JPS63306740A - Packet communication system - Google Patents

Abstract

PURPOSE:To attain the information transmission with high quality without deteriorating the transmission efficiency by providing a packet disassembly equipment so as to smooth the delay of each packet at the reception side and to detect a faulty packet. CONSTITUTION:A packet is inputted to a smoothing buffer 31 of a packet disassembly device 24 at the reception side from the sender side via a packet network 23 and the 1st and 2nd sequence number information 36 added to the packet is outputted to a CPU 32. When the packet transmission information 37 is generated in the 1st sequence number order in the CPU 32 and a faulty packet is detected depending on the difference from the 2nd adjacent sequence numbers to be arranged. That is, when the difference is a prescribed value, the presence of a faulty packet is detected and inputted again to the buffer 31 in addition to the information 37 and faulty information 38 is outputted to the input buffer 34. Then a coded signal 39 is outputted to the decoder 29 from the buffer 34. Thus, the information transmission with high quality is attained without deteriorating the transmission efficiency.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention relates to a packet7) communication method, and in particular uses a packet7) with a variable bit length so that the packets differ depending on the packet7) received within a switching network. Relating to a packet communication method in which the delay time is set to a constant value on the receiving side, and the receiving side is able to detect lost packets within the network or abnormal packets that have undergone a transfer delay of more than a certain value within the network. .

[Conventional technology]

Traditionally, for example, transmitting low pit rate voice using packets has not been widely used, despite the significant savings in transmission and switching costs compared to circuit switching. .

The reason for this is that transmitting encoded information in packets causes various problems as described below.

One of the problems is that transmitting packets through a packet switching system typically results in different delays for each packet. Delay variations are introduced at switching nodes within a packet-switched network. Unlike data, audio information must be decoded from a digital representation to an analog representation at the same rate as it was coded to achieve high quality audio reproduction. If a variable delay is introduced during the transmission of a voice packet, the above decoding cannot be performed without some compensation.

Conventionally, a method to solve this problem is to take advantage of the fact that conversational audio such as on a telephone has a sound part and a silent part, and generate packets only for the sound part of the sound.
Thereafter, no packets are transmitted during the period of silence, and after receiving the first packet of the sound part, the receiver delays that packet by a certain time interval (seconds), and after this initial delay, the packet is not transmitted after the first packet of the sound part. There is a method of reproducing the original analog audio by decoding the packets of the sound part at the same speed as the packet generation interval T (seconds).

FIG. 5 shows a time chart for this method. In FIG. 5, a is a voice signal, b is a coded signal obtained by encoding the voice signal by an encoder, C is a fixed-length transmission packet formed from the coded signal at a fixed time interval, and d represents a received packet received on the receiving side, and e represents a received bucket) d are arranged in a predetermined order and are reproduced packets that are sent for reproduction.

In the conventional example shown in FIG. 5, (td+D') time is defined as t4, the delay time for a packet to reach the receiver from the sending side, and D', the delay time for the received packet to be played back. Packet 2 that has been delayed by a certain value or more cannot be reproduced on the receiving side, and some kind of interpolation is required as necessary.

As explained above, in this conventional method, as shown in the coded signal diagram of FIG.
A b/sADPCM (adaptive differential PCM) method was used. In other words, even when the amount of significant information generated changes over time, it is encoded at a constant bit rate and transmitted in fixed length packets, which has the problem of large waste from the perspective of information transmission efficiency. Ta.

As a solution to this problem, on the transmitting side, a packet with a variable bit length is formed whose length is made variable according to the number of information bits generated within a fixed time interval T (seconds), and a specific modulo A packet communication method is being considered in which communication is performed using variable bit length packets, in which a sequence number consisting of a number is added and transmitted, and the receiving side arranges and reproduces the received packets according to the above sequence number (
For example, see Proceedings of the 1981 National Conference of the Institute of Electronics and Communication Engineers, Information Systems Division, 181 "Variable Qubit Coding Method Between Audio Blocks").

[Problem that the invention seeks to solve]

As explained above, in conventional packet communication systems, when significant information changes over time, it is possible to improve information transmission efficiency by using packets with variable bit lengths.

However, when using variable bit length packets, the sizes of the packets vary, so if a variable delay is introduced during the packet transmission period, the reception side , it becomes difficult to smooth the delay, and the occurrence of abnormal packets that are lost for some reason during the transmission, or that are delayed for more than a certain amount of time and cannot be sequentially reproduced increases. An important issue is accurately detecting the occurrence of abnormal packets and performing interpolation or retransmission.
Conventionally, this problem has not been sufficiently solved.

The purpose of the present invention is to solve the above problems.
An object of the present invention is to provide a packet communication system using variable bit length packets, which allows smoothing of delays and detection of abnormal packets on the receiving side.

[Failure to solve the problem]

The present invention has a block time of /f (seconds) (where f
is the sampling frequency, and k is the number of sampling points. )
Forms a variable bit length block in which the number of information bits in a packet is varied according to the number of information bits generated in the packet,
In the packet communication method, the blocks are collected at regular time intervals T (seconds) and required information bits are added to them to form packets with a fixed bit length, and these packets are used for communication. On the side, specify a first sequence number modulo m (m is a natural number) every /f (seconds) in the block time to the above packet, and specify to the above packet the first sequence number corresponding to a specific block of the blocks included in the packet. The packets are sent with the first sequence number modulo n (n is a natural number) specified in the order of the packets generated, and the receiving side sends the packets to the first sequence number that arrived. A certain amount of time (
Dm+R) (seconds) (where D is the allowable delay time per packet and R is the maximum delay time when forming a packet.) After that,
The present invention is characterized in that the packets are sequentially reproduced in accordance with the first sequence number, and abnormal packets are detected from the difference between the second sequence numbers of the adjacent packets arranged in the order.

Further, in the present invention, it is preferable to detect an abnormal packet when the difference in sequence numbers is 2.

[Effect]

On the receiving side, after a certain period of time (Dm+R) (seconds) after the arrival of the first packet or specific reference packet, the above packets are arranged according to the first sequence number, and the packets are arranged into blocks at every /f (seconds) of the block time. Disassemble and regenerate.

Here, since the fixed time (Dm+R) is the maximum delay time allowed for each packet, each packet, excluding abnormal packets, always arrives at the receiving side in sequence, and delay smoothing is performed correctly. Note that in the present invention, packets to be transmitted and received have a fixed bit length, so the variation in delay time from packet to packet is originally smaller compared to packets with variable bit length.

On the other hand, regarding the second sequence numbers of the arranged packets, the difference Δ between the sequence numbers of two adjacent packets is determined. However, the difference between number n and number 1 is 1. In this way, the difference Δ is always Δ=1 if no abnormal packet occurs, and Δ≠1 when an abnormal packet occurs, and the occurrence of the abnormal packet 7) can be detected.

In this case, Δ=2 occurs when the previous packet is lost and becomes an abnormal packet, and if necessary, interpolation can be performed using reproduced signals from previous and subsequent packets, or retransmission can be performed.

Furthermore, the case of Δ<l or Δ>2 may also occur, but the probability thereof is usually very small.

〔Example〕

Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 is a time chart showing one embodiment of the present invention, and FIG.
The figure is a diagram showing the delay characteristics of the packet.

First, FIG. 2 will be explained. In Figure 2, the horizontal axis represents time t (seconds), and the vertical axis represents packet delay probability.
3 shows a packet end-to-end delay probability distribution function in a packet communication system to which the present invention is applied. In Fig. 2, td is a fixed packet common delay time common to all received packages, such as transmission path delay in each packet network and node processing delay, and D is a fixed packet common delay time that is unique to each network. Delay time for each packet (Dm is the allowable delay time for each packet, and the probability at that time is Po)
It is. Further, R is the maximum delay time during block formation.

Here, on the receiving side, when the delay time for packets with a proportion of p0 (po≦1) to subside is (td+D,) (seconds), on the transmitting side, the delay time after the first packet or a specific reference packet is (seconds), the receiving side receives P with a delay time of (td+D,)(seconds).
. By adding a delay time of R (seconds), almost 100% of the packets arrive.

That is, on the receiving side, after the first packet or a certain reference packet arrives, a delay of (Dm+R) time is added, and the packets are subsequently broken down into blocks and reproduced at a timing of <k/f (seconds). By doing this, delay smoothing is performed and the digital signal is regenerated into an analog signal at the same speed as at the time of coding.

In Fig. 1, b' is a block that uses a variable length encoder to convert the information of the sampling points of the analog signal into the amount of significant information of the information source using, for example, a differential encoding method. It is configured by performing variable length encoding accordingly. And for each professional 7th player, /f during block time.
(seconds), a specified first sequence number of modulo 16 (n=16>) is attached.

C is a transmission packet, in which the above blocks are collected so that the total number of bits is less than a certain value, and necessary additional bits are added to make the bits of a certain length to form a packet of packet length T. .

Note that the maximum delay time allowed for packetization is R (seconds). That is, if the sum of the number of information bits included in R/ (k/f) blocks does not reach a certain value,
Additional bits (waste bits) (shaded areas in FIG. 2) are added, and one packet is composed of R/(k/f) blocks. Then, each packet is given a second sequence number of 1, 2, modulo 4 (m=4>) in the order of formation.
3.4.1.2, and add the first sequence number of the specific (in this case the first) block included in the packet as 1.6.10.12.3.12
Add like this.

d is the received packet, and each packet is (t, +D
) (seconds) of delay and are shown in the order of arrival. In this example, there are no abnormal packets, and the packets arrive normally and sequentially.

e' is the playback block sent out for playback, starting from the arrival of the first packet (or a particular reference packet) with an additional delay of (Dm+R) (seconds), and each block is sent out for playback according to the first sequence number. Blocks are arranged at block times of /f (seconds). In other words, packets arriving at the receiving side are buffered (Dm
+R) (seconds) later, it is divided into blocks every (k/f) seconds and played back, so excluding the influence of delay time for each packet, the digital signal is correctly played back into an analog signal at the same speed as when coding. .

On the other hand, when these reproduced packets are arranged, the difference Δ between the second sequence numbers of two adjacent packets is determined.

Note that this difference Δ is the n number of modulo n (4 in this example)
The difference between this and the next number 1 is 1. The second sequence numbers are in the order of 1.2.3.4.1-, and the values of the differences Δ determined as described above are all 1 as shown in FIG. That is, it can be seen that there are no abnormal packets. Note that when the difference Δ in sequence numbers is Δ=2, the immediately preceding packet is detected as an abnormal packet.

Therefore, according to this embodiment, a variable bit rate encoding method is essentially used to encode the information source, communication is performed using variable length packets without reducing information transmission efficiency, and delay smoothing is applied on the receiving side. In addition to detecting abnormal packets, if consecutive packets are generated every T (seconds) on the sending side and the sequence number difference Δ = 2, the previous packet was lost or delayed for more than a certain period of time and is discarded. High-quality information transmission is possible by interpolating the abnormal packet using signals reproduced from previous and subsequent packets or by arranging retransmission, as necessary.

Note that the probability that two or more immediately preceding packets become abnormal packets consecutively, that is, Δ=3 or more, is (1-P
In general, Po is set as a value very close to 1, which is considered a practically acceptable value even if there is a delay kerato, so (1-PO)3'' = 0, and in practice it is happens very rarely (e.g. Po
When =0.999, (1-PO)"!=iIO-"
).

In addition, the probability that n packets become abnormal packets and Δ=1 makes it appear as if there were no abnormal packets is (
1-P.

Furthermore, by using Δ=2 as the detection criterion for an abnormal packet, it is possible to prevent an increase in the packet buffer time, which would be a problem if Δ>2 or Δ<1, and a decrease in packet transmission efficiency can be avoided.

FIG. 3 is a block configuration diagram of a packet communication system to which an embodiment of the present invention is applied, and FIG. 4 is a block configuration diagram showing details of the packet disassembly device.

The packet communication system of this embodiment includes a variable length encoder 2
1 and a packet assembly device 22, and a receiving side including a packet disassembly device 24 and a decoder 25 are connected through a packet network 23.

Here, the variable length encoder 21 encodes the input analog human input signal with a variable bit length corresponding to the amount of information in a block time of /f (seconds), and outputs block b' in FIG. In the packet assembly device 22, this output block b' is input and a fixed time interval T (seconds) is input.
Collect blocks b′ for each block, add required additional bits as necessary to sequentially form a packet of fixed bit length T, and add the second sequence number modulo 4 and the block time every /f. The specified first sequence number of modulo 16 and the first sequence number added to the first block included in the packet are added to the header, for example, and sequentially sent to the packet network 23.

The packet disassembly device 24 includes a smoothing buffer 31, a CPU 32, a memory 33, an input buffer 34 for a lost recovery and decoder, and a clock supply device 35, as shown in FIG.

Packets transmitted from the transmitting side and arriving through the packet network 23 are input to the smoothing buffer 31 and buffered for a certain period of time (Dm +R) (seconds). During this time, the first and second sequence number information 36 added to the packet are input to the CPU 32 and further stored in the memory 33. Then, in the CPU U32,
The packet transmission information 37 is created in the order of the first sequence number, and the difference Δ between the second sequence numbers of adjacent packets is calculated using the second sequence number stored in the memory 33. When an abnormal packet is detected, , and outputs the abnormal packet information to the smoothing buffer 31 in addition to the packet sending information 37. In accordance with this packet transmission information 37, the smoothing buffer 31 outputs abnormality information as a transmission packet and abnormality information 38 at the same time as the transmission packet every T (seconds). The input buffer 34 divides the input packet into reproduction blocks e' every /f (seconds) and outputs the divided blocks as encoded signals 39 to the decoder 25. Also, depending on the abnormality information, predetermined interpolation is performed using the encoded signals of the packets before and after the abnormal packet, and either the encoded signal 39 for the abnormal packet is output or a retransmission request signal is output.

The decoder 25 reproduces and outputs the input encoded signal into the original analog signal. Note that the clock supply device 35 supplies necessary clocks to each part.

As described above using FIGS. 3 and 4, the present invention can be easily realized as an apparatus using known techniques.

〔Effect of the invention〕

As explained above, the present invention uses fixed bit length packets made up of blocks of variable bit length to achieve efficient information transmission by utilizing the burstiness of the information source, and smoothing the delay for each packet on the receiving side. By simultaneously detecting abnormal packets, it becomes possible to interpolate the information of the missing packet from previous and subsequent values or request retransmission as necessary, allowing high-quality information transmission without reducing transmission efficiency. There is an effect that can be done.

[Brief explanation of drawings]

FIG. 1 is a time chart showing one embodiment of the present invention. FIG. 2 is a characteristic diagram of the packet delay time distribution. FIG. 3 is a block diagram showing a packet communication system according to an embodiment of the present invention. FIG. 4 is a block diagram showing details of the packet disassembly device. FIG. 5 is a time chart showing a conventional example. 1 to 16...Sequence number, 21...Variable length encoder, 22...Packet assembly device, 23...
Packet a, 24...Packet disassembly device, 25...Decoder, 31...Smoothing buffer, 32...CPtJ, 33...Memory, 34...
- Input buffer, 35... Clock supply device, 36.
... Sequence number information, 37... Packet sending information, 38... Sending packet and error information, 39...
Encoded signal, D, D'...delay time for each packet, D,...allowable delay time for each packet, R...maximum delay time when forming a packet, T...fixed time interval,
a...Audio signal, b...Encoded signal, b'...Block, C...Transmission packet, d...Reception packet, e...Reproduction packet, e'...Reproduction block , f
...Sampling frequency, k...Number of sampling points, t, ...Packet common delay time. Patent Applicant Nippon Telegraph and Telephone Corporation Agent Patent Attorney Takashi Ide TD: BUG-2 TOTENTSUKENEE+MA D: PAP-2 TONI OTO M, 1, KAIMA Dm = 8 PAKE 2 DOB and Goenmachima Dai as an example. F13 figure tail 4 figure

Claims (2)

[Claims] (1) Variable bits that change the number of information bits in a packet according to the number of information bits generated within the block time k/f (seconds) (where f is the sampling frequency and k is the number of sampling points) This block is collected at a fixed time interval T (seconds) and the necessary information bits are added to it to form a packet with a fixed bit length, and this packet is used for communication. In a packet communication method that uses The first sequence number corresponding to a specific block of blocks and the second sequence number modulo n (n is a natural number) specified in the order of the packets generated are assigned and transmitted, and on the receiving side, the above A packet is stored for a certain period of time (D_m+R) (seconds) from the time the first packet or a specific reference packet arrives (where D_m is the allowable delay time per packet, and R is the maximum delay time when forming a packet.) The packet communication method is characterized in that the packet communication method is characterized in that the packets are subsequently reproduced in sequence according to the first sequence number, and abnormal packets are detected from the difference in the second sequence numbers of the adjacent packets that are arranged. (2) The packet communication method according to claim (1), in which an abnormal packet is detected when the difference in sequence numbers is 2.