CN1178055A - File server for multimedia file distribution - Google Patents

Abstract

Disclosed is a technique for distribution of multimedia files from file service over arbitrary telecommunication networks to consumers where those networks offer connections of a certain bandwidth for transmission. The consumers deliver requests for presentation of multimedia files and the file servers deliver those files thereof. According to the proposed method, the transmission rates of the multimedia files are smoothed by spreading the transmission over time leading to additional delays in the delivery of said information. Minium transmission rates for maximum values of the delays are calculated ,and these values are provided as control data for the transmission. These control data are appended to respective multimedia files, and the multimedia files are delivered to the receivers with respective addtional delays between the requests for presentation of the files, corresponding to the maximum delays. By the proposed method time critical files being encoded with a variable amount ..

Description

Be used for the file server that multimedia file distributes

The invention relates to and distributing multimedia file by file server on the information transmission network arbitrarily, said arbitrary network can be a wide area network (WAN), for example Broadband ISDN or internet (Internet), or a Local Area Network.It relates to all files to the time requirement harshness, and for example the video file of film and so on, audio presentation or teleconference are showed.These files are encoded with variable in time data volume.Specifically, The present invention be directed to " video request program (video-on-demand) " service, for example such in the interactive television of expection, although amount of coded data has given fluctuations there, but transmission quality remains unchanged, and the video frequency program of continuous playback program request is provided there.

The known file server in this environment is the digital video signal that compresses to client's playback on the net in telecommunication as a provider.Those clients need special decoder that digital video signal is converted to TV signal in party of one's own.These decoders also comprise a decoder buffer, and these decoders " are realized in the top setting box (Set-Top-Box) so-called usually.These vision signals as preset time length frame transmit.In order to realize simple management (Fig. 2), usually the data file is transmitted a predetermined constant bit rate is provided Internet resources.Because the size of the video compression frame that therefore compression when causes depends on scene content separately, so compressed video signal has intrinsic variable-digit speed.Owing to these reasons, the file server of distribution movie program has just run into such problem: promptly will transmit time dependent especially amount of information from server to the user in the given time.

In the U.S. Patent application No 288525 (number of documents RO9-93-60) that submitted on August 10th, 1994, disclosed a kind of traditional multi-medium data server by applicant.The composition of this server is: direct access storage device (DASD), one group of data buffer (each buffer is corresponding to an independent time interval that is associated with a kind of multimedia show) and a massage storage are used to store the multimedia show data.Data/address bus links to each other massage storage with this group data buffer, a bus switch by controller function is provided here.

Also be to have disclosed a kind of top setting box among the U.S. Patent application No.473315 (number of documents R09-95-042, submit to June 7 nineteen ninety-five) that transfers the applicant, be used for recipient according to the multimedia communications system of prior art.This device enables to receive in the locality and handle from the multimedia programming of remote multi-media file server reception.Local top setting box control system comprises a DASD, is applicable to the compression program section of the predetermined quantity that buffer memory is received from the multimedia file server.Particularly that invention is at a kind of new formatting method, and it provides timely broadcast and buffer memory to the program segment of receiving from server, so that watch selected multimedia programming as required.

Have again, in order to go up the video file that transmits compression, two kinds of compress modes of current consideration: fixed bit speed mode and variable-digit speed mode at digital-telecommunication link (for example link that turns round with known Asynchronous Transfer Mode (ATM)).These modes have been described among Fig. 2.A kind of approach that realizes constant bit-rate is to use a kind of special compression mechanism, is filled degree with so-called " closed-loop control " according to encoder buffer and regulates image quality in encoder.This respectively have other variable quantification of branch to cause variable quality in compression is handled.High or move that quality descends in the strong scene at image complexity.With this approach contrasts be, the problem of having avoided changing quality is certainly handled in a kind of compression with constant quantification, but it has produced the video file of variable bit rate in time, this is difficult to manage in network, because because the fluctuating of speed is the function of scene content, the feasible traffic that explosion type occurs.

The known technology of quoting from previously is published in September, 1994 in the open research association of the 4th express network collection of thesis that French Brest (Brest) is held, the author is Marcel Graf, be entitled as " traffic conditioner of VBR vision signal in atm end-system ", in the literary composition special concern a kind of traffic conditioner method, it can reduce the outburst of video signal flow significantly.

Use file server to be: in connecting with a kind of typical scenario that a wide area network (WAN) links to each other, user and each network provider negotiate a traffic contract, and this contract designated user injects the traffic of this network and what is called " service quality " that the user obtains from this network (QoS).User one side, the communication unit (cell) that the traffic conditioner function will send by buffer memory thus regulate the traffic of sending and can abide by contract.Traffic conditioner is a kind of means of control traffic outburst, but the transmission of real time data has added more restriction to traffic conditioner.Single frame of video must in time arrive receiver.In order to eliminate tremble (jitter) that introduces in the transmission course, receiver was temporary in a smoothing buffer device to them earlier before the frame of receiving is decoded.The traffic conditioner function is introduced additional trembling.This two class is trembled all to be needed to eliminate.

In the article of above-mentioned M.Graf, also analyzed to reduce outburst and to produce additional trembling and reached balance between the smooth delay less.The solution of being advised is the fluid model (fluidmodel) of video transmission, and it enables definite by trembling and smooth delay that adjusting is introduced, and at transmitter one side and the needed buffer capacity of receiver one side.Suitably the startup of delayed video decoder can realize trembling smoothly, and must avoid buffer buffer memory deficiency here, and the moment that promptly every frame is consumed can not be early than the moment in this frame arrival buffer.Illustrated that in this piece article the level and smooth additional delay of introducing of trembling is equivalent to the maximum that occurs in the frame series and trembles.Observation shows that for film or music record, transmission quality is much more important than the zero-time of each displaying.Because the music that writes down on video band on the startup video unit equipment or the magnetic tape station all has necessary operation, so the user has been accustomed to this initial delay of ten thousand.Therefore, the initial delay that reaches about 10 seconds is admissible.

Report on (1995) 2417 volumes and the 234-242 page or leaf at SPIE (SPIE) and to have delivered one piece of meeting article writing by Wu-Chi Feng and S.Sechrest, be entitled as " for the compressed video that sends record in advance employed smoothly and buffer memory ".In this piece article, disclosed a kind of similar solution route.Authors have proposed a kind of method of level and smooth gentle poke word video data, for use in sending the compressed video signal of record in advance.This article has also been paid close attention to such problem: if make video signal quality keep constant, then the transmission requirement network-adaptive of compressed video data is in the big fluctuating to bandwidth requirement.A kind of averaging of the smoothed video data of the use of advising allows to cause time-delay to realize smoothly to consume other resources.

For example another kind ofly method arranged earlier transferring the possession of also to have provided in the Canadian patent application sequence number 2130395 in applicant (number of documents RA9-93-027, on August 18th, 1994 submitted to) about the dispensing of multimedia video frequency file on WAN.The invention that the there is described is different with above-mentioned approach, the delay of attempting to reduce showing request and showing the time interior user of user between the two itself to be experienced.Its suggestion utilizes two kinds of point of departures: wherein a kind of user of being intermediate point interactional with it, it combines limited storage facility with the interactional high-speed traffic passage of user.Another kind is main point of departure, and it provides the file mass memory, and provides and the slower of intermediate point and more cheap communication port confession file dispensing.Intermediate point provides part temporary for some files.When receiving that at intermediate point a displaying from the user requires, intermediate point determines whether that this file is temporary by part.If then can obtain interrupted watching immediately at intermediate point.If not, then intermediate point determines transmit required time according to the transmission rate between file size and at 2.Described file server conception also can be applicable to the present invention.

Have again, disclosed the method that task is carried out arrangement of time of the multinode multimedia file server that is used for the distributed computer system in the U.S. Patent Application Serial Number of submitting to April 7 nineteen ninety-five 418542 (number of documents SA9-94-118) by applicant, this server is with the digitized forms storage and transmit film.In a kind of interconnected network a plurality of node links together.Each node has a processor, buffer, output port and communication path, and has the nonvolatile storage stored video data.Each node is served the interior a group beholder of special time period of a concrete film.What advised is the method that optimization is handled video request, so that film is stored in the buffer rather than must extracts from nonvolatile storage.The time scale of carrying out arrangement of time in this method is about a few minutes, and this is different from fully carries out time scale of the present invention.

The another kind of arrangement of time mechanism of TV programme that provides on demand discloses in U.S. Patent application No.287434 (literary composition retaining YO9-94-133), and this application was submitted on August 8th, 1994, also was to transfer the possession of in applicant.The arrangement of time device (scheduler) of being advised is performed formation at least one film keeps a candidate, and definite candidate performs the longest of request that maximum wait time is arranged in the request and allow the stand-by period, thereby its time yardstick is about a few minutes.In the time accessing separately the data flow ability, not immediately film to be arranged the time, but, allow that until just performing the longest of request that maximum wait time is arranged in the request stand-by period just begins the video performance before finishing the candidate by the performance of arrangement of time device delayed video.Have other " stream " in the meantime and add formation.

Disclosed solution route in the disclosed world in July nineteen ninety-five (PCT) patent application WO.9103112A number about the longest fault-tolerant time of wait.Utilize a kind of program transmission optimization mechanism,, pressed redundant sequence by the program of segmentation according to a kind of arrangement of time algorithm and transmit particularly for cable TV (TV) net.The corresponding response time of maximum duration that selection and user need wait for for the playback required programs.Guarantee that at the used arrangement of time algorithm of transmit stage receiver is accepted all program segment in such a way separately, this mode enables receiving side's real-time playback program.

The shortcoming of the first of several approach of Tao Luning is the arrangement of time algorithm that need carry out when file is showed in dispensing one in front.Also to when dispensing one file, carry out complicated algorithm and come the smoothed video data.The shortcoming of other algorithms is that they only send file with the fixing arrangement of time that precomputes.

Therefore, an object of the present invention is to provide a kind of multimedia file dispensing mechanism and a kind of multimedia file server, even it allows also constant transmission quality can be arranged under the situation that the difference that has the front to discuss by wide area network transmission data is required.In addition, these requirements should automatically be satisfied and not need a large amount of processing operations when dispensing, and should not require the user to do further interaction.A special problem that solves is to exist speed to rise and fall between the frame of the different scenes of the representative that is transmitted in selecting the video frequency program service as required.

The practice that the present invention addresses this problem is for showing that multimedia file provides an additional time delay, so as to handle since during these frames transmission in time dispersion cause frame to arrive to postpone at the receiver place.Different multimedia file respective delay value is a calculated off-line, promptly calculated before dispensing, and transmission control parameters separately is attached on each multimedia file; And a rate controlled mechanism arranged in server.Proposed Control Parameter guarantees to make time delay be adapted to the network environment, particularly transmission bandwidth that exists best.Proposed rate controlled mechanism is when producing the arrangement of video file transmission time when separately Control Parameter provides, and it prevents in receiver the buffer underfed or overflows.Time delay can be adapted to other characteristics of bearer network, for example transmission cost or recipient's memory capacity simply.Above-mentioned is to finish when separately the multimedia file of dispensing to network environment (as transmission rate) and/or to the initialization of the Control Parameter working point of user environment (as the encoder buffer size), and can under situation, be adjusted, thereby be transparent to the user without any user intervention.

Relation between additional period delay and the transmission rate can be calculated by for example file server, and the practice is to wanting the serviced file of displaying separately to scan.Because proposed fast algorithm, make the length of this process and each file that linear complex relationship be arranged, thereby to typical file only need be based on each file method of smooth transmission rate in time with a few minutes, this algorithm.The characteristics of that curve are that a large amount of peak values is arranged, and this is because different multi-media events (for example action sequence of film or complicated sound event) has different information contents.By disperseing data separately, it is moved in the transmission course later, those peak values are reduced.Thereby move according to separately time and to have introduced additional time of delay.

In a most preferred embodiment, delay/transmission data are stored in the beginning of showing file separately, promptly are attached to the file front, can take in the starting stage of showing like this.

By following detailed description and claim, and with reference to the accompanying drawings, will be clear that such or such characteristics, different aspect and advantage that the present invention compares with prior art to most preferred embodiment.These accompanying drawings are:

Fig. 1 is a schematic diagram, and the fluid model according to prior art video transmission in a video request program (VOD) network is described;

Fig. 2 is the comparison diagram of constant bit-rate compression and variable-digit speed compression;

Fig. 3 has described for the video file of being made up of monochrome frame, the path example in the fluid model shown in Figure 1;

Fig. 4 provides the transmission path example of seeing from VOD network receiver one side, is included as to avoid receiver buffer underfed or overflow and to border that the path adds;

Fig. 5 provides the overall picture according to multimedia file dispensing environment of the present invention;

Fig. 6 provides the most preferred embodiment according to a transmission unit of the present invention;

Fig. 7 provides the most preferred embodiment of credit exhibit (credit token) generator that comprises in the transmission unit according to Fig. 6;

Fig. 8 provides the most preferred embodiment of the segmenting unit that comprises in the transmission unit according to Fig. 6;

Fig. 9 provides and sends the table that contains control data that multimedia file is used according to the present invention; And

Figure 10 provides the program code that is used for an algorithm of definite receiver buffer sizes and initial time delay according to an embodiment of the invention and realizes.

The most preferred embodiment of the present invention that is described in more detail below is about " video request program (VOD) " network, there with the compressed digital video signal of video server playback.But in addition, method and the file server proposed according to the present invention also can be applied to the pure audio environment, and audio presentation unit, there is compressed into the variable length audio access unit that will be transmitted; Perhaps be applied in the multimedia remote communications field, teleconference for example, the video display in there will be transmitted to one group of participant.

Fig. 1 demonstrates the video transmission model that has traffic conditioner.Transmitter one side 1, video file is read from massage storage 2, sends by a rate control device then, and this rate control device is just regulated the traffic when file enters network.Delay and delay that this model is ignored in the network are trembled, so the output of rate control device is directly feed in the receiver buffer 4 in this model.It is buffered up to being read by Video Decoder 5 there.Also must additional buffer in the receiver of reality is used and synchronizer tremble by the delay that network causes with processing.Be noted that the present invention can be applicable to have the video file frame of constant or variable arrival intervals.

The meaning of traffic conditioner is that the transmission of coded message scatter in time, thereby arrive in receiver place information and to be delayed that (transmission information in advance is possible under video storage situation of the present invention, thereby the alleviation delay issue, but this is not for initial frame).So receiver must postpone an initial delay value D to video decode with respect to transmission is initial.

When sending data with fixed rate R ', rate controller alternately is in " spare time " state and " enlivening " state.Fixed rate is realized easily, when initialization speed is carried out a program control because only need.An example path according to Fig. 1 model is shown in Fig. 3.For simply, this model assumption traffic performance is for similar to fluid, and is limitlessly detachable.Receiver buffer fullness slope of a curve is corresponding to speed R '.Suppose that bulk ground takes out data to decoder from the receiver buffer, these data blocks are called " addressed location (AU) ", and suppose that these AU are taken out immediately from buffer.

Rate controlled is overflowed to prevent the receiver buffer by the vacant capacity of counter 420 supervision receiver buffers.When counter is pointed out to still have clearance spaces in the receiver buffer (counter＜0), rate controlled is in " enlivening " state, when when having living space occupied full (counter=0), rate controlled transfers " spare time " state to.

Rate controlled is refresh counter continuously.When it is in " enlivening " state, it with speed R ' to the counter decrement.When it knows that receiver has taken out data from buffer, just by taking out data volume to counter-increments.

By resolving coded video data stream, rate controller can obtain about when receiver takes out data from buffer information.Video data stream includes coded representation (being called video access unit VAU), voice data (being called audio access unit AAU) and the encode additional data (being called coding extra cell COU) of video data, and these data are referred to as data cell hereinafter.Be compressed the semanteme of video flowing by parsing, rate controller extracts about (a) data unit size and (b) information of time (with the form of time mark) that this data cell must decoded device decoding.When video flowing during according to ISO/IEC13818-1 standard code for example, these values obviously comprise, but they also can impliedly be encoded.Rate controller considers that receiver postpones D, is added to the D value on the time mark.

For the transmission compressed video data is separated rate controller with parse operation data unit size and time tag information are temporary in the buffer.For this purpose, it produces an exhibit (token) for each data cell, and it is to put into the exhibit buffer.An exhibit comprises data unit size value and a time mark.For refresh counter, rate controller by exhibit separately the indicated moment of time mark from the exhibit buffer, take out exhibit, and by data unit size value separately to counter-increments.

For operating system correctly, avoid receiver buffer reserves not enough or overflow, the video file that must basis be transmitted correctly be selected parameters R ', B and D.Video file be can the method check by explained hereunder be compressed and dependence B=f (R ') and D=f (R ') determined in advance, and can be with for example form event memory.When broadcast program, for example can utilize receiver buffer sizes B as this table of search index, thereby determine R ' and D value rapidly.

Fig. 4 provides the constraint of receiver buffer to rate controller output communication amount.In this coordinate system, the traffic (traffic of receiving when it equals to moment t under Fig. 1 model assumption) that is transmitted during to moment t is mapped to time t.Provide two constraint curves among the figure: the curve that is designated as " the receiver buffer overflows " is the upper limit of the traffic that must be able to not surpass, otherwise buffer overflows, similarly, the curve that is designated as " receiver buffer storage deficiency " is the lower limit that prevents buffer storage deficiency.The shape of curve is determined by video file: the order of a curve ladder takes out a data unit corresponding to decoder from buffer, the size of ladder is corresponding to the size of addressed location.As shown in FIG., the position of curve is determined by B parameter and D.

By constituting rate controller output communication discharge curve (among the figure shown in the thick line), can determine minimum for the required receiver buffer sizes of given speed R ' B.This traffic curve only depends on " the receiver buffer overflows " curve, because rate controller always attempts to make the receiver buffer to be full of.For becoming this curve, must when supposing the receiver buffer in initialization, the first step be filled fully.So traffic curve only depends on the shape of overflowing curve and the position of not depending on it.The practice of determining the B value is " the receiver buffer overflows " curve and the traffic is collinear moves down together as much as possible, till traffic curve contact " receiver buffer storage deficiency " curve.The distance of overflowing between the not enough curve of storage of curve is the B value.The B value also can be determined by algorithm, and good calculating overflowed curve and traffic curve at the maximum difference that overflows the curved edge place.

For given R ', can determine that by constituting initial buffer loading curve (as shown in phantom in FIG.) required receiver postpones the minimum of D.This initial buffer loading curve is that slope is the straight line of R ', and it represents the rate controller initialization afterwards until first work period that the receiver buffer is filled fully, has confirmed the previous supposition of doing.Since this straight line necessarily can not with " receiver buffer storage deficiency " curved intersection, so by move right as far as possible this curve until with the minimum of determining D till " storage deficiency " curve contacts.The D value is the distance of this straight line and t axle crosspoint and the decoding of first data cell.Can determine mouth by algorithm, calculate straight line and the crosspoint of t axle and the ultimate range of first data cell decoding of process " storage deficiency " curved edge.

Fig. 5 demonstrates the applied environment of proposed video server.In the present embodiment, the digitized video file is encoded, for example according to the ISO/IEC13818-1 standard code.For realizing mass conservation, these files are encoded with variable-digit speed.Video server 110 sends the compressed encoding digital of digital video data to client 160 (receiver) by network 150, and it is the video data decompress(ion) and convert the vision signal that can watch on display 190 to.Be stressed that, network can use any host-host protocol, except ATM, also has so-called synchrodata level (SDH) agreement, or synchronous optical network (SONET) agreement, also can be any mixing of each broadband integrated services digital network (B-ISDN) transmission and switching system.ATM provides optional bandwidth, promptly almost can connect with any bandwidth.The present invention regulates required transmission bandwidth according to Control Parameter, utilizes this fact of ATM with this.

Video server 110 comprises a control unit 120, a transmission unit 130 and a massage storage 140.Receiver one side, client provides a receiver buffer 170 and Video Decoder 180.For satisfying the request of client to the transmission of video program, control unit 120 reads the control data that is associated with particular video program from massage storage 1440.For the buffer sizes of given receiver buffer 170, control unit 120 is determined the operation of additional parameter with control transmission unit 130 and Video Decoder 180.Transmission unit 130 receives following parameter from control unit 120: receiver buffer sizes B, receiver postpone D and unit segmentation speed R.Network 150 receiving element segmentation speed R parameters.Video Decoder 180 receives receiver and postpones the D parameter.This latter two parameter also can for example transmit in the signal message at one.All parameters all are constant in the video file transmission course in the present embodiment.

Video server 110 uses the algorithm of suggestion to determine the needed transmission rate minimum peak of concrete video frequency program according to the size of receiver buffer.This peak rate is used to be established to the communication channel of receiver, and the practice is to keep this peak rate in network 150.Receiver 160 must the delayed videos decoding, to allow buffer 170 be filled when the initialization.Video server 110 uses the algorithm of proposing to determine the minimum delay according to the size and the selected peak rate of receiver buffer, and this value is offered receiver 160.So video server can use the different peak rates that are adapted to each receiver buffer capacity and postpone to come the same variable rate video program of playback.

In replayed section, video server or send data flow with selected peak rate perhaps is in " spare time " state.Closed-loop control analog receiver buffer fullness level in server and the switching between two states, thus avoid the receiver buffer to overflow.The reserves deficiency that parameter hardware circuit solution of being selected by proposed algorithm and initial delay time have guaranteed to avoid the receiver buffer.

When having determined that above-mentioned parameter and network 150 set up server 110 to the connection between the client 160, just begin to transmit video file (program).Transmission unit reads compressed video data and it is segmented into some unit for network 150 transmission from massage storage 140.Client 160 receives these unit and they is temporarily stored in the receiver buffer 170.Will wait for D chronomere after receiving first unit, Video Decoder 180 begins the compressed video data decoding and it is converted to vision signal to present to display 190 then.Note that common client computer 160 has additional buffer and uses additional delay to decode to eliminate trembling and helping of introducing in the network.

Transmission unit 130 shown in Figure 6 comprises a letter of credit deposits yields device 210, credit exhibit buffer 220, data buffer 230, segmenting unit 240, reaches system clock 250.Compressed video program data piece reads and offers letter of credit deposits yields device 210 from massage storage 140.Letter of credit deposits yields device 210 resolution data pieces are to generate data cell (DU) and to be credit exhibit of each data cell generation.A credit exhibit comprises a time mark, and on behalf of receiver, this mark will take out the time of this data cell from its buffer.Also comprise a credit value, the size of this value representative data unit.The credit exhibit is temporary in exhibit buffer 220.The data block of reading from massage storage 140 is temporary in data buffer 230.

Segmenting unit 240 sense data and it is segmented into atm cell from data buffer 230.Fragmentation procedure is by the credit exhibit control of reading from credit exhibit buffer 220.System clock 250 provides reference clock for segmenting unit 240.

Fig. 7 provides a most preferred embodiment of letter of credit deposits yields device 210.It comprises a what is called " decode time mark and big or small extractor " 310 (DTSE), adder 320 and combiner 330.DTSE 310 resolves compressed video flowing, and is that each data cell of finding in the stream is extracted its decode time mark and data unit size.These values are used to this data cell that the credit exhibit is provided.If the decode time mark is a receiver postpone D when being zero receiver from the receiver buffer, take out moment of data cell.

DTSE 310 is categorized as each data byte in the video flowing and belongs to one of three types of data cells:

(1) video access unit (VAU);

(2) audio access unit (AAU);

(3) be not included in the 3rd class part that all bytes in (1) or (2) are considered to data cell, be called coding extra cell (COU) hereinafter.

VAU and AAU are coded directly onto in the data flow with separately DTS (data time mark) according to the ISO/IEC13818-1 standard.When a VAU or AAU did not have DTS, then the extrapolated DTS of DTS value from the front came.For considering the system layer extra byte in the data, DTSE also sends a DTS and sizes values for it by a time interval COU of generation (it comprises all bytes that do not comprise in (1) and (2)) of rule.Its implementation can be like this: whenever running into one during according to the system-timing reference value of above-mentioned standard or program clock reference value in stream, just produce a COU, and use this clock reference value as DTS.

DTS postpones to offer adder with reference to D with the receiver from control unit, and it is the addition of these two values.Time mark that combiner produces addition and data cell (DU) size combinations form a credit exhibit together.

Segmenting unit 240 shown in Figure 8 comprises a what is called " sectionalizer and time identification marker " 410 (STS).Credit counter 420 and credit exhibit customer 430.STS 410 becomes atm cell being compressed video data segment.It has two states.At first state, the unit segmentation speed R that it provides by control unit divides to estimate and each unit (" enabling " state), otherwise is " spare time " state (" make can not " state).It is included in the time mark of representative system clock 250 currencys regularly and is compressed in the video flowing.Receiver uses these time marks to make its clock and file server synchronous according to the standard of quoting from previously.

The state of STS 410 is by credit counter 420 controls.The size of credit counter 420 tracking receiver buffer free spaces, and prevent that the receiver buffer from overflowing.As long as its Counter Value, just makes STS 410 work greater than zero.The credit exhibit makes this counter-increments, makes this counter decrement and be transmitted the unit.It is reset to zero when negative value occurring.The receiver buffer sizes B parameter that is provided by control unit 120 is provided its initial value.

Credit exhibit customer 430 receiving system clocks 250 and at them the specified moment of time mark is read credit token from credit token buffer 220 separately.When its consumed credits exhibit, it rises in value to credit counter 420 with the given DU sizes values of exhibit.When unit of cell size value segmentation that STS 410 usefulness send, just to credit counter 420 depreciations.

Control unit 120 uses follows the concrete control data that is compressed video frequency program, determines that according to receiver buffer sizes B receiver postpones D and unit segmentation speed R.

In the present embodiment, control data comprises a table (shown in Figure 9), is that unit segmentation speed R, receiver buffer sizes B and receiver delay D are provided with clauses and subclauses in the table.For example, in the table clauses and subclauses i provide a tlv triple (Ri, Bi, Di).The data communication channel bandwidth that network provides must be selected in one group of centrifugal pump mostly.In this case, the clauses and subclauses of table are advantageously corresponding to the discrete bandwidth values that is provided by network.Control unit is tabled look-up with the receiver buffer sizes B that is provided, to find the lowest element segmentation speed Ri with the clauses and subclauses Bi of being less than or equal to B.Then, it (Ri, Bi Di), provide Di to client to the transmission unit value of providing.

This table is compressed video frequency program by parsing and repeats exemplary algorithm off-line shown in Figure 10 and constitutes.For a given unit segmentation speed Ri, use this algorithm can determine that buffer sizes Bi and receiver postpone Di.Resulting result is stored in this table.This algorithm provides the output valve of DTSE.(DUDTS, DU size) value sequence of output at it be expressed as in the drawings (DTSk, Sk), k=0 ..., N-1.

Claims (13)

1. the method for distribution digital multimedia file in the multimedia file distribution environments is time dependent in the amount of information of multimedia file described in this environment, and the formation of described file allocation environment is:

A network (150), it provides the connection with certain bandwidth for transmitting described multimedia file;

The receiver (160) that a plurality of and described network links to each other, they send the request that will show described multimedia file; And

The file server (110) that a plurality of and described network links to each other, they store described multimedia file, and at shown in show the described multimedia file of request distribution;

Method described here is characterised in that the following step:

By disperseing the transmission rate that sends level and smooth described multimedia file of described information in time, this causes introducing additional delay in the described information transmission;

To described delay maximum value calculation minimum transmission rate, and provide the control data of described value as transmission;

To the additional described control data of each multimedia file;

With corresponding to the described separately displaying request of described maximum delay with show that the additional delay between the described file distributes described multimedia file to described receiver (160).

2. according to the method for claim 1, it is characterized in that described receiver (160) comprises that this machine buffer (170) is used for the described information of storage before showing multimedia file, the size of described here machine buffer is used to described calculating.

3. according to the method for claim 2, it is characterized in that:

Each file server is defined as the needed minimum peak transmission rate of specific multimedia file according to described receiver buffer sizes;

In described telecommunications network, set up and arrive the data communication channel of receiver separately by in described network (150), keeping described hardware circuit solution;

Described file server is determined the described minimum delay by described receiver buffer sizes and selected hardware circuit solution, and this value is offered described receiver;

Described receiver (160) postpones the decoding to described multimedia file when initialization, to fill up described receiver buffer.

4. according to any one method in the claim 1 to 3, it is characterized in that: described control data is by resolving described multimedia file, and by buffer " reserves deficiency/overflow " curve, calculates for given data rate R '.

Calculate minimum additional delay, the practice is to determine that all are the straight line of R ' and the intersection point of t axle with described buffer " reserves deficiency " curve intersection or the slope that contacts, and determine ultimate range between moment of this intersection point and consumption first data cell as the minimum delay, and

Calculate minimum receiver buffer sizes B, the practice is the minimum range of determining between described " overflowing " curve and described " reserves deficiency " curve, thereby an arrangement of time with peak rate R is suitable between described two curves.

5. according to any one method in the claim 1 to 4, it is characterized in that rate controller has been used in the transmission of multimedia file, it is with speed R checker between idle state and transmission operating state.

6. according to any one method in the claim 2 to 5, it is characterized in that:

Each receiver is to each its this machine buffer sizes of file server report;

Described network is informed to transmitting the transmission delay that connection stood that each multimedia file carries out from described file server to described receiver; And

Described file server is determined the operating point separately of the described multimedia file of transmission at described control data.

7. according to the file server of the distribution multimedia file of any one method in the aforementioned claim, it is characterized in that:

Store the device (140) of described digital multimedia file and described control data;

Determine the device (120) of transmission parameter for each multimedia file according to described control data;

Send the device (120,130) of described transmission parameter to network and receiver;

Send the device (130) of described multimedia file at described transmission parameter.

8. according to the file server of claim 7, it is characterized in that:

Store the massage storage (140) of described multimedia file and described control data;

Read the control unit (120) of the described control data of following described each multimedia file;

Transmission unit (130) at described each multimedia file of described control data dispensing.

9. according to the file server of claim 7 or 8, control data described here comprises described receiver buffer sizes, described hardware circuit solution and described additional delay.

10. according to Claim 8 or 9 file server, the feature of transmission unit described here is:

A letter of credit deposits yields device (210), it reads the data block of described multimedia file from described massage storage, resolve described data block obtaining data cell, and each data cell is produced a credit exhibit, and the formation of this credit exhibit is:

A time mark, it represents separately that receiver will take out the moment of described data cell from its buffer, and

A credit value, it represents the size of described data cell;

A credit exhibit buffer (220), its temporary described credit exhibit;

A data buffer (230), its temporary described data block of from described massage storage, reading;

A segmenting unit (240), its sense data and described data sementation is become junior unit from described data buffer according to described credit exhibit;

A system clock, it provides reference clock for described segmenting unit.

11. according to the file server of claim 10, letter of credit deposits yields device described here also has following feature:

A decode time marker extraction device (310), it is resolved described multimedia file and is decode time mark of each described data cell extraction;

A big or small extractor (310), it is resolved described multimedia file and extracts the size of data cell for each described data cell;

An adder unit (320), it is added together described decode time mark value and described control data;

A combiner unit (330) constitutes the credit exhibit to the size combinations of resulting time mark and described data cell together.

12. according to the file server of claim 10, segmenting unit described here also has following feature:

A sectionalizer unit (240), it is segmented into described junior unit to described multimedia file;

A credit counter (420), it controls described sectionalizer by the free space that keeps the tracking receiver buffer;

A credit exhibit customer (430), it receives described system clock, and utilizes described system clock by the pointed moment of time mark is read described credit exhibit from described credit exhibit buffer separately.

13. according to any one file server in the claim 7 to 12, its feature also is:

The definition multimedia file is to compile the device of described control data.

Images