JP2001203749A - Device and system for high efficiency data transmission - Google Patents

Abstract

PROBLEM TO BE SOLVED: To transmit a transmission unit in a forcedly short elapsed time, also to collectively transmit short data and to improve transmission efficiency by lowering the overhead of division transmission. SOLUTION: This system with a configuration which makes data from an inputting means redundant and encoded the and performs packet transmission, is provided with a buffer storing these data in the unit of a prescribed length, a data storing and forced transmission instructing means which performs force division encoding and makes a transmission instruction (S15) when a prescribed time passes from the initial time when the data are stored in an empty buffer (S14) and a data encoding dividing means which makes the data of a prescribed length from the buffer redundant and encoded.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention generally relates to a technique for improving the reliability of communication in a communication path for performing data communication using packets, and more particularly, to the time required for communication such as the Internet. When communication is performed using a network that is easy to transmit and packets are not always transmitted reliably, communication loss and fluctuation of required time are reduced, time is shortened, and communication quality and efficiency are improved. About techniques.

[0002]

2. Description of the Related Art As a technique for improving the reliability of communication in a communication path in which data communication is performed by using packets, generally, on the transmission side, data to be transmitted is made redundant, divided into a series of plural packets, and transmitted. However, by enabling the receiver to recover the data that the sender wanted to transmit by only receiving a part of a series of packets, some of the packets sent by the sender could not reach the receiver. Better techniques are used.

[0003] For example, a first conventional example shown in Japanese Patent Application Laid-Open No. Hei 8-228190 "Progressive error correction system and method" will be described with reference to FIG.

FIG. 20 is a block diagram showing an embodiment of the conventional example 1. In this figure, a transmitting apparatus 101 and a receiving apparatus 201 include a data terminal apparatus (DTE) 104, a data communication apparatus (DCE) 105, Data terminal equipment (DTE) 204 and data communication equipment (DCE) 205
Exchange data via each. The transmitting device 101 and the receiving device 201 are realized using common electronic components. The electronic components are microprocessor 110/2
10, a digital signal processor (DSP) 111/211 coupled to a microprocessor, and a controller 112/21 for synchronizing timing signals between these components.
Consists of two. The controller 112/212 also controls the microprocessor 110/210 and the digital signal processor 111/212 to ensure that packets are transmitted and received.
The monitoring of 211 is performed.

The microprocessor 110/210 comprises a central processing unit (CPU) 113/213, an EEPROM (electrically erasable programmable ROM) 114/214, and a RAM (random access memory) 115/215. The CPU 113 executes a program stored in the EEPROM 114, performs packetization (packet assembly), framing, and encoding of data for the transmission device 101, and forms N parity packets to form M parity packets. N packets are encoded by combining several packets selected from the packets, and the data stream composed of N + M packets is transmitted to the receiving device 201. Similarly, CPU2
13 executes a program stored in the EEPROM 214 to perform packet disassembly and decoding of data for the receiving device 201, and at least N packets out of N + M packets in the data stream are received without being destroyed. As far as possible, it operates to reconstruct N original packets. Note that the values of N and M do not need to be fixed, and can be dynamically set according to the number of errors detected by the receiving device 201.

In the technique shown in the first conventional example, M parity packets are generated for N packets, and N + M packets are transmitted from the transmitting apparatus 101 to the receiving apparatus 201 as a data stream. So that
The data stream cannot be transmitted from the transmitting device 101 to the receiving device 201 until N packets are collected on the transmitting device 101 side. Data terminal equipment (DTE)
If a packet arrives at 104 with a delay on the time axis, the packet that arrived earlier waits for a packet that arrives later until N packets are collected, and the transmission until all data arrives is delayed. There was a problem. When generating M parity packets from N packets, it is assumed that the size of each packet is uniform. It is necessary to make the length of each packet equal to the maximum size, or to the size obtained by adding a margin,
There is a problem that unnecessary traffic that is larger than the required number of packets N occurs.

As a technique for solving this problem, a second conventional example shown in Japanese Patent Application Laid-Open No. 9-64913, entitled "Data Communication Assurance Method" will be described with reference to FIG. FIGS. 21A and 21B are system configuration diagrams in a second conventional example, in which FIG. 21A is a block diagram illustrating a data communication system using the same conventional example, and FIG. 21B is an explanatory diagram illustrating a configuration of a memory in the conventional example. . In FIG. 21A, an information processing apparatus 300 includes a CPU 301 for performing an arithmetic process and a memory 30.
2 and a communication network interface (hereinafter, communication network I / F) 3
And a communication network 304 using a communication network I / F 303.
, And can communicate with another information processing apparatus 300.

[0008] FIG. 21B shows each information processing apparatus 300.
The configuration of the memory 302 of FIG.
10 and the received packet 320, three fragments 311 and 321 are provided, respectively, and restoration fragments 312 and 322 are provided. Here, the sizes of the fragments 311 and 321 and the restoration fragments 312 and 322 are
4 is smaller than the maximum data length that can be transmitted and received.
The three fragments 311 are generated by dividing the content of the transmission packet 310 into three, and the restoration fragment 31
2 is such that even if any one of the three fragments 311 or the restoration fragment 312 is lost, the contents of the transmission packet 310 can be restored.
11 is generated by a predetermined logical operation.

In the information processing apparatus 300, as described above, the fragment 311 and the restoration fragment 312 are generated from the transmission packet 310 and transmitted to the other information processing apparatus 300 via the communication network 304. By receiving all the fragments 321 or receiving any two of the three fragments 321 and the restoration fragment 322, the processing device 300 can restore the same information as the transmission packet 310 as the reception packet 320. it can. In the above description, the content of one transmission packet 310 is divided into three fragments 311. However, this is generally performed by N fragments 311.
The value of N can be dynamically set according to the fragment loss rate in the communication network 304.

In the technique shown in the second conventional example, unlike the first conventional example, data redundancy is performed based on one packet, so that the data transmission delay that may occur in the first conventional example is reduced. Will be resolved. However, in the second conventional example, when the size of a transmission packet 310 before being divided into fragments 311 is equal to or less than the maximum data length that can be transmitted and received on the communication network 304, information that can be transmitted as one fragment is It will be divided into many fragments and transmitted. In the second conventional example, for example, predetermined data is transmitted in a set division number. For this reason, when the present technology is used in a high-speed communication network with a high data transmission rate, which is becoming popular in recent years, if short data is always divided into short lengths, the overhead load for division in such a high-speed communication network increases. . In this way, the number of fragments greatly affects the traffic, so that the number of fragments to be transmitted greatly increases, so that wasteful time is required and there is a problem of transmission efficiency. The problem is to apply the present technology to IP packets generated based on a TCP / IP protocol used on the Internet, for example, on a 100 Mbps to 1 Gbps Ethernet LAN constructed using a switching hub, which is currently spreading. Can often easily occur.

[0011]

The problems to be solved by the prior art are summarized as follows. As shown in the first prior art, if the data contents are to be ensured, the redundancy is increased and the required number of packets is increased. It takes a long time to receive the data, and as shown in the second conventional example, there is a problem that the transmission efficiency is reduced when the data is divided into short data and transmitted.

SUMMARY OF THE INVENTION The present invention has been made to solve the above-described problems. Forcibly transmitting a transmission unit with a short elapsed time and transmitting short data at a time, a data transmission delay and a wasteful transmission are made. The generation of traffic is suppressed, the overhead time of divided transmission is reduced, the communication quality is improved, and efficient transmission can be performed.

[0013]

According to the present invention, there is provided a high-efficiency data transmitting apparatus, comprising: a buffer for storing data in a predetermined length unit in a configuration in which data from an input unit is redundantly encoded and transmitted; When a predetermined time elapses from the initial time at which data is stored from the empty state of the buffer, data accumulation / forced transmission instructing means for instructing transmission by forcibly dividing and encoding, and making predetermined-length data from the buffer redundant, Data encoding / division means for encoding is provided.

Still further, the data encoding / dividing means determines the number of data length divisions and the redundancy based on a predetermined length of the transmission data and a transmission probability obtained in an allowable transmission delay time after a predetermined time has elapsed. I did it.

Still further, the data storage / forced transmission instructing means dynamically changes the predetermined time according to the transmission probability obtained in the allowable transmission delay time.

Still further, a packet receiving means is provided to obtain delay time information and calculate a transmission probability.

Furthermore, an identifier indicating whether the data can be divided and combined is added to the input data, and when the data can be divided and combined, the data is stored in a predetermined length unit of the buffer, and the data is redundantly encoded.

Still further, an identifier for identifying a stream is added to the input data, and the buffer is stored in units of stream identifiers, and when the data in the buffer reaches a predetermined length, the data is redundantly encoded and transmitted.

A high-efficiency data transmission system according to the present invention comprises: a buffer for storing data from an input means in a unit of a predetermined length; and a predetermined time from an initial time at which data is stored when the buffer is empty. Via a communication path, a data storage / forced transmission instructing means for instructing transmission by forcibly dividing and encoding when the time elapses, and a data encoding and dividing means for making redundant and encoding predetermined length data from the buffer, It comprises a data receiving device provided with data decoding means for receiving data from the data transmitting device and immediately starting decoding when data of a predetermined redundancy is received.

[0020]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Embodiment 1 A system that suppresses packetization of data and suppresses the waiting time until the start of transmission to improve transmission efficiency will be described. FIG. 1 is a diagram showing a system configuration according to Embodiment 1 of the present invention. In FIG. 1, a data transmitting means 1 and a data receiving means 2 are connected by a communication path 3. The data transmission means 1 includes a data input means 4, a data storage / forced transmission instruction means 5, which is a new important element, a buffer 6, a data encoding division means 7,
The data receiving means 2 includes a packet receiving means 9, a data decoding means 10, and a data output means 11.

Next, the operation of the system according to the present embodiment will be described with reference to FIG. In FIG. 2 for explaining the operation, in the data transmission means 1, one or a plurality of data 20 received by the data input means 4 are temporarily stored in the buffer 6 according to the instruction of the data accumulation / forced transmission instruction means 5. The data accumulation / forced transmission instructing means 5 passes the data encoding / dividing means 7 when the time T has elapsed from the time when the oldest data among the data 20 stored in the buffer 6 is stored. The data encoding / dividing means 7 first divides the data 20 passed from the data accumulation / forced transmission instructing means 5 into N (N = 2 in FIG. 2) packets 21 and then divides the data into N packets 21 for redundancy. The original data 20 is decoded if the packet 21 holding M parity information (M = 1 in FIG. 2) can be obtained from any of the N packets 21 out of the N packets M 21. Encode and generate as possible. After that, the data coding division unit 7 transmits the data to the communication channel 3 using the packet transmission unit 8.

The N + M packets 21 transmitted to the communication path 3 are received by the packet receiving means 9 of the data receiving means 2 and passed to the data decoding means 10. N + M packets 21 transmitted by the packet transmitting means 8
Do not always arrive in the order in which they were sent.
The time required for the transmission of the data is not necessarily constant, and it is possible that some of the data may be lost and not passed to the data decoding means 10. Since the original data 20 can be decoded from the N packets 21, the data decoding unit 10 performs the data decoding process when any of the N packets 21 is passed from the data receiving unit 2. Do
The reconstructed data 21 is output using the data output means 11.

FIG. 3 is a flowchart showing the operation of the data storage / forced transmission instruction means 5 in the present embodiment. In the first processing step S10 of the data accumulation / forced transmission instructing means 5, first, when the data 20 is received from the data input means 4 (hereinafter, the description of the steps is omitted, S
11) The received data 20 is stored in the internal buffer 6 (S12), and the time at which it was received is temporarily stored (S13). Thereafter, when the time T has elapsed (S14
Yes) The data 20 accumulated in the buffer 6 is passed to the data encoding / dividing unit 7 (S15), the buffer 6 is emptied (S16), and the process returns to the process of receiving new data 20 from the data input unit 4 again (S11). . On the other hand, if new data 20 is received from the data input means 4 before the time T elapses (No in S14) (Yes in S17), the new data 20 is additionally stored in the internal buffer 6 until the required data amount is reached (S18). ).

The operation of the data encoding / dividing means 7 and the data decoding means 10 according to the present embodiment will be described with reference to FIGS. The data encoding / dividing means 7 can use various methods generally used for data redundancy. For example, when M = 1, a parity method using exclusive OR can be adopted. N
= 2, M = 1, and the operation when the parity method using the exclusive OR is adopted. In general, various units such as bits, bytes, and 4 bytes can be set as units for making data redundant or divided, and any of these units can be used. Since the effect of the present invention does not change even when the unit is adopted, the operation will be described below assuming that data is handled in byte units for easy understanding of the description.

FIG. 4 is an explanatory diagram of the operation of the data encoding / dividing means 7 based on a parity method using exclusive OR with N = 2 and M = 1. The data encoding / dividing means 7 first divides the data 20 instructed to be transmitted from the data accumulation / forced transmission instructing means 5 by N = 2 equal-size partial data A
22 and B22. At this time, data 20
If the size of N is not divisible by N = 2, the data 20 is divided after adding data of the value 0 corresponding to the number of bytes obtained by taking the remainder of N to the size of the data 20. Next, N = 2 partial data 2
2, the exclusive OR of the bits at the same position is calculated, and parity data P23 having the same size as the partial data is generated from the result (FIG. 4A). Then, the generated N = 2 pieces of partial data A 22 and B 22 and M =
For one piece of parity data P23, an ID number, the size of the entire data, and a value indicating the head position of each partial data with respect to the entire data (-1 in the case of parity data, a value indicating that). To generate a packet 21 (FIG. 4B).

FIG. 5 is an explanatory diagram of the operation of the data decoding means 10 corresponding to the data encoding / dividing means 7. Among the packets 21 passed from the packet receiving means 9, the packets 21 having the same ID number are subjected to the data decoding processing.
When N = 2 packets 21 having the same ID number among the received packets 21 are both packets 21 holding the partial data 22, the partial data A 22 and B 22 held by both packets 21 are simply combined. Thus, the data 20 is restored (FIG. 5A). On the other hand, if one of the N = 2 packets 21 having the same ID number is the packet 21 holding the partial data 22 and the other one is the packet 21 holding the parity data 23, the packet held by the packet 21 The exclusive OR of the bits at the same position of the data A 22 and the parity data P23 is obtained,
The data 20 is restored by reproducing the partial data B 22 that has not been received from the result and then combining the received partial data A 22 with the reproduced partial data B 22 (FIG. 5B).

Since the system of the present embodiment operates as described above, for example, if the probability that a packet arrives from the packet transmitting means 8 to the packet receiving means 9 within the time Td in the communication path 3 is 0.9, the data Without redundancy
The probability that the entire data will arrive within the time Td when sent in one packet will be 0.81, whereas the entire data will be within the time Td by performing N + 1, M = 1 and 2 + 1 redundancy. The probability of arrival increases to 0.972. In addition, the processing contents of the data transmitting means 1 and the processing contents of the data receiving means 2 are the same as those of the conventional examples 1 and 2 except that the data storage / compulsory transmission instructing means 5 holds data for a certain time.
Is not particularly complicated as compared with the processing contents of Therefore, the processing time is not much different from that of the prior art, and is sufficiently short with respect to the packet communication delay time in the communication path 3 and the time during which data stays in the buffer 6 can be suppressed to T at most. The time required for the data input to the means 4 to be obtained from the data output means 11 is at most approximately the sum of the communication delay time on the communication path 3 and the time T instructed by the data accumulation / forced transmission instructing means 5. The time required for data input to the data input means 4 to be obtained from the data output means 11 with a probability of 0.972 is represented by time T
Data transmission delay can be suppressed by assuring d + T or less.

In this embodiment, the data input means 4
Is temporarily stored in the buffer 6 according to the instruction of the data accumulation / forced transmission instructing means 5 and then passed to the data encoding / dividing means 7 where the data is encoded and packetized. The number of packets to be generated is significantly reduced as compared to the case where encoding and packetization are individually performed, and the generation of useless traffic is suppressed. Furthermore, in the present embodiment, the data added to make the packet size uniform can be at most N-1 bytes, which is a value sufficiently smaller than the entire data size. Can be suppressed.

In the above description, the redundancy method used in the data encoding / dividing means 7 and the data decoding means 10 is as follows.
N = 2, M = 1 and the parity system using the exclusive OR is operated. However, even when N> 2, the parity system using the exclusive OR can be adopted by the same processing. The same effect as in the present embodiment can be obtained. When M> 1, for example, a Reed-Solomon method can be used as a method corresponding to general values of N and M, or the same effect can be obtained by adopting another method. When N = 1 and M = 1, redundancy by duplication can be adopted, and the same effect can be obtained.

Further, in the above, the data encoding / dividing means 7
Is an operation of first dividing the data 20 passed from the data accumulation / forced transmission instructing means 5 into N pieces of partial data 22 and then generating M pieces of parity data 23 by a redundancy process. First, the passed data 20 may be subjected to redundancy processing into data that can be divided into N + M packets 21, and then divided into N + M packets 21.
In general, the same effect can be obtained even if a similar method of generating N + M packets 21 is adopted so that the original data can be restored if any N packets 21 are obtained.

FIG. 6 is a flowchart showing the operation of another data storage / forced transmission instructing means 5b in the present embodiment. The operation of the data storage / forced transmission instructing means 5b for forcibly transmitting data according to the length of the buffer 6 in addition to the time T will be described with reference to FIG. Processing S20 to S1 of data accumulation / forced transmission instructing means 5 in the present embodiment
The loop of FIG. 6 operates similarly to FIG. Even if the time T has not elapsed (No in S14), if the size of the data 20 stored in the buffer 6 is equal to or larger than the predetermined value L (Yes in S21), the data 20 stored in the buffer 6
Is passed to the data encoding division unit 7 (S15), the buffer 6 is emptied (S16), and the process returns to the process of receiving new data 20 from the data input unit 4 (S11). On the other hand, before the time T elapses (No in S14) and the buffer 6
If the size of the data 20 stored in the data input unit 4 is smaller than the fixed value L (No in S21), when new data 20 is received from the data input unit 4 (Yes in S17), the new data 20 is added to the internal buffer 6. It is stored (S18).

That is, the size required for the buffer 6 is
The sum can be limited to the sum of the constant value L and the maximum size of the data 20 passed at one time from the data input means 4, and the size of the data 20 passed to the data encoding / dividing means 7 is also limited to the above-mentioned size at the maximum. Therefore, the efficiency of the processing is improved.
Further, when the frequency of data input to the data input means 4 is high, the data transmission processing is performed even if the fixed time T does not elapse.

If the unit for compiling data as one packet is short, the overhead load increases and a division loss occurs as in the second conventional example. If it is too long, it takes time to complete the transmission. On the other hand, data is not always of an appropriate length. In the present embodiment, the buffer length is determined as the optimum packet length, and data is collected in the unit. FIG. 7 is a flowchart showing the operation of still another data storage / forced transmission instructing means in the first embodiment. In the following description, operations other than the data accumulation / forced transmission instructing means 5 are the same as those shown in FIGS. 1, 2, 4, and 5 shown in the first embodiment, and a description thereof will be omitted.

In the flow of FIG. 7, the data input means 4
The buffer length of the data 20 received in step (1) is set to a predetermined optimal division length, and transmission is started when data is accumulated in the buffer. Specifically, in FIG. 7, the processing starts from the processing S30 of the data accumulation / forced transmission instructing means 5c.
(S11), and if the buffer 6 is empty (Yes in S31), the time when the data 20 is received is temporarily stored (S13). Then, the data 20 received up to the empty size of the buffer 6 is stored in the buffer 6 (S31). As a result, if the size of the data 20 stored in the buffer 6 exceeds the predetermined value L (Yes in S21), or if the size of the data 20 stored in the buffer 6 does not exceed the predetermined value L In both cases (No in S21), if the time T has elapsed from the stored time (Yes in S14), the data 20 accumulated in the buffer 6 is passed to the data encoding and dividing means 7 (S15), and the buffer 6 is emptied ( S16).

If there is a portion of the received data 20 which has not yet been stored in the buffer 6 (Yes in S33)
s) The part of the data 20 that has not yet been stored in the buffer 6 is the data 2 newly received by the data input means 4.
0 (S34) and the process is repeated again (S3
1). On the other hand, all received data 20 are stored in buffer 6
(No in S33), if there is new data 20 from the data input means 4 (Yes in S17), the data 20 is newly received from the data input means 4 (S1).
1) Continue the process. Then, all the received data 20 is stored in the buffer 6 (No in S33), and if there is no new data 20 from the data input means 4 (S1).
(No in 7), the process returns to the determination as to whether the time T has elapsed from the stored time (S14).

As described above, the size required for the buffer 6 can be limited to the fixed value L, and the size of the data to be passed to the data encoding / dividing means 7 can be limited to the above-mentioned size at the maximum. Therefore, it is possible to appropriately limit the size of the work area required for the processing of the data accumulation / forced transmission instructing means 5 and the data encoding / dividing means 7.
For this reason, the processing efficiency is improved, and when the frequency of data input to the data input unit 4 is high, the data transmission processing is performed even if the fixed time T does not elapse.

Embodiment 2 It is desirable that the length of the data transmission unit be variable depending on the congestion degree and quality of the transmission path. If the number of errors is large, the data length of the transmission unit may be shortened. If the number of errors is small, the data may be lengthened to avoid division loss.
At this time, a time limit is provided to shorten the completion time. FIG.
9 is an operation flowchart according to Embodiment 2 of the present invention. That is, in the data encoding and dividing means 7,
It shows a method of adjusting the values of N and M when encoding and dividing the data 20 passed from the data accumulation / forced transmission instructing means 5d into N + M packets 21. The system configuration of the present embodiment and the operation of the data encoding / dividing means 7 except for the method of adjusting the values of N and M are the same as those in FIGS. I do.

Starting from this processing S40, first, based on the maximum allowable delay time Th and its achievement probability Ph, the data storage / forcible transmission instructing means 5 stores the data 20 in the buffer 6.
In consideration of the maximum time T during which the packet stays, the distribution of the packet arrival rate with respect to the communication delay
The arrival rate P of packets received within Th-T
d is obtained (S41). Next, from the size Ld of the data 20 passed from the data accumulation / forced transmission instructing means 5 and the maximum data size Lp that can be included in the packet 21 transmitted to the communication path 3 by the packet transmitting means 8, Ld /
The minimum division number Nmin satisfying Lp ≦ Nmin is obtained (S4
2). Then, N ≧ Nmin, M ≧ 1, Ph ≦ Σ ^M _{i = 0} ( _{N + M} C
_i · (1-Pd) ⁱ · Pd ^{N + Mi} ) and N / (N
+ M) is set as large as possible, and N and M are set as N + M as small as possible (S43). Here, Σ ^M _{i = 0} ( _{N + M} C _i · (1−Pd) ⁱ · P
d ^{N + Mi} ) is an expression indicating the probability that N + M packets 21 are transmitted and N or more packets 21 are received within the time Th-T. When the influence of the packet size is larger than the number of packets as traffic to the communication path 3, the evaluation of N / (N + M) is more important than the evaluation of N + M. When the influence of the number of packets is greater, N + M than the evaluation of N / (N + M)
The values of N and M are determined with emphasis on the evaluation of. Thereafter, based on the values of N and M obtained by the above processing, the data encoding / dividing means 7 encodes and divides the data 20 passed from the data accumulation / forced transmission instructing means 5 to generate a packet 21.

Since the system according to the present embodiment operates as described above, the distribution of the packet arrival rate with respect to the communication delay time on the communication path 3 and the size of the data 20 passed from the data storage / forced transmission instructing means 5 And the size of the largest data that can be included in the packet to be sent,
The present invention satisfies the allowable maximum delay time Th and its achievement probability Ph, suppresses the generation of useless traffic, and improves the quality and efficiency of data communication.

The distribution of the packet arrival rate with respect to the communication delay time on the communication path 3 used above can be obtained by separately measuring, for example, the communication characteristic of the communication path 3 separately from the processing of the above embodiment. Alternatively, data transmission means 1
From the packet transmission means 8 which is a component of the data reception means 9 which is a component of the data reception means 2
1 is sent, the packet receiving means 9 sends each packet 2
1 is measured, the statistical information obtained from the measurement result is transmitted from the data receiving means 2 to the data transmitting means 1, and the distribution of the packet arrival rate with respect to the communication delay time in the communication path 3 used as described above. And the above processing may be performed.

FIG. 9 is a flowchart showing the operation of adjusting the upper limit time T for storing data 20 in another data storage / forced transmission instructing means 5e according to the second embodiment. The data accumulation / forced transmission instructing means 5e performs the following processing based on the allowable maximum delay time Th and its achievement probability Ph. However, in the following, in the data encoding / dividing means 7, the values of N and M when encoding and dividing the data 20 passed from the data accumulation / forced transmission instructing means 5 into N + M packets 21 are predetermined. This will be explained as follows.

Starting from this processing S50, first, Ph = Σ^M
_{i = 0}(_{N + M}C_i・ (1-Pd)ⁱ・ Pd^{N + Mi}Pd that satisfies
Is obtained (S51). Where Σ^M _{i = 0}(_{N + M}C_i・ (1-
Pd)ⁱ・ Pd^{N + Mi}) Indicates the arrival rate of one packet as Pd
When N + M packets are transmitted, N or more packets are transmitted.
The probability that the ket will arrive. Next, communication delay on the communication path 3
From the distribution D (τ) of the packet arrival rate to the delay time, D
Find t satisfying (Th−t) ≧ Pd, and store the value in data storage.
Maximum for storing data 20 by product / forced transmission instructing means 5e
Time T is set (S52).

As described above, based on the distribution of the packet arrival rate with respect to the communication delay time in the communication path 3, the allowable maximum delay time Th and the achievement probability Ph are satisfied, and the data storage / forced transmission instructing means 5 By accumulating data for as long as possible, data encoding efficiency is improved.

In the above description, when the data encoding / dividing means 7 encodes and splits the data 20 passed from the data accumulation / forced transmission instructing means 5 into N + M packets 21, the values of N and M are Although it has been described in advance that this is done in advance, it may be dynamically adjusted in accordance with the time T, and the effect of the present invention can be further enhanced.

Further, as the distribution of the packet arrival rate with respect to the communication delay time in the communication path 3 used above, substantially the same effect can be obtained by using the reception information of the receiving circuit provided on the transmission side instead of the arrival information from the other party. Is obtained. FIG. 10 is a diagram showing the configuration, in which the first data transmitting means A 1b, the first data receiving means A 2b paired with the first data transmitting means A 1b, and the first data transmitting means A 1b Second data receiving means B for transmitting and receiving data via communication path 3
2c and the data receiving means B 2c.
And a second data transmitting means B 1c for transmitting and receiving data via the communication path 3 and the data receiving means A 2b. Data input means A 4 / B 4c, data accumulation / forced transmission instructing means A 5 / B 5c, and buffer A 6 / B, which are components of data transmitting means A 1b and B 1c.
6c, data encoding / dividing means A 7 / B 7c, packet transmitting means A 8 / B 8c, and data receiving means A
Packet receiving means A, which is a component of 2b / B 2c
9 / B 9c, data decoding means A 10 / B 10
c, and the data output means A 11 / B 11c are the same as those in FIG. 1, and a detailed description thereof will be omitted.

In this system, the packet receiving means A 9 in the data receiving means A 2 b is replaced by the data transmitting means B 1
c sent from the packet transmitting means B 8c
The communication delay time of each packet is measured for each packet. Then, the statistical information obtained from the measurement result is presented to the data coding division unit A7 of the data transmission unit A1b. The data encoding / dividing unit A 7 of the data transmitting unit A 1 regards the statistical information presented from the packet receiving unit A 9 as a distribution of a packet arrival rate with respect to a communication delay time of the transmitting side on the communication path 3 and performs data transmission. It adjusts the values of N and M when encoding and dividing the data 20 passed from the accumulation / forced transmission instructing means A5 into N + M packets 21.

As described above, the data transmitting means A 1 does not receive the statistical information on the communication delay time of each packet from the data receiving means B 2 of the communication destination, but receives the communication delay time of the communication path 3 from its own receiving side. Can be obtained, and appropriate N and M can be obtained in accordance with the communication quality of the communication path 3.

The concept of the present embodiment can be applied to the case where the data transmitting means 1 communicates with a plurality of data receiving means 2. That is, FIG. 11 is a configuration diagram when the data transmitting unit 1d communicates with a plurality of data receiving units. The data transmitting unit 1d transmits and receives data to and from the data receiving unit A 2e via the communication path A 3a. , And simultaneously transmits and receives data to and from the data receiving means B 2f via the communication path B 3b. It should be noted that the data input means 4, the data storage / forced transmission instructing means 5f, the buffer 6, the data encoding and dividing means 7, the packet transmitting means 8, and the data receiving means A2 / eB 2f which are components of the data transmitting means 1d.
Packet receiving means A 9 / B 9,
The data decoding means A 10 / B 10 and the data output means A11 / B 11 are the same as those in FIG.

In this configuration, the data encoding / dividing means 7b is connected to the data receiving means A 2e via the communication path A3.
When transmitting and receiving data, the maximum data size Lpa that can be included in the packet 21 transmitted by the packet transmitting means 8 to the communication path A3 and the packet arrival rate with respect to the communication delay time in the communication path A3. Using the arrival rate Pda of the packets received within the communication delay time Th-T obtained from the distribution, the data 20 specified by the data storage / forced transmission instructing means 5f is converted into N + M packets 2
When the values of N and M are adjusted when encoding and dividing into 1, and when data is transmitted / received to / from the data receiving means B2 via the communication path B3, the packet transmission means 8 transmits the data via the communication path B3.
Packet received within the communication delay time Th-T determined from the maximum data size Lpb that can be included in the packet 21 to be transmitted and the distribution of the packet arrival rate with respect to the communication delay time on the communication path B3. The N and M values are adjusted when the data 20 specified by the data storage / forced transmission instructing unit 5f is encoded and divided into N + M packets 21 using the arrival rate Pdb.

In this system, the communication path A 3a /
Since an operation is performed to obtain an appropriate N and M according to the communication characteristics of each of the B 3b and perform communication, the effect of improving the quality and efficiency of data communication can be enhanced.

Embodiment 3 FIG. A description will be given of a case where transmission of short data is waited until a predetermined buffer length is reached to avoid division loss. FIG. 12 is a flowchart showing the operation of the data storage / forced transmission instruction means 5 in the embodiment of the present invention. The system configuration in the present embodiment is the same as FIG.

In this embodiment, the data input means 4 receives the data 20 as a packet and stores it in the buffer 6 as it is. Data storage / forced transmission instruction means 5
g starts with the process S60, first receives the data 20 (S11), and then if the buffer 6 is empty (S11).
(Yes of 31) The received time is stored (S13).
Further, the type of the received data 20 is checked (S6).
1). If the received data 20 is a packet that cannot be divided and combined (No in S61), the buffer 6
Is checked, and if all the data 20 can be stored in the buffer 6 (Yes in S62), the received data 20 is stored in the buffer 6 (S63). On the other hand, if the received data 20 is a packet that can be divided and combined (Yes in S61), the packet stored in the buffer 6 is checked, and if a combineable packet is in the buffer 6 (Yes in S64), The received data 20 is stored in the buffer 6 while being combined with the same packet up to the empty size of the buffer 6 (S65).
No. 61) The data 20 is stored in the buffer 6 as a single packet up to the empty size of the buffer 6 (S3).
2).

Here, a packet that can be divided and combined is a packet that holds data in the form of a stream described below that does not need to store a data boundary and that holds continuous data belonging to the same data stream. ,
Non-combinable packets are packets that hold data in stream format but belong to different data streams, or hold data that is not continuous, or datagrams that need to maintain data boundaries. This is a packet holding format data. Regarding the determination of whether or not to divide a packet and the division and combination of packets,
This will be described with reference to FIGS.

FIG. 13 is an explanatory diagram showing an example of the internal structure of a packet and the operation of dividing and combining the packet. In this figure, description will be made using an IP (Internet Protocol) packet defined by RFC791 and generally used for communication on the Internet. FIG. 13A is an explanatory diagram showing the structure of an IP packet.
Is composed of an IP header 26 and IP data 27. I
The P header 26 is an area for holding attribute information of the IP data 27, and includes version information of the IP protocol used in the IP packet 25, the length of the IP header 26,
A flag indicating information such as the overall length of the P packet 25, a packet ID for identifying the packet, and whether or not the packet can be divided;
A division offset value indicating a relative position with respect to a packet before division when divided, protocol information on data carried as IP data 27, IP packet 2
Source IP address indicating the source of No. 5, IP packet 2
5 destination IP addresses indicating the destinations. The flag of the IP header 26 includes the IP packet 25
Whether or not to permit division of the IP packet 25
However, if the packet has already been divided, information such as whether there is subsequent data is included. The protocol of the IP header 26 may be a TCP (Transmission Control Protocol) packet or a UDP (User Datagram Protocol) as information indicating the type of protocol applied to the data carried as the IP data 27.
col) packet, or information indicating another data stream such as a packet of another protocol. The IP data 27 is the data to be transmitted by the IP packet 25 itself.

The method of dividing and combining IP packets is R
Specified by FC791. The outline is described below. In the IP packet 25, the version number, the packet ID, the flag, the transmission source I included in the IP header 26
Whether or not the division / combination is possible can be determined based on the P address and the destination IP address. If the flag of the IP header 26 indicates that the IP packet 25 can be divided, the IP packet 25 is permitted to be divided. FIG.
FIG. 4B is an explanatory diagram showing how to divide the IP packet 25. In the figure, the IP packet X 25 is divided into two IP packets A 25 and B 25.

In the figure, the data storage / forcible transmission instructing means 5g of the data transmitting means 1 instructs and the data encoding / dividing means 7 performs the division procedure roughly as follows.
First, IP data X 2 included in IP packet X 25
7 at an appropriate position, and the first half is divided into IP data A 2
7. The latter half is referred to as IP data B27. Next, based on the IP header X26 included in the IP packet X25, I
A P header A 26 and an IP header B 26 are generated.
At this time, the IP header A 26 and the IP header B 26
Most of the information is the same as the IP header X26 except for the following part. That is, in the IP header A 26, the flag is set so that there is always subsequent data after the division, and the entire length of the IP packet is set to be reduced by the amount of the divided IP data B27. On the other hand, in the IP header B 26, the presence / absence of the subsequent data indicated by the flag is set to the same value as the presence / absence of the subsequent data indicated by the flag of the IP header X 26, and the total length of the IP packet is set by the divided IP data A27. Is set to decrease by
Further, the value of the division offset is also divided into the IP data A 2
It increases by seven. Finally, the generated IP header A
26 and IP data A 27 are combined into an IP packet A 25, and similarly, IP headers B 26 and I
By combining the P data B 27, the IP packet B 2
5 is assumed.

Next, how to combine the IP packets 25 will be described with reference to FIG. FIG. 13C is an explanatory diagram showing how the IP packets 25 are combined. In the figure, two IP packets A 25 and two IP packets B 2
5 are combined to generate an IP packet Y25. The condition under which two IP packets A 25 and IP packet B 25 can be combined is simply that both are one IP packet 2
5 is a continuous packet divided from 5,
Specifically, when the following condition is satisfied, two IP packets A
25 and the IP packet B 25 can be combined. First, it is necessary that the version numbers, the source IP address, the destination IP address, the protocol, and the packet ID of the IP header A 26 and the IP header B 26 match, and that the flags both indicate that the division is possible. This condition indicates that both packets are generated by dividing one original packet. Next, the values of the division offsets are compared, and the flag of the younger one (here, IP header A 26) indicates that there is subsequent data after division, and the value of the division offset and the length of IP data A Must match the division offset of the other IP header B26. This condition is a packet in which the IP packet A 25 and the IP packet B 25 hold continuous data among a plurality of packets generated by dividing the original packet, and the IP packet A 25 is in the first half. Indicates that data is retained. When these conditions are satisfied, the two IP packets A 25 and B 25 can be combined.

In the figure, the procedure of combining performed by the data output means 11 after data is received by the data transmission means 2 is roughly as follows. In this example, it is known that the IP packet A 25 holds the first half of the data by the determination of whether or not the two IP packets A 25 and B 25 can be combined, so the IP packet A 25 is included in the IP packet A 25 first. IP data A 27 as the first half, I
The IP data Y 27 is generated by connecting the IP data B 27 included in the P packet B 25 with the latter as the latter half. Next, the IP header A included in the IP packet A 25
The IP header Y 26 is generated from the IP header B 26 included in the IP packet 26 and the IP header B 26 included in the IP packet B 25. At this time, most of the information of the IP header Y 26 is the same as the IP header A 26 and the IP header B 26, but the following part is different. That is, the IP packet length of the IP header Y 26 is set based on the size of the new IP data Y 27, and the presence or absence of the subsequent data in the flag of the IP header Y 26 is the same value as the presence or absence of the subsequent data in the flag of the IP header B 26. Is set, and the IP header Y
The same value as the division offset of the IP header A 26 is set as the division offset of the IP header A 26. Finally, the generated IP header Y 26 and IP data Y 27 are combined into an IP packet Y 25.

Based on the above, S32, S65, S in FIG.
How to store the data 20 in the buffer 6 in 63 will be described with reference to FIG. FIG. 14 is an explanatory diagram for explaining how to store the data 20 in the buffer 6 in the present embodiment. In the figure, the packet passed from the data input means 4 as the data 20 is an IP packet X 25, and
It is assumed that at least several IP packets 25 are stored.
Note that the size (IP packet length) of the entire IP packet 25 can be determined by examining the IP header 26 of the IP packet 25. For this reason, when storing a plurality of IP packets 25 as the data 20 of the buffer 6, each IP packet 2 can be stored even if a break between packets is not explicitly held.
5 in the buffer 6 based on the IP packet length of
, Each IP packet stored in the buffer 6 can be separated and taken out.

When several IP packets 25 are already stored in the buffer 6, there are four methods for storing the data 20 given to the buffer 6, as shown in FIGS. Hereinafter, as the data 20 in the buffer 6, the IP packet A 25 and the IP packet B 25
The following description assumes that two IP packets 25 are stored, but this can be handled in the same way even when an arbitrary number of IP packets 25 are stored.

FIG. 14A shows the state of the data 20 included in the data 20.
This is an operation when the P packet X 25 cannot be combined with any of the IP packets A 25 and B 25 already stored in the buffer 6 and the size thereof is smaller than the free area of the buffer 6. This operation corresponds to the process of S63 in FIG.
That is, when the data 20 is a packet that cannot be divided and combined and all the data 20 can be stored in the buffer 6, or in the process of S32, that is, when the data is a packet that can be divided and combined but there is no packet that can be combined in the buffer 6. Corresponds to the case where all the data 20 can be stored. In this operation, following the IP packets A and B25 already stored as the data 20 in the buffer 6,
The entire IP packet X 25 included in the passed data 20 is stored in the buffer 6.

FIG. 14 (b)
The P packet X 25 cannot be combined with any of the IP packets A 25 and B 25 already stored in the buffer 6, and the size thereof is larger than the free space of the buffer,
This is an operation when the IP packet X25 can be further divided. This operation corresponds to the process of S32 in FIG. 12, that is, when data is a packet that can be divided and combined but there is no packet that can be combined in the buffer 6, and when a part of the data 20 can be stored. In this operation, first, the IP packet X25 included in the passed data 20 is divided into an IP packet X1 25 and another IP packet X2 25 having a size that can be stored in the buffer 6, and then the data 20 in the buffer 6 is divided. Newly generated IP following IP packets A25 and B25 already stored as
The packet X125 is stored in the buffer 6.

FIG. 14 (c) shows the data included in the data 20.
P packet X 25 can be combined with any of IP packets A 25 and B 25 already stored in buffer 6,
In addition, this is an operation in a case where the size after the combination is completely within the buffer 6. This operation is performed in the process of S65 in FIG. 12, that is, when data is a packet that can be divided and combined and there is a packet that can be
Is applicable when all can be saved. Here, it is assumed that IP packet X 25 can be combined with IP packet B 25. In this case, in this operation, the IP packet A 25 already stored as the data 20 in the buffer 6,
Of the B 25, the IP packet B 25 and the IP packet X 25 included in the passed data 20 are combined to generate a new IP packet B + X 25, and the generated IP packet B + X 25 is replaced with the IP packet B 25 Save to buffer.

FIG. 14 (d) shows the I
P packet X 25 can be combined with any of IP packets A 25 and B 25 already stored in buffer 6,
This is an operation when the size after the combination does not fit in the buffer 6. This operation corresponds to the process of S65 in FIG. 12, that is, when data is a packet that can be divided and combined and there is a packet that can be combined in the buffer 6, and when a part of the data 20 can be stored. Here I
It is assumed that the P packet X 25 can be combined with the IP packet B 25. In this case, in this operation, first, the IP packet X 25 included in the passed data 20 is stored in the buffer 6.
Is divided into an IP packet X125 and another IP packet X2 25 having a size that can be stored in the buffer 20. Then, among the IP packets A 25 and B 25 already stored as the data 20 in the buffer 6, the IP packet B 25 The newly generated IP packet X1 25 is combined to generate a new IP packet B + X1 25, and the generated IP
The packet B + X1 25 is stored in a buffer instead of the IP packet B25.

It should be noted that FIG. 14C and FIG.
In the description of the operation shown in (d), the IP packet X 25
Can be combined with the IP packet B 25, but this can be handled in the same manner even with the IP packet A 25. Further, the data 20 in the buffer 6 includes a plurality of IPs.
When the packet 25 is stored, the given IP packet X25 or the IP packet X1 25 generated by dividing the IP packet X25 and one of the IP packets 25 stored in the data 20 in the buffer 6 IP
After combining the packet 25, the newly generated combined IP packet 25 and other IPs remaining in the buffer 6
The packet 25 can be combined again, and even if this processing is repeated as much as possible, it can be handled in the same manner as described above.

The data storage / compulsory transmission instructing means 5 is the
As the process of S32, S65 or S63 of 2, after storing the data 20 in the buffer 6 as described above,
The process continues as follows. That is, if the size of the data 20 stored in the buffer 6 is equal to or larger than the fixed value L (Yes in S21), or if the size of the data 20 stored in the buffer 6 is smaller than the fixed value L ( S
21) If the time T has elapsed from the storage time (Yes in S14), the data 20 stored in the buffer 6
Is passed to the data encoding division means 7 (S15), and the buffer 6 is emptied (S16). If the received data 20 is an indivisible packet (No in S61) and the entire data 20 cannot be stored in the buffer 6 (No in S62), the entire data 20 can be stored in the buffer 6. For this purpose, the data 20 once stored in the buffer 6 is passed to the data encoding / dividing means 7 (S15).
The buffer 6 is emptied (S16).

After that, if there is a portion of the received data 20 that has not been stored in the buffer 6 even if the received data 20 is an indivisible packet and could not be stored in the buffer 6 (Yes in S33). ), The part of the data 20 not stored in the buffer 6 is regarded as the data 20 newly received by the data input means 4 (S
34), the process is repeated again (S31). On the other hand, all the received data 20 is stored in the buffer 6 (S
No. 33), new data 20 from the data input means 4
If there is (Yes in S17), a new data input means 4
(S11), and the process is continued.
Then, all the received data 20 is stored in the buffer 6 (No in S33), and if there is no new data 20 from the data input means 4 (No in S17), it is determined whether the time T has elapsed from the stored time. Return to judgment (S1
4).

Following the above-described processing in which the data storage / forced transmission instructing means 5g instructs the buffer 6, the data encoding / dividing means 7 converts the data specified by the data storage / forced transmission instructing means 5g into N + M packets 21. The data is encoded and divided, and transmitted to the packet receiving unit 9 of the data receiving unit 2 using the packet transmitting unit 8. The packet 21 received by the packet receiving means 9 is decoded by the data decoding means 10 and passed to the data output means 11. In the above description, the details of the operations of the data encoding / dividing unit 7, the packet output unit, the packet receiving unit 9, and the data decoding unit 10 are the same as those in the first embodiment, and therefore, the description of the detailed operations is omitted. The data output unit 11 checks the boundary of the packet from the data 20 passed from the data decoding unit 10, divides the packet into packets, and outputs the data 20.

In the present embodiment, since the system operates as described above, data belonging to different data streams can be collectively encoded. The processing efficiency can be improved because the size can be limited. When data belonging to different data streams is input to the data input means 4,
The frequency of data input is higher and the amount of input is larger than when a single data stream is handled. Therefore, as compared with the case where only data belonging to a single data stream is handled, the chances of performing data transmission processing even if the fixed time T does not elapse are increased. In this way, the elapsed time until transmission is reduced, and data communication efficiency is improved.

Also, two or more IP packets 25
Is larger than the sum of the sizes of the IP packets 25 before the combination by the IP header 26 × n.
(N: the number of combined IP packets minus 1). Accordingly, FIG.
The operation shown in (c) and FIG. 14 (d) is different from the operation shown in FIG. 14 (a) and FIG.
7 can be stored in the buffer 6 substantially more. That is, when the data accumulation / forced transmission instructing means 5 stores the IP packet 25 received as the data 20 from the data input means 4 in the buffer 6,
If possible, the operation of storing the packets in the buffer 6 while combining them is more than once from the data accumulation / forced transmission instructing means 5 to the data encoding / dividing means 7 as compared with the case where such processing is not performed. IP data 27 can be passed.

Although an example of packet division / combination in the IP (Internet Protocol) has been described above, it is also possible to perform packet division / combination based on a higher-level protocol. For example, TCP defined in RFC793
(Transmission Control Protocol) will be described with reference to FIG.

FIG. 15 is an explanatory diagram showing the internal structure of a TCP packet. TCP packet is generally IP packet 2
5 and its IP data 27 is TCP
The header 28 and the TCP data 29 are held. In this figure, the details of the IP header are the same as in FIG. However, the protocol of the IP header 26 is TCP
Fixed to The TCP header 28 is an area for storing attribute information of the TCP data 29, and includes a source port number, a destination port number, a sequence number,
CK number, TCP header length, etc.

[0073] TCP is a protocol for handling a data stream.
The CP data 29 includes the transmission source I included in the IP header 26.
From the data transmission end determined by the P address and the source port number included in the TCP header 28 to the data reception end determined by the destination IP address included in the IP header 26 and the destination port number included in the TCP header 28 continuously. It can be considered as part of a stream of data sent.
Which part in the flow of the TCP data 29 corresponds to the data can be determined by the sequence number in the TCP header 28 and the size of the TCP data. Therefore, the transmission source I held in the IP header 26
P address, destination IP address, and TCP header 28
Can be regarded as belonging to the same data stream if the source port number and the destination port number held by the two TCP packets belong to the same data stream, and whether the two TCP packets hold continuous data in the data stream Can be determined from the sequence number held by each TCP header 28 and the size of the TCP data. Therefore, for the TCP packet, it is possible to divide and combine the packets by performing a process similar to the operation of dividing and combining the IP packets 25 shown in FIGS. 13B and 13C. is there.

Due to the characteristics of the protocol, the IP packet 25 is generally handled as a “data block” having a finite size. For this reason, packets can be combined or re-divided as if they were data streams inside a "data chunk", but cannot be combined or re-divided beyond that range. This means that the packet ID can be combined only between the same IP packets. On the other hand, a TCP packet is a part of the “data flow” having an infinite size, and includes a data transmitting end (a source IP address and a source port number) and a data receiving end (a destination IP address and a destination). Port numbers) are the same,
At the IP level, which is a lower level protocol of TCP, different “blocks of data”, that is, data divided into IP packets corresponding to different packet IDs can be combined or re-divided. For this reason, packet division and combination at the TCP level rather than at the IP level can reconstruct packets more flexibly, so that the efficiency of data communication can be further improved.

Embodiment 4 FIG. 16 is a system configuration diagram according to the fourth embodiment. In FIG. 16, a data transmitting means 1g transmits and receives data to and from a data receiving means A 2g via a communication path A3, and at the same time, a communication path B
3 to send and receive data to and from the data receiving means B 2h. The data transmission sources A1 30a / A2 30 transmit data to the data reception sources A1 31a / A2 31b via the data transmission means 1g and the data reception means A 2g, respectively.
30b designates a data stream A1 3 to the data input means 4.
2a / A2 32b, and the data 20 is passed from the data output means A 11a to the data stream A1 32a / A.
Data 20 is received in the form of 232b. Here, the data stream is, for example, a virtual communication path 3 for transmitting data between the data transmission source A1 and the data reception source A1. Similarly, data transmission source B1 30c / B2 30d
Are data transmitting means 1g and data receiving means B, respectively.
Data receiving source B1 31c / B2 31 via 2h
d to transmit data to the data transmission source B1 30c
/ B2 30d passes the data 20 to the data input means 4 in the form of a data stream B1 32c / B2 32d, and the data receiving source B1 31c / B2 31d sends the data stream B1 3 from the data output means B 11b.
The data 20 is received in the form of 2c / B2 32d.
A buffer A 6q for the data receiving means A 2g and a buffer B 6b for the data receiving means B 2h exist in the data transmitting means 1g, and these are stored by the data storing / forcible transmission instructing means 5, and The data is passed to the data encoding / dividing means 7b. The other packet transmitting means 8 which is a component of the data transmitting means 1g, and the packet receiving means A 9a / B 9b and the data decoding means A 10a / which are other components of the data receiving means A 2a / B 2b. B 10b is the same as FIG.

Next, the operation of the data storage / forced transmission instructing means 5h in the present embodiment will be described with reference to FIG. Starting from the main processing S70, the data 20 is received (S11), and the destination of the data 20 is checked (S7).
1). If the destination of the data 20 is the data receiving source A1 31
a or A2 31b, the received data 20 is stored in the buffer A while leaving the information indicating the destination of the data 20.
6a (S72). Next, if the storage time of the oldest data 20 stored in the buffer A 6 exceeds the time T (Yes in S73), the buffer A 6a
The data 20 stored in the buffer A 6 is passed to the data encoding / dividing means 7 as being addressed to the data receiving means A 2g.
a is emptied (S74). Then, the data input processing (S1
Return to 1). On the other hand, if the destination of the data 20 is the data receiving source B1 31c or B2 31d, the data 20
Is stored in the buffer B 6b while keeping the information indicating the destination of the data (S75). Next, if the storage time of the oldest data 20 in the buffer B 6b exceeds the time T (S76).
Yes), the data 20 is addressed to the data receiving means B 2h and passed to the data encoding / dividing means 7, and then the buffer B
6b is emptied (S77). Then, the data input processing (S
Return to 11).

The data encoding / dividing means 7 converts the data 20 passed from the data accumulation / forced transmission instructing means 5h into N +
It is encoded and divided into M packets 21 and the data receiving means A 2 using the packet transmitting means 8 according to the destination.
g or data transmission means B 2h. The packet 21 received by the packet receiving means A9 of the data receiving means A 2g is decoded by the data decoding means A10 and passed to the data output means A11. Similarly, the received packet 21 to the data receiving means B2 is decoded by the data decoding means B10, and the data is output.

The data output means A 11 outputs the data 20
Of the data 2 addressed to the data receiving source A131a
0 is the data reception source A1 31a and the data reception source A
The data 20 addressed to 231b is passed to the data receiving source A2 31b. Similarly, the data output means B 11 outputs the data 20
The data 20 addressed to the data receiving source B1 31c is stored in the data receiving source B1 31c, and the data 20 addressed to the data receiving source B2 31d is output in the data receiving source B2 31
pass to d.

The method of checking the destination of the data 20 in the data storage / forced transmission instructing means 5h, the data output means A11, and the data output means B11 in this embodiment will be described with reference to FIG. Hereinafter, the data stream A132a / A2 32b / B1 in the present embodiment
32c / B2 32d is TCP (Transmission Control
It is assumed that the data 20 received by the data storage / forced transmission instructing means 5h from the data input means 4 or the data 20 output by the data output means A11 and B11 is a TCP packet. .

FIG. 15 is an explanatory diagram showing the internal structure of a TCP packet. As described in the above embodiment, the TCP data 29 held by the TCP packet
From the transmission end of data determined by the source IP address included in the P header 26 and the source port number included in the TCP header 28, the destination is determined by the destination IP address included in the IP header 26 and the destination port number included in the TCP header 28. It can be considered as a part of a series of data flows sent continuously to the data receiving end. Here, the destination IP address included in the IP header 26 generally indicates a difference between computers that send TCP packets, and the destination port included in the TCP header 28 indicates a difference between communication targets operating in the same computer. For this reason, without impairing generality, in the present embodiment, the data receiving means A 2g using the destination IP address
/ B 2h, and using the combination of the destination IP address and the IP port address, the data receiving source A1 31a
/ A2 31b / B1 31c / B2 31d. Thereby, the destination of the data 20 can be determined based on the destination IP address and the destination port number included in the TCP packet.

Based on the above manner, the data storage / forced transmission instructing means 5h checks the destination IP address contained in the TCP packet received from the data input means 4 as the data 20, and the received data 20 is transmitted to the data receiving means A 2g. Via the data receiving source A1 31a
Alternatively, the data should be sent to the data receiving source A2 31b or the data receiving source B2h via the data receiving means B2h.
It is determined whether the data should be sent to 1 31c or the data receiving source B2 31d. As a result, the data storage / forced transmission instructing means 5 h transmits the data 2 received from the data input means 4.
If 0 is addressed to the data reception source A1 31a or A2 31b, a save instruction is issued to the buffer A 6a (S in FIG. 17).
72) On the other hand, if the data receiving sources B1 31c or B2
If it is addressed to 31d, it is stored in buffer B 6b (FIG. 1).
7 S75).

Similarly, the data output means A 11
Is the data 20 passed from the data decryption means A 10
By examining the destination port number included in the TCP packet received as the
It is determined whether it is addressed to 1a or A2 31b, and output to each. The operation of the data output unit B11 is the same, and the destination port number included in the TCP packet of the data 20 is checked to determine whether the destination port number is the data reception source B131c or the data reception source B2.
It knows whether it is to 31 and outputs it to each destination.

Since the system according to the present embodiment operates as described above, for example, the data transmission source A1 30a transmits to the data reception source A1 via the same communication path A3.
Data 20 handled on data stream A1 32a to 31a and data 20 handled on data stream A2 32b from data source A2 30b to data source A2 31b can be stored in the same buffer A6. Becomes As a result, regardless of data belonging to different data streams, the data encoding / dividing means 7 can collectively encode the data, thereby improving the efficiency of encoding and transmission.

FIG. 1 shows a specific image of the data transmitting means 1, the data receiving means 2, and the communication path 3 described above.
FIG. In the figure, two communication devices A 40a, B
40b, the Internet 4 using the communication I / FB connected to the packet transmitting means 8 and the packet receiving means 9;
2 using the communication I / F A connected to the data input means 4 and the data output means 11.
And two premises LANs using the Internet 42
It operates to perform a relay process of data communication between the 43. Either of the communication devices A and B operates as the data transmitting unit 1 and the other operates as the data receiving unit 2. Internet 42
Is the communication path 3.

The communication devices A 40a and B 40b are connected to the communication I / O
As described above, the communication path B3 connected to the FB has an effect of improving the quality and efficiency of the data communication. Two private LANs 4 via the Internet 42
Between 3, the communication delay is large, the communication time varies, and data loss is likely to occur. By applying to this,
Higher quality and more efficient transmission is possible.

The division / combination of data of the same data stream described in the third embodiment can be applied to applications. FIG. 19 is a diagram showing the system configuration.
And a first data communication unit 46 and a second data communication unit 47. The second data communication unit 47 has the data transmission unit 1 and the data reception unit 2 and is connected to the communication path 3.

The operation of each component is as follows. The application 45 exchanges data with the first data communication means 46 for data communication. The first data communication means 46 operates the second data communication means 47 as an alternative to the communication path 3 and transmits / receives packets to / from the second data communication means. The second data communication unit 47 regards the transmission and reception of the packet from the first data communication unit 46 as the transmission and reception of data, and transmits and receives the packet to and from the communication path 3 by the operation of the data transmission unit 1 and the data reception unit 2. . The internal operations of the data transmitting means 1 and the data receiving means 2 are as already described, and the detailed description is omitted. Thus, the second
The quality and efficiency of data communication on the communication path 3 is improved by the data communication means 47, and the general data communication means such as a TCP / IP protocol stack is used as the first data communication means 46. It is possible to perform data communication as if connected using a higher quality communication path 3 without altering the communication path.

[0088]

As described above, according to the present invention, the time during which data is stopped in the buffer is limited, and further, data division,
Since the encoding length is variably set by the channel information, the time until the transmission is completed can be shortened, and since the transmission is performed at a predetermined length, the overhead load is prevented from increasing, so that the data can be transmitted with high quality and high efficiency. effective.

[Brief description of the drawings]

FIG. 1 is a diagram showing a system configuration according to a first embodiment of the present invention.

FIG. 2 is a diagram illustrating an operation of the system according to the first embodiment.

FIG. 3 is an operation flowchart mainly showing data accumulation / forced transmission instruction means according to the first embodiment;

FIG. 4 is a diagram for explaining an operation of a data encoding and dividing unit according to the first embodiment.

FIG. 5 is a diagram illustrating an operation of a data decoding division unit according to the first embodiment.

FIG. 6 is an operation flowchart mainly showing another data accumulation / forced transmission instructing unit in the first embodiment.

FIG. 7 is a diagram illustrating still another data storage and operation according to the first embodiment.
It is an operation | movement flowchart mainly on a forced transmission instruction | indication means.

FIG. 8 is a diagram illustrating data storage and operation according to the second embodiment of the present invention.
It is an operation | movement flowchart mainly on a forced transmission instruction | indication means.

FIG. 9 is an operation flowchart mainly showing another data accumulation / forced transmission instructing unit in the second embodiment.

FIG. 10 is a diagram showing another system configuration according to the second embodiment.

FIG. 11 is a diagram showing still another system configuration according to the second embodiment.

FIG. 12 is an operation flowchart mainly showing data accumulation / forced transmission instruction means according to Embodiment 3 of the present invention.

FIG. 13 is a diagram illustrating an internal structure of a packet and packet division / combination according to the third embodiment.

FIG. 14 is a diagram illustrating a storage combination in which a buffer stores data according to the third embodiment.

FIG. 15 is a diagram showing an internal structure of a TCP packet.

FIG. 16 is a diagram showing a system configuration according to a fourth embodiment of the present invention.

FIG. 17 is an operation flowchart mainly showing data accumulation / forced transmission instruction means in the fourth embodiment.

FIG. 18 is a diagram showing an example of a specific connection form to which the present invention is applied.

FIG. 19 is a diagram showing an example in which the present invention is applied to an application.

FIG. 20 is a configuration diagram of a data transmission system as a first conventional example.

FIG. 21 is a configuration diagram of a data transmission system as a second conventional example.

[Explanation of symbols]

1, 1b, 1c, 1d, 1g Data transmission means, 2, 2
a, 2b, 2c, 2e, 2f, 2g, 2h Data receiving means (A), (B), 5, 5b, 5c, 5d5e, 5
f, 5g, 5h Data storage / forced transmission instruction means, 6,
6a, 6b Buffers (A), (B), 7, 7b Data coding division means (A), (B), 10 data decoding means (A), (B), S14 Elapsed time check step.

Claims (7)

[Claims] In a configuration in which data from an input means is redundantly encoded and transmitted in a packet, a buffer for storing the data in units of a predetermined length, and a first buffer for storing data from an empty state of the buffer. Data storage / forced transmission instructing means for instructing transmission after a predetermined time has elapsed from the time, and data encoding / dividing means for making the predetermined length data from the buffer redundant and encoding. A high-efficiency data transmission device characterized by the above-mentioned. 2. The data encoding / dividing means determines a data length division number and a redundancy based on a predetermined length of transmission data and a transmission probability obtained in an allowable transmission delay time after a predetermined time has elapsed. The high-efficiency data transmission device according to claim 1, wherein: 3. The high-efficiency data transmission according to claim 1, wherein said data accumulation / forced transmission instruction means dynamically changes a predetermined time according to a transmission probability obtained in an allowable transmission delay time. apparatus. 4. The high-efficiency data transmission device according to claim 2, further comprising packet receiving means, wherein delay time information is obtained, and a transmission probability is calculated. 5. An apparatus according to claim 1, wherein an identifier indicating whether or not division / combination is possible is added to the input data, and if division / combination is possible, the data is stored in a predetermined length unit of a buffer, and is redundantly encoded. The high-efficiency data transmission device according to claim 1. 6. An input data is added with an identifier for identifying a stream, the buffer is stored in units of the stream identifier, and when the data in the buffer reaches a predetermined length, the data is redundantly encoded and transmitted. The high-efficiency data transmission device according to claim 1, wherein: 7. A buffer for storing data from an input means in units of a predetermined length, and when a predetermined time elapses from an initial time when data is stored from an empty state of the buffer,
A data transmission device having data accumulation / forced transmission instructing means for instructing transmission to perform forced division encoding, and data encoding / dividing means for making the predetermined length data from the buffer redundant and encoded; And a data receiving device having data decoding means for immediately starting decoding when data having a predetermined redundancy is received upon receiving data from the data transmitting device.