3GP dash design principles and standards

Dynamic adaptive streaming ove HTTP (DASH) design principles and standards
3GP dash design principles and standards

Thomas stockhammer Qualcomm


Abstract

This article provides an in-depth understanding of the dash standard documents adopted by 3GPP and MPEG.

In particular, the 3GPP version provides a standard description of the media representation-segment format and transmission protocol.

In addition, it attaches the information provided by the dash client to create a stream Service Information Description for users.

The dash solution provides multiple service types (such as on-demand video, live video, and time shifting live video ),

Support for multiple functions (such as adaptive bit rate switching, multi-language support, AD insertion, special effects, and DRM)

And different configuration options.



Figure 1 Video will account for 66 percent of global mobile

Data traffic by 2014


1. Introduction

Mobile devices are increasingly familiar with network access.

Data Display: video data accounts for 66% of mobile data (as shown in Figure 1 ),

At the same time, mobile users expect a better experience for video quality, user interaction, and special effects.

The entire ecosystem includes content providers, network operators, service providers, device manufacturers, and technology providers.

These requirements must be met.

Therefore, affordable and mature technologies will be adopted first.


Traditional circulation often uses a stateful protocol.

For example, RTSP: when the client connects to the upstream server, the server will track the status of the client until the client is disconnected.

Generally, there is frequent communication between the client and the server.

After a session is established between the client and the server, the server sends the content of the continuous stream packet in the form of UDP and TCP transmission.

On the contrary, HTTP is stateless.

If the HTTP client requests a certain amount of data, the server sends the corresponding data to respond, and then the transaction ends.

Each HTTP request is considered a complete and independent one-time transaction.


In addition, for stream, progressive download can use the standard HTTP web server for media transmission.

Clients that support HTTP can locate media files by submitting byte requests to the web server.

Disadvantages of progressive download include:

I). When the gradual download starts, the user decides to stop watching bandwidth waste.

Ii). It is not really bit rate adaptive.

Iii). The live video service is not supported.

Dash overcomes the shortcomings of stream and progressive download Based on RTP/RTSP.


2. Design Principles

HTTP-based incremental download has good market adaptability.

Therefore, HTTP-based streams are moved closer to HTTP-based progressive downloads as much as possible,

However, we need to consider the defect of the progressive download mentioned above.



Figure 2 example media distribution architecture

Figure 2 shows a possible HTTP-based stream-based media delivery architecture.

Media representation processing usually generates segments, which contain different encoding versions of one or more components of media content.

These segments are typically hosted on one or more media source Servers Based on the media representation description (MPD.

The media source server is more suitable for HTTP servers,

Therefore, all communications with the server are based on the HTTP protocol (as shown in the figure ).

Based on this MPD metadata (it describes the relationship between segments and how to form a media representation ),

The client uses http get or get-like methods to request segments.

In this way, the client completely controls the stream session,

For example, it manages regular requests and smooth playback of segments, possible bit rate adjustment or other attributes,

For example, you can change the properties or user preferences of a device.


Most hierarchical media distributions require Server clusters to process connections from all independent clients.

HTTP-based CDN has been successfully used to provide web pages, reducing the pressure on the source server and reducing the download latency.

This system usually consists of a distributed cluster of page buffer proxy and a request redirect cluster.

Considering the scale, coverage, and the reliability of the HTTP-based CDN system, it is advantageous to set up a stream service on an existing device.

This can reduce capital and operating expenses, reduce and eliminate node resource allocation.

The intermediate HTTP server/Cache/Proxy shown in principle 2.

For the above reasons, you must select HTTP as the stream service transmission protocol.


3GPP also chooses HTTP-based stream service for the above reasons.

Based on the dash solution released by MPEG, The 3gp-dash specification is proposed.



3. 3GPP adaptive HTTP stream

Which of the following figure 3 shows the tenth version of 3GPP adaptive HTTP flood from ipvp-dash,

The 3gp-dash instruction document is displayed,

This document provides the following content:

A standard definition of media representation. A media representation is a structured dataset that the dash client uses to describe the accessed data.

The format definition of segments, which can be uniquely referenced by HTTP-URL.

The definition of the transmission protocol used to transmit segments, that is, HTTP/1.1.

. Dash uses the provided information to create a standard description of the stream service for the user.



Figure 3 solution overview-paip-dash


The 3GPP dash defines two levels:

Section I. 12.2 provides the data Encapsulation Format for media segments

Section II. 12.4 provides framework instances.


3g-dash supports multiple services:

. VOD

. Broadcast of live media

. Time shifting live broadcast