JPH0310542A - Voice packet composing method - Google Patents

Abstract

PURPOSE:To shorten a voice packet composing time by using a packet composing means so as to use a sampled voice signal at the same sampling point of the same sampling clock on plural voice communication lines thereby forming a prescribed length of packet. CONSTITUTION:Voice signals of 20 channels are sampled simultaneously with an 8kHz sampling clock by using A/D converters 10, 11...12 and each 8-bit PCM signal of the 20 channels is stored in registers 13, 14...15 for each 125mus. Then a packet composing circuit 2 fetches the 8-bit PCM signal into the buffer memory 16 from the registers 13, 14...15 for a time of 125mus to latch the succeeding PCM signal in the registers 13, 14...15. Then the PCM signal fetched in the buffer memory 16 is used to compose the packet.

Description

[Detailed Description of the Invention] [Summary] This invention relates to a method for assembling voice packets when voice is packetized and multiplexed for transmission in a system where there are multiple voice communication lines from a source station to a destination station. The purpose of the present invention is to provide a voice packet assembling method that can shorten the assembling time of a predetermined length using audio signals sampled at the same sampling point of the same sampling clock of a plurality of voice communication lines using a packet assembling means. Configure it to be a packet.

[Industrial Application Field] The present invention relates to an improvement in a voice packet assembly method when voice is packetized and multiplexed for transmission in a system in which a plurality of voice communication lines exist from a source station to a destination station.

Voice communications have strict limits on the delay time required for transmission, and the current public M? 4 (voice) is designed based on the propagation delay time of 30 ms or less.

On the other hand, when audio is transmitted in packet form, the following delay time occurs within the network, and it is desired to reduce this delay as much as possible.

■Audio packet assembly time (accumulation time to collect a certain length of audio samples into a packet), ■Packet sending time (time required to send a packet from the station to the transmission path), ■Packet transmission time (time to propagate through the transmission path) ), ■Intra-network delay time fluctuation absorption time (Since the packet transmission system is based on store-and-forward, the waiting time at the switching center changes, and the time from when a packet is input to the network until it arrives at the other station is constant) Packet disassembly time (processing time to extract the sample from the packet).

Among the above delay time elements, the packet transmission time (3) is determined by the packet length and transmission speed, and the only way to shorten this is to shorten the packet.

However, the packet length cannot be longer than a certain range depending on the system due to the balance with the header length, which becomes an invalid transmission time.

The packet transmission time in (2) is determined by the transmission distance and cannot be shortened.

The intra-network delay time fluctuation absorption time (2) is determined by the intra-network delay time fluctuation width, and cannot be shortened.

Therefore, it is desired to be able to shorten the time required for assembling and disassembling one packet.

An example of a communication system to which the present invention is applied is as shown in FIG.
BX), 40, 41, and 42 are packet multiplexing devices (hereinafter referred to as MUX), which
There is a 20-channel (hereinafter referred to as CH) voice communication line between PBX 30 and PBX 32, and a LOCH voice communication line between PBX 30 and PBX 32.

There are also 30 voice communication lines between PBX30 and MUX40, and MUX40 has 20 channels for PBX31 and PB
The 10CR for X32 is time-division multiplexed and transmitted through one transmission path.

Furthermore, there are 20 voice communication lines and 10 voice communication lines between the MUX 41 and the PBX 31, and between the MUX 42 and the PBX 32, respectively.

In this case, it is desired to shorten the time for assembling and disassembling fixed-length packets using MTJX40, 41.42.

(Prior art) For example, from PBX30 to PBX31 in Figure 6,
In order to assemble packets for 0CH, conventional technology assembles packets of a fixed length for each CH, so for example, when the packet length is 64 bytes (information length excluding header), audio The sampling clock of 8
Assuming KHz and 8-bit PCM, it takes 8 ms to accumulate 64 samples and assemble a packet.

[Problem to be solved by the invention]

As explained above, in the conventional technique, a packet of a fixed length is assembled for each CH, so there is a problem that it takes a long time to assemble the packet.

The present invention can shorten the time required to assemble a packet of a certain length.
The purpose is to provide a voice packet assembly method.

[Means to solve the problem] FIG. 1 is a diagram showing the principle of the present invention.

As shown in FIG. 1, in a system in which there are a plurality of voice communication lines from a source to a destination station, when the voice of the plurality of voice communication lines is packetized and multiplexed transmitted, the packet assembling means 1 Audio signals sampled at the same sampling point of the same sampling clock on a plurality of audio communication lines are used to form packets of a predetermined length.

(Function) According to the present invention, for example, PBX 30 and PBX in FIG.
Assume that there are 20 audio communication lines as shown in 31, and the two channels were sampled at the same sampling time using the same sampling clock.
When a 0CH audio signal is used, a packet is assembled using 20CH audio signals sampled at - degrees, as shown in Figure 1 (A), and the packetization time is reduced to form packets of the same length. It becomes 1/20,
It is possible to shorten the time required to assemble voice packets.

〔Example〕

FIG. 2 is a block diagram of a packet assembling unit on the transmitting side according to an embodiment of the present invention, FIG. 3 is a block diagram of a packet disassembling unit on the receiving side according to an embodiment of the present invention, and FIG. 4 is a block diagram of a packet disassembling unit on the receiving side according to an embodiment of the present invention. FIG. 5 is a diagram showing the packet format, and is a diagram showing an example of occurrence frequency distribution with respect to sound pressure.

In Figure 2, for example, the audio signal of 20C) is converted into an analog
Digital converter (hereinafter referred to as A/D converter) 10,1
1. -...12, simultaneously sampled with 8KHz sampling clock, and 8-bit PC of each of 20 channels.
M signals are sent to registers 13, 14 .
...Keep it at 15.

The packet assembly circuit 2 includes registers 13, 14,
...From 15, the 8-bit PCM signal is converted to 125
During μs, data is loaded into the buffer memory 16 and registers 13, 14 . ...15 to hold the next PCM signal.

Then, a packet is assembled using the PCM signal taken into the buffer memory 16. In this case, in order to facilitate packet assembly and disassembly, the packet length is set to 60 samples, and the time of 3 sampling clocks is 125 μ5
One packet is assembled in 3=375 μs.

That is, the time is 1/20 of the time required to assemble each ICH.

In addition, considering the case of 8KHz sampling clock and 8-bit PCM audio, the audio sample value H"00゛'~H
The occurrence frequency of each "FF" (hexadecimal code of sound pressure with respect to amplitude according to μ law) has a constant distribution depending on the speaker, as shown in FIG. 5(A).

In other words, the frequency of occurrence of the maximum sound pressure H "" F
” occurs as shown in the figure, and when we calculate the frequency of occurrence with respect to sound pressure, we find that the frequency of occurrence differs depending on the sound pressure (Figure 5 (A))
The result is a curve as shown in .

Also, considering a packet multiplexed with 20 channels, 200
Speaker characteristics are averaged, sound pressure distribution is emphasized,
When the frequency distribution with respect to the sound pressure is determined, a curve as shown in FIG. 5(B) is obtained in which the frequency of occurrence varies depending on the sound pressure.

Therefore, by applying variable-length code conversion that corresponds to this frequency distribution and assigns short codes to sound pressures that occur frequently and long codes to sound pressures that occur less frequently, the packet length can be reduced. The average value can be reduced.

Therefore, the code converter 17 sequentially extracts samples from the buffer memory 16 and performs variable length code conversion.

A header indicating the source, destination, amount of information, etc. is added to the information equivalent to one packet code-converted by the code converter 17 by the header adding unit 18 and output.

The format of the packet is shown in FIG. 4, and the information section in this case contains information of 3 samples CHI to CH2O, for a total of 60 samples.

In addition, how many samples are loaded in one packet?
It may be determined at the time of system design, or an approximate packet length may be defined so as to satisfy a certain overhead, and determined in consideration of the number of channels to be multiplexed.

The packet disassembly unit on the receiving side is as shown in FIG. Performs the conversion and stores the buffer memory 26
Store in.

In the case of 20 channels, the stored samples are simultaneously processed by each of the 20 channels according to the 8 KHz sampling clock.
Transferred to H-compatible registers 23, 24°, 25,
120. 21.・
...It is converted into an analog signal at 22 and output.

That is, the packet disassembly time is also 1/20 of the disassembly time required to disassemble one packet formed by the conventional ICH.

(Effects of the Invention) As explained in detail above, according to the present invention, the time required for assembling and disassembling a packet can be reduced.

[Brief explanation of drawings]

FIG. 1 is a diagram of the principle of the present invention; FIG. 2 is a block diagram of a packet assembling unit on the transmitting side according to an embodiment of the present invention; FIG. 3 is a block diagram of a packet disassembling unit on the receiving side according to an embodiment of the present invention; FIG. 4 is a diagram showing the packet format of the embodiment of the present invention, and FIG.
FIG. 6 is a diagram showing an example of occurrence frequency distribution with respect to sound pressure, and FIG. 6 is a block diagram showing an example of a communication system. In the figure, 1 is a packet assembly means, 2 is a packet assembly circuit, is a packet disassembly circuit, 0 to 12 are analog-to-digital converters, and 3 to 15.2
3 to 25 are registers, 6.26 is a buffer memory, 7 is a code conversion section, 8 is a header addition section, 0 to 22 is a digital/analog conversion section, 7 is a code inverse conversion section, 8 is a header → removal section, 0 to 32 are voice exchanges, and 0 to 42 are packet multiplexers. The combined pressure of the two Hm and pCH → of the present invention! i! Ritu 1st Ri & CH warm up → Kenko 1 case against Kimida 5 outbreak! Solid solution shown as R outside city 5 Figure

Claims (1)

[Scope of Claims] In a system in which a plurality of voice communication lines exist from a source station to a destination station, when the voice of the plurality of voice communication lines is packetized and multiplexed transmitted, the packet assembling means (1): A voice packet assembling method characterized in that voice signals sampled at the same sampling point of the same sampling clock of the plurality of voice communication lines are used to form packets of a predetermined length.