KR100758514B1 - Information server and method of arranging carousel information - Google Patents

Abstract

In order to insert additional files in sequence in a transport stream representation of a carousel of files, the transport stream representation is preconfigured and a payload list of transport stream representations during configuration. Insertion points are specified, a transport stream payload is calculated for the additional file, and the transport stream payload is inserted into the transport stream representation at the next insertion point, and the carousel is output to output the files of the carousel. Each page of is designated with a reference name, one or more file addresses are specified for each reference name, an attribute is assigned to each reference name to specify the characteristics of the output, the pages of the carousel are referred to sequentially, and each Information server where files specified by addresses are output according to attributes associated with pages server).

File Carousel Transport Stream, Information Server

Description

Information server and method of arranging carousel information}             

1 is a diagram showing a broadcast network in which the present invention can be realized.

2 illustrates an execution section of an information server in which the present invention can be realized.

3 illustrates an arrangement of information servers in which the present invention may be implemented.

4A and 4B illustrate an information server in which the present invention can be implemented.

5 shows an object / data carousel.

6 illustrates an array of conceptual time slots for prioritizing data operations.

7 shows the configuration of a DVB section from files.

8 illustrates a dependency structure for data transformations.

Description of the main parts of the drawing

20: host, 8: mux

10: MOD, 24: Saver

22: control and data, 22: API server

28: Carousel builder, 34: SI tables manager

30: packetizer, 32: driver

The present invention relates to an information server and a method of arranging carousel information, in particular, when additional files are inserted in random order in a transport stream representation of a carousel of files, and a nature of a carousel is specified. When possible.

It is proposed to provide an information server to provide information on the transport stream of the selected channel. For example, various data files may be provided in a transport stream of a specific channel to broadcast video data and the like.
It is now proposed that a number of different data files are transmitted sequentially in the manner of a carousel. However, the present application recognizes that some flexibility is required for the page content of the carousel. In particular, it is proposed according to the invention that it should be possible to insert a file into another predetermined carousel of files. Unfortunately, due to the transformation occurring in the payload of the transport stream from the original carousel files, it is not possible to simply insert additional payload information without disturbing the entire transport stream.

According to the present invention, a method is provided for inserting an additional file in any order into a transport stream representation of a carousel of files, wherein the transport stream representation is preconfigured, wherein the method comprises: a payload list of transport stream representations during configuration Specifying insertion points in,

Calculating a transport stream payload for the additional file;

Inserting the transport stream payload into the transport stream representation at the next insertion point.

According to the present invention, there is provided an information server for providing information to an output transport stream, said information server comprising:
A carousel builder constituting transport stream payloads for the transport stream representation of the carousel of files,

A payload builder for constructing a transport stream payload for an insert file inserted into a carousel of files,

The carousel builder specifies the insertion points in the transport stream representation while constructing the payload so that the payload of the insertion file can be inserted at the next available insertion point.

In this way, insertion points allow additional payloads to be inserted into the output transport stream without disturbing the overall content of the transport stream. Moreover, insertion points can be provided at relatively short intervals through the carousel so that additional files can be inserted relatively quickly.

Preferably, the information server further comprises a multiplexer / packetiser which periodically consults the payloads calculated by the carousel builder and constructs transport stream packets for transmission, the multiplexer Refer to the insertion point and insert the payload of the insert file at the next available insertion point.

In this way, no significant amount of recalculation is required, and the multiplexer / packetizer simply inserts the required payload information at the next available insertion point.

The information server may comprise at least one stream generator for generating a transport stream, the stream generator comprising at least one carousel builder and multiplexer as specified above.

Preferably, the multiplexer forms MPEG2 transport stream packets from the payload calculated for transmission, the file is DSMCC objects, and the carousel builder constitutes a complete set of DSMCC sections for the object carousel. .

It is possible for the user to want to specify various pages of the carousel in other ways. For example, as described above, some files may require immediate insertion, for example when they have just received an update. Similarly, some pages of the carousel may include multiple versions such that successive versions are output at each rotation of the carousel.

According to the present invention, there is provided a method for outputting files of a carousel, said method comprising:

Designating each page of the carousel by a reference name,

Specifying one or more file addresses for each reference name,

Assigning an attribute to each reference name to characterize the output;

Sequentially referring to the pages of the carousel and outputting a file identified by addresses according to characteristics associated with each page.

According to the invention, there is also provided an information server for outputting files of a carousel, said information server comprising:
A register that designates each page of the carousel by reference name, specifies one or more file addresses for each reference name, and assigns an attribute to each reference name to specify the characteristics of the output for that page,
And a carousel builder that sequentially references the pages of the carousel and outputs files identified by addresses according to attributes associated with each page.

In this way, the user can easily specify the structure of the carousel.

Preferably, the attributes are a normal attribute indicating that only the file addressed for each page in each rotation of the carousel is output, and successive files are output from the file addresses listed for each page in each rotation of the carousel. Contains a periodic attribute indicating that

The properties also indicate that the file indicated by the file address is not only output when the carousel reaches each reference name, but also that the file indicated by the file address for that reference name is also output each time the contents of the file address are updated. Contains an attribute.

In this way, each page of the carousel can be easily specified not only for its contents, but also for how it is output during each rotation of the carousel.

Earlier, in order to prioritize processing resources or transmission bandwidth, tasks or data files are set to values indicating their relative importance. For example, in data transmission, one data source may be allocated 80% of the available bandwidth, while another data source may be allocated the remaining 20%. Similarly, when a processor needs to execute a number of different tasks, certain tasks may be allocated at a certain percentage of the processing time available.

Although this system works well on systems where tasks or data sources do not change often, this method is computationally inefficient in systems that are regularly reconfigured. In particular, when data sources or tasks are added or deleted, the system must recalculate the percent allocation for bandwidth or processing time.

The present application recognizes first that it is necessary to assign priorities to data operations in such a way that the relative priorities of the data operations are maintained so that they do not need to be recalculated even if the data operations are added or deleted.

A method is provided for prioritizing data operations, such as processing or transmission, the method comprising:

Constructing a periodic array of a predetermined number of conceptual time slots,

Assigning each data operation to a frequency value for an array of conceptual time slots,

Registering each data operation with the array in a series of consecutive conceptual time slots into a space determined according to a retrieval value;

Successively referring to each of the conceptual time slots of the array in which one or more data operations are registered, and executing the data operation whenever a data operation is registered and selected at the current conceptual time.

In addition, a system is provided for prioritizing data operations, such as processing or transmission, the system comprising:

 A periodic array of a predetermined number of conceptual time slots,

An input that specifies each data operation as a retrieval value for an array of time slots,

Register each data operation from the series of consecutive conceptual time slots to the array in a space determined according to the retrieval value, refer to each of the conceptual time slots of the array in which one or more data operations are registered, and also the data operation is conceptual. Control for initiating a data operation each time it is registered and selected in accordance with advances in conventional time.

In this way, priorities of data operation are all set for the conceptual time slots of the array. If an entry for one data operation occurs every nth entry and an entry for another data operation occurs every mth entry, clearly, the relative of the two entries, even if the actual allocation for the first two data operations is reduced. The occurrence remains the same whether or not an entry is made for another data operation. Thus, the method and system allow for the relative prioritization of various data operations without the need to recalculate their overall shared content when other data operations are added or deleted.

Advantageously, the system may comprise a stream generator for assembling transport stream packets from the selection of a data source, a recovery value is assigned to each data source, and the stream generator is controlled according to control. Select a data source from the data for the load.

By prioritizing the data sources to transmit in the transport stream, it is possible to allocate specific data sources to the transport stream with more bandwidth. When data from a data source is transmitted repeatedly, it can be used to control the relative number of times data from a particular data source is transmitted.

By this, other data sources with different priorities can be easily selected or removed without changing the relative transmission bandwidth already assigned to the data source used.

Preferably, information servers that provide information to the output transport stream, such as DVB / MPEG2, may include such a system to implement this method. The information server is preferably for producing a DVB compatible output stream.

When executing data operations, although the proportion of processing time or transmission bandwidth is important, some data operations are more time dependent in that they require immediate action. For example, when a processor processes video display and word processing operations, it is even more important that the video processing operation is performed before the word processing operation. Similarly, in an information server that has allocated only a small percentage of bandwidth for the transmission of news items, it is very important that certain news reports be allocated to bandwidth immediately rather than waiting for the planned allocation of bandwidth.

The need for prioritizing data operations in this way was first recognized by the present application.

Methods and systems are proposed that prioritize data operations, such as processing or transmission, where each data operation is assigned a potential value that indicates a relative urgency for that data operation, and the selection of the next data operation to be executed at least sets the potential values. Determined on the basis of

In this way, the processor or transmitter may decide to perform the most important data operation next.

An information server is proposed that provides information in a transport stream of a selected channel. For example, various data files may be provided in the transport stream of a particular channel to broadcast video data and the like. Many different data files are preferably transmitted sequentially through the carousel. In addition, it is desirable that each pile of the carousel should be changed from time to time. Unfortunately, there is no direct correspondence between the original file and the transport stream packet because of file conversion through the protocol stack forming the transport stream packet. Thus, replacing individual files results in more serious changes in the transport stream.

A method may be provided for constructing a transport stream from a carousel of a file, the method comprising:

Converting data from a file into transport stream packets via a protocol stack;

For each file used in the transformation, it stores a dependency link that represents the transformed data resulting from the transformation, and for any translated data used in the transformation to form further transformed data, in the carousel And when the file changes, storing the dependent link representing the additionally converted data so that the dependent link represents the converted data requiring recalculation as a result of the change.

In addition, an information server for outputting a carousel of files as a transport stream is provided, the information server comprising:

At least one carousel builder that transforms data from the file to form a transport stream packet through a protocol stack;

A series of registers,

For each file used in the transformation, stores a dependent link representing the converted data resulting from the transformation, and for any transformed data used in the transformation to form further transformed data, when the file in the carousel changes And a subordinate mechanism for storing the subordinate link representing the additional converted data such that the subordinate links stored in the register represent the converted data requiring recalculation as a result of the change.

In this way, when a file is changed or replaced, a simple structure is provided that represents only a fraction of the overall change for the transport stream that requires recomputation. In other words, when the file changes, it is not necessary to recalculate the transforms for the entire transport stream.

For large carousels, the recalculation of the corresponding entire transport stream takes an unacceptable amount of time. However, by identifying only a portion of the transport stream that requires recalculation, it is possible to provide an information server that can quickly update individual files in a carousel.

Preferably, the register forms part of a dependent network comprising a dependency node, the dependent node based on the primary node based on the file of the carousel and the data transformed in successive layers of the protocol stack. It includes the computed node.

The subordinate node preferably stores pointers in a particular portion of the bit stream that represents the carousel of the file. Preferably, the subordinate link represents what is stored as part of the subordinate network and how one part of the bitstream is used to calculate another part of the bitstream, and the pointers act as content criteria for the bitstream.

In this way, the subordinate network provides a computational direct mechanism for the information server to identify the bitstream portions required for insertion into the transport stream. Following the dependency structure, the pages in the original file can be easily described.

Preferably, since the carousel has a DSMCC object carousel, the content reference pointers of the subordinate nodes point to a specific sub-part of the bit stream representation of the DSMCC object carousel, and the main nodes are the content reference and directory object of the DSMCC file. directory object).

Once again, this provides a computational direct system to identify appropriate data for insertion into transport stream packets.

At least one of the files of the carousel has different versions of that file, so that subsequent versions of the carousel broadcast are transmitted with different versions. In this case, another at least one carousel builder preferably converts the data of each of the versions, and the dependent structure stores the dependent links for each version, so that successive versions of transport stream packets are stored in each of the carousel broadcasts. It is selected when rotating.

In this way, at each turn of the carousel broadcast, the information server does not need to recalculate the required payload and transport stream packets. The dependency structure allows the information server to simply change the pointer for different versions, so the output transport stream changes easily with each rotation of the carousel.

It is known to take video and / or audio data from various sources and arrange them into transport stream packets such as MPEG2 to transmit as separate channels of the transport stream.

It is now proposed to transmit various types of information in such transport streams. The information may include images, Java applications, and the like.

An information server is provided for providing information to an output transport stream, the information server comprising:

An API server for receiving control and data from an external host controller,

A storage for storing at least the data,

At least one stream generator for assembling transport stream packets from the data under control of an API server.

In this way, the information server can be provided as a separate complete unit, for example, which can be easily multiplexed with other similar information servers or providers of other transport streams for video information. Through the API server, the information server can be controlled and configured by an external host computer. Information sent by the information server can be transferred from the host controller to the storage of the information server. Similarly, the stream generator can insert the selected information into the selected channel of the transport stream as required by the host controller under the control of the API server. If it is required to transmit information repeatedly or periodically to send other information, the payload already calculated for the transport stream can also be stored by the stream generator, so that transport stream data need not be assembled at every transmission. .

Advantageously, each stream generator comprises at least one carousel builder constituting a series of payloads for periodic transmission in said transport stream.

The carousel builder may configure DSMCC object carousels and / or DSMCC data carousels and may construct DSMCC sections from input data files.

In this way, a series of payloads can be prepared and sent periodically in the channel of the transport stream. If more than one carousel builder is provided, the transport stream channel may be provided with carousels of a number of different pages.

Preferably, each stream generator comprises a multiplexer / packetizer that arranges the payload into transport stream packets and multiplexes them to form a transport stream.

In particular, when various data sources are provided to the stream generator, for example if there are multiple carousel builders and / or there is a transfer of raw data from the store, the various pages are multiplexed together to transport stream packets. Can be provided.

Preferably, by simply changing the Operating System Abstraction Layer, the API server and at least one stream generator interface with one of a plurality of operating systems so that the information server can operate with other operating systems. It includes an operating system abstraction layer.

Thus, the information server can be configured in any of a number of operating systems such as IRIX, LINUX, SOLARIS, Windows NT, and the like. The information server can be based on a platform such as C ++, but by changing only the operating system abstraction, it is possible to change the operating system without changing other structures of the information server.

Preferably, the stream generator includes a driver abstraction layer that interfaces with one of a plurality of drivers for the DVB encoder, such that by simply changing the driver abstraction layer, the information server can be used with another driver.

In this way, by changing the driver abstraction layer, it is possible to change the driver in which the information server is used without changing other parts of the information server.

The invention is more clearly understood from the following description, given by way of example only with reference to the accompanying drawings.

The following description relates to an information server to which the present invention can be applied. In particular, a design is proposed for a DVB compliant data server, referred to as an information server, capable of generating a DVB compliant transport stream comprising DSMCC data and object carousels. In addition, the information server will support dynamic update of content inside the DSMCC data and object carousel while it is streamed out.

1 shows a part of a DVB system including an information server. However, the information server may also be included in other DVB networks, such as cables or ground, and may also be included in other non-DVB networks.

The play-out section 2 comprises a number of video / audio sources 4 as well as live sources 6. From these sources, data, for example in the form of MPEG2 transport stream packets, is supplied to the multiplexer 8 and output from the play-out section 2. The multiplexed data stream is for example supplied from the satellite 12 to the modulator 10 for transmission. The signal is received by the set-up box 14 of the television 16. Signals from a number of other modulators 10 ', 10 ", etc. may also be supplied and transmitted to satellite 12.

In this system, it is also possible to provide the information server 18. Like video / audio sources 4, 6, information server 18 provides data to multiplexer 8, for example in the form of MPEG2 transport stream packets, to be inserted into a data stream for transmission. The information server is used to broadcast any form of data, but specifically to broadcast other data services in the form of MHEG5 applications, HTML pages, Java applications, IRD software upgrades, and DSMCC objects of data carousels. do. As shown, the information server 18 is provided with a host 20 which is used to control the operation of the information server 18 along with the content and characteristics of the data output from the information server 18.

The information server is controlled by the host computer using a control network such as TCP / IP over ethernet. The generated transport stream is passed to receivers (PCs, STBs, etc.) via some stream network, such as Ethernet, DVB ASI, DVB LVDS, or ATM.

2 shows an overall design of an information server.

API server 22 receives control and data from an external host. The data is transferred to the storage unit 24 for continuous retrieval. At least one stream generator 26 is provided for outputting the MPEG2 data stream. Thus, in operation, the API server receives data, for example a series of images, and passes the data of these images to storage 24. Each stream generator 26 generates a stream of data under the control of the API server 22 based on the description of the carousel. In particular, the API server 22 receives control from an external host to properly control the stream generator 26. As will be described later, stream generator 26 may include a carousel builder to transfer data from reservoir 24 in the correct order.

The purpose of an API server is to provide a software interface that can remotely access a host computer (eg, a PC) via TCP / IP. The software interface is implemented using the RPC protocol and allows the user to select object and data carousels in a manner that is independent of the transport bit stream but rather focuses on the conceptual structure of the carousel starting from the content to be broadcast by the carousels. Provides users of the information server with a set of software abstractions that they allow to specify.

In addition, the API server manages content data given to the content disk of the information server.

From an architectural point of view, an API server is realized with a single connection process that includes two main modules: an "RPC Layer" and an "Implementation". The "RPC layer" manages control network connections with one or more host computers. Each connection is called a session. It receives RPC requests from the host computer session, serializes those requests, checks whether they are valid, determines the data (arguments) for the requests, and sets the appropriate entry point in the "API execution". Extract, receive the result (or error) from the API execution, sort the received results into an RPC response, and finally pass the response back to the host computer. API execution provides a set of functions that implement the functionality of an information server API. It interacts with the content disk, writes / reads data to this disk, and checks the validity of the variable in each API call.

Sending streams from the stream generator 26 to the multiplexer of the playout section 2 can take a number of different forms. Streaming is proposed to be provided to the information server in a separate unit. Thus, streaming can be selected from, for example, DVB-ASI via PRISM board (manufactured by Sony Digital Network Solutions of Japan), DVB-LVDS, ATM, or Ethernet.

The information server is proposed to be offered in a standard 4U 19 inch rack-mount. The control network between the host computer and the information server can be, for example, Ethernet or ATM.

3 illustrates how one host computer is used to provide any number of information servers, each generating multiple streams.

4A illustrates an information server in more detail with reference to only one stream generator. The API server 22 includes an RPC (Remote Procedure Call) protocol that allows the API to be accessed by the host computer. It is also possible to run with IIOP as control APIs become more popular. Thus, the API server 22 receives data and control from the remote host computer. API server 22 also runs under an operating system abstraction layer or OSAL. This allows the information server to run on a number of different operating systems. In particular, it is only necessary to change OSAL to adapt the information server to run on other operating systems.

The stream generator 26 may output raw transport stream data. In particular, under the command of the API server 22, the stream generator 26 simply retrieves data from the storage 24 and outputs the data without knowing its contents. However, stream generator 26 may configure carousels, such as DSMCC object carousels and DSMCC data carousels. Thus, stream generator 26 includes at least one carousel builder 28.

4b illustrates a stream generator. The stream generator is responsible for dynamically designing the transport bit stream starting from the conceptual description of the data / object carousel. The stream generator's architecture consists of multiple threads running simultaneously. Here are the threads:

1. The SGMain thread is responsible for communicating with the API server. It receives instructions from the API server to start or stop the object carousel, to drop the carousel from the running stream, or to update some content inside the running stream. In addition, the SGMain thread is also responsible for managing other threads. This means controlling the start / stop of these threads and their synchronization.

2. The Carousel Builder (CB) thread is responsible for organizing the DSMCC sections that make up the object carousel (binary). The CB thread receives a carousel description from the SGMain thread. Once the DSMCC sections are configured, they are delivered to the Mux thread as payloads (see below). The CB thread also communicates with the SI thread to insert the appropriate service information into the tables managed by the SI thread.

3. The SI thread is responsible for creating and updating service information (SI) tables. It dynamically constructs the MPEG2 sections that make up these SI tables and passes those sections to the Mux thread as payloads. Currently two tables, PAT and PMT, are managed by the SI thread. The CB thread receives requests to the CB thread to preserve program numbers in the PMT, to preserve PIDs for the carousel, and to add or delete descriptions for the PMT needed to identify the carousels.

4. The Mux thread receives payloads from the CB and SI threads, and multiplexes the payloads into MPEG2 transport stream packets. In addition, the Mux thread communicates with the physical device driver to pump out transport stream packets over the physical interface Ethernet, ASI, ATM.

Each component in the stream generator (SGMain, SI, Carousel Builder, Mux) is a separate thread that runs concurrently with other threads. SI, Carousel Builder operates in "push" mode. They compute some of the transport bit streams and forward them to Mux. Mux threads operate in "pull" mode. As soon as data is available from either the SI or carousel builders, it begins to pull this data and multiplex the payload into transport stream packets.

Each carousel builder 28 may construct a carousel of information, for example, as shown in FIG. 5. The carousel builder properly configures the transport stream data to transmit a series of DSMCC files in rotation. For example, each page as shown in FIG. 5 may have MHEG or HTML files or Java applications. As a result, the stream generated by the stream generator contains each of these pages in turn.

In addition, it is also possible for a carousel builder to construct sub-carousels such that, in each cycle of the main carousel of pages, the particular page itself is cycled through a series of sub-pages.

Since each output stream may include a number of different carousels, a number of other carousel builders 28 may also be provided. Multiplexer / packetizer 30 arranges the data in an output stream from various carousel builders 28.

As described in Figures 4A and 4B, the stream generator 26 also uses an Operating System Abstraction Layer. Moreover, it uses a Driver Abstraction Layer. In this way, in the operating system, the output driver can be changed without changing the configuration of the entire information server. Instead, you simply need to change the driver implementation.

The driver 32 sends the transport stream data to the DVB encoder board.

Each carousel builder 28 operates on its own data, so the multiplexer processes the schedule between the various carousels of the transport stream.

Carousel builders effectively have their own memories. That is, since the pages of the carousel are constructed from the data in the storage 24, the transport stream data of those pages can be stored by carousel builders. In this way, when the page of the carousel changes under the direction of the API server from the host computer, there is no need to recalculate the data of the entire carousel and only recalculate the changed card. Even if the data of one page is dependent on another page, it is proposed to include the dependency structure again so that it is necessary to recalculate only the data for pages requiring change, not the entire pages.

Each carousel builder 28 reads content and descriptions, assembles a DSMCC object carousel, and prepares the appropriate DSI, DII, or DDB sections.

SI tables manager 34 is for managing PAT / PMT sections, program numbers, and PIDs. The multiplexer / packetizer 30 is for dividing the sections into packets and assembling the packets into slots for the driver 32.

As described above, the information server preferably comprises an operating system abstraction layer OSAL. It may be object oriented C ++ based on a class library. OSAL is proposed to provide an API that abstracts some of the essential operating system services offered by various operating systems, such as Window 95, Window NT, UNIX, LINUX, and so on. OSAL abstracts threads, semaphores, files, directories, processors, file system independent pathnames, and inter-process message queues. It is proposed to provide (abstraction).

The OSAL thread abstraction supports the creation and destruction of independent execution units within the address space of a single process. Thread scheduling follows a simple priority scheme based on the number of discrete priority classes.

OSAL semaphore abstraction supports associated wait / signal operations, intra-process, bounded and non-bounded counting semaphores that can be used to synchronize the execution state of other threads.

The OSAL file / directory abstraction provides a simple virtual file system with hierarchical directory structures and read / write, executable / directory permissions. File system independent pathnames provide an operation system independent naming scheme that can be mapped to other physical file system organizations. Operations are provided to create / delete a file, read / write byte sequences to / from files, locate a particular byte number, and retrieve the file's size. Operations through a directory include creation / deletion and iteration through configuration files / directories.

The OSAL processing abstraction may support the concept of code / data segments and virtual address space that need to be part of the address space. Actions may be provided to start / stop processing and to query the status of the processing (execution, termination).

The OSAL message queue abstraction may provide a means for performing inter / process communication based on a byte oriented first-in first-out message queue. Operations may be provided to put or retrieve variable length byte strings from a queue, and to check the total number of messages / bytes available in each queue.

Each of the OSAL abstractions can be supported by another powerful kind of C ++ handle class. Processing classes provide an object-oriented view of the operating system abstraction. Due to the powerful typing application written on top of OSAL, the API benefits from strategic compile time type checking. Examples of treatment classes represent operating system independent entries, and operating system services are represented by methods on the treatment class. Execution of the processing class is lightweight and efficient across all platforms that support the ANSI C ++ standard because the implementation and implementation use a low-level API that is specific to the system (Window 95, Window NT, Unix). It is portable. In most cases, low-level APIs are mapped directly onto services supported by the underlying operating system so that the overhead added by OSAL is minimal.

OSAL class libraries and APIs are a cross-platform platform of DVB-compliant data broadcast servers that support the delivery of HTML, Java, MHEG applications, and other data through DSMCC objects and data carousels. It can be used to form cross-platform implementations. The thread abstraction allows multiple threads to generate DSMCC object carousel bit streams from content data, (b) a dedicated thread to generate DVB service information, and (c) multiplex DSMCC sections and table data to transport stream packets. Can be used to support the thread. The semaphore abstraction is used to perform synchronization between these threads. File and directory abstractions are used to implement access to persistent content in the server's file system. The process abstraction is used to execute one process dedicated to network communication and another process to handle the generation of transport streams leaving the server. The message queue abstraction is used to carry out the communication between the network stack and the generation of the bit stream. Using OSAL class libraries and APIs, data broadcast server application code is executed unmodified on each operating system on which OSAL is currently supported.

As described above, each stream generator 26 may have a number of carousel builders 28. Sometimes it may be desirable for pages of one carousel to be sent more frequently than pages of another carousel. To this end, the multiplexer / packetizer 30 may be presented in transport stream packets from one carousel more often than another. For example, if there are only two carousels and one has three times the weight of the other, when filling transport stream packets, the multiplexer / packetizer assigns 75% weight to one carousel, from the other carousel. Each time a packet is filled with data, three packets are filled with data from the carousel.

Unfortunately, this system is computationally expensive when changes are made to the characteristics of the data transmitted in the transport stream. For example, in the example given above, if a third carousel is added with similar importance to a carousel of sub-criticality already used, the system may have 60%, 20%, and 20% of the transmission time for the three carousels, respectively. The% given should be recalculated. In other words, whenever the importance of the carousel changes or the carousel is added or deleted, the rate executed by the multiplexer / packetizer 30 must be recalculated.

In the information server, a new algorithm for the scheduling of packets is proposed.

With reference to FIG. 6, a data array for adjusting the importance of various data sources is described.

It is arranged into a series of conceptual time slots t. Each time slot t in the array allows registration of one or more data sources, for example carousels, raw data from storage 24, and the like. By assigning each data source to a relative priority based on conceptual retrieval, various data sources can be easily prioritized.

In the example described, the data source 1 is assigned a priority of "4", so it is generated every four conceptual time slots. Similarly, data source 2 is given priority "3" and data source 3 is given priority of "2". The multiplexer / packetizer 30 sequentially forms the MPEG packets by following a designated data source along the array and ignoring empty conceptual time slots. For example, according to the data source shown in FIG. 6, this fills the packet with data from the data source 1, then fills the packet with data from the data source 2, and then from the data source 3. Fill the packet with data, then fill the packet with data from data source 2, then fill the packet with data from data source 3, and then fill the packet with data from data source 1. In this way, data from the data source 1 is transmitted three times every four times transmitted from the data source 2 and every six times transmitted from the data source 3.

By using an array as described above, no recalculation is required when a data source is added or deleted or when the priority of the data source changes. For example, when data source 3 is deleted, the relative priority between data sources 1 and 2 is maintained. Similarly, if the fourth data source is sent at a very high priority, for example allocation "1", the relative priorities and retrieval, even if the data sources 1, 2, 3 are all sent in the previous half times, They are the same. Thus, a system that prioritizes data operations becomes more efficient in computation.

By using this system in the stream generator 26 of the described information server, it is easy to allocate bandwidth to various data sources, such as various carousel builders 28. However, this may equally apply to allocating processing time in the CPU. In particular, various tasks requiring execution may be assigned priorities in the same manner as bandwidth is allocated.

The ability to obtain resources such as processing or bandwidth is proposed not necessarily based on the proportional sharing of resources as well as the latency of initial access. That is, some processing may require action more immediately than others. For example, low latency allocation of CPU processing time is important to ensure smooth operation of multimedia applications. If two processes, for example word processing and video display, are allocated at a predetermined rate of processing time, it is more important that the video display is given access to more immediate processing.

To account for this need, additional parameters may be assigned to each resource consumer or data operation, whether this is data processing or data transmission. This parameter is intended to define the latency of resource access.

Thus, since each source or data operation is given a latency value, multiple resource consumers are compared to determine which one should be done first, or similarly, many to see which one must be sent first. Resource consumers are compared. In the case of the information server discussed above, if there is very important information that should be transmitted and received as soon as possible, very high latency values can be given to the appropriate data source, so that this data is provided by the multiplexer / packetizer 30. It is inserted into the data stream as soon as possible.

In carousel builders 28 described above, carousel builders each configure a plurality of DVB sections to be supplied to multiplexer / packetizer 30. These sections contain data representing the carousel of that particular carousel builder. The information server or carousel builder itself is preferably capable of changing or replacing one or more pages within the carousel. However, as described above, once carousel builder 28 constructs the appropriate carousel from the data in reservoir 24, the data is not recalculated and is simply stored for cyclical transmission. To change a page or file in a carousel, it is possible to recalculate all data for that carousel. However, this may cause an undesirable delay before a new carousel is sent.

7 illustrates a process of data conversion performed by the carousel builder. The data files have headers added to them to form BIOPs as part of the DSMCC object carousel. These are grouped together to form modules for the DSMCC data carousel and then divided into DVB sections for delivery to the multiplexer.

As will be explained, if file 1 changes, BIOP file 1 also changes with the results of the modules and sections. However, the BIOPs for files 2 and 3 remain unchanged. Therefore, after changing to file 1, the carousel builder does not need to recalculate the BIOPs for files (2) and (3).

In order to enable the carousel builder 28 to avoid unnecessary calculations, a dependent mechanism is proposed. By using this mechanism, it is possible to quickly update the bit stream representation of the DSMCC object carousel using only its updates to its contents, ie files, directories and the like.

The dependent structure is arranged in the carousel builder by a series of nodes and pointers. In particular, the dependent network comprises two kinds of entities: (a) dependent nodes that store pointers called (a) content references in a particular sub-part of the bit stream representation of the DSMCC object carousel, and (b) consists of dependency links that indicate how one set of content criteria is used to calculate another set of content criteria. The subordinate node itself has two types, namely the main nodes, including (1) the content criteria for the configuration contents of the DSMCC object carousel (eg, DSMCC file and directory objects), and (2) the content criteria of the other nodes. There is a computed node containing the transformations for these fields. This transformation is applied to a set of content criteria, resulting in a set of new content criteria indicating new portions of the DSMCC object carousel bit stream.

Because the carousel builder organizes the appropriate data, the dependent links can result in a portion of the DSMCC object carousel bit stream (eg one module) being part of another bit stream (eg the same compressed content of the same module). Stored to indicate the method. Dependent links are made when the transformation is performed by the computed nodes of the slave network. In particular, when a transformation of a dependent node uses a content criterion associated with another node (primary or computed node), a dependency relationship exists between the node used and the node calculating the transformation. Dependent links are stored for the node used to indicate the node it was used for.

Dependent links allow to quickly determine the set of transforms for which the dependent mechanism needs to be reapplied to recalculate a new bitstream when the contents of the object carousel change, ie when the data of the primary nodes change.

When the update is run, some DSMCC file and directory objects are changed. As a result, the major nodes involved in these changes also change. For the principal nodes where the change has occurred, the dependent links associated with these principal nodes are followed to represent any calculated result node. If a calculated node is reached, the dependent link of that calculated node is used to move back to any further calculated node. All computed nodes that come up in this way are cleared along with the forward dependency link. As a result, the subordinate network frees specifically linked computed nodes. Therefore, the carousel builder repeats the original calculation process for those calculated nodes that are currently empty. As a result, when the computed nodes apply variants to recalculate a new set of content criteria, the dependent links are once again formed from the referenced node.

If the primary nodes do not change or the computed nodes are not deleted, the subsequent computed node may simply refer back to the unchanged primary or computed node. Thus, if the computed node refers only to the main computed nodes that have not changed, it can be used directly without reapplying the transform. As long as the content criteria upon which the result depends do not change, the result can be reused directly in the updated bit stream without having to recalculate it. Carousel parts that are not affected by the update are already stored in the subordinate network, so the subordinate network allows for quick updating of the contents of the DSMCC object carousel.

In this way, a dependent mechanism can be used to construct the bit stream representations of the DSMCC sections that make up the object carousels. The bit stream representation consists of a list of content criteria indicating different sections. Content criteria are passed to the multiplexer, which packetizes the DSMCC section into an MPEG transport stream. This mechanism can be used to recalculate modified DSMCC sections when some of the configuration parts of the DSMCC object carousel (eg, files and directories) change. It is also possible to apply the technique to DSMCC data carousels.

This is illustrated in FIG. 8. In particular, by changing the main node n4 and referring to the dependent link, it is clear that the content of the calculated node n7 must also be changed. By referring to the dependent link for computed node n7, it is clear that the content of computed node n8 should also be changed. For main node n4 containing the content of deleted computed nodes n7, n8 and new content criteria, computed node n7 is recalculated using the content criteria of main nodes n4 and n5. . The content criteria for computed node n8 are then recalculated from the content criteria of computed nodes n6 and n7. It can be seen that this processing does not need to recalculate the content criteria for the calculated node n6.

For each stream generator 26 of the information server, it is desirable to provide additional flexibility in updating the DSMCC file object transmitted in the transport stream.

As discussed above, this stream generator 26 includes one or more carousel builders 28 and a multiplexer / packetizer 30 to construct the result of an MPEG transport stream. Furthermore, the slave network is made up of a process of calculating the required DSMCC sections of the DSMCC object carousel, called payloads, which constitute the bit stream at the DSMCC level.

It is desirable to provide "hot insertion" of DSMCC file objects. This is out-of-band insertion of the DSMCC file object as fast as possible in the broadcast bit stream.

As described above, once the required DSMCC sections of the DSMCC object carousel have been calculated for the carousel, the DSMCC sections are simply stored and periodically retransmitted. Therefore, it is not desirable to simply hot insert the DSMCC file object into the objects of the carousel since the DSMCC level bit stream needs to be recalculated. On the other hand, simply inserting DSMCC sections for hot insertion into an already formed bit stream will probably contaminate the transformation of the transmission of the DSMCC object carousel.

The multiplexer loops repeatedly through the list of calculated payloads and dynamically divides the payloads into MPEG2 transport stream packets for transmission. In order to allow "hot insertion" of DSMCC file objects, "insertion points" are specified in the payload list for the complete set of DSMCC sections for the object carousel. These "insertion points" are markers indicating where the multiplexer can safely insert additional payload for out-of-band DSMCC file objects.

Thus, the payload for out-of-band update is calculated and passed to the multiplexer. The multiplexer then waits until the next available marked "insertion point" and inserts the out-of-band payload. This effect delays the transmission of the rest of the carousel data but does not interfere with its contents.

As discussed above, it is desirable to provide periodic updates of a single DSMCC file object. That is, in each cycle of the carousel broadcast, a different version of the special DSMCC file may be transmitted.

For the slave network described above, periodic updates are made by having nodes separate from the slave network used to calculate the payload for each version of the DSMCC object. The list of complete payloads is given to the multiplexer, but this list is internally organized in such a way that the multiplexer conveys the next payload value each time it requests it in the object carousel of the next cycle. This mechanism is very effective because the pointer simply points to the appropriate payload, rather than calculating an updated version of the DSMCC object for each cycle of the carousel.

Selection of the DSMCC file object for broadcast order, broadcast count, and hot update is preferably performed by the user. To do this, it is suggested that data be input in the manner of the table given below.

Table 1

    Reference name        Characteristics              Full name     Page 1       normal   / sports / intro     Page 2       Miracle   / news / item 1   / news / item 2   / news / item 3   / news / item 4     Page 3       Each enemy   / gambling / horses     Page 2       Miracle   / news / item 5   / news / item 6

 As can be seen, the table consists of three item columns. The first column lists the reference name of the DSMCC file object as handled by the broadcast receiver. The list of complete reference names constitutes the virtual namespace handled by the receiver. The second column specifies the attributes for the item referenced by the reference name. Three attributes are proposed to be provided: normal, periodic and immediate. These are explained later. The third column specifies a list of full names, in other words, the physical file name.

The "normal" attribute means that the referenced page will always output the listed file. On the other hand, for the "cyclic" attribute, the table will contain one or more full names, and in each cycle of the carousel, the next full name listed for that page will be included. The "immediate" attribute usually has the same effect as the "normal" attribute in that a single file will be sent for that page. However, if the referenced page has an "Instant" attribute, as soon as an update is received for that page, the page will be inserted with "hot insertion" but will not reappear until its normal position in the carousel.

In the example given above, the carousel configuration table specifies a logical carousel consisting of three pages called page 1, page 2, page 3. The broadcast order is designated by the first column, namely-page 1-page 2-page 3-page 2 and then page 1 again. However, in the periodic page, each cycle of the carousel takes the next item from the list of full names (third column). As a result, from a content point of view, the carousel order is / sports / intro-/ news / item1-/ gambling / horses-/ news / item 5-/ sports / intro-/ news / item 2-/ gambling / horses-/ news / item 6-/ sports / intro-/ news / item 3-/ gambling / horses-/ news / item 5-/ sports / intro-/ news / item 4-/ gambling / horses-/ news / item 6 ... to be.

Following the entry of data for the carousel configuration table, the appropriate carousel builder calculates the required DSMCC section according to the dependent network discussed above. In particular, the page specified by the reference name is set up with major nodes. If a page is designated as periodic, the carousel builder sets up identical duplicate pages corresponding to each periodic object. The carousel builder then calculates the appropriate modules for that page. Preferably, when calculating a module, the same periodic pages are assigned to separate modules, so that recalculation is kept to a minimum in each period. Similarly, pages with an "instant" attribute are assigned to separate modules to facilitate insertion of pages when needed.

It is possible to provide an information server to provide information to the transport stream of the selected channel. For example, various data files may be provided in a transport stream of a particular channel to broadcast video data and the like.

Claims (10)

delete delete delete delete delete delete In the information server 18 for outputting the files of the carousel A register defining each page of the carousel by a reference name, defining one or more file addresses for each reference name, and assigning an attribute to each reference name to define an attribute of the output for that page; And And a carousel builder (28) for sequentially referring to the pages of the carousel and outputting the files identified by the addresses according to attributes associated with each page. The method of claim 7, wherein The attributes include a normal attribute indicating that only the file indicated by the address for each page is output for each round of the carousel, and in each cycle of the carousel, And a cyclic attribute indicating that a sequential file is output from the file addresses listed for the page. The method of claim 8, The attributes are not only outputted the file indicated by the file address when the carousel reaches the respective reference name, but also the file for that reference name whenever the contents of the file address are updated. And an immediate attribute indicating that the file indicated by the address is also output. In the method of outputting the files of the carousel, Defining each page of the carousel by a reference name; Defining one or more file addresses for each reference name; Assigning an attribute to each reference name to define the nature of the output; And Sequentially referring to the pages of the carousel, and outputting the files identified by the addresses according to attributes associated with the respective pages.

Images