KR100460529B1 - Filtering system and method for streaming of data over wireless network - Google Patents

Abstract

The present invention relates to a data transmission and reception system and a method for data streaming in a wireless data communication network, and to analyze the characteristics of MPEG compression and the characteristics of the video to reduce the loss of video as much as possible while implementing a low bandwidth and adaptively It is an object of the present invention to provide a data transmission / reception system and method for data streaming in a wireless data communication network, characterized in that it is filtered, and designed to give a minimum load to the streaming server for data processing at the time of filtering.

The present invention is a system for data compression and reproduction for compressing and transmitting and reproducing video and video data from a remote server via a data communication network. The present invention provides various data from a remote server via a wireless data communication network. A subscriber station having a wireless communication module embedded therein for reproducing; QoS leveling that implements a low bandwidth to the subscriber station and sets the overall data rate based on the response to the client's data rate according to MPEG frames by analyzing the characteristics of MPEG compression and video with reduced video loss. The QoS filter performs adaptive filtering with various parameters by referring to the current QoS level, and parsing and frame extraction of MPEG data must be performed. It is characterized by consisting of a server for adjusting the overall data rate by applying different QoS filters according to the shape of the macroblock forming the frame.

According to the present invention, it is possible to adjust the overall data rate by applying different QoS filters according to the parsing and frame extraction of MPEG data, the characteristics of each extracted frame, and the type of macroblock that forms the frame. It works. In addition, the server performs QoS leveling that sets the overall data rate based on the response to the client's data rate according to the transmitted MPEG frame, so that the QoS filter filters as various parameters by referring to the current QoS level. The overall load rate can be maintained, and the overall load can be reduced by deleting MPEG frames based on the response to the packet error rate of the client, and the maximum QoS can be guaranteed by applying the appropriate QoS filter. By applying QoS filter to guarantee real-time by (Streaming), it is possible to get maximum effect without burdening server operation.

Description

Data transmission / reception system and method for data streaming in wireless data communication network {FILTERING SYSTEM AND METHOD FOR STREAMING OF DATA OVER WIRELESS NETWORK}

The present invention relates to a data transmission / reception system and a method for data streaming in a wireless data communication network, and more particularly, to perform streaming in consideration of the compression and image characteristics of data in an environment of a wireless data communication network while complying with the MPEG standard. The present invention relates to a data transmission / reception system and method for data streaming in a wireless data communication network.

As is well known, the most important requirement in wired / wireless data streaming is that the moving picture data must be continuously reproduced in a limited bandwidth, and the loss of original data should be as small as possible.

However, current MPEG is a standard created for storing a large amount of multimedia data through digital media and has a design that does not consider actual streaming at all.

For that reason, new methods for streaming are constantly being proposed, and the developed technologies are being used, but the specifications are not published, and they are not superior to MPEG data in terms of quantity and quality. Thus, MPEG streaming has been studied in a number of areas, some of which have been realized. However, due to the characteristics that still require high bandwidth and reliability, it is difficult to extend the area to the wireless area.

The prior art of MPEG streaming on the wire has been its main focus on how to lower high bitrates for transmission. In such a study, a typical streaming server / client structure on a wired line transmits MPEG bitstreams by layer, and deletes or deletes a specific picture using a characteristic of MPEG composed of each picture. Picture).

Clearly, these methods have succeeded in lowering the bandwidth in their results, but they have neglected much in the area of MPEG video playback. In addition, most of the areas in which research on a wired network are inaccessible on a wireless network having different characteristics as a lack of preparation for a large change in bandwidth and a high error rate, which are not considered important in the past.

In addition, the importance of the configuration of the filter is also to create a low bandwidth (Bandwidth), but the actual loss of compressed data is inseparable from the quality, so only for the implementation of low bandwidth (Bandwidth) The related art, which has been studied, has a problem in that the compression quality is lowered, and low quality video data is reproduced in terms of reproduction of original data. In particular, in the case of moving picture data, there is a problem that the disconnection phenomenon due to the high error rate is still present during data reproduction.

The present invention has been made in view of the above-described state of the art, and in order to reduce the loss of images as much as possible while implementing a low bandwidth, the characteristics of the MPEG compression and the characteristics of the image are analyzed and adaptively filtered according to the characteristics thereof. It is an object of the present invention to provide a data transmission / reception system and method for data streaming in a wireless data communication network, which is designed to apply a minimum load to a streaming server for data processing.

1 is a view schematically showing the configuration of a data transmission and reception system for data streaming in a wireless data communication network according to an embodiment of the present invention;

2 is a view showing a data flow in a data transmission and reception system for data streaming in a wireless data communication network according to an embodiment of the present invention;

3 is a detailed structural diagram of a subscriber station for processing data transmitted from a data transmission / reception system for data streaming in a wireless data communication network according to an embodiment of the present invention;

4 is a detailed structural diagram showing an MPEG streaming server of a data transmission / reception system for data streaming in a wireless data communication network according to an embodiment of the present invention;

FIG. 5 is a detailed structural diagram showing a QoS filter in the MPEG streaming server shown in FIG. 4;

6, 7, 8, and 9 are flow charts for each layer of MPEG data processed by a data transmission / reception system for data streaming in a wireless data communication network according to an embodiment of the present invention, and illustrating the respective process steps;

10 and 11 are views for explaining the operation of the QoS filter in the data transmission and reception system for data streaming in a wireless data communication network according to an embodiment of the present invention;

12 is a block diagram of a frame and a macroblock processed in a data transmission / reception system for data streaming in a wireless data communication network according to an embodiment of the present invention;

13 is a view showing an input form to a DCT coefficient processing unit in a data transmission / reception system for data streaming in a wireless data communication network according to an embodiment of the present invention;

14 is a view showing a correlation between I, B, P frames of MPEG in a data transmission and reception system for data streaming in a wireless data communication network according to an embodiment of the present invention;

15, 16, and 17 are views illustrating a state of use of a frame removing unit in a data transmission / reception system for data streaming in a wireless data communication network according to an embodiment of the present invention;

18 illustrates macroblock components of I, B, and P frames of MPEG in a data transmission / reception system for data streaming in a wireless data communication network according to an embodiment of the present invention;

19 is a diagram illustrating equations and parameter levels for smv , dmv , and pI in a data transmission / reception system for data streaming in a wireless data communication network according to an embodiment of the present invention;

20 is a table according to parameter determination in a QoS leveler of a data transmission / reception system for data streaming in a wireless data communication network according to an embodiment of the present invention;

21 is an operation table of a filter parameter determiner of a data transmission / reception system for data streaming in a wireless data communication network according to an embodiment of the present invention;

22 is a flowchart illustrating a signal flow of an MPEG parser in a data transmission / reception system for data streaming in a wireless data communication network according to an embodiment of the present invention;

23A and 23B are flowcharts illustrating a frame extracting unit and a frame removing unit in a data transmission / reception system for data streaming in a wireless data communication network according to an embodiment of the present invention;

24 is a flowchart illustrating a signal flow of an MV analyzer in a data transmission / reception system for data streaming in a wireless data communication network according to an embodiment of the present invention;

25 is a flowchart illustrating a DCT coefficient processor in a data transmission / reception system for data streaming in a wireless data communication network according to an embodiment of the present invention;

26 is a flowchart illustrating a signal flow of a QoS leveler in a data transmission / reception system for data streaming in a wireless data communication network according to an embodiment of the present invention;

27 is a flowchart illustrating a signal flow of a filter parameter determiner in a data transmission / reception system for data streaming in a wireless data communication network according to an embodiment of the present invention.

* Description of the symbols for the main parts of the drawings *

500: server, 502: receiver,

504: QoS leveler, 506: QoS filter,

508: MPEG data, 510: MPEG parser,

512: buffer, 514: transmitter.

In order to achieve the above object, according to a preferred embodiment of the present invention, in a system for data compression and playback for compressing, transmitting and playing video and video data from a remote server via a data communication network, a wireless data communication network A subscriber station having a wireless communication module embedded therein for receiving and reproducing various data from a remote server; QoS leveling that implements a low bandwidth to the subscriber station and sets the overall data rate based on the response to the client's data rate according to MPEG frames by analyzing the characteristics of MPEG compression and video with reduced video loss. The QoS filter performs adaptive filtering with various parameters by referring to the current QoS level, and parsing and frame extraction of MPEG data must be performed. There is provided a data transmission / reception system for data streaming in a wireless data communication network, comprising a server for adjusting the overall data rate by applying different QoS filters according to the shape of a macroblock forming a frame.

Preferably, the server comprises an MPEG parser for applying the MPEG data requested by the subscriber; A buffer for storing data through the MPEG parser; A receiving unit for receiving network reception information feedback from the subscriber; A QoS leveler for determining the GoS level in consideration of the packet loss rate and the transmission rate through the information received through the receiver; A QoS filter for performing filtering by applying a parameter generated based on the level determined by the QoS leveler; Provided is a data transmitting / receiving system for data streaming in a wireless data communication network comprising a transmitter for transmitting a QoS filtering result to a subscriber station.

Preferably, the QoS filter comprises a frame extracting unit for extracting the MPEG data stored in the buffer to separate into frames; A frame remover which removes a B frame or a B or P frame according to a parameter set in the frame separated by the frame extractor; A deletion queue for storing contents of the frame removing unit to delete the frame; An MV analyzer for calculating a result according to MV, MC, Intra, and Coded for each MPEG frame when the B and P frames are not deleted; A DCT coefficient deletion unit for deleting coefficients generated by discrete cosine change (DCT) and quantization based on the analysis contents through the MV analysis unit; A filter parameter determination unit that sets an operation parameter of the frame elimination unit as the information QoS level in the QoS leveler, and determines a parameter of the DCT coefficient deleting unit which performs the actual QoS filter by adding the QoS level and the result from the MV analyzer; The DCT coefficient deleting unit for adding DCT coefficients for an intra or coded macroblock for each frame to a deletion queue to delete the DCT coefficients according to a parameter set in the filter parameter determining unit; In the frame elimination unit, the stream elimination unit performs the actual elimination from the stored elimination queue in which the execution result of the frame elimination unit and the DCT coefficient elimination unit are combined, and then recombines them into a complete MPEG bit stream to satisfy the standard MPEG bit stream. Provided is a data transmission / reception system for data streaming in a wireless data communication network.

Preferably, the MPEG parser reads MPEG data, detects a start code of a GOP divided into an I frame and the next I frame, and stores the start code in the buffer in units of GOPs. A filtering method for streaming is provided.

Meanwhile, the present invention provides a filtering method of a data transmitting / receiving system comprising a subscriber station, a wireless data communication network, and a server to efficiently transmit compressed data and perform high quality reproduction. Making a request; The server transmitting the corresponding data to the subscriber terminal; Transmitting, by the subscriber terminal, information about a reception state of data from a server to the server as feedback; Provided is a filtering method for data streaming in a wireless data communication network, wherein the server is configured to perform data processing based on information on a reception state.

Preferably, the data processing of the server may include reading an MPEG bitstream from storage in which MPEG data is stored through the MPEG parsing; Transmitting a GOP header to a buffer in units of GOPs; Extracting, by the frame extracting unit, the GOP layer bitstream stored in the buffer in units of pictures; Adding the extracted image to the deletion queue through the frame removing unit and notifying that the extracted image is deleted; Generating a bitstream of a slice layer in which each slice is divided into respective macroblocks through an MV analyzer in the bitstream of the image layer; A process of synthesizing the results according to MV, MC, Intra, and Coded according to the characteristics of each frame with reference to the image header; Performing QoS filtering on DCT coefficients of each macro block in a DCT coefficient deleting unit including a limit pass filter and a multi-divider pass filter; Storing the filtered DCT coefficients in a deletion queue and deleting the filtered DCT coefficients; As a result, a bitstream of a new slice layer with reduced information is generated; Recombining the generated standard MPEG bitstream into a buffer; The filtering unit provides a filtering method for data streaming in a wireless data communication network, which comprises a step of transmitting to a subscriber terminal for each image unit.

Preferably, the server is provided with a filtering method for data streaming in a wireless data communication network, characterized in that for analyzing the data log file to determine the characteristics of the data communication network for the subscriber station.

Preferably, during the QoS filtering process, a feedback process is performed based on a result of recombination of the packet transmitted for each frame into a frame again; Determining whether a stream received by the stream determination unit of the subscriber station is present and performing an error check on the frame; Transmitting the number of lost frames and the number of frames in error as feedback; A filtering method for data streaming in a wireless data communication network is provided, wherein the QoS leveler determines a received content as a percentage and specifies a basic filter parameter value according to the defined content.

Preferably, the parameter specifying process reduces the overall bandwidth by first adjusting the filter parameter value upward when the number of lost frames increases, removing the frame when a continuous frame loss occurs, and removing the frame when the error rate increases. A filtering method for data streaming in a wireless data communication network is provided, further comprising the step of lowering the level of the frame removing unit again when the level of the error rate is lowered and the error rate is lowered.

Preferably, QoS filtering comprises the steps of performing an MPEG parsing process; A frame extraction process; Removing a specific frame from the extracted frames; Analyzing the motion vector; DCT coefficient removal process; A leveling process of determining a level of QoS; There is provided a filtering method for data streaming in a wireless data communication network comprising a process of determining a frame parameter.

Preferably, the MPEG parsing process includes: determining whether to perform the first execution by an algorithm of the MPEG parser; Determining whether the first 32 bits are a sequence start code; If the first 32 bits are not a sequence start code, deeming that it is not an MPEG file and performing error processing for a subscriber's request; If each first 32 bit is a sequence start code, detecting each code of the MPEG stream as a 32 bit code; Determining whether the detected code is a sequence start code; In the case of a sequence start code, processing a header, initializing a GOP counter for one MPEG stream, and initializing an interleaving flag used in the frame removing unit according to a GOP counter; Determining whether the detected code is a sequence termination code; In the case of the sequence termination code, ending the role of all streaming servers for one request when the MPEG stream is terminated; Determining whether the detected code is a GOP start code; In the case of a GOP start code, processing a header, incrementing a GOP counter according to a new GOP, and initializing a frame counter according to each GOP; Determining whether the detected code is an image start code; In the case of the picture start code, when one frame is processed, 32 bits are read again from the stream and the above process is repeatedly performed until the sequence is completed. A filtering method is provided.

Preferably, the frame extraction process includes incrementing a counter for a frame; Recording the current position and displaying it as an image start; Reading subsequent data and skipping until the next start code is reached; Displaying the position as the end of the image; Initializing various variables required for the operation of the QoS filter; Moving to the frame removing unit; Calculating a value of dmv, smv, pi after the frame removing unit is performed; Performing filtering to remove actual DCT coefficients; There is provided a filtering method for data streaming in a wireless data communication network, characterized in that the stream alignment is performed to reflect all the modifications in the MPEG stream.

Preferably, the process of removing a specific frame from the extracted frames may remove one frame according to its level, or process all macroblocks divided by slices through an MV analysis unit to perform dmv, smv, pI for image analysis. Generating a value of; Determining whether a frame is removed from the frag string; In the case of erasing a frame, recording only the positions of start and end in the deletion queue; Determining whether the frame is a slice header unless the frame is to be deleted; Processing each header in the case of the slice header; The actual macroblock is provided with a filtering method for data streaming in a wireless data communication network, characterized in that the call to the MV analysis unit, the process is performed repeatedly.

Preferably, the analyzing of the motion vector comprises the steps of: first determining whether a motion vector exists; Increasing the number of macroblocks having the motion vector if the motion vector exists; Increasing the number of intra macroblocks if there is no motion vector; determining an f code and a motion vector; obtaining and summing hsv and vsv; Recording the position of the macro block; Skipping to END_OF_BLOCK; Determining whether all macroblocks of one slice have been processed; There is provided a filtering method for data streaming in a wireless data communication network, characterized in that the process is repeated until all macroblocks of one slice are processed.

Preferably, the DCT coefficient elimination process includes first converting an actual filter value according to each parameter level; Reading the positions of the macroblocks stored in the MV analyzer one by one; Terminating if the position of the macroblock is an image end position indicating the end of the frame; Thereafter, determining whether the selected filter is a low pass filter or a multipass filter; Reading a DCT coefficient if the selected filter is a low pass filter; Incrementing a counter for the DCT coefficients; Determining whether the counter of the DCT coefficient is equal to the number specified by the parameter level; Using the DCT coefficients up to that point, adding the position to the deletion queue, and deleting the DCT coefficients thereafter; Continuing to read the buffer; Determining whether the buffer is END_OF_BLOCK; Since the up to the position becomes the end of the DCT coefficient to be deleted, a filtering method for data streaming is provided in the wireless data communication network, which comprises a process of deleting the position by adding the position to the deletion queue.

Preferably, when the DCT coefficient is removed using a multi-pass filter, dividing the actual filter value according to the parameter level into a numerator and a denominator; Reading the DCT coefficients; Incrementing the counter; Recording the position of the current DCT coefficients in a deletion queue; Determining whether the buffer is equal to END_OF_BLOCK; If the same, terminating the processing; Otherwise, determining whether the processing of the DCT coefficient corresponding to the denominator has been read; Repeating retrieving a location from the queue; When END_OF_BLOCK is terminated, a filtering method for data streaming in a wireless data communication network is provided. The method includes deleting all positions remaining in a queue.

Preferably, the leveling process of determining the level of QoS comprises: reading a network log received from a subscriber station; Calculating an average of the two previous values to obtain a current frame loss rate; Setting each weight to an approximate value, for example, 0.3, 0.6, 1.0 for the second previous value, the first previous value, and the current value; Determining whether two of the three data are above average; Obtaining a difference between the two data above the average and the average and multiplying the weights; Determining whether one piece of data lower than the average is current data; Multiplying the difference from the mean by 0.5 if one data lower than the mean is current data; If there are two sub-average data, multiplying the difference between the mean and the value by a weight if less than the mean; If one data higher than the mean is a current value, then multiplying the difference from the mean by 0.5; In order to obtain a frame error rate, a filtering method for data streaming is provided in a wireless data communication network, which comprises a process of calculating an average of three previous values.

Preferably, the step of determining the frame parameter comprises the steps of initializing a variable; The process of determining the parameter by the frame loss rate, for example, when the frame loss rate is less than 5%, less than 20%, 20% to 40%, greater than 40% to determine the parameter and ; When the frame loss rate is 20% to 40%, determining whether the frame removing unit level is greater than 2; Setting to 1 if greater than 2, and decreasing the level to 1 if not greater than 2; If the frame loss rate is greater than 40%, determining whether the frame removing unit level is less than 2; Setting the frame removing unit level to 2 when it is small and increasing it by 1 when it is larger than 2; When the frame error rate is smaller than 10%, the frame error rate is divided into 10% to 20% and larger than 20%, and determining a corresponding parameter; Setting as 1 if less than 10% and the previous frame remover level is not 0; In the case of 10% to 20%, setting as 2 when the previous frame removing unit level is smaller than 2; If it is greater than 20%, a filtering method for data streaming is provided in a wireless data communication network, characterized in that if the previous frame removing unit level is smaller than 3, it is set to 3, and if it is not, it is increased by one. .

EMBODIMENT OF THE INVENTION Hereinafter, this invention is demonstrated in detail with reference to drawings.

First, the present invention has a first object of reducing the overall data rate as a QoS filter for MPEG data as a method for overcoming low bandwidth on a wireless network. To achieve the object, first of all, parsing and frame extraction of MPEG data is performed. (Frame Extracting) should be done, and the total data rate is adjusted by applying different QoS filters according to the characteristics of each extracted frame and the type of macroblock forming the frame.

At this time, the server performs QoS leveling that sets the overall transmission rate based on the response to the client's transmission rate according to the transmitted MPEG frame, and the QoS filter filters as various parameters by referring to the current QoS level. To maintain the overall baud rate.

Also, based on the response to the packet error rate of the client, it also serves to reduce the overall load by deleting MPEG frames.

The server's response to these two responses is independent, and the results of those responses must be complex to reflect the current network conditions.

In addition, the present invention analyzes the characteristics of MPEG moving images reproduced in real time by focusing on the quality of MPEG data, which is overlooked in the existing methods, and applies the most appropriate QoS filter, not the filter focusing only on the existing bandwidth. By guaranteeing the maximum quality and applying QoS filter to guarantee the real time according to streaming, the server operation does not burden the server operation and can get the maximum effect.

1 is a diagram schematically showing the configuration of a data transmission and reception system for data streaming in a wireless data communication network according to an embodiment of the present invention.

Referring to this, reference numeral 100 denotes a portable computer device which is one of the user's mobile terminals, and 200a to 200n denotes a mobile terminal of a user capable of cellular phone, PCS, and other satellite communications.

In addition, 300 denotes a repeater for data relay between the mobile terminals 100 and 200a to 200n and a server to be described later, 400 denotes a data communication network for transmitting various wireless data, and 500 denotes a maximum image while implementing low bandwidth. It shows a QoS (Quality of Service) server that analyzes the characteristics of MPEG compression and the characteristics of video in the state of reducing loss of the signal, and performs adaptive filtering according to the characteristics.

2 is a diagram illustrating a data flow in a data transmission / reception system for data streaming in a wireless data communication network according to an embodiment of the present invention.

2 is a flowchart of network data between the subscriber station 1000 and the server 500. When the subscriber station 1000 makes a request (1) for MPEG multimedia data to the MPEG streaming server 500, the server 500 transmits the corresponding data to the subscriber station 1000, and the subscriber station 1000 feeds back information on the reception state of the data from the server 500 as a feedback. To 500.

3 is a detailed structural diagram of a subscriber station for processing data transmitted from a data transmission / reception system for data streaming in a wireless data communication network according to an embodiment of the present invention.

Referring to this, as a detailed structural diagram of the subscriber station 1000, MPEG data (A) transmitted from the server 500 is received by the receiver (Receiver: 1010) of the subscriber station 1000 to receive a buffer (Receive Buffer). It is stored in the (1020) and serves to modify the received data through the standard MPEG playback unit 1060, such as error checking and correction, frame generation through the stream justifier (1030) without any abnormality and the result To the play buffer (Play Buffer: 1040).

The standard MPEG playback unit 1060 receives corresponding MPEG data from a play buffer 1040 and plays the received MPEG data. When the network correction information such as error correction result and loss, error, and transmission rate is transmitted by the stream justifier (1030), the network log sender (1050) feeds this information back to the server in a unit of 1 second ( (B).

4 is a detailed structural diagram illustrating an MPEG streaming server of a data transmission / reception system for data streaming in a wireless data communication network according to an embodiment of the present invention.

Referring to this, MPEG data 508 requested by a subscriber as a detailed structure diagram of a server is stored in a buffer Buffer 512 through an MPEG parser 510. The MPEG parser 510 reads the MPEG data 508 to detect a start code of a group of pictures (hereinafter referred to as a GOP) divided into an I frame and the next I frame, and then stores the GOP in a buffer (Buffer: 512). Store each in units.

When the network reception information B from the subscriber is received by the receiver 502, the information is transmitted to the QoS leveler 504. Here, the QoS level is set in consideration of the packet loss rate and the transmission rate. This is applied as a parameter to the actual QoS filter (Filter) 506. The QoS filter (506) performs filtering according to its own parameters, and the result is stored in the buffer (Buffer: 512) and sent to the sender (Sender: 514). Transmits the contents of the buffer Buffer 512 to the subscriber station 1000.

FIG. 5 is a detailed structural diagram showing a QoS filter in the MPEG streaming server shown in FIG.

Referring to this, MPEG data A stored in the buffer 512 is divided into MPEG frames through a frame extractor 600.

Frames separated by the Frame Extractor 600 are placed in a Delete Queue 604 to delete B frames or B and P frames according to parameters set through the Frame Dropper 602. Save the contents.

In addition, when the B and P frames are not deleted, the MV analyzer 608 calculates a result of MV, MC, Intra, and Coded for each MPEG frame. After completing the MV analysis unit, a DCT coefficient discarder (606) for deleting coefficients generated by a discrete cosine transform (DCT) and quantization, which is the basis of MPEG compression, is provided. Perform.

The filter parameter determiner 610 sets an operation parameter of the frame dropper 602 as the information QoS level of the QoS leveler 504, and sets the QoS level and the MV analyzer. The result of 608 is determined to determine the parameters of the DCT coefficient canceler 606 that performs the actual QoS filter.

The DCT coefficient canceler 606 sets DCT coefficients for an intra or coded macroblock for each frame in the filter parameter decider 610. Add to Delete Queue (604) to delete according to parameter.

Thereafter, the stream arranger 612 deletes the result of performing the frame dropper 602 and the DCT coefficient deleter 606 by the frame dropper 602. Queue: 604 performs actual deletion and then recombines the complete MPEG bitstream (Citstream: C) to associate each frame with a header to satisfy the standard MPEG bitstream.

Hereinafter, MPEG bitstream management and MPEG QoS filtering according to an embodiment of the present invention will be described in detail with reference to the accompanying drawings.

6, 7, 8, and 9 are flowcharts illustrating layers of MPEG data and respective process steps processed by a data transmission / reception system for data streaming in a wireless data communication network according to an embodiment of the present invention.

Referring to this, MPEG Parser 510 reads an MPEG bitstream from storage in which MPEG data is stored and transmits the MPEG bitstream to the buffer 512 in GOP units starting from the GOP header. The frame extractor 600 extracts the GOP layer bitstream stored in the buffer 512 in picture units. The extracted picture is notified by the frame dropper (602) in addition to the deletion queue (Delete Queue) 604, if necessary, to be deleted.

In the bitstream of the picture layer of FIG. 7, each slice is generated in the bitstream of the slice layer separated by each macroblock in the MV analyzer 608. The result of MV, MC, Intra, and Coded is synthesized according to the characteristics of each frame.

Then, the DCT coefficient deletion unit 606, which consists of a Limit Pass Filter and a MultiDivider Pass Filter, filters the DCT coefficients of each macroblock and deletes them. Queue: 604). The result is a bitstream of a new slice layer with reduced information.

The stream arranger 612 shown in FIG. 8 actually deletes and removes the frame dropper 602 of FIG. 9 and the DCT coefficient deleter 606 of FIG. This is done by referring to Delete Queue (604), and generates a null frame, recombines it into a standard MPEG bitstream, and stores it in the buffer. Thereafter, the sender 514 performs network transmission to the subscriber station 1000 for each image unit. This is received through the receiver 1010 of the subscriber station 1000.

The receive buffer shown in FIG. 9 (Receive Buffer: 1020) stores a bitstream in units of images from a server. The Stream Justifier: 1030 performs a quick error check from the Receive Buffer 1020, detects an errored packet, replaces it with a null frame, and resets it to a standard MPEG bitstream that can be reproduced. The standard MPEG playback unit 1060 can be played back without error by storing it in the playback buffer 1040.

10 and 11 illustrate an operation of a QoS filter in a data transmission / reception system for data streaming in a wireless data communication network according to an embodiment of the present invention.

Referring to this, as shown in FIG. 10, when the DCT result 10-1 and the quantization result 10-2 for an arbitrary image are viewed, a total of 11 DCTs including DC coefficients and 10 AC coefficients are included. The coefficients appear. The result of the quantization (10-3) according to the zigzag scan is composed of the final MPEG bitstream (10-4) through Huffman Coding.

FIG. 11 is a diagram illustrating two QoS filtering methods used in the present invention. The QoS filter 506 is based on DCT coefficient removal and is designed in such a way as to guarantee minimal load on the operation of a server. It became.

The first is a limited pass filter, whose parameter is given by 4. This uses only the first four coefficients, and the coefficients considered to be the unshaded portion of the matrix 11-1. The result 10-3 in FIG. 10 is derived from the result 11-4 after filtering, in which the length is reduced by 31 bits from 61 bits to 30 bits.

In addition, the second filter is a multi-divider pass filter, in which the parameter is 2, and coefficients considered to use only ½ of the total number of coefficients are not shaded on the matrix 11-2. It becomes a part that is not. The result (11-5) was a 26 bit reduction from 61 bits to 35 bits. The characteristics and commonalities of the two filters are as follows.

Limited pass filters use a fixed number of DC coefficients on all macroblocks. This can have a great effect on the reduction of the overall bitstream size.

The multi divider pass filter has a relatively large reduction in the number of DC coefficients in each macro block, and a relatively small reduction in a macro block having a small DC coefficient. This filter outperforms the limiting pass filter in terms of image quality.

Both of the two filters have a common method of removing the DCT coefficients and can apply different parameters according to the type of each macroblock.

12 is a block diagram of a frame and a macroblock processed in a data transmission / reception system for data streaming in a wireless data communication network according to an embodiment of the present invention.

Referring to this, an MPEG frame is composed of an I frame 12-1, a P frame 12-2, and a B frame 12-3. Hereinafter, macroblock configuration diagrams 12-4 and 12-5 of each frame are described. 12-6), the shaded portion includes the DCT coefficient, that is, the portion to which the QoS filter 506 is applied.

First, referring to the macroblock configuration diagram 12-4 for an I frame, only the intrablocked intrablock (12-7) macroblock exists in the I frame 12-1, which corresponds to QoS filtering. do.

Referring to the macroblock diagram 12-5 of the P frame 12-2, the intra MC + intra macroblock 12-8 and the no MC + intra macroblock 12-9 macroblocks are used for QoS filtering. Among the non-intra blocks, all coded MC + non Intra + Coded (12-10) and no MC + non Intra + Coded (12-12) macroblocks are QoS filtering.

The macroblocks as described above all have DCT coefficients and correspond to QoS filtering. In addition, in the macroblock configuration diagram 12-6 of the B frame 12-3, the intra macroblock 12-14 and the non Intra + Coded 12-15 also include DCT coefficients. Apply.

FIG. 13 is a diagram illustrating an input form of a DCT coefficient processor in a data transmission / reception system for data streaming in a wireless data communication network according to an embodiment of the present invention.

Referring to this, it is divided into an intra macroblock of an I frame, an intra macroblock of a B, P frame, and a non-intra coded macroblock (B, P).

FIG. 14 is a diagram illustrating a correlation between I, B, and P frames of MPEG in a data transmission / reception system for data streaming in a wireless data communication network according to an embodiment of the present invention.

Referring to this, the frame removing unit 602 for deleting a specific frame is operated based on the relationship as shown in FIG. 14. The P frame uses forward prediction, starting from the previous I or P frame. The B frame is encoded using forward prediction from a previous I or P frame and forward prediction from a subsequent I or P frame. Therefore, frame deletion, which is the operation of the frame remover 602, occurs from the first B frame to the next P frame, and then to the entire GOP. The operation level description of the frame removing unit 602 is described with reference to FIGS. 15, 16, and 17.

15, 16, and 17 are diagrams illustrating a state of use of a frame removing unit in a data transmission / reception system for data streaming in a wireless data communication network according to an embodiment of the present invention.

Referring to this, the frame removing unit 602 of FIG. 15 performs the function as a method of gradually increasing deletion by placing each drop level. Numeral 15-1 shows the first frame deletion as an operation of the level 1 of the frame dropper 602.

The first frame to be deleted is the first B frames following the I frame of the GOP, which is closest to the I frame, which is a completed screen, and thus has the advantage of having the greatest effect of deletion and the least reduction of screen quality.

(15-2) is a level 2 operation, in which deletion of the B frame deleted in (15-1) and the last B frame in the GOP occur simultaneously. This is because a new I frame appears in the next GOP, so there is less screen quality due to deletion, and thus deletion is more effective.

(16-3) and (16-4) of FIG. 16 are level 3 operations, after the first and last B frames of the GOP are deleted, the B frames between the P frame and the P frame from the beginning of the remaining B frames. Will be deleted. After (16-4), all B frames are deleted from the GOP, and only the I and P frames constitute the GOP.

17 (17-5) deletes all P frames in the operation of level 4, so that the GOP has only one image stopped as only one I frame.

The frame remover 602 deletes a frame in each GOP. In this case, the GOP is a set of consecutive frames, and if only the frames of the same position are deleted each time, discontinuity continues to occur at the same position, thereby degrading the feeling of the entire image.

In order to prevent this, the frame removing unit 602 operates as an interleaving method for the levels 1,2,3.

IBBPBBPBBPBB IBBPBBPBBPBB IBBPBBPBBPBB

If there are three consecutive GOPs as described above, the interleaving application method according to each level is as follows.

Level 1:

I □□ PBBPBBPBB I □□ BPBBPBBP IBBPBBPBBP □□

Level 2:

I □□ PBBPBBP □□ IBBP □□ P □□ PBB I □□ PBBPBBP □□

Level 3:

I □□ P □□ PBBP □□ I □□ PBBP □□ P □□ I □□ P □□ PBBP

18 is a diagram illustrating macroblock components of I, B, and P frames of MPEG in a data transmission / reception system for data streaming in a wireless data communication network according to an embodiment of the present invention.

Referring to this, as described above, the filtering is performed in the actual DCT coefficient deleting unit 606. The parameter of the filter is based on the parameter through the QoS level input of the QoS leveler. The actual filtering is enabled by offsetting the value via (Analyzer: 608).

The MV analyzer 608 analyzes the offset of the parameter as a statistic of the motion vector of each macroblock. The average size of the motion vector smv and dmv for the consistency of the direction of the motion vector are determined. We compute three statistics: the percentage of intra coded blocks, pI . dmv, smv and pI are processed for one frame (P, B), which is the best data to estimate the motion and change of the image.

The average size smv of the motion vectors is summed up when the motion vector exists in the macroblock, and in this case, the forward / reverse f-code (1993 ISO / IEC, Range of Only motion vectors in half-pel and integer pel resolution [487] were calculated to reduce the load and improve the processing efficiency.

The direction coherence value of motion vector, dmv, is also used as a sign bit for macroblocks to measure how consistent the direction of motion vectors in the entire frame is. If the ratio is high, it is used to estimate the exact change of MV. Also, dmv is used to prevent misjudgment caused by the cancellation of MV values of two regions having opposite directions.

pI can be also calculated at the same time to extract smv above as a percentage of intra-coded macroblock on a frame occupies the entire macroblock as a percentage of the Intra Coded Block.

19 is a diagram illustrating equations and parameter levels for smv , dmv , and pI in a data transmission / reception system for data streaming in a wireless data communication network according to an embodiment of the present invention.

19 is a parameter level for the equations for smv, dmv, and pI and the DCT coefficient filter. In the case of dmv , the sign of each motion vector with respect to the total number of macroblocks including the motion vector is different. It is the ratio of the number difference. The reason for the weighting of 4/7 and 3/7 is that it includes the ratio difference of the motion vectors according to the average aspect ratio 4: 3.

The parameter level of the DCT coefficient filter is the level of the filter determined by the QoS leveler 504 and the filter parameter decider 610. The level of the LPF / MDPF for one level is correlated with each other. In case of converting to MDPF while using LPF level 3, the same filter is applied as level value 3.

For level 0 the LPF is 64, the number of DCT coefficients that can be passed through the filter. Since there can be up to 64 DCT coefficients for a macroblock, this is actually the same as no filter applied.

Similarly, MDPF invalidates the filter because the divider is 1 and 100% of the DCT coefficient is passed. The level then rises from 1 to 5, reducing the number of DCT coefficients passed through the filter.

LPF shows a sharp decrease from 64 to 16, which is set as the starting point of the effect in MPEG, and MDPF also varies from 75% (¾) to 25% (¼).

20 is a table according to parameter determination in a QoS leveler of a data transmission / reception system for data streaming in a wireless data communication network according to an embodiment of the present invention.

Referring to this, feedback is provided based on a result of recombination of the packet transmitted by the frame into a frame in the subscriber station 1000. A stream justifier 1030 of the subscriber station 1000 determines whether a received frame exists and performs an error check of the frame.

The number of lost frames and the number of frames in which errors occur are transmitted as feedback. The QoS leveler 504 determines the received content as a percentage and specifies basic filter parameter values according to the defined content. The QoS Leveler 504 ensures maximum picture quality without the use of filters in the case of losses of less than 5%.

However, if the number of lost frames increases, first increase the filter parameter value to reduce the overall bandwidth. In case of continuous loss of frames, frame removal is instructed. If the error rate increases, the level of the frame dropper 602 is increased. If the error rate is lowered, the level of the frame dropper 602 is lowered again.

21 is an operation table of a filter parameter determiner of a data transmission / reception system for data streaming in a wireless data communication network according to an embodiment of the present invention.

Referring to this, FIG. 21smvWowpI, dmvThe operation table of the filter parameter deciding unit (610) for setting a parameter offset value according to, wherein the basic parameter is determined by the QoS leveler (504)smvWowpI, dmvTo By reference The offset of the increase and decrease of the parameter is determined.

smv has a value from 0 to 7, and in many motion vectors, the f-code is less than 2, and the smv value of 0.0 is 1/4 for the P frame and 1/4 of the entire B frame. Will account for 2/5 of the total. And 80% of the total concentration is below 2.0. If smv is higher than 2.0, it can be regarded as an image with considerable movement.

If smv is less than 1.0, the range of motion vectors stays in a small range, which means that the estimation of motion vectors is applied in close proximity, which is estimated to be a moving or slow moving image, and is a coded macroblock (rather than an intra macroblock). Coded MacroBlock) is often used.

Due to the nature of slow moving images, intra macroblocks in I frames are often used for motion compensation, so the limited pass filter is applied, keeping the parameters of Intra (I) as they are, and relatively small and no significant changes in the image. Intra (P, B) parameters are adjusted upward.

Similarly, since the size of the coded macroblock is small, the upward adjustment of the parameter does not significantly affect the quality of the image. In case of 1.0 ~ 2.0 that smv shows proper movement, MDPF is used rather than LPF to maintain image quality, and general filtering is performed as (3,3,4).

The QoS leveler 504 may be ignored except when there is no change in the set value. In addition, if smv has a high value of 2.0 or higher, it is assumed that the image moves at a very high speed, and the parameters of intra macroblocks of I, P, and B, which are relatively less important, are upwardly adjusted and coded for compensation of fast movement. The parameters of the macroblocks are adjusted downwards to ensure effective dynamic compensation and smooth display through them.

PI indicating the intra macroblock coming percent of each frame is a value capable of estimating a change in the image, if pI of the P frame, the forward prediction (Forward Prediction) a due to the properties of using intuitively know a temporal change of an image You will have the advantage. In addition, it has a higher value than the B frame, and the change range is very large, and the response to the change of the image is very sensitive.

On the other hand, the B frame has a relatively lower value than the P frame and its variation is not wide. pI is less than 0.15 (15%) in the case of P frame, which is almost half of the total, evenly distributed from 10 to 70% and even between 70% to 100%.

However, in the case of B frames, since the use of bidirectional prediction, the number of intra blocks is greatly reduced, and 80% of the total is located within 5%. If the pI value is more than 0.05 (5%), it can be classified as a special case, and only 0.30 (30%) or more shows only a slight distribution.

With these two pI values, you can set specific conditions and apply additional filter parameters when the conditions are met. If smv is less than or equal to 0.1, P-pI is less than 3%, B-pI is less than 0.1%, and dmv is more than 0.7, it can be estimated that the image is still unchanged. In this case, in the case of MPEG, it is possible to see an adverse effect such as the screen flickering and contouring due to motion compensation for still images. Intra (I) uses MDPF to process the slow image, which is easy to feel the change in image quality, while Intra (P, B) is rarely used. If it is used to show a characteristic that does not, it does not filter because it shows the actual screen change.

In the case of Coded MacroBlock, when the image is frozen, most of the screen change is processed as Intra (P, B), so the inverse effect of Motion Compensation is the maximum filtering to see the effect of performance. Set the parameter.

If the dmv is less than 0.7, it is reasonable to assume that the smv is generated by offsetting the average value because the motion is bisected rather than guessed as a still image. If smv of the screen is less than 0.2, P-pI is less than 7%, B-pI is less than 0.3%, dmv is more than 0.6, this is the case where the change of the image is relatively small and shows slow motion. Can be guessed.

In this case, the MC and coded macroblocks occupy more than Intra (I) and Intra (P, B) because of the small movement of the screen.Intra (I) and Intra (P, B) adjust the parameters upward. Since coded macroblocks have a high MV dependent ratio, their size is smaller than the average coded macroblock.

If the intra macroblock has excellent image reconstruction, the image quality is not severely deteriorated even if the parameter is adjusted downward. If smv is less than 0.2, P-pI is more than 30% and B-pI is more than 5%, it can be estimated to be very slow and very changeable image. In this case, dmv is not considered because a large portion of the screen is processed intra in consideration of a large portion of the MC. In this case, MDPF is used to set Intra (I)> Intra (P, B)> encoded macroblocks to 3 (½), 4 (⅓) and 5 (¼), respectively. In the case of all frames, it shows the largest size. Even if the filter divider is low, sufficient performance and image quality can be guaranteed.

If smv is more than 2.0, P-pI is less than 7%, B-pI is less than 0.3% and dmv is more than 0.6, it can be estimated as a very fast and very little image. In this case, the screen may be fixed, moving in one direction, moving out of the screen, or sudden camera shake. In particular, the MDPF is used to encode Intra (I), Intra (P, B), All of the macroblocks have the effect of sufficient filter through 5 (¼), 5 (¼), 5 (¼) setting and have the characteristic of the image which is most ignored in visual response in all frames.

If smv is more than 2.0, P-pI is more than 30%, B-pI is more than 5%, the frame is estimated to be much changed and fast moving, so the frame depends more on Intra than MC. However, in the case of non-intra coded macroblocks, the size becomes larger than the average coded macroblocks, as opposed to the above.

Therefore, Intra (I) and Intra (P, B) are adjusted down, and the parameters of the relatively coded macroblock are adjusted up. In the case of a lot of change and a lot of movements, the error of one wrong macroblock grows to the full screen. Therefore, coded macroblocks should be treated more importantly than Intra (I) and Intra (P, B). The reason for this is that dmv is not used because the processing method is the same even if the direction is the same for the image showing the fast movement and the change.

Finally, if smv is more than 3.5, P-pI is more than 50% and B-pI is more than 10%, it can be estimated as a very fast moving image with very fast movement, and it is difficult to show a sophisticated screen regardless of orientation. It is a video. In this extreme case, even MPEG, which actually contains a lot of information, may not show an accurate picture until an I frame appears. Therefore, Intra (I) shows certain information on the screen as it is without using a filter to prevent the reduction of the overall image quality due to filtering.Intra (P, B) does not have a great advantage in image quality due to the characteristics of the image. Run to the maximum.

In the coded macroblock, it is difficult to see a large difference in the image under consideration unless the image is stopped. Therefore, the average value of MPDF can be specified without significant change, so that the appropriate capacity can be reduced and the proper screen quality can be maintained.

22 is a flowchart illustrating a signal flow of an MPEG parser in a data transmission / reception system for data streaming in a wireless data communication network according to an embodiment of the present invention.

First, the algorithm of the MPEG Parser (510) determines whether the first execution (Step 1: ST-1), and if it is the first time to open the actual MPEG stream (Step 2: ST-2), It is determined whether or not 32-bit is a sequence start code (step 3: ST-3).

If the first 32 bits are not the sequence start code, it is regarded as not an MPEG file and error processing for the subscriber's request is performed (ST-4).

On the other hand, when the first 32 bits are a sequence start code, each code of the MPEG stream is detected as a 32 bit code (step 5: ST-5), and it is judged whether or not the detected code is a sequence start code ( Step 6: ST-6), in the case of the sequence start code, processes a header, initializes a GOP counter for one MPEG stream, and is used in the frame dropper 602. The interleaving flag is initialized in accordance with the GOP counter (step 7: ST-7).

It is determined whether the detected code is a sequence end code (step 8: ST-8). In the case of the sequence end code, when the MPEG stream is terminated, all streaming servers for one request ( The role of the streaming server 500 is terminated (step 9: ST-9).

It is determined whether or not the detected code is a GOP start code (step 10: ST-10). In the case of the GOP start code, the process also processes a header and increments a GOP counter according to a new GOP, and frames according to each GOP. Initialize the counter (Step 11: ST-11).

It is determined whether or not the detected code is a Picture Start Code (Step 12: ST-12). If the code is a Picture Start Code, the code is moved to a part for processing an image. After the control is passed to the frame extractor 600, after one frame is processed, 32 bits are read from the stream again and the above process is repeatedly performed until the sequence is completed (step 13: ST). -13).

23A and 23B are flowcharts illustrating a frame extracting unit and a frame removing unit in a data transmission / reception system for data streaming in a wireless data communication network according to an embodiment of the present invention.

First, FIG. 23A illustrates an operation of a frame extractor 600. First, the frame extractor 600 increments a counter for a frame as an overall part of processing one frame (step 20). ST-20) The current position is recorded and displayed as the start of the image (Step 21: ST-21).

Subsequently, subsequent data is read out and skipped until the next start code is reached (Step 22: ST-22). Then, the position is displayed as a picture end (step 23: ST-23), the various parameters required for the operation of the QoS filter are initialized (step 24: ST-24), and the frame removing unit 602 (Step 25: ST-25).

After the frame removing unit 602 is performed , the values of dmv, smv, and pi are calculated (step 26: ST-26), and filtering is performed to remove actual DCT coefficients (step 27: ST-27). ), The stream arranging unit (Stream Arranger) reflecting all the modification results in the MPEG stream is completed, thereby completing processing for one frame (step 28: ST-28).

FIG. 23B illustrates an operation of the frame dropper 602. The frame dropper 602 removes one frame according to its level, or all macroblocks divided by slices. ) Is processed through the MV analyzer (Analyzer: 608) to generate the values of dmv, smv, pi for image analysis.

As an operation of the frame dropper 602, a GOP counter is used for interleaving, and it is determined whether the frame is removed from the flag string based on the GOP counter (step 30: ST). -30).

In the case of erasing a frame, only the positions of the start and the end are all recorded in the deletion queue (DeleteQueue 604) and subsequently deleted from the MPEG stream (step 31: ST-31).

If the frame is not deleted, it is determined whether it is a slice header (step 32: ST-32), and if it is a slice header, each header is processed (step 33: ST-33), and the actual macro The block calls the MV analyzer 608 and repeats the process (step 34: ST-34).

24 is a flowchart illustrating a signal flow of an MV analyzer in a data transmission / reception system for data streaming in a wireless data communication network according to an embodiment of the present invention.

Referring to this, the processing of the MV analyzer 608 calculates the presence or absence of a motion vector, the number of intra codings, and determines the size of the motion vector to calculate data of image analysis. .

First, it is first determined whether a motion vector exists (step 40: ST-40), and if the motion vector exists, the number of macroblocks having the motion vector is increased (step 41: ST-41). If does not exist, the number of intra macroblocks is increased (step 42: ST-42).

Then, the f code and the motion vector are determined (step 43: ST-43), and hsv and vsv are calculated and summed (step 44: ST-44). Then, the position of the macroblock is recorded for the DCT coefficient deleting unit 606 (step 45: ST-45).

After that, the process skips to END_OF_BLOCK (step 46: ST-46). The above processes determine whether all macroblocks of one slice have been processed (step 47: ST-47), and all macroblocks of one slice. Repeat until this is processed.

25 is a flowchart illustrating a DCT coefficient processor in a data transmission / reception system for data streaming in a wireless data communication network according to an embodiment of the present invention.

Referring to this, as the DCT coefficient canceling unit 606, a low pass filter and a multi-division pass filter are included, which is initially determined according to each parameter level. The actual filter value is converted (step 50: ST-50).

Then, the positions of the macroblocks stored by the MV analyzer (Analyzer 608) are read one by one (step 51: ST-51), and the process ends if the image end position indicates the end of one frame (step 52). ST-52). Thereafter, it is determined whether the selected filter is a low pass filter or a multipass filter (step 53: ST-53).

If the selected filter is a low pass filter, the DCT coefficient is read (step 54: ST-54), the counter for the DCT coefficient is increased (step 55: ST-55), and this is specified by the Parameter Level. It is judged whether or not it is a number (step 56: ST-56), and if it is a specified number, the DCT coefficient up to that time is used and the position is added to the deletion queue 604 so that subsequent DCT coefficients are deleted. (Step 57: ST-57).

Subsequently, the buffer is continuously read (step 58: ST-58), and it is determined whether the buffer is END_OF_BLOCK (step 59: ST-59), and the position is the end of the DCT coefficient to be deleted. Is added to the deletion queue 604 for deletion processing (step 60: ST-60).

In case of using the Multi Divide Pass Filter, the actual filter value according to the parameter level is divided into numerator and denominator (step 61: ST-61). (Step 62: ST-62).

In addition, in the case of the multi-pass filter, since the DCT coefficients are deleted as a ratio, it is not immediately known whether the current DCT coefficients are deleted or preserved, and thus the position is recorded using a queue (step 63: ST-63). ).

Similarly, it is determined whether or not the buffer is equal to END_OF_BLOCK (step 64: ST-64), and if it is the same, the processing is terminated, and whether the processing of the DCT coefficient corresponding to the denominator is not identical and is read out. (Step 65: ST-65), since the DCT coefficients corresponding to the number of molecules are confirmed to be preserved at that time, it is repeated to remove the position from the queue (Step 66: ST-66). Step 67: ST-67, Step 68: ST-68, Step 69: ST-69).

As a result, when the END_OF_BLOCK ends, all positions remaining in the queue are subject to deletion (step 70: ST-70, step 71: ST-71).

FIG. 26 is a flowchart illustrating a signal flow of a QoS leveler in a data transmission / reception system for data streaming in a wireless data communication network according to an embodiment of the present invention.

Referring to this, the QoS leveler 504 is a part of determining a parameter for inferring a current state and applying a change to a filter as network feedback from the subscriber station 1000.

In this case, the focus is that if the response is too slow for the purpose of generating a value that is not too slow or too fast depending on the network condition, the processing of the corresponding filter is delayed when the network condition is bad, and the network condition has recovered. In this case, the time to restore the original state is delayed.

On the other hand, when the response is fast, the filter parameter is constantly changed according to the network conditions, which is constantly changed, so that the subscriber does not see a consistent video and obtains only the constantly changing video.

Therefore, in the present invention, the previous two data and the current data have three types, weighted according to each time difference, and detecting the overall change according to the position of the current data according to the average and calculating a value accordingly.

First, the network log received from the subscriber station 1000 is read (step 80: ST-80), and the average of the current frame loss rate and the two previous values is calculated (step 81). Step: ST-81).

Each weight is set to the approximate values 0.3, 0.6, and 1.0 corresponding to ⅓, ⅔, 1 (step 82: ST-82).

When the average is generated according to three data, there are always two data which are higher or lower than the average value based on the average value and one data that is the opposite. In this case, it is determined whether the data of two identical personalities are above the average (step 83: ST-83), and if the three data are larger than the average (step 84: ST-84), The difference is obtained and the weight is multiplied (step 85: ST-85).

On the other hand, it is determined whether one piece of data that is lower than the average is the current data (step 86: ST-86). If one piece of data that is lower than the average is the current data, the difference from the average is multiplied by 0.5 ( Step 87: ST-87).

The final result subtracts the mean of the sum of the product of the difference and the weight from the mean and the sum of the weights, and adds that value if one is lower than the mean so that the current value has a higher impact on the final result.

In the opposite case, if there are two values smaller than the mean (step 88: ST-88), if it is smaller than the mean, the difference between the mean and the value is multiplied by the weight (step 89: ST-89) and is higher than the mean. If the value is present (step 90: ST-90), the difference from the mean is also multiplied by 0.5 (step 91: ST-91).

Then, it is determined whether the sum of the frame loss rates is greater than two (step 92: ST-92, step 93: ST-93).

In the end, the Frame Loss Rate calculates the average from the previous data to the current data to compensate for the difference from the average, and imposes a contribution based on the distance from the average of all three data to estimate the overall flow. In addition, this result has the effect of increasing the contribution according to the final result when the current value is different from the mean so that the current state is impaired due to the mean according to the previous data and the response is not slow. -94).

Taken together, the average prevents slow response to changes in current network conditions, and also eliminates pop-up values such as noise in the middle to prevent many immediate changes.

Frame Error Rate is used by averaging the previous three data. Frame Error Rate is directly related to the function of Frame Dropper 602, so the result can be known as an average, and it should be used simply because abrupt change should be avoided (Step 95: ST-95). .

27 is a flowchart illustrating a signal flow of a filter parameter determiner in a data transmission / reception system for data streaming in a wireless data communication network according to an embodiment of the present invention.

Referring to this, the filter parameter determiner 610 is for determining the actual parameter of FIG. 20 based on the frame loss rate and the frame error rate determined at the QoS level.

In the figure, the F.D level represents the frame removing unit 602 level, LPF (Limit Pass Filter) represents a limited pass filter, and MDPF (Multi Divider Pass Filter) represents a multi-pass filter.

First, while processing the frame loss rate (Frame Loss Rate) is initialized a variable so that the result does not affect the frame error rate (Frame 100: ST-100), the frame loss rate (Frame Loss Rate) According to (Step 101: ST-101), the determined parameter is determined (Step 102: ST-102, Step 103: ST-103, Step 104: ST-104, Step 108: ST-108).

If the frame loss rate is 20% to 40%, it is basically determined whether or not the frame removing unit level is greater than 2 (step 105: ST-105), and if the frame loss rate is higher than 2, it is set to 1 ( Step 106: ST-106) If the level is not greater than 2, the level is decreased by 1. (Step 107: ST-107).

If the frame loss rate is greater than 40%, it is determined whether the frame remover level is less than 2 (step 109: ST-109), and if it is small, the frame remover level is set to 2 (the second). Step 110: ST-110, if it is greater than 2, increase by one (step 111: ST-111).

In these two cases, the set parameters allow the user to know the value before the frame remover level is changed in the frame loss rate when increasing or decreasing the previous result in the frame error rate.

Then, according to the frame error rate (step 112: ST-112), if less than 10%, divided into 10% ~ 20%, greater than 20% and determine the corresponding parameter (step 113) ST-113, step 116: ST-116, step 119: ST-119) If less than 10% and the previous frame dropper level is not 0 (step 114: ST-114) (1st step ST-115).

In the case of 10% to 20%, if the previous frame dropper level is less than 2 (step 117: ST-117), the value is set as 2 (step 118: ST-118).

If it is greater than 20%, if the previous frame dropper level is less than 3 (step 120: ST-120), it is set to 3 (step 121: ST-121), otherwise it is increased by 1. (Step 122: ST-122).

In addition, although a data transmission / reception system and method for data streaming in a wireless data communication network according to an embodiment of the present invention have described data processing centering on MPEG in this embodiment, it is evident that it can be applied to all kinds of various multimedia.

Meanwhile, the data transmission / reception system and method for data streaming in a wireless data communication network according to an embodiment of the present invention are not limited to the above embodiments but various modifications can be made without departing from the technical gist of the present invention.

As described above, a data transmission / reception system and method for data streaming in a wireless data communication network according to the present invention include parsing and frame extraction of MPEG data and macros constituting frames and characteristics of each extracted frame. According to the type of the block, it is possible to adjust the overall data rate by applying different QoS filters. In addition, the server performs QoS leveling that sets the overall data rate based on the response to the client's data rate according to the transmitted MPEG frame, so that the QoS filter filters as various parameters by referring to the current QoS level. The overall load rate can be maintained, and the overall load can be reduced by deleting MPEG frames based on the response to the packet error rate of the client, and the maximum QoS can be guaranteed by applying the appropriate QoS filter. By applying QoS filter to guarantee real-time by (Streaming), it is possible to get maximum effect without burdening server operation.

Claims (18)

A system for data compression and playback for compressing, transmitting and reproducing video and video data from a remote server via a data communication network, A control unit which receives and reproduces various data from a remote server via a wireless data communication network, generates a transmission rate and an error rate data of the transmission data, and automatically feeds back to the server that transmitted the data, and receives and transmits the compressed data therein. A subscriber station having a wireless communication module for transmitting the error rate information; An MPEG parser for authorizing MPEG data requested by a subscriber to the subscriber station; A buffer for storing data through the MPEG parser; A receiving unit which receives network reception information feedback from the subscriber; A QoS leveler for determining the GoS level in consideration of the packet loss rate and the transmission rate through the information received through the receiver; A QoS filter for performing filtering by applying a parameter generated based on the level determined by the QoS leveler; A data transmission / reception system for data streaming in a wireless data communication network, characterized in that the server consisting of a transmission unit for transmitting the QoS filtering result to the subscriber station. delete 2. The apparatus of claim 1, wherein the QoS filter comprises: a frame extractor for extracting MPEG data stored in the buffer and separating the MPEG data into frames; A frame remover which removes a B frame or a B or P frame according to a parameter set in the frame separated by the frame extractor; A deletion queue for storing contents of the frame removing unit to delete the frame; MV analysis unit for calculating the results according to the MV and MC, Intra, Coded for the B, P MPEG frame when the B, P frame is not deleted; A DCT coefficient deleting unit for deleting coefficients generated by discrete cosine change (DCT) and quantization based on MPEG analysis according to the analysis contents; A filter parameter determination unit that sets an operation parameter of the frame elimination unit as the information QoS level in the QoS leveler, and determines a parameter of the DCT coefficient deleting unit which performs the actual QoS filter by adding the QoS level and the result from the MV analyzer; Discrete cosine change (DCT) and quantization, which are the basis of MPEG compression, according to DCT coefficients for macroblocks that are intra or coded for each frame, and the analysis contents of the MV analysis unit. The DCT coefficient deleting unit for adding the generated coefficients to the deletion queue to delete the coefficients generated by " In the frame elimination unit, the stream elimination unit performs the actual elimination from the stored elimination queue in which the execution result of the frame elimination unit and the DCT coefficient elimination unit are combined, and then recombines them into a complete MPEG bit stream to satisfy the standard MPEG bit stream. Data transmission and reception system for data streaming in a wireless data communication network, characterized in that consisting of. The wireless data receiver of claim 2, wherein the MPEG parser reads MPEG data, detects a start code of a GOP divided into an I frame and the next I frame, and stores the start code in a buffer in a GOP unit. Data transmission and reception system for data streaming in a communication network. A filtering method of a data transmission / reception system comprising a subscriber station, a wireless data communication network, and a server to efficiently transmit compressed data and to perform high quality reproduction. The subscriber station making a request for MPEG multimedia data to an MPEG streaming server; The server transmitting the corresponding data to the subscriber terminal; Transmitting, by the subscriber terminal, information about a reception state of data from a server to the server as feedback; The data transmission and reception method for data streaming in a wireless data communication network, characterized in that the server is a process for performing data processing based on the information on the reception status. The method of claim 5, wherein the data processing of the server comprises: reading an MPEG bitstream from a storage in which MPEG data is stored through the MPEG parsing; Transmitting a GOP header to a buffer in units of GOPs; Extracting, by the frame extracting unit, the GOP layer bitstream stored in the buffer in units of pictures; Adding the extracted image to the deletion queue through the frame removing unit and notifying that the extracted image is deleted; Generating a bitstream of a slice layer in which each slice is divided into respective macroblocks through an MV analyzer in the bitstream of the image layer; A process of synthesizing the results according to MV, MC, Intra, and Coded according to the characteristics of each frame with reference to the image header; Performing QoS filtering on DCT coefficients of each macro block in a DCT coefficient deleting unit including a limit pass filter and a multi-divider pass filter; Storing the filtered DCT coefficients in a deletion queue and deleting the filtered DCT coefficients; As a result, a bitstream of a new slice layer with reduced information is generated; Recombining the generated standard MPEG bitstream into a buffer; And transmitting the data from the transmitter to the subscriber terminal for each image unit in the wireless data communication network. The method of claim 5, wherein the server analyzes a data log file to determine characteristics of the data communication network for the subscriber station. 7. The method of claim 6, further comprising: during the QoS filtering process, performing feedback based on a result of recombination of the packet transmitted by frame into a frame by the subscriber station; Determining whether a stream received by the stream determination unit of the subscriber station is present and performing an error check on the frame; Transmitting the number of lost frames and the number of frames in error as feedback; And determining, by the QoS leveler, the received content as a percentage and specifying a basic filter parameter value according to the defined content. 9. The method of claim 8, wherein the parameter specifying process reduces the overall bandwidth by first adjusting the filter parameter value upward when the number of lost frames increases, and when the loss of subsequent frames occurs, removing the frames and increasing the error rate. The method of transmitting and receiving data for data streaming in a wireless data communication network, further comprising: increasing the level of the frame removing unit and lowering the level of the frame removing unit again when the error rate is lowered. 7. The method of claim 6, wherein QoS filtering comprises: performing an MPEG parsing process; A frame extraction process; Removing a specific frame from the extracted frames; Analyzing the motion vector; DCT coefficient removal process; A leveling process of determining a level of QoS; A method for transmitting and receiving data for data streaming in a wireless data communication network, characterized in that the step of determining the frame parameter. The method of claim 10, wherein the MPEG parsing process comprises: determining whether to execute the first time using an algorithm of an MPEG parser; Determining whether the first 32 bits are a sequence start code; If the first 32 bits are not a sequence start code, deeming that it is not an MPEG file and performing error processing for a subscriber's request; If each first 32 bit is a sequence start code, detecting each code of the MPEG stream as a 32 bit code; Determining whether the detected code is a sequence start code; In the case of a sequence start code, processing a header, initializing a GOP counter for one MPEG stream, and initializing an interleaving flag used in the frame removing unit according to a GOP counter; Determining whether the detected code is a sequence termination code; In the case of the sequence termination code, ending the role of all streaming servers for one request when the MPEG stream is terminated; Determining whether the detected code is a GOP start code; In the case of a GOP start code, processing a header, incrementing a GOP counter according to a new GOP, and initializing a frame counter according to each GOP; Determining whether the detected code is an image start code; In the case of the picture start code, when one frame is processed, 32 bits are read again from the stream and the above process is repeatedly performed until the sequence is completed. How to send and receive data. 11. The method of claim 10, wherein the frame extraction process comprises: incrementing a counter for a frame; Recording the current position and displaying it as an image start; Reading subsequent data and skipping until the next start code is reached; Displaying the position as end of the image; Initializing various variables required for the operation of the QoS filter; Moving to the frame removing unit; Calculating a value of dmv, smv, pi after the frame removing unit is performed; Performing filtering to remove actual DCT coefficients; A method of transmitting and receiving data for data streaming in a wireless data communication network, characterized in that the stream alignment is performed to reflect all the modifications to the MPEG stream. The method of claim 10, wherein the removing of a specific frame from the extracted frames comprises removing one frame according to its level, or processing all macroblocks divided by slices through an MV analyzer to perform a dmv for image analysis . generating a value of smv, pi ; Determining whether a frame is removed from the frag string; In the case of erasing a frame, recording only the positions of start and end in the deletion queue; Determining whether the frame is a slice header unless the frame is to be deleted; Processing each header in the case of the slice header; The actual macroblock is a method of transmitting and receiving data for data streaming in a wireless data communication network, characterized in that consisting of the process of repeatedly calling the MV analysis unit. The method of claim 10, wherein the analyzing of the motion vector comprises: determining whether a motion vector exists for the first time; Increasing the number of intra macroblocks if a motion vector exists; If the motion vector does not exist, increasing the number of macroblocks having the motion vector; determining an f code and a motion vector; obtaining and summing hsv and vsv; Recording the position of the macro block; Skipping to END_OF_BLOCK; Determining whether all macroblocks of one slice have been processed; A method of transmitting and receiving data for data streaming in a wireless data communication network, characterized by repeating the process until all macroblocks of one slice are processed. 11. The method of claim 10, wherein the DCT coefficient removing step comprises: initially converting an actual filter value according to each parameter level; Reading the positions of the macroblocks stored in the MV analyzer one by one; Terminating if the position of the macroblock is an image end position indicating the end of the frame; Thereafter, determining whether the selected filter is a low pass filter or a multipass filter; Reading a DCT coefficient if the selected filter is a low pass filter; Incrementing a counter for the DCT coefficients; Determining whether the counter of the DCT coefficient is equal to the number specified by the parameter level; Using the DCT coefficients up to that point, adding the position to the deletion queue, and deleting the DCT coefficients thereafter; Continuing to read the buffer; Determining whether the buffer is END_OF_BLOCK; And a process of deleting the position by adding the position to the deletion queue, since the up to the position becomes the end of the DCT coefficient to be deleted. 16. The method of claim 15, further comprising: dividing a value of an actual filter according to a parameter level into a numerator and a denominator when removing a DCT coefficient using a multipass filter; Reading the DCT coefficients; Incrementing the counter; Recording the position of the current DCT coefficients in a deletion queue; Determining whether the buffer is equal to END_OF_BLOCK; If the same, terminating the processing; Otherwise, determining whether the processing of the DCT coefficient corresponding to the denominator has been read; Repeating retrieving a location from the queue; A method of transmitting / receiving data for data streaming in a wireless data communication network, characterized in that, when END_OF_BLOCK is terminated, deleting all remaining positions in the queue. 11. The method of claim 10, wherein the leveling process of determining the level of QoS comprises: reading a network log received from a subscriber station; Calculating an average of the current Frame Loss Rate and the two immediately preceding values; Setting each weight to an approximate value of 0.3, 0.6, 1.0; Determining whether two of the three data of the same nature are above average; For each of the three data, obtaining a difference from the average and multiplying the weights by the weight; Determining whether one piece of data lower than the average is current data; Multiplying the difference from the mean by 0.5 if one data lower than the mean is current data; When there are two values smaller than the mean, multiplying the difference between the mean and the value by a weight if less than the mean; If the value higher than the mean is a current value, multiplying the difference from the mean by 0.5; A method of transmitting and receiving data for data streaming in a wireless data communication network, characterized in that the process of calculating the average of the previous three values to obtain a frame error rate (Frame Error Rate). The method of claim 10, wherein the determining of the frame parameter comprises: initializing a variable; When the frame loss rate is 20% to 40%, determining whether the frame removing unit level is greater than 2; Setting to 1 if greater than 2, and decreasing the level to 1 if not greater than 2; If the frame loss rate is greater than 40%, determining whether the frame removing unit level is less than 2; Setting the frame removing unit level to 2 when it is small and increasing it by 1 when it is larger than 2; When the frame loss rate is less than 10%, the process is divided into 10% to 20% and greater than 20%, and determining a corresponding parameter; Setting the value as 1 when the level is less than 10% and the previous frame remover level is not 0; In the case of 10% to 20%, setting as 2 when the previous frame removing unit level is smaller than 2; If greater than 20%, if the previous frame removing unit level is less than 3, set to 3, otherwise it is increased by one.