscriber
      issues SUBSCRIBE or FETCH requests to obtain content.

   Original Publisher:  The endpoint that creates and first publishes a
      track's content.  The original publisher is responsible for the
      media encoding, object payload, and track properties.

   End Subscriber:  The final consumer of a track's content.  An end
      subscriber decodes and renders the media.

Westerlund & Sarker      Expires 1 January 2027                 [Page 8]
Internet-Draft                MoQ Overview                     June 2026

   A single endpoint may act in multiple roles simultaneously.  In
   conferencing, each participant is both an original publisher (for
   their own media) and an end subscriber (for other participants'
   media).

3.2.  Relays

   Relays are intermediary nodes that are both publishers (to downstream
   subscribers) and subscribers (to upstream publishers).  They
   terminate MOQT sessions on each side independently — a relay is not a
   transparent proxy.

   Relays provide:

   *  Fan-out: A single upstream subscription serves multiple downstream
      subscribers requesting the same track.

   *  Caching: Objects can be stored and served to subscribers that join
      later, without fetching from the original publisher again.

   *  Geographic distribution: Relays can be deployed close to
      subscribers to reduce latency.

   Relays are content-agnostic: they forward object payloads without
   interpreting or modifying them.  This enables end-to-end encryption
   where relays cannot access media content, while still performing
   their routing and caching functions based on object metadata (track
   names, group IDs, priorities, properties).

   Relays are enforcement points for access and authorization.  In many
   deployments that have any restriction on access of the content the
   relay is expected to be the node that receives proof of authorization
   and will be responsible to verify and periodically re-verify this
   proof before delivering content.

   Relays are also not necessarily application-agnostic when it comes to
   deployment and operation.  The configuration of relay topologies,
   authorization policies, caching strategies, and resource limits are
   application-specific concerns.  A relay serving a live streaming
   platform operates differently from one supporting a conferencing
   service, even though the protocol mechanics are identical.  Moreover,
   even though relays need not parse media payload formats in order to
   forward content, they may still apply application-aware policy based
   on visible metadata, naming conventions, and authorization state.

   For further discussion of relay operational considerations, including
   denial-of-service resilience, see [I-D.englishm-moq-relay-dos].

Westerlund & Sarker      Expires 1 January 2027                 [Page 9]
Internet-Draft                MoQ Overview                     June 2026

3.3.  Deployment Topologies

   MOQT supports several deployment topologies:

   Point-to-point:  A publisher connects directly to a subscriber with
      no intermediate relay.  Suitable for contribution ingest, AI
      produced dissemination, or testing.

    Publisher <--> Subscriber

   Single relay:  A publisher connects to a relay, which serves one or
      more subscribers.  The relay provides fan-out and caching.

    Publisher <--> Relay <--> Subscriber(s)

   Relay tree:  A publisher connects to an origin relay, which fans out
      to edge relays closer to subscribers.  This is the typical CDN
      deployment model.

    Publisher <--> Origin Relay <--> Edge Relay(s) <--> Subscriber(s)

   Relay mesh:  Relays interconnect with multiple peers, subscribing and
      publishing bidirectionally between each other.  Each relay both
      receives content from and forwards content to its neighbouring
      relays.  This topology suits conferencing and multi-publisher
      scenarios where content flows in multiple directions
      simultaneously and there is no single origin.

    Publisher A <--> Relay 1 <----> Relay 2 <--> Subscriber C
                      ^    ^       ^    ^
                      |     \     /     |
                      |      v   v      |
                      |     Relay 3     |
                      |       ^         |
                      v       |         v
            Subscriber B  Publisher D  Subscriber E

   In all topologies, each hop is an independent MOQT session.  Relays
   manage subscriptions on each session independently and can aggregate
   multiple downstream subscriptions into a single upstream
   subscription.

   MOQT supports graceful session migration: a relay that needs to shut
   down sends a GOAWAY message directing its peers to reconnect to a
   different server.  The peers then establish new sessions and re-issue
   their subscriptions, allowing relay maintenance or replacement, but
   continuity depends on object availability, timing, and application/
   session policy.

Westerlund & Sarker      Expires 1 January 2027                [Page 10]
Internet-Draft                MoQ Overview                     June 2026

4.  Data Model

   MOQT defines a hierarchical data model for organising media content.
   Understanding this model is essential to working with any part of the
   MoQ protocol suite.  The data model is defined in Section 2 of
   [I-D.ietf-moq-transport] and summarised below for convenience.

4.1.  Hierarchy

   The data model has four levels:

    Track
      └── Group (join point)
            └── Subgroup (stream-mapped unit)
                  └── Object (smallest addressable payload)

   Track:  A named, time-ordered sequence of groups.  Tracks are the
      unit of subscription — a subscriber requests a track by its Full
      Track Name and receives objects from that track.

   Group:  A temporal collection of objects within a track.  A group
      represents a join point: a subscriber can begin receiving content
      at any group boundary without needing prior groups.  For video, a
      group typically corresponds to a Group of Pictures (GOP) starting
      with a random access point.

   Subgroup:  A sequence of objects within a group that are sent on a
      single QUIC stream.  Objects in a subgroup share priority and have
      dependency relationships consistent with in-order delivery.
      Subgroups enable the priority mechanism to favour certain objects
      (e.g., base-layer frames) over others within the same group.

   Object:  The basic data element — an addressable, immutable sequence
      of bytes.  Each object is uniquely identified by its Full Track
      Name, group ID, and object ID.  Objects within a group are ordered
      by object ID.

4.2.  Track Naming

   Every track is identified by a Full Track Name consisting of two
   parts:

   Track Namespace:  An ordered tuple of one or more fields ( up to 32)
      (e.g., "example.com", "meeting123", "alice").  The namespace
      provides hierarchical structure that relays use for prefix-based
      routing and subscription matching.

   Track Name:  A byte sequence identifying a specific track within the

Westerlund & Sarker      Expires 1 January 2027                [Page 11]
Internet-Draft                MoQ Overview                     June 2026

      namespace (e.g., "video-hd", "audio", "catalog").

   The combination of namespace and track name is unique within a given
   scope.  A scope is defined by the server or set of servers that a
   client connects to — specifically, those sharing the same host and
   path in the connection URI.  Within a scope, no two distinct tracks
   can have the same Full Track Name, which means it can serve as a
   cache key at relays.

4.3.  Properties

   Tracks and objects can carry additional metadata called Properties.
   Properties are visible to relays and can influence how content is
   distributed.  There are two scopes of properties — track-wide and
   per-object — and within either scope, individual properties can
   optionally be marked as immutable to support end-to-end
   authentication.

   Track Properties:  Metadata associated with an entire track.
      Examples include default publisher priority, group order
      preference, maximum cache duration, and delivery timeouts.  Track
      Properties are communicated in control messages (PUBLISH,
      SUBSCRIBE_OK, FETCH_OK).

   Object Properties:  Per-object metadata carried in object headers.
      MOQT defines the framework for object properties, including gap
      indicators (Prior Group ID Gap, Prior Object ID Gap).  Media-
      specific properties such as timestamps and frame markings are
      defined by LOC and MSF for use with encoded audio/video samples.

   Immutable Properties:  Either track or object properties may be
      placed inside an Immutable Properties container.  Once set by the
      original publisher, these cannot be modified or removed by relays.
      This enables end-to-end authentication: subscribers can verify
      that immutable values have not been altered in transit.

   Properties are registered in IANA registries and use a Key-Value-Pair
   encoding.  Relays that do not understand a property must forward it
   unchanged.

4.4.  Object Immutability and Cacheability

   A fundamental property of MOQT objects is immutability: once an
   object is published, its payload must never change.  Two received
   objects with the same Full Track Name, group ID, and object ID must
   contain identical bytes.

   This guarantee enables:

Westerlund & Sarker      Expires 1 January 2027                [Page 12]
Internet-Draft                MoQ Overview                     June 2026

   *  Caching at relays without cache-invalidation complexity.

   *  Deduplication when the same object arrives from multiple upstream
      sources.

   *  End-to-end integrity verification — a subscriber can detect if an
      object has been tampered with.

   An object can transition from "exists" to "does not exist" (e.g.,
   cache expiry), but its contents cannot be modified.  The protocol
   signals object status (normal, end-of-group, end-of-track) to
   communicate lifecycle information.

5.  MOQT: The Transport Protocol

   Media over QUIC Transport (MOQT) [I-D.ietf-moq-transport] is the
   publish/subscribe protocol that carries objects between publishers
   and subscribers.  MOQT treats all object payloads as opaque byte
   sequences — it provides delivery, prioritisation, and caching without
   any knowledge of the content being carried.  This section provides an
   overview of its key mechanisms.

5.1.  Session Establishment

   MOQT runs over either native QUIC or WebTransport:

   *  Native QUIC: The client connects using the TLS Application-Layer
      Protocol Negotiation (ALPN) identifier "moqt" and identifies the
      server using a "moqt://" URI.  The authority and path are
      communicated via Setup Options.

   *  WebTransport: The client establishes a WebTransport session over
      HTTP/3, using an HTTPS URI derived from the moqt:// URI.

   Both provide stream multiplexing and secure transport, though the
   deployment model differs — particularly in browser-based environments
   where only WebTransport is available.  Applications should treat them
   as functionally similar substrates with some environment-specific
   constraints.

   After the transport connection is established, each endpoint opens a
   unidirectional control stream and sends a SETUP message.  The SETUP
   exchange negotiates the protocol version and extensions (Setup
   Options) before any other messages are exchanged.

Westerlund & Sarker      Expires 1 January 2027                [Page 13]
Internet-Draft                MoQ Overview                     June 2026

5.2.  Publishing and Subscribing

   MOQT provides several mechanisms for publishers and subscribers to
   exchange content:

   SUBSCRIBE:  A subscriber requests newly published objects for a
      track.  The publisher responds with SUBSCRIBE_OK and begins
      forwarding matching objects as they are produced.  Subscription
      filters (Largest Object, Next Group, AbsoluteStart, AbsoluteRange)
      control which objects are delivered.

   PUBLISH:  A publisher initiates a subscription by pushing content to
      a subscriber.  This inverts the usual direction — the publisher
      offers a track rather than waiting for a subscription request.

   FETCH:  A subscriber requests a specific range of previously
      published objects.  Unlike SUBSCRIBE, which delivers objects as
      they are produced going forward, FETCH retrieves objects that
      already exist.  A subscriber that joins a track mid-stream
      typically needs both: a SUBSCRIBE to receive new objects, and a
      FETCH (called a Joining Fetch) to retrieve the recent objects
      needed to start decoding — for example, the beginning of the
      current video group of pictures.

   Namespace Discovery:  Publishers advertise available namespaces via
      PUBLISH_NAMESPACE.  Subscribers discover content by sending
      SUBSCRIBE_NAMESPACE (to learn about matching namespaces) or
      SUBSCRIBE_TRACKS (to receive PUBLISH messages for tracks within
      matching namespaces).  Relays use namespace prefix matching to
      route subscriptions to the appropriate upstream publishers.

   All subscription interactions follow a state machine: Idle → Pending
   → Established → Terminated.  Either endpoint can terminate a
   subscription — the subscriber via cancellation, the publisher via
   PUBLISH_DONE.

5.3.  Priorities and Delivery

   When network capacity is insufficient to deliver all subscribed
   content, MOQT's priority system determines what gets delivered first:

   *  Subscriber Priority: Set per-subscription by the subscriber to
      express relative importance among its own subscriptions.

   *  Publisher Priority: Set per-object by the publisher to express
      relative importance within a track (e.g., base layer vs.
      enhancement layer).

Westerlund & Sarker      Expires 1 January 2027                [Page 14]
Internet-Draft                MoQ Overview                     June 2026

   *  Group Order: Ascending (oldest first) or descending (newest first)
      delivery of groups within a subscription.

   Under congestion, lower-priority content may be delayed or dropped
   entirely.  Two timeout mechanisms support controlled data loss:

   *  Object Delivery Timeout: If an object cannot be sent within this
      duration after its first byte was produced, it is discarded.

   *  Subgroup Delivery Timeout: If a completed subgroup's data is not
      fully acknowledged within this duration, the stream is reset —
      meaning any undelivered data on that stream is abandoned and will
      not be retransmitted.

   Objects can be sent on QUIC streams (reliable, in-order within the
   stream) or as QUIC datagrams (unreliable, low-latency).  The choice
   is the Object Forwarding Preference, set per-object by the original
   publisher.

   Client side Adaptive Bitrate (ABR) is possible by either ending a
   subscription and replace it with another track that is another
   represenation at another bit-rate than the previous one.  Steering
   these changes to be done at the next group boundary.

   The WG is working on sender side ABR where the publisher uses its
   understanding of the congestion situation to select the set of tracks
   to be published to the subscriber.

5.4.  Extensibility and Session Management

   MOQT is designed for evolution without breaking existing
   implementations:

   Setup Options:  Negotiated per-session in the SETUP exchange.
      Unknown options are ignored, allowing new features to be
      introduced without breaking older endpoints.

   Message Parameters:  Per-message metadata included in control
      messages.  Parameters are peer-to-peer and are not forwarded by
      relays.

   Properties:  Per-track and per-object metadata that is forwarded and
      cached by relays, unlike Message Parameters.  Properties use IANA-
      registered type codes.  Unknown properties must be forwarded
      unchanged.

   GREASE (Generate Random Extensions And Sustain Extensibility):  Reserved

Westerlund & Sarker      Expires 1 January 2027                [Page 15]
Internet-Draft                MoQ Overview                     June 2026

      code points that implementations may send to exercise their peers'
      handling of unknown values.  This ensures that deployed software
      does not break when new extensions are introduced.

   GOAWAY supports graceful session migration and can reduce disruption
   during relay maintenance or replacement, but continuity depends on
   object availability, timing, and application/session policy.

6.  Streaming Formats

   MOQT is content-agnostic — it transports opaque object payloads
   without interpreting them.  A streaming format defines how media
   content is packaged into MOQT objects and how subscribers discover
   and select tracks.  Streaming formats bridge the gap between raw
   media encodings and the MOQT data model.

6.1.  Role of a Streaming Format

   A streaming format specifies:

   *  How encoded media samples are mapped to MOQT objects and groups.

   *  A catalog format describing available tracks and their properties.

   *  Rules for time-alignment and ABR switching between tracks.

   *  Timeline information mapping MOQT group/object IDs to media
      presentation time.

   *  Any additional metadata tracks (events, logs, metrics).

   Without a streaming format, a subscriber receiving objects would have
   no way to discover what content is available, how to decode it, or
   how to synchronise multiple tracks.

6.2.  MSF: MOQT Streaming Format

   The MOQT Streaming Format (MSF) [I-D.ietf-moq-msf] is the primary
   streaming format defined for MoQ.  It uses LOC (Low Overhead
   Container) [I-D.ietf-moq-loc] packaging, where each encoded media
   sample (audio frame or video frame) is placed in a separate MOQT
   object.  This one-sample-per-object mapping allows the transport to
   make independent delivery decisions for each frame — for example,
   prioritising base-layer video frames over enhancement layers, or
   dropping an expired frame without affecting the delivery of
   subsequent frames.

   MSF defines:

Westerlund & Sarker      Expires 1 January 2027                [Page 16]
Internet-Draft                MoQ Overview                     June 2026

   Catalog:  A JSON track named "catalog" that describes all available
      tracks in a namespace.  The catalog includes codec information,
      bitrates, resolutions, language, and relationships between tracks
      (render groups for synchronised playback, alternate groups for ABR
      switching).  Delta updates allow efficient signalling of track
      additions and removals.

   Media Timeline:  An optional track that maps MOQT group/object IDs to
      media presentation timestamps and wallclock times.  This enables
      seeking and synchronisation.  A template mechanism supports
      regular patterns without requiring an explicit timeline track.

   Event Timelines:  Optional metadata tracks carrying time-synchronised
      application data (e.g., sports scores, ad insertion markers, GPS
      coordinates).

   Publish Tracks:  Tracks defined in the catalog to which the
      subscriber can publish data back (e.g., QoE metrics, logs).

   MSF supports content with target latencies ranging from real-time
   (under 500ms) to standard live (several seconds) to video-on-demand.

6.3.  CMSF: CMAF-Compliant Streaming Format

   CMSF [I-D.ietf-moq-cmsf] extends MSF to support CMAF (Common Media
   Application Format) packaged media.  It is designed for workflows
   that use ISO-BMFF containers and existing DRM systems.

   Key differences from base MSF:

   *  Objects contain CMAF Chunks rather than raw codec samples.  A CMAF
      Chunk is the elementary media segment structure used by ISO-BMFF-
      based streaming (the container format behind DASH and HLS fMP4),
      consisting of metadata and media data boxes.

   *  Initialization segments (CMAF Headers) are carried in the catalog
      as base64-encoded inline data.

   *  Content protection uses Common Encryption (CENC) with commercial
      DRM systems (Widevine, PlayReady, FairPlay) rather than the
      application-layer encryption defined by MoQ Secure Objects.

   *  SAP-type event timelines signal stream access points for seeking.

   CMSF targets deployments that need interoperability with existing
   DASH/HLS ecosystems and hardware-based content decryption modules.

Westerlund & Sarker      Expires 1 January 2027                [Page 17]
Internet-Draft                MoQ Overview                     June 2026

6.4.  Layering

   The relationship between the components is:

    +------------------+
    | Application      |
    +------------------+
    | Streaming Format |  <-- MSF or CMSF
    | (catalog, rules) |
    +------------------+
    | Media Container  |  <-- LOC or CMAF
    +------------------+
    | MOQT             |  <-- transport protocol
    +------------------+
    | QUIC / WebTrans  |
    +------------------+
    | UDP / IP         |
    +------------------+

   The streaming format layer defines the media semantics.  The
   container layer defines how individual samples are serialised.  MOQT
   provides the delivery machinery.  Applications select the appropriate
   combination for their needs.

7.  Security Architecture

   MoQ employs a layered security model.  Each layer addresses different
   threats and operates at a different scope.

7.1.  Hop-by-Hop Security

   MOQT runs over QUIC or WebTransport, both of which use TLS 1.3 for
   the underlying connection.  This provides:

   *  Confidentiality and integrity of all data on each hop.

   *  Server authentication (and optionally mutual authentication).

   *  Protection against on-path attackers between adjacent nodes.

   However, hop-by-hop security does not protect content from relays.  A
   relay terminates the TLS session and has access to all MOQT message
   contents, including object payloads, unless additional end-to-end
   protection is applied.

Westerlund & Sarker      Expires 1 January 2027                [Page 18]
Internet-Draft                MoQ Overview                     June 2026

7.2.  End-to-End Object Encryption

   MoQ Secure Objects [I-D.ietf-moq-secure-objects] provides end-to-end
   authenticated encryption of media object payloads.  It is based on
   the SFrame mechanism ([RFC9605]).

   The scheme:

   *  Encrypts the object payload and optional encrypted properties.

   *  Authenticates the group ID, object ID, Full Track Name, and
      immutable properties.

   *  Derives per-track keys from a shared base key using HKDF (a
      standard key derivation function that deterministically produces
      cryptographic keys from input material).

   *  Forms a unique nonce (a value used exactly once) for each object
      from the group ID and object ID, ensuring that no two objects are
      encrypted with the same parameters.

   Relays can still route and cache objects based on unencrypted
   metadata (track names, group IDs, priorities, object properties) but
   cannot access the media content.  The architecture separates metadata
   required for routing/caching from metadata that can be authenticated
   or encrypted end-to-end.  The exact protected elements depend on the
   object security scheme and application profile.

   Key distribution is out of scope for the MoQ specifications —
   applications use external key management systems (e.g., MLS-based key
   distribution for conferencing [RFC9750],[RFC9420], or out-of-band
   provisioning for streaming).

7.3.  Authorization

   In general, MOQT defines where authorization material can be carried
   in protocol exchanges, while the semantic interpretation and
   verification procedure depend on the selected authorization scheme
   and the relay or publisher’s local policy.

   MOQT provides a generic AUTHORIZATION TOKEN parameter that can be
   included in control messages (SUBSCRIBE, FETCH, PUBLISH,
   PUBLISH_NAMESPACE, and others).  Both publishers and subscribers use
   this single mechanism to present credentials; relays verify the token
   before forwarding content or establishing upstream subscriptions.
   Tokens can be registered with session-scoped aliases to avoid
   retransmitting large values on every message.

Westerlund & Sarker      Expires 1 January 2027                [Page 19]
Internet-Draft                MoQ Overview                     June 2026

   The working group has defined two authorization schemes that operate
   over this mechanism:

   *  Privacy Pass Authentication [I-D.ietf-moq-privacy-pass-auth] —
      provides privacy-preserving, unlinkable authorization using
      Privacy Pass tokens.  It supports fine-grained access control
      through namespace and track name matching rules, and defines
      mechanisms for continuous re-authorization over long-lived
      sessions — either by obtaining multiple tokens in advance (batched
      issuance) or by exchanging a validated token for fresh ones
      directly with the relay (reverse issuance flow).

   *  Common Access Tokens (C4M) [I-D.ietf-moq-c4m] — provides a bearer
      token scheme using structured tokens with claims that express
      authorization scope.  C4M follows a model familiar from HTTP-based
      systems.

   The choice between schemes depends on application requirements:
   Privacy Pass prioritises subscriber anonymity, while C4M offers
   simpler integration for deployments with existing identity
   infrastructure.

   The streaming format catalog can signal authorization requirements
   per-track (via the authInfo field in MSF), allowing subscribers to
   discover what credentials they need before attempting to subscribe.

7.4.  Privacy Considerations

   MoQ's relay-based architecture means that intermediaries necessarily
   observe certain metadata in the course of routing and caching
   content.  Key privacy considerations include:

   *  Relay visibility of metadata: Track namespaces, track names, group
      IDs, and object properties are visible to every relay in the
      delivery path.  This metadata can reveal information about content
      and participants even when payloads are end-to-end encrypted.

   *  Traffic analysis: Object sizes, timing patterns, and subscription
      behaviour can allow relays or on-path observers to infer what
      content is being consumed or who is communicating.

   *  Subscriber unlinkability: Without privacy-preserving authorization
      (such as Privacy Pass), relays can correlate a subscriber's
      identity across multiple sessions or subscriptions.

Westerlund & Sarker      Expires 1 January 2027                [Page 20]
Internet-Draft                MoQ Overview                     June 2026

   MOQT provides mechanisms that applications can use to mitigate these
   risks, including padding (to obscure traffic patterns) and token
   schemes that support unlinkable authentication.  However, achieving
   strong privacy requires deliberate application design — for example,
   using generic namespace structures and consistent object sizes.

8.  Defining a MoQ Application

   MOQT, streaming formats, container formats, and security mechanisms
   are composable building blocks.  A complete MoQ application is
   defined by specifying which components to use and how they are
   configured.  This set of choices is also what creates an
   interoperability point: two implementations that make the same
   choices can exchange content without additional negotiation.

8.1.  Application Design Choices

   Defining a MoQ application requires decisions in each of the
   following areas:

   Transport version:  Which MOQT version the application requires, as
      identified by the protocol's ALPN string (the identifier used
      during TLS connection setup to select the application protocol).
      This determines the available protocol features.

   Streaming format and catalog:  How media or data is packaged into
      MOQT objects, and how available tracks are described to
      subscribers.  An application may adopt an existing format (MSF,
      CMSF) or define a new one.

   Container and codecs:  What container format wraps the encoded
      samples (e.g., LOC, CMAF) and which codecs are supported.

   Security:  Whether end-to-end encryption is required (and which
      scheme), and which authorization scheme(s) are used to control
      access.

   Relay deployment model:  How relays are deployed and operated — tree,
      mesh, or hybrid — and what trust and authorization policies apply
      between nodes.

   Extensions:  Any mandatory MOQT extensions (Properties, Setup
      Options) that endpoints must support.  Applications that need
      relay-visible metadata beyond what MOQT defines can register new
      Properties or use the application-reserved type ranges.

Westerlund & Sarker      Expires 1 January 2027                [Page 21]
Internet-Draft                MoQ Overview                     June 2026

   If existing components do not cover an application's needs, the MoQ
   framework supports extension without modifying the base protocol: new
   streaming formats can define their own catalog structures and
   packaging rules, new Properties can be registered via IANA, and new
   event timeline types can be defined for synchronised metadata.

8.2.  Existing Application Definitions

   The MoQ working group has defined two complete application
   specifications that illustrate the design choices above:

   MSF (MOQT Streaming Format):  Uses MOQT + LOC packaging + a JSON
      catalog.  Targets real-time and live streaming with LOC's minimal
      per-sample overhead.  Supports ABR switching via time-aligned
      alternate groups, media timelines, event timelines, and publish
      tracks for subscriber-to-platform communication.  Defined in
      [I-D.ietf-moq-msf].

   CMSF (CMAF-compliant Streaming Format):  Extends MSF with CMAF/ISO-
      BMFF packaging and Common Encryption (CENC) for DRM integration.
      Targets deployments that interoperate with existing DASH/HLS
      ecosystems and hardware content decryption.  Defined in
      [I-D.ietf-moq-cmsf].

   Both specifications make concrete choices for every area listed in
   Section 8.1, producing fully interoperable application definitions.

8.3.  Reading Guide

   The following table maps common goals to the relevant specifications:

Westerlund & Sarker      Expires 1 January 2027                [Page 22]
Internet-Draft                MoQ Overview                     June 2026

    +==============================+==================================+
    | Goal                         | Specification                    |
    +==============================+==================================+
    | Understand the wire protocol | [I-D.ietf-moq-transport]         |
    | and session mechanics        |                                  |
    +------------------------------+----------------------------------+
    | Build a live streaming or    | [I-D.ietf-moq-msf]               |
    | conferencing application     |                                  |
    | using LOC                    |                                  |
    +------------------------------+----------------------------------+
    | Use CMAF packaging with DRM  | [I-D.ietf-moq-cmsf]              |
    +------------------------------+----------------------------------+
    | Add end-to-end encryption    | [I-D.ietf-moq-secure-objects]    |
    +------------------------------+----------------------------------+
    | Implement privacy-preserving | [I-D.ietf-moq-privacy-pass-auth] |
    | authorization                |                                  |
    +------------------------------+----------------------------------+
    | Implement bearer-token       | [I-D.ietf-moq-c4m]               |
    | authorization                |                                  |
    +------------------------------+----------------------------------+
    | Understand the LOC container | [I-D.ietf-moq-loc]               |
    | format                       |                                  |
    +------------------------------+----------------------------------+
    | Operate relays and           | [I-D.englishm-moq-relay-dos]     |
    | understand DoS resilience    |                                  |
    +------------------------------+----------------------------------+

                                  Table 1

9.  Security Considerations

   The MoQ security model is described in Section 7.  The key residual
   risks are:

   Relay Trust:  Relays see all MOQT metadata and, without E2E
      encryption, all object payloads.  Deployments must carefully
      consider which entities operate relays and what authorization they
      require.

   Cache Poisoning:  If a relay accepts objects from an unauthorized
      publisher, it may serve corrupted content to subscribers.  Relay
      implementations must verify publisher identity and authorization
      before caching.

   Traffic Analysis:  Even with E2E encryption, object sizes, timing
      patterns, and track names can reveal content characteristics.
      Padding and generic naming mitigate but do not eliminate this
      risk.

Westerlund & Sarker      Expires 1 January 2027                [Page 23]
Internet-Draft                MoQ Overview                     June 2026

   Authorization Token Replay:  Token replay protection is delegated to
      the specific token scheme.  Implementations must select schemes
      with appropriate replay resistance for their threat model.

   Amplification:  A single SUBSCRIBE can trigger unbounded data flow
      from a publisher or relay.  Rate limiting, per-session resource
      caps, and the EXCESSIVE_LOAD error code provide defences.  See
      [I-D.englishm-moq-relay-dos] for detailed analysis.

10.  IANA Considerations

   This document has no IANA actions.

11.  Contributors

   The following persons have contributed to this document:

   *  Gwendal Simon

12.  Acknowledgments

13.  Informative References

   [RFC9000]  Iyengar, J., Ed. and M. Thomson, Ed., "QUIC: A UDP-Based
              Multiplexed and Secure Transport", RFC 9000,
              DOI 10.17487/RFC9000, May 2021,
              <https://www.rfc-editor.org/info/rfc9000>.

   [RFC9605]  Omara, E., Uberti, J., Murillo, S. G., Barnes, R., Ed.,
              and Y. Fablet, "Secure Frame (SFrame): Lightweight
              Authenticated Encryption for Real-Time Media", RFC 9605,
              DOI 10.17487/RFC9605, August 2024,
              <https://www.rfc-editor.org/info/rfc9605>.

   [RFC9420]  Barnes, R., Beurdouche, B., Robert, R., Millican, J.,
              Omara, E., and K. Cohn-Gordon, "The Messaging Layer
              Security (MLS) Protocol", RFC 9420, DOI 10.17487/RFC9420,
              July 2023, <https://www.rfc-editor.org/info/rfc9420>.

   [RFC9750]  Beurdouche, B., Rescorla, E., Omara, E., Inguva, S., and
              A. Duric, "The Messaging Layer Security (MLS)
              Architecture", RFC 9750, DOI 10.17487/RFC9750, April 2025,
              <https://www.rfc-editor.org/info/rfc9750>.

Westerlund & Sarker      Expires 1 January 2027                [Page 24]
Internet-Draft                MoQ Overview                     June 2026

   [I-D.ietf-moq-transport]
              Nandakumar, S., Vasiliev, V., Swett, I., and A. Frindell,
              "Media over QUIC Transport", Work in Progress, Internet-
              Draft, draft-ietf-moq-transport-18, 12 May 2026,
              <https://datatracker.ietf.org/doc/html/draft-ietf-moq-
              transport-18>.

   [I-D.ietf-moq-msf]
              Law, W. and S. Nandakumar, "MOQT Streaming Format", Work
              in Progress, Internet-Draft, draft-ietf-moq-msf-01, 2 June
              2026, <https://datatracker.ietf.org/doc/html/draft-ietf-
              moq-msf-01>.

   [I-D.ietf-moq-cmsf]
              Law, W., "CMSF- a CMAF compliant implementation of MOQT
              Streaming Format", Work in Progress, Internet-Draft,
              draft-ietf-moq-cmsf-01, 3 June 2026,
              <https://datatracker.ietf.org/doc/html/draft-ietf-moq-
              cmsf-01>.

   [I-D.ietf-moq-secure-objects]
              Jennings, C. F., Nandakumar, S., and R. Barnes, "End-to-
              End Secure Objects for Media over QUIC Transport", Work in
              Progress, Internet-Draft, draft-ietf-moq-secure-objects-
              00, 2 March 2026, <https://datatracker.ietf.org/doc/html/
              draft-ietf-moq-secure-objects-00>.

   [I-D.ietf-moq-privacy-pass-auth]
              Nandakumar, S., Jennings, C. F., and T. Meunier, "Privacy
              Pass Authentication for Media over QUIC (MoQ)", Work in
              Progress, Internet-Draft, draft-ietf-moq-privacy-pass-
              auth-02, 2 March 2026,
              <https://datatracker.ietf.org/doc/html/draft-ietf-moq-
              privacy-pass-auth-02>.

   [I-D.ietf-moq-c4m]
              Law, W., Lemmons, C., Simon, G., and S. Nandakumar,
              "Authorization scheme for MOQT using Common Access
              Tokens", Work in Progress, Internet-Draft, draft-ietf-moq-
              c4m-01, 18 June 2026,
              <https://datatracker.ietf.org/doc/html/draft-ietf-moq-c4m-
              01>.

Westerlund & Sarker      Expires 1 January 2027                [Page 25]
Internet-Draft                MoQ Overview                     June 2026

   [I-D.ietf-moq-loc]
              Zanaty, M., Nandakumar, S., and P. Thatcher, "Low Overhead
              Media Container", Work in Progress, Internet-Draft, draft-
              ietf-moq-loc-02, 15 March 2026,
              <https://datatracker.ietf.org/doc/html/draft-ietf-moq-loc-
              02>.

   [I-D.ietf-webtrans-overview]
              Kinnear, E. and V. Vasiliev, "The WebTransport Protocol
              Framework", Work in Progress, Internet-Draft, draft-ietf-
              webtrans-overview-12, 2 March 2026,
              <https://datatracker.ietf.org/doc/html/draft-ietf-
              webtrans-overview-12>.

   [I-D.englishm-moq-relay-dos]
              English, M., Pardue, L., Sharma, A., and I. Swett,
              "Denial-of-Service Considerations for Media over QUIC
              Relay Deployments", Work in Progress, Internet-Draft,
              draft-englishm-moq-relay-dos-00, 2 March 2026,
              <https://datatracker.ietf.org/doc/html/draft-englishm-moq-
              relay-dos-00>.

Authors' Addresses

   Magnus Westerlund
   Ericsson
   Email: magnus.westerlund@ericsson.com

   Zaheduzzaman Sarker
   Nokia
   Email: zaheduzzaman.sarker@nokia.com

Westerlund & Sarker      Expires 1 January 2027                [Page 26]