WO2012146098A1

WO2012146098A1 - Method and corresponding system for storing and playing streaming media

Info

Publication number: WO2012146098A1
Application number: PCT/CN2012/072606
Authority: WO
Inventors: 赵宇; 刘继年; 孙健; 王芳; 陈光亮
Original assignee: 中兴通讯股份有限公司
Priority date: 2011-04-27
Filing date: 2012-03-20
Publication date: 2012-11-01
Also published as: CN102761524B; CN102761524A

Abstract

A method and corresponding system for storing and playing streaming media, realizing the storage and playing of streaming media based on a Digital Video Broadcasting Tier (DVB Tier), enabling users to achieve the expected consistent effect when the playing relevant to trick mode is performed. The method for storing comprises: a streaming media server segments the media content when the latter is stored; each segment includes one or more subsegments; the streaming media server generates a subsegment index atom for each subsegment, and the subsegment index atom includes: the sampling number information, the sampling offset information, the sampling length information and the sampling time information. The storage system includes a segmenting device and a subsegment index atom generation device. This invention can quickly locate the starting and ending position of the data frame which is needed to be played, save the network traffic, reduce the user's waiting time during the playing, and improve user experience.

Description

Stream media storage, playing method and corresponding system

Technical field

The invention relates to the field of multimedia, in particular to a streaming media storage and playing method and a corresponding system. Background technique

The traditional streaming service is based on RTSP (Real Time Streaming Protocol) / Real Time Transport Protocol (RTP). However, due to the shortcomings of RTSP/RTP in crossing the firewall, the HyperText Transfer Protocol (HTTP) has been developed in the past two years to provide streaming services, namely HTTP Streaming (streaming) services. In the MPEG (Moving Pictures Experts Group) standards organization, it is called DASH (Dynamic Adaptive Streaming over HTTP), which is based on hypertext transfer protocol. In recent years, the broadcasting service and the mobile service have gradually merged, and different multimedia contents and services are transmitted on different networks. MPEG2-TS (MPEG2 Transport Stream) for broadcasting is also widely used on the Internet. Similarly, MPEG2-TS is also used in DASH services.

Since the HTTP protocol is essentially a text transfer protocol, it is not suitable for real-time streaming media transmission. In order to facilitate the functions of live playback (for time-shifted services) and viewing while downloading, the program content is usually physically or logically divided into a number of small segments, as described in FIG. The media presentation description (MPD, Media Presentation Description, or program index table) information such as the index of the program index table, the segment of the body segment, and the like, each time one or more segments of the media file are downloaded for playback, and then the downloaded The data is decoded and played. However, for the fast forward and rewind function, the usual method is to download the corresponding segment according to the MPD, and then find the corresponding frame to play in the segment file. In this way, the client will download a lot of redundant data, which seriously affects the smoothness of playback and the network bandwidth. Summary of the invention

In the current technology, a common practice is to define a segment index atom. When performing a trick mode, for example, during fast forward, the client proceeds according to the address information of the key frame recorded in the segment index. Access keyframes for fast forward results. However, a fatal flaw in this method is that the fast forward multiple does not match the actual situation, especially in the case where the key frames are sparse, the fast forward effect and the desired effect will have a large gap.

A new D VB Tier framework (Digital Video Broadcasting Tier framework) is defined in Appendix D of the European Telecommunications Standards Institute 101 154. In the video coding, each data frame is divided into multiple tiers according to the dependency, and the tier high data frame depends on the tier low data frame. However, under the framework of DVB Tier, how to play the trick mode related to the tier architecture has not been defined.

The technical problem to be solved by the present invention is to provide a streaming media storage and playing method and a corresponding system, which realizes storage and playback of streaming media based on DVB Tier, so that the user can obtain the same effect as expected when performing playback related to the trick mode. .

To solve the above technical problem, the present invention provides a streaming media storage method, including: a streaming media server, when saving media content, segmenting the media content, each segment including one or more sub-segments, The streaming media server generates a sub-segment index atom for each sub-segment, and saves the sub-segment index atom, where the sub-segment index atom includes: a sample number information, a sample offset information, a sample length Information and sample time information.

Preferably, in the sub-segment index atom:

The number of pieces of data indicates the number of data frames indexed by each level in the current sub-segment;

The sample offset information indicates an offset of a storage start address of each indexed data frame in the current sub-segment with respect to the current sub-segment first address;

The sample length information indicates a storage length of each indexed data frame in the current sub-segment;

The sample time information represents a time offset of each indexed data frame in the current sub-segment relative to the first sample data frame in the current sub-segment.

Preferably, the sub-segment index atom further includes bit number identification information indicating the current required number of bits of the sample offset information and the sample length information.

To solve the above technical problem, the present invention also provides a streaming media playing method, including: In any of the following scenarios, the terminal downloads a data frame according to the content of the sub-segment index atom downloaded from the streaming server, and plays the downloaded data frame: a scene entering the trick mode from the normal play mode, from the trick mode The scenario of entering the normal play mode and the scene of the time shifting on-demand mode; the sub-segment index atom includes: the number of pieces of information, the sample offset information, the length information of the sample, and the time information of the sample.

Preferably, when the terminal enters the trick mode from the normal play mode, the step of the terminal downloading the data frame according to the content of the sub-segment index atom downloaded from the streaming server includes:

After receiving the command to enter the trick mode, the terminal determines the starting data frame;

Determining, by the terminal, the segment information of the segment index according to the segmentation index in the pre-downloaded media representation description (MPD) to the streaming media server to download the segment to which the sub-segment of the start data frame belongs The highest level of the data frame to be played in the sub-segment in the trick mode, traversing the sub-segment index atom, searching for the index after the starting data frame, and the level is less than the highest level of the indexed The data frame downloads the searched indexed data frame in the order of the searched time information of the indexed data frame that is searched.

Preferably, the starting data frame is the first data frame after the current data frame and having the lowest level. Preferably, when the terminal enters the normal play mode from the trick mode, the step of the terminal downloading the data frame according to the content of the sub-segment index atom downloaded from the streaming server includes:

After receiving the command to exit the trick mode, the terminal searches for the lowest indexed data frame after the current data frame in the sub-segment index atom, according to the recorded in the sub-segment index atom. The sampled offset information and the sample length information of the indexed data frame are requested to be downloaded to the streaming server for downloading the data frame and subsequent data frames.

Preferably, when the terminal enters the time-shifted location-on-demand mode, the step of downloading the data frame by the terminal according to the content of the sub-segment index atom downloaded from the streaming server includes:

After receiving the command to enter the time shifting on-demand mode, the terminal determines the sub-segment corresponding to the time according to the time-shifted on-demand time, and sends the address information of the segment index according to the pre-downloaded media representation description (MPD) to the streaming server. Requesting to download a segment index of the segment, and obtaining a sub-segment index atom of the sub-segment from the segment index; Traversing the sub-segment index atom, searching for the lowest-level indexed data frame after the time-shifted on-demand time, according to the sample-like data frame recorded in the sub-segment index atom Transmitting information and sample length information, requesting the streaming server to download the sample data frame and subsequent data frames.

In order to solve the above technical problem, the present invention also provides a streaming media storage system, including a segmentation device and a sub-segment index atom generating device, wherein:

The segmentation device is configured to: when the media content is saved, segment the media content, each segment includes one or more sub-segments;

The sub-segment index atom generating means is configured to: generate a sub-segment index atom for each sub-segment, and save the sub-segment index atom, the sub-segment index atom includes: a number of pieces of information, 釆Sample offset information, sample length information, and sample time information, where:

In order to solve the above technical problem, the present invention further provides a streaming media playing system, which is located on the terminal side, and includes a downloading device and a playback device, wherein:

The downloading device is configured to: download a sub-segment index atom from the streaming media server, and when the terminal is in any one of the following scenarios, download the data frame according to the content of the downloaded sub-segment index atom: The scene enters the scene of the trick mode, the scene from the trick mode into the normal play mode, and the scene that enters the time shift positioning on-demand mode; the sub-segment index atom includes: the number of pieces of information, the offset information of the sample, the length of the sample Information and sample time information; The playback device is configured to: and play the downloaded data frame.

Preferably, the downloading device includes a determining start data frame module, a first sub-segment index atom acquisition module, and a download data frame module, where:

The determining the start data frame module is configured to: when the terminal is in a scene that enters the trick mode from the normal play mode, after receiving the command to enter the trick mode, determining a start data frame; the first sub-segment The index atom acquisition module is configured to: request, according to the address information of the segment index in the pre-downloaded media representation description (MPD), the streaming media server to download the segment index of the segment to which the sub-segment in which the start data frame belongs, from Obtaining, in the segment index, a sub-segment index atom of the sub-segment;

The download data frame module is configured to: determine a highest level of a data frame to be played in the sub-segment in the current trick mode, traverse the sub-segment index atom, and find the level after the start data frame, and The indexed data frames that are smaller than the highest level are indexed, and the searched indexed data frames are sequentially downloaded in accordance with the sampled time information of the found indexed data frames.

Preferably, the starting data frame is the first data frame after the current data frame and having the lowest level. Preferably, the downloading device includes a searching module and a first downloading module, where:

The searching module is configured to: when the terminal is in a scene from the trick mode to the normal play mode, after receiving the command to exit the trick mode, searching for the lowest level after the current data frame in the sub-segment index atom Indexed data frame of the index;

The first downloading module is configured to: request, according to the sample offset information and the sample length information of the indexed data frame recorded in the sub-segment index atom, to download the search to the streaming server The indexed data frame and the subsequent data frames.

Preferably, the downloading device includes a second sub-segment index atom acquisition module and a second download module, where:

The second sub-segment index atom acquisition module is configured to: when the terminal is in the time-shifted on-demand mode, after receiving the command to enter the time-shift positioning on-demand mode, determine the time according to the time-shift on-demand time Corresponding sub-segment, requesting, by the streaming media server, to download the segment index of the segment according to the address information of the segment index in the pre-downloaded media representation description (MPD), The second download module is configured to: traverse the sub-segment index atom, and find the lowest-level indexed data frame after the time-shifted on-demand time, according to the record recorded in the sub-segment index atom The sample offset information and the sample length information of the sample data frame are requested to the streaming server to download the data frame and the subsequent data frame.

By using the method in the embodiment of the present invention, for the fast forward and fast rewind operations of the TS (Transport Stream) file in the DASH scenario, the entire segment of the file is not downloaded, and only the data frame to be played is downloaded, that is, the data frame can be quickly located. To the start and end of the data frame that needs to be played, it not only saves network traffic, but also greatly reduces the waiting time during user playback, and does not affect between various trick modes, and between normal mode and trick mode. Smooth switching greatly enhances the user experience.

BRIEF abstract

Figure 1 is a schematic diagram of fast-forward frame fetching in the tier frame;

2 is a flow chart of realizing fast forward using ssix atoms defined by the present invention in an embodiment of the present invention; FIG. 3 is a flowchart of realizing fast forward and normal play using ssix atoms defined by the present invention in an embodiment of the present invention;

4 is a flow diagram of time shift positioning playback using the ssix atom defined by the present invention in accordance with an embodiment of the present invention.

Preferred embodiment of the invention

In order to solve the above technical problem, the present invention defines a sub-segment index atom, including: sample number information, sample offset information, sample length information, and sample time information, where:

1) a number of pieces of information, used to indicate the number of data frames indexed by each level in the current sub-segment;

2) sampling offset information, used to indicate an offset of a storage start address of each indexed data frame in the current sub-segment with respect to a current sub-segment first address; 3) sample length information, used to indicate the storage length of each indexed data frame in the current sub-segment;

4) Sample time information, used to indicate the time offset of each indexed data frame in the current sub-segment relative to the first sample data frame in the current sub-segment.

By providing the number of pieces of information, the streaming server can set an index for each frame or one frame every few frames. The implementation is flexible, especially when the index is several frames, the overhead can be reduced, for example, a 24 s, 144 A minute movie, a full trick mode index (ie index of all frames) has an overhead of 1.6 Mbytes, and if one frame is indexed every 4 frames, the overhead of 406 Kb will be reduced.

In addition to the above four pieces of information, the sub-segment index atom may further include bit number identification information, which is used to indicate the current required number of bits of the sample offset information and the sample length information. The bit identification information can conveniently control the number of bits (16-bit or 32-bit) occupied by the sample offset and the sample length, which can further reduce the network bandwidth.

Generally, the streaming media server segments the media content when saving the media content, each segment includes one or more sub-segments, and the streaming media server generates a sub-segment index atom for each sub-segment, and The sub-segment index atom is saved, and the sub-segment index atom includes: sample number information, sample offset information, sample length information, and sample time information. Preferably, the sub-segment index atoms may further include bit number identification information.

The above media content is provided by the content provider to the streaming server.

After the streaming server generates and saves the sub-segment index atom of the media content, in any of the following scenarios, the terminal downloads the data frame according to the content of the sub-segment index atom downloaded from the streaming server, and plays the downloaded Data frame: The scene that enters the trick mode from the normal play mode, the scene that enters the normal play mode from the trick mode, and the scene that enters the time shift positioning on-demand mode.

• When the terminal enters the trick mode from the normal play mode, it specifically includes:

A1. After receiving the command to enter the trick mode (for example, fast forward or rewind), the terminal determines the start data frame;

A2. The terminal requests, according to the address information (for example, URL information) of the segment index in the pre-downloaded media representation description (MPD), to download, by the streaming server, the sub-segment in which the start data frame is located. The starting data frame is preferably the first data frame after the current data frame and having the lowest level, since the lowest level data frame does not depend on any other data frame.

A3, the terminal determines a highest level of the data frame to be played in the sub-segment in the current trick mode, and traverses the sub-segment index atom, and searches after the start data frame, and the level is smaller than The highest-level indexed data frame downloads the searched indexed data frame in the order of the searched time information of the indexed data frame that is searched.

Determining the highest level of the data frame to be played is determined according to the parameters of the current trick mode. Specifically, the highest level of the frame to be played in the sub-segment is determined according to the multiple of fast forward or fast reverse.

• When the terminal enters the normal play mode from the trick mode, it specifically includes:

The current data frame is the data frame that is played when the exit of the trick mode command is received.

The above sub-segment index atom is downloaded when the terminal is in the trick mode.

Preferably, the terminal can continue the current trick mode play during the search process, that is, maintain the fast forward or fast reverse state.

• When the terminal enters the time-shifted positioning in-demand mode, it specifically includes:

C1, after receiving the command to enter the time shift positioning on-demand mode, the terminal determines, according to the time shift on-demand time, the sub-segment corresponding to the time point, according to the address information of the segment index in the pre-downloaded media representation description (MPD) The streaming server requests to download the segmentation line of the segment where the sub-segment is located

Determining the segment corresponding to the time point according to the time shift on-demand time comprises: calculating a presentation time (relative time) of the time-shifted on-demand time, and searching for the sub-segment in which the presentation time is located according to the MPD.

C2, traversing the sub-segment index atom, searching for the lowest-level indexed data frame after the time-shifting on-demand time, according to the 数据 data frame recorded in the sub-segment index atom The sample offset information and the sample length information are requested to be downloaded to the streaming server for downloading the data frame and subsequent data frames. The present invention will be described in detail below with reference to the drawings and specific embodiments.

Example 1

In this embodiment, the MPEG2-TS is used as a media file storage format on the streaming media server as an example. The streaming media server logically segments the media file, and each media segment corresponds to a segment index, media segmentation, and The segmentation index has a URL in the MPD, that is, each segment and its segmentation index can be accessed through the MPD. The segmentation index contains sidx box (segmented index atom) and ssixbox (subsegmented index atom).

In this embodiment, the definition of the ssix atom is as follows:

Aligned ( 8 ) class SubsegmentlndexBox extends FullBox ( ssix ' , version, flags ) { unsigned int (32 ) subseg_count;

Unsigned int (8) subseg_level_count;

For ( i=l; i <= subseg_count; i++)

For ( j =1; j <= subseg_level_count; j ++) {

Unsigned int (32 ) accumulated_level_size;

If ( accumulated— level — size == 0 ) {

Unsigned int (24) num_samples;

Unsigned int (8) bytes—in— unit

For ( k=l; k <= num_samples; k++ ) {

If ( bytes— in— unit == 0 ) {

Unsigned int (32) sample— of f set

Unsigned int (32) sample— size

} else {

Unsigned int (16) sample— of f set— short 7 unsigned int (16) sample— size— short

}

If ( flags & 0x01 ) {

Unsigned int (32) sample— time

}

}:

Among them, the bold part is new content, and the semantics of the newly added field are explained as follows:

Num— samples : that is, the number of pieces of information indicating the number of indexed data frames belonging to level j;

Bytes—in— unit: The number of bits of the identification information, indicating whether the offset and length are expressed in 16 bits or 32 bits. The value is units of bytes. In this implementation, when units of bytes= 0, the offset and The length is represented by 32 bits; Preferably, the number of bits occupied by the sample_offset_sic can be extended by the bytes-in-unit, which can be calculated by using units of bytes, that is, the number of bits occupied by the offset information is equal to sample_offset- Short*bytes—in— unit; Similarly, the number of bits occupied by the offset length information is equal to sample—size—shor*bytes—in—unit.

Sample-offset: indicates the offset from the stored start address of the indexed data frame S[i, k] relative to the first address of the sub-segment using the offset information represented by 32 bits;

Sample_ size : The length information of the sampled data frame S[i , k] indicated by the 32 bits indicates the storage size of the sampled data frame.

Sample—offset— short : The offset information of 16 bits is used to indicate the offset of the storage start address of the indexed data frame S[i , k] relative to the first address of the sub-segment i.

Sample— size—shorter: The length information of the 16-bit representation indicates the storage length of the indexed data frame S[i , k];

Sample_ time: the time information of the sample, indicating the time offset of the indexed data frame S[i , k] relative to the first data frame in the sub-segment j, the time offset is stored in the sidx The unit of timebase in the atom is the same.

Example 2

This embodiment is an example of entering the trick mode from the normal play mode, and the trick mode is described as fast forward. As shown in Figure 2, it includes:

Step 201: The terminal receives the command to enter the fast forward mode, and determines the highest level of all the data frames to be played according to the fast forward multiple;

Referring to Figure 1, the highest level in the data frame to be played can be determined according to the fast forward multiple.

Step 202: Continue normal play until the lowest level data frame S1 is played, and obtain the segment to which the sub-segment in which the S1 frame belongs;

Step 203: Download a segment index corresponding to the segment, and parse the index to obtain a sub-segment index primitive;

Step 204, it is determined whether all the data frames smaller than the highest level after the S1 frame in the sub-segment are downloaded to the terminal, and if yes, step 208 is performed, otherwise step 205 is performed; Step 205: Obtain offset information and length information of the next frame data according to a sequence of sampling times.

Step 206: Request the frame from the streaming server according to the obtained offset information and length information terminal;

Step 207, the terminal decodes and displays the frame, and returns to step 204;

Step 208, moving to the next segment that meets the fast forward multiple, and returning to step 203.

釆 Use the above method to speed up the search for data frames. The above process is merely an example, and other implementations may be presented by those skilled in the art in accordance with the inventive concept.

Example 3

This embodiment is an example of transitioning from the trick mode to the normal play mode, and the trick mode is described as fast forward. As shown in Figure 3, it includes:

Step 301: Receive a command of fast forward to normal play;

Step 302: Obtain a last data frame when fast forwarding;

Step 303, determining whether the level of the frame is 1, if yes, then jump to 305, otherwise, jump to step 304;

Step 304, obtaining information of the next indexed data frame in the sub-segment index atom, returning to step 303;

Step 305: Obtain offset information and length information of the frame in the sub-segment index atom. Step 306: Request, to the streaming media server, all data frames of the segment after the frame (including the frame);

Step 307: The terminal normally plays the downloaded data frame.

Step 308, downloading all segmentation data when the next segment is used;

Step 309, normal play. Current process.

Example 4 This embodiment is an example of entering the time shift positioning on-demand mode. As shown in Figure 4. The live stream in this example has only a video stream, that is, there is only one track in the stream. include:

Step 401: After receiving the command to enter the time shift positioning on-demand mode, obtain the time shift on-demand time; Step 402, determine whether the time-shifted on-demand time is UTC (World Standard Time), and if yes, skip to step 404, otherwise skip to Step 403;

Step 403: Convert the time shift on-demand time to the UTC time.

Step 404, obtaining a start UTS of the TS file from the MPD;

Step 405: Calculate a presentation time of the time shift on-demand time, that is, a time offset;

Step 406, traversing the initial sampling time of each segment recorded in the MPD, and calculating which segment the playback point falls in;

Step 407: Obtain a URL of a segment index corresponding to the segment from the MPD, and download the segment index, and obtain a sub-segment index atom from the segment index;

Step 408, traversing the sub-segment index atom, and finding a data frame with a level of 1 and a play time greater than or equal to a time-shifted on-demand time;

Step 409: Obtain offset information and length information of the frame in the sub-segment index atom, and download all data frames after the frame (including the frame);

Step 410: Normally play the downloaded data.

Step 411, the next segment downloads the entire segment data into a normal playback process. Current process.

The invention is suitable for various scenes such as local play, on-demand multimedia files for multi-branch playback and the like.

One of ordinary skill in the art will appreciate that all or a portion of the above steps may be performed by a program to instruct the associated hardware, such as a read only memory, a magnetic disk, or an optical disk. Alternatively, all or part of the steps of the above embodiments may also be implemented using one or more integrated circuits. Correspondingly, each module/unit in the foregoing embodiment may be implemented in the form of hardware, or may be implemented in the form of a software function module. The invention is not limited to any What is the combination of specific forms of hardware and software.

It is a matter of course that the invention may be embodied in various other forms and modifications without departing from the spirit and scope of the invention.

Industrial applicability

Claims

Claim

1. A streaming media storage method, comprising:

The streaming media server segments the media content when saving the media content, each segment includes one or more sub-segments, and the streaming media server generates a sub-segment index atom for each sub-segment, and saves the The sub-segment index atom includes: a sample number information, a sample offset information, a sample length information, and sample time information.

2. The method of claim 1 wherein

In the sub-segment index atom:

3. The method according to claim 1 or 2, wherein

The sub-segment index atom further includes bit number identification information indicating the current required number of bits of the sample offset information and the sample length information.

4. A streaming media playing method, comprising:

In any of the following scenarios, the terminal downloads a data frame according to the content of the sub-segment index atom downloaded from the streaming server, and plays the downloaded data frame: a scene entering the trick mode from the normal play mode, from the trick mode The scenario of entering the normal play mode and the scene of the time shifting on-demand mode; the sub-segment index atom includes: the number of pieces of information, the sample offset information, the length information of the sample, and the time information of the sample.

5. The method of claim 4, wherein

When the terminal enters the trick mode from the normal play mode, the terminal according to the streaming media The steps of downloading the data frame of the content of the sub-segment index atom downloaded by the server include:

6. The method of claim 5, wherein

The starting data frame is the first data frame after the current data frame and having the lowest level.

7. The method of claim 4, wherein

When the terminal enters the normal play mode from the trick mode, the step of the terminal downloading the data frame according to the content of the sub-segment index atom downloaded from the streaming server includes:

8. The method of claim 4, wherein

When the terminal enters the time-shifted location-on-demand mode, the step of the terminal downloading the data frame according to the content of the sub-segment index atom downloaded from the streaming media server includes:

After receiving the command to enter the time shifting on-demand mode, the terminal determines the sub-segment corresponding to the time according to the time-shifted on-demand time, and sends the address information of the segment index according to the pre-downloaded media representation description (MPD) to the streaming server. Requesting to download a segment index of the segment, and obtaining a sub-segment index atom of the sub-segment from the segment index;

Traversing the sub-segment index atom, searching for the lowest-level indexed data frame after the time-shifted on-demand time, according to the sample-like data frame recorded in the sub-segment index atom Transmitting the information and the length information, requesting the streaming server to download the data frame and After the data frame.

9. A streaming media storage system comprising a segmentation device and a sub-segment index atom generating device, wherein:

10. The system of claim 9, wherein

11. A streaming media playback system, located on the terminal side, comprising a downloading device and a playback device, wherein:

The downloading device is configured to: download a sub-segment index atom from the streaming media server, and when the terminal is in any one of the following scenarios, download the data frame according to the content of the downloaded sub-segment index atom: The scene enters the scene of the trick mode, the scene from the trick mode into the normal play mode, and the scene that enters the time shift positioning on-demand mode; the sub-segment index atom includes: the number of pieces of information, the offset information of the sample, the length of the sample Information and sample time information;

The playback device is configured to: and play the downloaded data frame.

12. The system of claim 11 wherein:

The downloading device includes a determination start data frame module, a first sub-segment index atom acquisition module, and a download data frame module, wherein:

13. The system of claim 11 wherein:

14. The system of claim 11 wherein:

The downloading device includes a searching module and a first downloading module, where:

The first downloading module is configured to: request, according to the sample offset information and the sample length information of the indexed data frame recorded in the sub-segment index atom, to download the search to the streaming server The indexed data frame and the subsequent data frame.

15. The system of claim 11 wherein

The downloading device includes a second sub-segment index atom acquisition module and a second download module, where: the second sub-segment index atom acquisition module is configured to: In the scenario of the broadcast mode, after receiving the command to enter the time shift positioning on-demand mode, the sub-segment corresponding to the time is determined according to the time-shifted on-demand time, according to the address of the segment index in the pre-downloaded media representation description (MPD) Information requesting the streaming server to download the segmentation index of the segment, from

The second download module is configured to: traverse the sub-segment index atom, and find the lowest-level indexed data frame after the time-shifted on-demand time, according to the record recorded in the sub-segment index atom The sample offset information and the sample length information of the sample data frame are requested to be downloaded to the streaming media server to download the sample data frame and the subsequent data frame.