INTERNATIONAL ORGANIZATION FOR STANDARDIZATION
ORGANISATION INTERNATIONAL NORMALISATION
ISO/IEC JTC1/SC29/WG11
CODING OF MOVING PICTURES AND ASSOCIATED AUDIO

ISO/IEC JTC1/SC29/WG11 N1559
MPEG97
21 February 1997

Source: DSM-CC Sub-Group
Status: Approved
Title: Answers to the DSM-CC FAQ Version 1.0

DSM-CC FAQ Version 1.0

 

1. About DSM-CC (overall)
1.1 Are there any articles or tutorial kits about DSM-CC ?
1.2 I'm confused. Isn't DSM-CC a video stream VCR-like control protocol?
1.3 What organizations have adopted DSM-CC?
1.4 What's the difference between DSM-CC and DAVIC protocol?
1.5 Is there a reference implementation of DSM-CC (U-N, U-U, download, carousel)?
1.6 Can I use just one element of the standard, or must I implement all the protocols and interfaces?
1.7 Why is the DSM-CC specification document so large?
1.8 What are the different connection types required in DSM-CC? How is low bandwidth accommodated for each of the connections?
1.9 Is DSM-CC applicable to markets other than video dialtone? What are the applications that can take advantage of DSM-CC?
1.10 Is DSM-CC only used with set top boxes?
1.11 What is the relation between DSM-CC and TINA?
1.12 Does Internet make DSM-CC redundant with RSVP, RTSP, RTP and SCP?
1.13 What companies are developing DSM-CC implementations?
1.14 The Session Resource Manager is a key element of DSM-CC and a single point of failure. What kind of back-up precautions can be taken on the SRM to secure safe operation of the system?
1.15 What is the minimum system setup when using DSM-CC? (Low bandwidth system, no upstream channel etc.)
1.16 What are the procedures for conformance testing DSM-CC?

2. About DSM-CC U-N Session Control
2.1 What is the scope of the session protocol?
2.2 What is DSM-CC session ? What is the purpose of DSM-CC session control?
2.3 What is the difference between the DSM-CC session control and the call control in telecommunication networks?
2.4 What set of problems does it solve that aren't solved by other protocols such Q.2931, RSVP, H.245?
2.5 What is the relation between the DSM-CC AddResource operation and the set-up/connection operation of existing resource control protocols e.g., (Q.2931, RSVP)?
2.6 What is meant by DSM-CC allows the use of heterogeneous networks?
2.7 There is only one SRM in the DSM-CC U-N reference model. The situation where the Client and the Server reside in domains of different SRMs seems to be ignored. Can DSM-CC U-N globally scale up?
2.8 Is the session protocol valid for broadcast scenarios?

3. About DSM-CC U-N Configuration
3.1 What is the difference between BOOTP or DHCP and DSM-CC UN Config?

4. About DSM-CC U-U
4.1 What is meant by DSM-CC allows the use of heterogeneous networks?
4.2 Why does DSM-CC U-U use CORBA? Do all the DSM-CC protocols require CORBA, IDL, RPC?
4.3 What is the relationship of the object classes of the Service Domain (User-User) to specific object domains like CORBA, Java, and COM? Does the client have to implement a complete Object Request Broker?
4.4 What is the scope of the Service Domain interfaces? What functions does it provide?
4.5 Is the interfaceof the Service Domain valid for broadacst scenarios?

5. About DSM-CC Download
5.1 How does the download protocol differ from download protocols such as Trivial File Transfer Protocol?

6. About DSM-CC SDB-CCP
6.1 What is the relation between the DSM-CC SDB-CCP protocol and the DAVIC ZAP protocol?

7. About DSM-CC Section (Transport)

 

About DSM-CC (overall)

Are there any articles or tutorial kits about DSM-CC ?

<In English>
Vahe Balabanian et al, Digital Storage of Media Command and Control Protocol Applied to ATM IEEE JSAC Vol. 14 No. 6, August 1996
Vahe Balabanian et al, An Introduction to Digital Storage Media Command and Control Protocol (DSM-CC) IEEE Communication, November 1996.
DSM-CC Tutorial

<In Japanese>
Masa Kawashima, ITE Mag. (reference to be made available)
Masa, Kawashima IEICE-J Mag. (reference to be made available)

 

I'm confused. Isn't DSM-CC a video stream VCR-like control protocol?

DSM-CC (Digital Storage Media Command and Control) work began early on to provide VCR like functions as an annex to MPEG-2 TS ISO/IEC 13818-1. However, it subsequently expanded in MPEG-2 ISO/IEC 13818-6 to provide a Client-Network-Server paradigm for selection, access and control of distributed video sources. The ISO/IEC 13818-6 now includes VCR like functions in the Stream Control within a larger set of function shown below: - User-Network Client Configuration
- User-Network Session protocol
- User-User Directory, Stream control, File access
- Interactive and broadcast download
- Broadcast object carousel
- Switched digital broadcast channel change protocol

While all the above functions can be implemented together with some obvious synergism, a subset of the functions can also be implemented with other non-DSM-CC functions. In other words DSM-CC IS is a tool-kit.

What organizations have adopted DSM-CC?

Since DSM-CC IS is a tool-kit, an industry consortium can choose to implement any subset of DSM-CC IS. The solutions of the problems in those organizations are mirrored in the functions selected from DSM-CC.

  1. DAVIC has adopted:
    - User-Network Client Configuration
    - User-Network Session protocol
    - User-User Directory, Stream control, File access
    - Interactive and broadcast download
    - Broadcast object carousel
    - Switched digital broadcast channel change protocol
  2. DVB has adopted:
    - User-Network Session protocol
    - User-User Directory, Stream control, File access
    - Interactive and broadcast download
    - Broadcast object carousel
  3. ATM Forum:
    - Audiovisual Multimedia Services (AMS) document makes use of the User-Network
    Session protocol to control the AMS session.
  4. ITU-T SG16 group working on the Audio-Video-Graphics conferencing specification shows its operation with DSM-CC in Annex A of T.131
  5. ATSC (Advanced Television Systems Committee) is expected to adopt DSM-CC
  6. FIPA (Foundation for Intelligent Physical Agents) in its Layer Model uses DSM-CC under ORB and TRADER
  7. TINA-C has shown interest because of common technology with DSM-CC, but no formal relationship has been established

What's the difference between DSM-CC and DAVIC protocol?

DAVIC is a consumer of standards, that selects specific pieces and provides restrictions for interoperability. DSM-CC IS as a standard covers the needs of a universe of DSM-CC implementers through:

  1. options
  2. implementers defined fields for extensions
  3. DSM-CC fields the scope of which is left to implementers
  4. values for parameters to be selected by implementers, such as timers

DAVIC as a group of implementers that seek interoperability has chosen a set of options from the DSM-CC IS specification, e.g., carriage of DSM-CC U-N session commands over UDP and TCP. DAVIC also extended the resource descriptors in DSM-CC IS and added network dependent variables. For example DAVIC Client ID, specified in DSM-CC IS as any 20 byte ISAP field, to be E.164 NSAP with the HO-DSP containing the IP address for the stream control. The preceding are just but few examples.

Is there a reference implementation of DSM-CC (U-N, U-U, download, carousel)?

No reference implementations exist at the present time, however an example implementation has been graciously made available as shown below:

In Jan. 24, 1997, GCL (Graphics Communication Laboratories) announced that it is offering free Distribution of DSM-CC Software, compliant to the ISO/IEC 13818-6 International Specification. The software is in executable binary files, without source codes. The following can be downloaded free of charge from the GCL web page:
- DSM-CC Server System Software,
- DSM-CC Client Lib. + MHEG-5 Engine Software

The ensuing conditions apply:
- Expire on April 24, 1997
- No warranty
- No reverse engineering etc.

Please visit
http://www.gctech.co.jp/

Can I use just one element of the standard, or must I implement all the protocols and interfaces?

The ISO/IEC 13818-6 DSM-CC IS has assembled together a number of highly synergistic functional groups. Although each one can be implemented separately, value can be gained from their synergy in terms of reduced cost of implementation, interoperability and portability. The functional groups are:
- User-Network Client Configuration
- User-Network Session protocol
- User-User Directory, Stream control, File access
- Interactive and broadcast download
- Broadcast object carousel
- Switched digital broadcast channel change protocol

 

Why is the DSM-CC specification document so large?

Many other standards have each of the equivalent parts to DSM-CC specified in separate specifications. If all of these specifications are brought together which may take considerable effort by the user, the user not only is left with a bigger document but may encounter potential duplications and controversies that need to be resolved. It is left to implementers to find his/her way through the multiplicity of the standards in order to integrate the specifications into one implementation and ensure interoperability with other such vendors who use the same specifications. The experts that specified DSM-CC spent their time and effort in order to bring all aspects of multimedia services into one specification and facilitate the work of the potential DSM-CC implementers.

What are the different connection types required in DSM-CC? How is low bandwidth accommodated for each of the connections?

There are three types of connections required in DSM-CC, a) for Session control messages, b) for video streams c) for control of the video streams and access to other applications. The U-U messages on the other hand could cover a range based on the application used and the QoS chosen by the end-user.

  1. The DSM-CC U-N messages are out-of-band control. They are comparable in length to Q.2931 messages and do not require broadband connections for their operation.
  2. For the DSM-CC U-U video bitstream, the MPEG-2 coding dictates the QoS for the connections. No limitation on bandwidth is inherent to the DSM-CC User-User specification.
  3. For the DSM-CC U-U RPC, a light CDR is defined in the DSM-CC User-User specification to optimize the use of the bandwidth.

 

Is DSM-CC applicable to markets other than video dialtone? What are the applications that can take advantage of DSM-CC?

The DSM-CC IS was targeted to meet the video dialtone market. However, the concepts used in DSM-CC satisfy a wider range of multimedia services.

The applications that can take advantage of DSM-CC are:

Video on Demand

Data broadcast

Broadband Internet access

AudioVideoGraphic Conferencing using ITU-T T.120 recommendations

And others

The DSM-CC work was extended as part of the MPEG-4 activity ISO/IEC 14496-6, with the DSM-CC Multimedia Integration Framework (DMIF). DMIF will expand upon the current DSM-CC standard (ISO/IEC 13818-6) to enable the convergence of interactive, broadcast and conversational multimedia into one specification which will be applicable to set tops, desktops and mobile stations. DMIF encompasses the following directions:

  1. An end-to-end ìsessionî across multiple network provider implementations
  2. Integration with multiple network technologies, such as ATM and the Internet
  3. Integration with specific object domains, such as CORBA and JAVA
  4. Multiple devices participating as peers within the same session
  5. Fully symmetric consumer and producer operations within a single device
  6. The scheduling and real-time switching/multiplexing of bit streams

 

Is DSM-CC only used with set top boxes?

The benefits realized by DSM-CC for set top boxes such as, U-N Configuration, small footprint session control and lightweight download, in order to cope with their limited resources, can equally benefit PC desktops or PCTVs, devices such as Network Computers and devices such as Web-TVs. DSM-CC adds negligible overhead to such devices.

DSM-CC is also used in cable modems as defined in DAVIC 1.2 specification for broadband Internet access.

 

What is the relation between DSM-CC and TINA?

Although there is no formal relationship between DSM-CC and TINA, DSM-CC and TINA are built on similar precepts. DSM-CC U-U and TINA use IDL for common object concepts defined by OMG. The session concepts in DSM-CC U-N and TINA are similar, however TINA does not provide the protocol specifics or the state machines.

 

Does Internet make DSM-CC redundant with RSVP, RTSP, RTP and SCP?

No, DSM-CC is not made redundant. DSM-CC works in conjunction with these protocols.

There are two main components of DSM-CC, U-N and U-U. Equivalence can be drawn between U-N and RSVP on the one hand and U-U and RTSP, RTP and SCP on the other hand.

RSVP is not able to account for the stacks above which IP packets are carried. DSM-CC U-N on the other hand accounts for the layers of transport above which IP packets are carried as network connection resources. In addition DSM-CC can account for network processing resources which are used, such as multicast servers and translators. Since DSM-CC U-N can be implemented independently from DSM-CC U-U it can operate with RTSP, RTP and SCP.

RTSP provides the equivalent of the DSM-CC U-U directory resolve and stream control while RTP and SCP provide the equivalent of MPEG-2 TS and allows for the selection of a transport within the confines of connections between the sender and the receiver (or receivers for multicast). In comparison DSM-CC U-U allows for far more capable stream control through the use of Normal Play Time (NPT) state diagrams. DSM-CC also allows for a request for transport connection between the receiver and any sender it identifies for reasons of geographical proximity to the receiver or for load balancing with off hour traffic troughs in less used locations. The result of both leads to an efficient use of underlying connection resources and video server ports.

What companies are developing DSM-CC implementations?

The companies that are developing DSM-CC implementations are:
Bellcore

What is the minimum system setup when using DSM-CC? (Low bandwidth system, no upstream channel etc.)

This question can be answered in many parts. First of all DSM-CC consists of several functional groups. While all the functions can be implemented together with some obvious synergism, a subset of the functions can also be implemented with other non-DSM-CC functions. For example the SDB (Switched Digital Broadcast) CCP (Channel Change Protocol) can be implemented with or without DSM-CC U-N Session messages. In the latter case it may be considered a minimal implementation while in the former it is a more DSM-CC involved implementation.

  1. The DSM-CC Session U-N is broken down to core or extended portions. All implementations must provide the core portion. In that sense the minimal DSM-CC implementation consists of the core implementation.
  2. The DSM-CC U-U is broken down to the following portions:
    1. Core consumer
    2. Enhanced consumer
    3. Core producer
    4. Enhanced producer

The minimal implementation consists of the core consumer portion.

  1. Within the core consumer, the minimal implementation of DSM-CC consists of broadcast download, and broadcast object carousel, with no upstream bandwidth and DSM-CC session U-N.

 

What are the procedures for conformance testing DSM-CC?

The conformance testing document ISO/IEC 13818-10 for DSM-CC IS is being prepared by the DSM-CC group. The Preliminary Working Draft (PWD) is targeted for February 21, 1997 and the International Standard (IS) is targeted for July 1998. Organizations that have shown interest in conformance testing are Bellcore, Columbia University, Jupiter and NIST.

About DSM-CC U-N Session Control

What is the scope of the session protocol?

A DSM-CC session groups all the resources used in an instance of a service. This is accomplished by tagging the resources with the network wide unique session ID. Consequently, the logging of all the resources for later use in billing is made possible. The session ID tag also facilitates the disposal of the resources when the session is released so that they can be made available to other service instances. Occasionally the resources are disposed off when a service is suspended in an active session and reestablished when it is resume later. In some situations when a service is changed by a Client, the release of some critical resources may be delayed for a short time to allow for the resources to be used by the Client in the new service without incurring the penalties for a new connection establishment. This occurs in situations where a Client suspends a given service and moves to another service. A connection is reused if the resourceís QoS matches or is larger than the QoS required by the new service.

To a user DSM-CC allows the delivery of multimedia over a guaranteed end-to-end QoS irrespective of the transport technology the user is using. Thus it allows the end-user the choice of the transport technology and media within the locality the service is provided which best suits his/her budget.

At an age where multiplicity of transport networks dominate, DSM-CC becomes an essential component of multimedia business. Each user chooses the level of service and pays for what he/she needs, each service provider gets paid for the resources used within a session.

  1. DSM-CC U-N allows the use and logging of heterogeneous resources made available to a session by network providers. This allows the proper collection of the revenue and its disbursement to the network providers. Although the choice of a specific resource is not visible to a user, the user implicitly participates in its choice through service selection.
  2. DSM-CC U-N enables Internet Service Providers (ISP) provide broadband access to their clients irrespective of their chosen access technology The ISP gateway router can be connected to ATM or Frame Relay and the clients can use over-the-air broadcast, PSTN,PLMN,NISDN,BISDN,ADSL,HFC,FTTC etc., .
  3. To broadcasters, DSM-CC IS provides unique features for broadcast download and broadcast object carousel data operation.

 

What is a DSM-CC session ? What is the purpose of DSM-CC session control?

DSM-CC session is an instance of network service which has, as attributes, the parties (users) involved with the service and the resources used for the service.

What is the difference between the DSM-CC session control and the call control in telecommunication networks?

The DSM-CC session control may sound similar to the call control in B-ISDN because a call has a cap-ability of grouping connections. However we would highlight the following two major differences.

  1. In the DSM-CC session control, resources are not limited to network connections and anything that is dynamically allocated and de-allocated can be treated as resources. Examples of such resources are processing power of a service unit (IWU, SDB server, MCU), and a network address, such as IP address.
  2. The DSM-CC session control may be used in a heterogeneous network environment where a stream of data may traverse different kinds of transport networks (ATM, ethernet, satellite/terrestrial/cable digital video broadcast) and different access networks, See section 2.4.

What set of problems does it solve that aren't solved by other protocols such Q.2931, RSVP, H.245?

  1. Q.2931 does not allow heterogeneous connections. DSM-CC User-Network Session protocol allows an ATM connection to be one connection segment within an end-to-end heterogeneous connection containing segments of other transport technologies such as HFC and LAN. In the case of an end-to-end ATM connection however, DSM-CC associates a specific VC, among established VCs, to specific Client and Server application interfaces at both ends e.g., for viewing and control of a movie.
  2. RSVP does not allow the use of non-IP networks or heterogeneous connections. DSM-CC User-Network Session protocol allows the transfer of data without necessarily using IP. DSM-CC allows connection segments within an end-to-end heterogeneous connection to contain difference transport technologies such as ATM, HFC, LAN etc. (Editor to establish html link to Section 2.5)
  3. H.245 provides a single session and allows only uniform end-to-end connections. DSM-CC User-Network Session protocol uses globally unique session Ids which aggregate all the resources used in a service instance for use in subsequent billing. It allows heterogeneous connections containing segments of other transport technologies such as HFC and LAN. Through potential resource descriptors specific to H.245, DSM-CC could bind logical channels to applications and make use of H.245 for symmetric negotiation of capabilities. DSM-CC can also bring to an H.245 workstation the capability to operate over heterogeneous connections, use of network processing resources and regular update of application codes through download.

 

What is the relation between the DSM-CC AddResource operation and the set-up/connection operation of existing resource control protocols e.g., Q.2931, RSVP?

As the purpose of DSM-CC session control is to associate resources for the same service, the actual protocols used for the allocation/de-allocation of each resource are outside the scope of DSM-CC. The DSM-CC AddResource operation is used just to bind (or register) resources to a session. The actual allocation of resource should be deferred to resource-specific protocols such as Q.2931 and RSVP. Resource descriptors are used DSM-CC U-N to describe the connections required. These resources are extendible to other networks than presently specified in the DSM-CC U-N specification.

What is meant by DSM-CC allows the use of heterogeneous networks?

DSM-CC relies on end-to-end significant association tags to preserve the thread of continuity of a stream across networks of different transport technologies. Thus the Server view of a transport represented by its stacks can be entirely different from the Clientís. This is exemplified by different resource types and IDs at each end. By having the same association tag at both ends, a server can indicate which resource the Clientís application interface must be connected to. This method is called Tap (application interface)-to-associationTag (connection resource interface) binding.

The Tap identifies the application interface and the associationTag identifies the connection resource interface. These two are joined together through the ConnBinder. A given DSM-CC U-U object may have a number of Tap-associationTag couplings specified in a ConnBinder. This is sent from the Server to the Client in order for it to perform the necessary bindings with the applications required at both ends of connections before the application operation begins.

Two DSM-CC concepts come into play.

There is only one SRM in the DSM-CC U-N reference model. The situation where the Client and the Server reside in domains of different SRMs seems to be ignored. Can DSM-CC U-N globally scale up?

Yes the SRM domain can be globally scaled up. The DSM-CC U-N reference model has SRM as one logical entity. It is possible to distribute agents terminating messages from users and let them interwork to form the one logical SRM. DSM-CC IS treats this interworking as implementation details of SRM. But there is an activity to specify this interworking in the MPEG/DSM group, which will realize the networking of SRMs of different network providers. This specification will form a part of the next version of DSM-CC, called DMIF (DSM-CC Multimedia Integration Framework)

 

The Session Resource Manager is a key element of DSM-CC and a single point of failure. What kind of back-up precautions can be taken on the SRM to secure safe operation of the system?

In the DSM-CC IS specification SRM is identified as a set of functions covering Session and Resource Management and User configuration. The DSM-CC IS also specifies interfaces between the Client and the Server to the SRM and uses the Specification and Description Language (SDL) to provide an exhaustive representation of their use, including exception cases. The realization of the SRM functions is left to a particular vendor or implementers group. For example assuming that the SRM functions are implemented through SRM nodes, it is possible to reroute traffic from a node which has failed to another node with a minimum impact on the service.

Is the session protocol valid for broadcast scenarios?

Yes, through carousel, under download in section 5.

About DSM-CC U-N Configuration

What is the difference between BOOTP or DHCP and DSM-CC U-N Config?

The config protocol, like BOOTP and DHCP, is a protocol which configures other protocols. The protocols differ in scope. The BOOTP and DHCP protocols configure state for other Internet protocols, such as for IP itself. The DSM-CC config protocol initializes the state of the DSM-CC session protocol.

The role of the DSM-CC U-N config protocol is to create the precondition for session set-up. Any other configuration protocol is equally acceptable as long as the first session preconditions can be met.

About DSM-CC U-U

Why does DSM-CC U-U use CORBA? Do all the DSM-CC protocols require CORBA, IDL, RPC?

The Service Domain (User-User) section of the standard adopts CORBA concepts. The other sections of the standard, for example config, session, and download, are protocols which do not adopt object concepts. The specifications of the Object Management Group provide three elements on which the design of the Service Domain builds. The three elements are:

What is the relationship of the object classes of the Service Domain (User-User) to specific object domains like CORBA, Java, and COM? Does the client have to implement a complete Object Request Broker?

Since the service interfaces are specified in Interface Definition Language, the process to create the client side object which is the surrogate for the remote service is to:

The client does not have to provide a full Object Request Broker. It provides a package, in its favorite implementation language, which emits interoperable protocol.

What is the scope of the Service Domain interfaces? What functions does it provide?

The key service objects visible to the consumer Client are:

There is also a Generic Service class on which applications can build. It provides functions, such as access control and life cycle, common to all services. There are also interfaces within the Service Domain to install new services or content.

Is the interface of the Service Domain valid for broadacst scenarios?

Since the object interfaces do not expose the transport mechanism, the same interfaces are valid for broadcast scenarios and interactive scenarios. (The client discovers the mechanism through the protocol state found inside the object references. The object reference, however, does not expose the state.) The architecture anticipates the design of applications which scale to both scenarios.

Why does DSMCC-UU adopt the object oriented paradigm to perform simple stacks like directory browsing, file retrieval, and stream control? Why does the IETF instead provide simple protocols, for example the RTSP for stream control?

The object design, to be specific the Interface Definition Language, automates the protocol design. It provides, after compilation, both the (implementation language specific) interface which encapsulates the protocol, plus the protocol itself. The application can program to the interface, in the application implementation language, rather than the protocol. The protocol itself is simple, as the functions it transports are simple.

The objective was also to support simple application development. The application treats invocations on the service objects just like subroutine calls. This requires Remote Procedure Call semantics. If other protocols were to support these semantics, it would be necessary to interpose code between the interface and the protocol stack, such as TCP/IP or UDP/IP, below. This would yield protocol design stacks with comparable footprints to the object design stacks.

 

About DSM-CC Download

How does the download protocol differ from download protocols such as Trivial File Transfer Protocol?

Compared to other download protocols, DSM-CC Download is designed for lightweight and fast operation in order to meet the needs of devices that contain limited memory. The DSM-CC download operates over heterogeneous connections and is applied to a number of network models one of which is the broadcast model with no upstream channel. The mechanisms used in download are:

 

About DSM-CC SDB-CCP

What is the relation between the DSM-CC SDB-CCP protocol and the DAVIC ZAP protocol?

The DAVIC ZAP protocol is one instance of the DSM-CC SDB-CCP protocol. As SDB-CCP does not assume any specific transport network, a SDB-CCP message has a field to carry some transport network specific parameters. DAVIC ZAP details the format of this field to use SDB-CCP in a baseband ATM access network.

About DSM-CC Section (Transport)

(To be provided)