JPH02288443A - Voice packet multiplexing device on reception side - Google Patents

Abstract

PURPOSE:To attain the reduction of packet abolition by reducing the overflow quantity of a buffer memory by performing write regulation when the working ratio of a buffer exceeds a first threshold value. CONSTITUTION:The buffer memory 12 stores a separated voice information in an area corresponding to each output trunk channel, and also, reads it out after prescribed delay time, and the voice information signal of a read out time division multiplex serial signal is multiplex-separated with the output of a multiplexer 17, and is sent to each output trunk channel. Furthermore, a buffer working ratio detecting means 18 detects the working ratio of the buffer memory 18, and supplies it to a write regulation means 14a, and the write regulation means 14a regulates the write of the voice information signal on the buffer memory 12 when a detected buffer working ratio exceeds a prescribed first threshold value. Thereby, the abolition of the packet due to the overflow of the buffer memory can be reduced.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a voice packet multiplexing receiving side device connected to a high-speed digital line, a dedicated line, etc. in a packet switching network.

[Prior art] Regarding voice packet communication, see the document "Packetized voice communication" (Tatsu Tsuda, Journal of the Institute of Electronics and Communication Engineers, Vol. 62, No. 2, p.
D182-184. February 1979). This document describes a device with one audio channel, but there is also an audio packet multiplexing device that multiplexes a plurality of audio channels and performs audio packetization.

The transmitting side device of the audio packet multiplexing device time-division multiplexes the audio information signals input from a plurality of input/drunk channels, and performs digital audio insertion (Digital 5pee).
ch Interpolation: DSI)
When this significant audio information signal reaches a predetermined block length, a header signal is added to it to form an audio packet, and the signal is sent to the receiving side via the transmission path. Send to device.

The audio packet multiplexing receiving device decomposes the received audio packet into a header signal and an audio information signal, and after absorbing delay fluctuations for the audio information signal, sends it out to a predetermined output trunk channel. This receiving side device selects an output trunk channel based on information in the header signal.

To be more specific, in conventional voice packet multiplexing receiving devices, voice packets received via a transmission path such as a high-speed digital line or a dedicated line in a packet switching network are separated into voice information signals and header signals in a packet decomposition unit. Disassemble. The header signal is converted from a logical channel number on the header signal to an incoming trunk channel number in an assigned header decoding section. This incoming trunk channel number corresponds to the output trunk channel number, and the allocation processor determines an area of the buffer memory in which to write the voice information signal according to the output trunk channel number, and writes the voice information signal to that area. . The audio information signal stored in the buffer memory by this writing is read out from the buffer memory after a certain period of time has elapsed in order to absorb delay fluctuations according to instructions from the allocation processor. This read audio information signal is sent to each output trunk channel via a demultiplexer.

[Problem M to be solved by the invention] However, in the conventional voice packet multiplexing receiving side device, the voice information signal output from the packet decomposition unit is
Based on the incoming trunk channel number of the header signal, it was unconditionally written to the finite capacity buffer memory. Therefore, in a temporary overload state due to a sudden increase in voice packet communication traffic, the packet queue in the buffer memory increases and the delay time temporarily increases. Therefore, the number of packets discarded due to packet overflow in the finite-length buffer memory increases temporarily. As a result, a significant deterioration in call quality occurred.

Furthermore, conventionally, the audio information signal stored in the buffer memory is continuously outputted from the buffer memory after a certain fixed delay time has elapsed to absorb delay fluctuations. Therefore, if the delay time fluctuates rapidly due to a sudden increase or decrease in voice packet communication traffic, this delay variation cannot be absorbed, resulting in instantaneous interruptions due to underruns or packet discards due to overruns, resulting in voice talk spurts. Discontinuities sometimes occurred in (sound parts). As a result, in this case as well, the voice communication quality was significantly degraded.

The present invention has been made in consideration of the above points, and even in overload or high load conditions of communication traffic of temporal voice packets, it is possible to maintain voice conversation without deteriorating voice call quality. It is an object of the present invention to provide a voice packet multiplexing receiving side device that is capable of performing communication effectively by using the voice packet multiplexing device.

[Means for Solving the Problems] In the first aspect of the present invention, a voice packet that receives multiplexed voice packets sent from a voice packet multiplexing sending device and transmitted through a transmission path in a packet switching network. The multiplex receiving side device is composed of the following means.

That is, packet disassembly means separates a multiplexed received audio packet into a header signal and an audio information signal, and a logical channel number included in the separated header signal is assigned to an output trunk channel corresponding to an incoming trunk channel. The audio information signal separated by the ladder decoding means and the packet decomposition means is stored in an area corresponding to each output trunk channel, and the stored audio information signal is transferred after a predetermined delay time has elapsed. read later,
Based on the output trunk channel number outputted from the buffer memory that outputs the audio information signal consisting of successive time division multiplexed serial signals and the allocation header decoding means, an allocation status table is created and the allocation status of the buffer memory is controlled. It comprises an allocation processor and a demultiplexer that demultiplexes the time division multiplexed serial signals output from the buffer memory and sends the demultiplexed signals to each output trunk channel.

The above-mentioned means are basic means for the receiving circuit, and the first aspect of the present invention is characterized by the provision of the following means. That is, a buffer usage rate detection means for detecting the usage rate of the buffer memory, and a write regulation that restricts writing of the audio information signal to the buffer memory when the detected buffer usage rate is equal to or higher than a predetermined first threshold value. It is distinctive in that it has a means.

In the second aspect of the present invention, each basic means in the voice packet multiplexing receiving device is provided in the same manner as in the first aspect of the present invention. This second invention is characterized in that the following means are provided. That is, statistical distribution detection means calculates the average and variance of instantaneous interruptions based on a plurality of instantaneous interruption information in the time-division multiplexed serial signal output from the buffer memory, and the buffer memory an additional delay time calculation means for calculating a delay time to be added in the above, and a readout restriction means for restricting reading of the audio information signal from the buffer memory when the calculated additional delay time is equal to or greater than a predetermined second threshold value. It is distinctive in that it has been established.

[Operations] The basic operations of both the first and second inventions are as follows.

The packet decomposition means separates the multiplexed received audio packet into a header signal and an audio information signal, provides the header signal to the allocated header decoding means, and provides the audio information signal to the buffer memory. The allocation header decoding means converts the logical channel number included in the separated header signal into an output trunk channel number corresponding to the incoming trunk channel and provides the converted output trunk channel number to the allocation processor. The allocation processor creates an allocation status table based on the output trunk channel number and controls the allocation status of the buffer memory.

As a result, the buffer memory stores the separated audio information signals in an area corresponding to each output trunk channel, reads out the stored audio information signals after a predetermined delay time, and reads out the time division multiplexed serial The audio information signal consisting of the signal is output to the demultiplexer. The demultiplexer demultiplexes the time division multiplexed serial signal output from the buffer memory and sends it to each output trunk channel.

In the first aspect of the present invention, in addition to the above-mentioned basic effects, the following effects occur. The buffer usage rate detection means detects the usage rate of the buffer memory and provides it to the write regulation means,
The write restriction means restricts writing of the audio information signal into the buffer memory when the detected buffer usage rate is equal to or higher than a predetermined first threshold. This reduces packet discards due to buffer memory overflow.

In addition to the above-mentioned basic functions, the second aspect of the present invention operates as follows. The statistical distribution detection means calculates the average and variance of the instantaneous interruption length based on the plurality of instantaneous interruption information in the time division multiplexed serial signal outputted from the buffer memory, and provides the calculated average and variance to the additional delay time calculation means. , the delay time to be added in the buffer memory is calculated based on the detected average and variance. The read restriction means restricts output of the audio information signal from the buffer memory when the additional delay time calculated by the additional delay time calculation means is equal to or greater than a predetermined second threshold. As a result, -temporal interruptions in talk spurts are reduced.

[Example] Hereinafter, an example of the present invention will be described in detail with reference to the drawings.

FIG. 1 is a functional block diagram showing this embodiment.

In FIG. 1, an audio packet from a transmission path 10 on which multiple channels are multiplexed is given to a packet disassembly unit 11, which converts the audio packet into an audio information signal (time division multiplexed serial signal). The audio information signal is provided to the buffer memory 12, and the header signal is provided to the assigned header decoding section 13.

The header signal is decoded into a logical channel number on the header signal by the allocation header decoding section 13, and then converted into an output trunk channel number and sent to the allocation processor 14.

The buffer memory 12 is a storage unit that inputs significant audio information signals, stores the audio information signals, and outputs them after a predetermined period of time to absorb delay fluctuations occurring in a packet switching network or the like. The inside of the buffer memory 12 is modeled as a single queue, so even if the input audio information signals are stored in different areas for each trunk channel, the order of the input audio information signals will not be reversed, and the output audio information signals will not be reversed. Temporal order is compensated.

The time-division multiplexed serial signal outputted from the buffer memory 12 and in which the time order is preserved has noise generated by the noise generation section 16 inserted into the silent section (pause) by the speech synthesis section 15, and then sent to the demultiplexer 17. given to. The reason for inserting noise is to prevent the listener from feeling uncomfortable if there is no background noise in the silent section.

The demultiplexer 17 is given a switching control signal from the allocation processor 14, and in response to this switching control signal, the demultiplexer 17 receives a time division multiplexed serial signal (audio information signal ) and outputs audio information signals to each output trunk channel CHI to CHn.

Here, the time division multiplexed serial signal which is the output of the buffer memory 12 is transmitted to n output trunk channels CI (1 to 1).
The audio information signal output to CHn is a signal in which n time slots are consecutively arranged serially. That is, each time slot from the first time slot to the nth time slot corresponds to the first trunk channel to the nth trunk channel, respectively.

The period from the first time slot to the nth time slot is called one frame length.

For example, when encoding an 8-bit PCM signal (audio information signal) with a sampling period of 125 μs, the length of one frame obtained by multiplexing is also 125 μs, and the encoding speed of the audio information signal is 64 Kbit/S. becomes. In practice, P
CM primary group high-speed digital line is 1.544Mb i
Since the transmission speed is t/s, the number n of time slots per frame is 24, that is, 24 trunk channels can be secured.

FIG. 2 shows a time chart for the audio information signal of the audio packet.

The audio packet shown in FIG. 2 (A) received via the transmission path 10 consists of a header signal and an audio information signal, of which the audio information signal is stored in the buffer memory 12 as shown in FIG. 2 (B). written to.

One frame of audio information signal written in this way (1
The audio information signal inserted into the information field of each audio packet is read out from the buffer memory 12 so as to have a period of one frame, that is, 125 μs, as shown in FIG. 2(C). Figure 2 shows the processing start points aligned at (A) to (D> in the same figure.In practice,
The processing starting point is different). This read time-division multiplexed serial signal is demultiplexed by the demultiplexer 17, and becomes as shown in FIG. 2 (D>).

Note that FIG. 2(D) also shows preprocessing of one statistical distribution detecting section, which will be described later. That is, the statistical distribution detection section 19
In order to calculate statistics for each channel, preprocessing is performed to multiplex and demultiplex serial signals before background noise synthesis. As shown in the figure, the preprocessed output of one statistical distribution detection section per channel of this serial signal results in talk spurts (sound parts) and pauses (silent parts) being arranged in a discrete manner.

The above is the basic configuration of the audio packet multiplexing receiving device that extracts and demultiplexes audio information signals from audio packets related to audio information signals that have been serialized by time division multiplexing. In addition, a configuration for controlling the delay time of the buffer memory 12, etc. is provided.

The buffer usage rate detection unit 18 calculates the usage rate β of the buffer memory 12 at every fixed time T8. That is, this buffer usage rate detection unit 18 detects the storage capacity Q of the audio information signal every fixed time TB, and determines the buffer usage rate which is the ratio of the detected storage capacity Q to the maximum storage capacity Qll1aX of the buffer memory 12. β is calculated and notified to the allocation processor 14.

Further, the n-channel statistical distribution detection section 19 detects the amount of statistical distribution of the audio information signal for each channel based on the time division multiplexed serial signal output from the buffer memory 12. As an amount of statistical distribution, for example, the length L of an instantaneous interruption within a talk spurt (instantaneous interruption due to underrun)
Calculate the mean E and variance ■ of F. The average E (LF) and the variance (LF) are determined by the following equations.

ILF)=ΣLFi/NA...(
1) V (LF)=Σ(Lri-IF, (LF)>
2÷NA (2) Here, NA is a constant time expressed discretely.

The mean and variance thus calculated are provided to the allocation processor 14. The allocation state control unit 14a of the allocation processor 14 controls the buffer memory 12 based on the buffer usage rate β given from the buffer usage rate detection unit 18 and the statistics E(LF) and V(LF) given from the statistical distribution detection unit 19. Determine the control state for. Such a control signal is given to the memory control section 20, and the memory control section 20 directly controls the buffer memory 12.

FIG. 3 shows in more detail the control principle configuration of this embodiment for the buffer memory 12 in FIG. 1.

In FIG. 3, a write control section 20W and a read control section 2O
R constitutes the memory control section 20 in FIG.

The write control section 20W performs write control on the buffer memory 12 based on the output of the allocation state control section 14a. The first feature of this embodiment is that writing to the buffer memory 12 is restricted when a predetermined condition is satisfied.

If the buffer usage rate β output from the buffer usage rate detection unit 18 is smaller than the first predetermined threshold value, a normal write control signal is given from the allocation state control unit 14a (not shown), and the packet The audio information signal from the decomposition unit 11 is written into the buffer memory 12 without write restrictions. On the other hand, when the buffer usage rate β is greater than or equal to the first predetermined threshold (high load state), a write restriction signal is given to the write control unit 20W to restrict writing to the buffer memory 12.

The reason why writing is restricted in this way is that although the restricted audio information signal is discarded, this restriction reduces the overflow of the buffer memory 12 and reduces packet discards when viewed as a whole. This is because it can be done.

The read control section 2OR performs read control for the buffer memory 12 based on the output of the allocation state control section 14a. The second feature of this embodiment is that reading of the buffer memory 12 is restricted when a predetermined condition is satisfied.

Based on the statistics output from the statistical distribution detection unit 19,
The allocation state control unit 14a calculates an additional delay time, and if this additional delay time is smaller than a second predetermined threshold d, the allocation state control unit 14a provides a normal read control signal (not shown). ), the audio information signal from the packet disassembly unit 11 is not read-out restricted and is stored in the buffer memory 12.
A delay time is added and read out. On the other hand, when the additional delay time is equal to or greater than the second predetermined threshold value 6 (overload state), a read regulation signal is given to the read control unit 20W to regulate reading from the buffer memory 12.

The reason for restricting reading in this way is that although there is a risk of overflow occurring, this restriction reduces momentary interruptions due to underruns in the buffer memory 12, and reduces packet discards (underruns) when viewed as a whole. This is because it is possible to reduce the number of packets discarded due to instantaneous interruptions.

Note that the additional delay time is determined as follows. The additional delay time is expressed by the following formula using the functions f(>, g(). Here, Do is the constant delay time at the initial setting,
Dj is a delay time for absorbing delay fluctuations.

D2 Do -to Dj
...(3) Dj = f (E
(LF)) +g (V (LF))...(4) Note that the functions f(> and g(>) are increasing and decreasing functions as LF increases and decreases, and both instantaneous interruptions are normally distributed N (E(LF),■(
LF)).

Buffer memory 12 by the allocation processor 14 mentioned above
The channel allocation control for is shown in a flowchart as shown in FIG. Note that FIG. 4 shows a flowchart for one trunk channel.

In FIG. 4, first, it is determined whether or not there is a voice packet, and if there is no voice packet, no control is performed on the buffer memory 12 (step 100).

If a voice packet is input, then it is determined whether the buffer usage rate β is greater than or equal to the usage rate threshold (step 101). If the usage rate is lower than the threshold, that is, if the load is high, the audio information signal of the trunk channel is discarded without being written to the buffer memory 12, and the write control unit 20W is instructed to restrict writing. The process ends (step LO2, 103).

If the buffer usage rate β is normal, the trunk channel is registered in the internal allocation status table, and the audio information signal of the trunk channel is written into the buffer memory 12 (steps 104 and 105). Thereafter, the statistical distribution detection unit 19 takes in the statistics, calculates the additional delay time, and determines whether the additional delay time is equal to or greater than the delay time value d (steps 106 to 108).

If the calculated additional delay time is smaller than the threshold d, it is determined that the state is normal, and the audio information signal of the trunk channel is stored in the buffer memory 12 after the additional delay time has elapsed from the time of writing. After that, a new allocation instruction is issued and the process ends (step 109.1).
10).

On the other hand, if the additional delay time is equal to or greater than the threshold value 6, the audio information signal of the relevant trunk channel is discarded without being read from the buffer memory 12, and the read control unit 2 instructs the OR to restrict reading, and ends the process. (step 111
．． 112).

FIG. 5 shows a time chart regarding delay fluctuations of voice packets.

A voice packet on the transmission path output of the transmitting device is transmitted within the packet switching network, undergoes transmission delay and delay variation, and reaches the transmission path input of the receiving device. The received voice packet is added with a fixed delay time DO at the time of initial setting,
Further, a delay time Dj for absorbing delay fluctuations is added, and a reproduced audio signal is output.

Figure 5 (A1) to (A3) are the transmitter outputs during normal operation;
Inputs to and outputs from the buffer memory 12 on the receiving side are shown.

Figure 5 (B1) to (B3) are the transmitter side outputs during abnormal operation,
Inputs to and outputs from the buffer memory 12 on the receiving side are shown. In this case, a momentary interruption occurs due to underrun.

As a result, the statistical amount output from the statistical distribution detection unit 19 changes and the average becomes large), and the calculated additional delay time may become equal to or greater than the threshold value 6. When the darkness value is 6 or more, reading from the buffer memory 12 is restricted,
From this point on, the number of instantaneous interruptions due to underrun will decrease.

Therefore, according to the above embodiment, the buffer usage rate is the first
Since writing is restricted when the threshold value is exceeded, it is possible to reduce the overflow amount of the buffer memory and reduce packet discards.

In addition, the additional delay time is calculated based on the average and variance of the instantaneous interruption length of the talk spurt, and readout is restricted when the additional delay time is greater than or equal to the second threshold. It is possible to reduce packet discards by reducing instantaneous interruptions.

In addition, although the above-mentioned example showed the case where the number of trunk channels was 24, the present invention is not limited to this.

Furthermore, the configuration for inserting background noise into silent portions may be omitted.

Further, in the above description, an image structure including a writing restriction configuration and a successive output restriction configuration is shown, but even if it has at least one of the configurations, a special effect can be achieved compared to the conventional one.

[Effects of the Invention] As described above, according to the first aspect of the invention, writing is restricted when the buffer usage rate is equal to or higher than the first threshold, thereby reducing the overflow amount of the buffer memory. This can reduce packet discards.

Further, according to the second aspect of the present invention, the additional delay time is calculated based on the average and variance of the instantaneous interruption length of the talk spurts, and when the additional delay time is equal to or greater than the second threshold, readout is restricted. As a result, it is possible to reduce temporal interruptions in talk spurts and reduce packet discards.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing an embodiment of the audio packet multiplexing receiving device according to the present invention, FIG. 2 is a time chart of the audio information signal of the audio packet, and FIG. 3 is a diagram showing the main parts of FIG. 1. A block diagram showing the detailed configuration, Figure 4 is the first
FIG. 5 is a flowchart showing the allocation control processing of the allocation processor shown in FIG. DESCRIPTION OF SYMBOLS 11...Z packet decomposition part, 12... Buffer memory, 13... Allocation header decoding part, 14... Allocation processor, 14a... Allocation state control part, 17... Demultif. Lexa, 18... Buffer usage rate detection unit, 19...
Statistical distribution detection section, 20... Memory control section, 20W... Write control section, 2OR... Read control section. Occurrence of instantaneous interruption due to C7y9'-run during abnormal operation) Time chart for delay fluctuations in audio (C7y9'-run) Fig. 5

Claims (2)

[Claims] (1) Voice packet multiplexing A voice packet multiplexing receiving device receives multiplexed voice packets sent from a transmitting device and transmitted through a transmission path in a packet switching network. packet disassembly means for separating a packet into a header signal and an audio information signal; and assigned header decoding means for converting a logical channel number included in the separated header signal into an output trunk channel number corresponding to an incoming trunk channel. and accumulating the audio information signal separated by the packet disassembly means in an area corresponding to each output trunk channel,
a buffer memory for reading out the accumulated audio information signal after a predetermined delay time and outputting an audio information signal consisting of the read time division multiplexed serial signal; and an output trunk channel number output from the allocated header decoding means. an allocation processor that creates an allocation status table and controls the allocation status of the buffer memory based on the above; and a demultiplexer that demultiplexes the time division multiplexed serial signal output from the buffer memory and sends it to each output trunk channel. , a buffer usage rate detection means for detecting the usage rate of the buffer memory, and a writing device that restricts writing of the audio information signal to the buffer memory when the detected buffer usage rate is equal to or higher than a predetermined first threshold value. 1. A voice packet multiplexing receiving side device, comprising: regulating means. (2) Voice packet multiplexing A voice packet multiplexing receiving side device that receives multiplexed voice packets sent from a transmitting side device and transmitted through a transmission path in a packet switching network receives multiplexed audio. packet disassembly means for separating a packet into a header signal and an audio information signal; and assigned header decoding means for converting a logical channel number included in the separated header signal into an output trunk channel number corresponding to an incoming trunk channel. and accumulating the audio information signal separated by the packet disassembly means in an area corresponding to each output trunk channel,
a buffer memory for reading out the accumulated audio information signal after a predetermined delay time and outputting an audio information signal consisting of the read time division multiplexed serial signal; and an output trunk channel number output from the allocated header decoding means. an allocation processor that creates an allocation status table and controls the allocation status of the buffer memory based on the above; and a demultiplexer that demultiplexes the time division multiplexed serial signal output from the buffer memory and sends it to each output trunk channel. , statistical distribution detection means for calculating the average and variance of instantaneous interruption lengths based on a plurality of instantaneous interruption information in the time division multiplexed serial signal output from the buffer memory; additional delay time calculation means for calculating a delay time to be added in the memory; and readout regulation means for regulating reading out of the audio information signal from the buffer memory when the calculated additional delay time is equal to or greater than a predetermined second threshold. An audio packet multiplexing receiving side device comprising: