JPH04356832A - High efficiency digital multiplexing transmission system - Google Patents

Abstract

PURPOSE:To make compatible the improvement of communication quality and the improvement of multiplexing efficiency by transmitting only the complementary information of a core information part when each sound channel is silent and transmitting data containing an additional information part when the sound channel has sounds. CONSTITUTION:A sound encoding means 20 executes the encoding of sound information to be inputted to each sound channel separately for the core information part and the additional information part. A multiplexing processing part 22 multiplexes only the complementary information required for synchronizing an operation mode with a coder at a communicating destination in respect to the sound channel, for which a silent block detects silence, and first multiplexes the core information in respect to the sound channel, for which no silence is detected, to the digital slot of a fixed length repeated in a fixed cycle. Afterwards, the additional information part of each sound channel not to detect the silence at the remaining part of the fixed length digital slot is multiplexed from the weight part, and the additional information part disabling multiplexing due to of the lack band width is abandonned.

Description

[Detailed description of the invention]

0001

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a high-efficiency digital multiplex transmission system in a multiplexing device or exchange for transmitting multiplexed signals over a digital line.

0002

2. Description of the Related Art FIG. 1 shows an example of a voice multiplexing device or switching system, and FIG. 2 shows an example of a multimedia multiplexing device or switching system. In FIG. 1, voice data from a telephone terminal 1 is input to a relay transmission network 3 via a voice multiplexer/switch 2, and the voice data is transmitted to a receiving phone via a voice multiplexer/switch 4 on the receiving side. It is transmitted to terminal 5. Here, relay transmission network 3
includes digital trunk lines and relay voice multiplexers/switches.

In FIG. 2, from the telephone terminal 1 to the PBX 6
, from the data terminals via the packet switch 7 , and from the TV conference terminals via the video switch 8 to the multimedia multiplexer/switch 9 , is input to the relay transmission network 10 for reception. On the receiving side, a PBX 12, a packet switch 13, or an image switch 1 is connected via a multimedia multiplexer/switch 11.
4 is output.

[0004] In such audio multiplexing devices, multimedia multiplexing devices, etc., time division multiplexing (TDM) is generally used.
) formula is used, but Digital Speech Interpolation (DSI) is used as a statistical multiplexing method to improve transmission efficiency.
) multiplexing system, digital data interpolation (
DDI) multiplexing method is also used. The DSI method is a method that detects silent sections of voice and transmits only the voiced sections, and the DDI method is a method that further complements and multiplexes information from data communication media such as packet switches that have waiting times into the silent sections of voice. It is.

[0005]

3 and 4 are examples of conventional band allocation systems for TDM multiplexing and statistical multiplexing. Figure 3 shows the TDM multiplexing method. In this method, a fixed band is allocated to each audio channel, and the efficiency of the audio encoder is pursued from the perspective of how to improve the sound quality within the given band. . On the other hand, in the statistical multiplexing method shown in FIG. 4, bands are accommodated among a plurality of audio channels, thereby expanding the equivalent band per channel and improving transmission quality.

FIG. 5 shows a band allocation method in the DSI method. In this method, by detecting silent sections, bandwidth allocation is flexible depending on whether there is a sound or no voice, and the bandwidth is increased and the sound quality is improved when the number of voice calls is large.

As described above, for example, in the conventional TDM multiplex system, even when the number of voice calls is small, a band is allocated assuming the maximum number of calls. In other words, the quality remains constant at the maximum number of calls, and there is a problem in that the average bandwidth cannot be expected to increase by adjusting the bandwidth according to the voice level.

[0008] Furthermore, in the DSI method, no information is sent during the silent section due to the detection of the silent section. In other words, the band is 0
This will improve transmission efficiency, but in the event of congestion, the encoded information for one audio channel will either be sent in its entirety or will be completely discarded.
There is a problem in that audio quality deteriorates significantly in channels where all channels are discarded.

[0009] The present invention always multiplexes the core information part for guaranteeing the minimum sound quality for the audio channel, for example.
Improving communication quality and multiplexing in statistical multiplexing by sequentially multiplexing parts of the additional information necessary to obtain desired sound quality starting from the highest priority parts and minimizing communication quality deterioration even during congestion discards. The aim is to simultaneously pursue improvements in efficiency.

0010

Means and Effects for Solving the Problems FIGS. 6 to 12 are block diagrams of the principle of the present invention. 6 to 9 are block diagrams showing the principle of a high-efficiency digital multiplex transmission system in a multiplexing device or exchange that transmits multiplexed signals on a digital line.

In FIG. 6, the audio encoding means 20 is, for example, an ADPCM Embedded
It is an audio encoder that separates audio input information into a core information part and an additional information part and encodes it. The core information part guarantees the lowest sound quality through its transmission, and this information is compressed and encoded using, for example, difference information.

[0012] The additional information section needs to be transmitted together with the core information section in order to obtain the desired sound quality, and can be discarded in stages, for example every 10 bits, depending on the priority to be transmitted when discarding the information. For example, the difference between the predicted value of the core information and the input signal is supplementary information that is encoded by a simple PCM quantizer and sent.

The silent interval detection means 21 detects silent intervals in order to utilize the silent intervals of audio input information for transmitting other information, as in the conventional statistical multiplexing method. Furthermore, the multiplexing processing means 22 performs a process of processing a voice channel in which silence is detected by the silent section detection means 21 among the plurality of voice channels to be multiplexed on the digital line between the encoder of the communication partner and the voice channel. Only the auxiliary information (side information) necessary for synchronizing the operation mode is first transmitted, and for voice channels for which silence has not been detected, the core information described above is first transmitted to a fixed-length digital slot that is repeated at a constant cycle. Multiplex.

[0014] Thereafter, the multiplexing processing means 22 multiplexes the additional information section of each audio channel for which silence has not been detected into the remaining portion of the fixed length digital slot, but in the multiplexing, the priority to be transmitted is Multiplexing is performed in stages starting from the part with the highest weight, that is, the part with the heaviest weight, for example, every 10 bits, and the additional information part that cannot be multiplexed due to insufficient bandwidth is discarded.

FIG. 7 is a principle block diagram of a multiplexing method based on the transmission required bandwidth for each audio channel, and the required bandwidth determining means 23 performs each transmission of audio input information for each of a plurality of audio channels to be multiplexed on a digital line. Determine the required bandwidth.

The multiplexing processing means 22 discards the additional information part within the required band determined by the required band determining means 23 when discarding the additional information part of each audio channel for which silence has not been detected by the silent section detecting means 21. Multiplexing for each audio channel is performed by discarding bits in stages, for example, every 10 bits, starting from the portion with the lower transmission priority, that is, the portion with lighter weight.

FIG. 8 is a block diagram of the principle of a multiplexing method according to call detection results. In the figure, a call detection means 24 detects whether each of the plurality of voice channels to be multiplexed is in a call state, that is, whether the call is in progress or the call is disconnected. The multiplexing processing means 22 also includes information necessary for synchronization with the encoder of the communication partner, that is, side information, for the voice channel for which the call detection means 24 has detected a call disconnection state. multiplexing the voice channels of the call without transmitting any encoded information.

[0018] As a method of call detection, a multiplexer or a switch transmits signaling information for a voice channel by a signal send (SS) signal and a signal receive (SS) signal.
R) If the signal is being transmitted as a signal, the SS signal and SR signal are monitored and the audio channel is determined based on the time during which the logical sum or logical product value of the two signals is constant. A method is used to detect a call-in-call or call-dropped condition.

FIG. 9 is a principle block diagram of a multiplexing system when a multiplexing device or switch is connected to a label multiplexing network such as a packet network or an ATM network. In the figure, a multiplexed frame outputted from a multiplexing processing means 22 is packetized by a packet assembling means 25 either as it is or by being divided into packets and outputted to a label multiplexing network. Note that although FIG. 9 shows a form in which a packet assembly means 25 is added to FIG. 6, FIG.
It is also possible to add similar information to FIG.

In the present invention, as described above, the audio information for each audio channel is separated into a core information section and an additional information section, and encoded by the audio encoding means 20. FIGS. 13 and 14 are explanatory diagrams of classification of audio transmission information in the present invention. FIG. 13 shows the separation of the core information section and the additional information section. The core information section is compressed and encoded using differential information, etc., and consists of core information to guarantee the lowest sound quality and side information that is auxiliary information necessary for synchronizing the operation mode between encoders. Ru.

[0021] Further, the additional information section is supplementary information in which, for example, the difference between the predicted value of the core information and the input signal is encoded and sent by a simple PCM quantizer, and the bits of the core information are sequentially sent from the bit with the heaviest weight. When discarding due to congestion, bits with lighter weights are discarded in stages, starting from the top. Then, when there is a voice, the core information section and the additional information section are transmitted, and when there is no voice, only the side information of the core information section is transmitted, and in FIG. 8, when the call is disconnected, nothing is transmitted.

FIG. 14 is an explanatory diagram of the data transmission request range in each of FIGS. 6 to 8. In the same figure, of the core information part, side information is 10 bits, core information is 20 bits.
Total of 30 bits, additional information is 1 depending on its weight
Assuming that each 0 bit is a total of 50 bits, in the principle block diagram shown in Figure 6, when there is a sound, a total of 80 bits are transmitted: 30 bits of core (strictly speaking, the sum of core information and side information) and 50 bits of additional information. , when there is no sound, side information 1
Only 0 bits are transmitted. Furthermore, when discarding congestion, 10 bits of the additional information of each voice channel are sequentially discarded starting from the bits with the lightest weight from the data during voice activity. As a result, when there is sound, the transmission band of each channel ranges from 30 bits for the core when there is sound to 80 bits including all additional bits, and when there is no sound, the transmission band is 10 bits only for side information.

In the principle block diagram shown in FIG. 7, the required bandwidth of each channel when there is sound is set to an arbitrary value every 10 bits between 30 bits for the core and 80 bits including all additional information. The additional bits are sequentially discarded 10 bits at a time when congestion is discarded, with the value of the requested bandwidth as the initial value, and bandwidth arbitration is performed.

In the principle block diagram of FIG. 8, the call detection means 2
Even 10 bits of side information are not transmitted to the voice channel in which call interruption has been detected, and no encoded information is transmitted at all. When a call is connected, only 10 bits of side information is available when there is no sound, and data ranging from 30 bits of only the core to 80 bits including all additional information when there is a call.
Transmitted for each channel.

Next, in the present invention, transmission compression of various control data is also performed in order to further improve transmission efficiency. For example, when transmitting signaling information for a voice channel as signal send (SS) and signal receive (SR) signals as described above, a flag indicating whether or not to transmit an SS signal is transmitted in units of a fixed transmission period. By adding it to the frame and transmitting it,
Transmission compression of signaling information in an interval where the SS signal does not change is performed.

[0026] Furthermore, when accommodating a data channel and transmitting a level signal for status confirmation with a communication partner as a remote signaling (RS) signal, a signal indicating whether or not an RS signal is being transmitted is also used. A flag is added to the frame and transmitted, thereby compressing the transmission of the RS signal.

As described above, in the present invention, depending on the presence/absence/silence of each audio channel, for example, only the side information of the core information part is transmitted when there is no sound, and the data including the additional information part is transmitted when there is sound. Furthermore, when discarding congestion, bits of the additional information section are discarded in order of weight, starting from the lowest, thereby improving communication efficiency and quality, and also compressing the transmission of various control data.

[0028]

Embodiment FIGS. 15 to 17 are explanatory diagrams of the basic configuration and operation of a first embodiment of the transmission system of the present invention. In the basic configuration block diagram of FIG. 15, the embodiment includes a speech encoder 30 that encodes input speech information from the telephone side, a silence detection section 31 that detects a silent section of the input speech information, and a detection result of the silence detection section 31. It consists of a multiplexer (MUX) 32 that multiplexes the output of the audio encoder 30 and outputs the multiplexed signal to the line side. The information used by the multiplexer 32 for audio multiplexing is the encoded data output from the audio encoder 30 (the band for each channel is fixed in advance, for example, 80 bits) and the output from the silence detector 31. The active/silent status of each audio channel. This silence detection can be performed by the audio encoder 30 without independently providing the silence detector 31, or can be detected by the multiplexer 32 using the output of the audio encoder 30.

The audio encoded data length Vi' actually transmitted for each audio channel (indicated by the subscript i) is determined by the following equation. W′≧Σ(VDFi ×(Vi′−Vs))+
N×Vs

・・・・・・(1) However, Vi ≧Vi ′≧Vmin
・・・・・・(2
) Here, Vi' is actually the number of bits allocated to each channel by band arbitration processing to be described later, and corresponds to the band of each channel. The band of a certain channel is, for example, the number of bits allocated to that channel multiplied by the number of frames in 1 sec, and is the channel transmission bit rate. In addition, in order to guarantee the minimum required sound quality, the number of accommodated lines is determined in accordance with the following formula at the line design stage.

W′≧N×Vmin
・・・・・・(3
) The meanings of the symbols used in these formulas are as follows. W′: Number of bits available on the line side Vi
: Voice encoded data length (value determined as a fixed value in advance) Vs : Number of bits of voice encoded data that need to be transmitted even during silence (side information) Vi ′ : Voice code or data length to be transmitted VDFi
: Voice presence flag N : Number of channels to accommodate Vmin : Encoded data length required for transmission with guaranteed minimum sound quality Also, as a condition for the audio encoder 30, encoding is performed in stages between Vi and Vmin. It is necessary to encode data in a format that allows data to be deleted, and for example, an embedded encoder using an adaptive differential (AD) PCM method is used.

FIG. 16 shows the state of the band and voice quality in the first embodiment. By detecting a silent section and transmitting only side information to the silent voice channel, it is possible to avoid deterioration in bandwidth and quality due to an increase in the number of voice calls. Furthermore, even when congestion is discarded, there is a very low possibility that all data for a certain channel will be temporarily discarded, and bandwidth reduction during congestion can be minimized.

In the quality diagram of FIG. 16, DSI in "improvement by DSI" represents statistical multiplexing, and means improvement in sound quality due to an increase in the average allocated bandwidth. FIG. 17 is an example of multiplexing processing at the time of congestion discard in the first embodiment. In the figure, channels #2 and #4 are silent, and only side information is sent to these channels. When congestion is discarded, for example, the additional bits 5 of each channel in the active state are sequentially discarded in the order of the channel number, starting with the additional bit 5 of channel #0, and the additional bit 4 of each channel in the active state is discarded sequentially starting from channel #0. By discarding up to 4 additional bits, all the remaining data will be multiplexed, and the multiplexing order will be multiplexing the side information of each channel, that is, ■
The process starts from 2 and ends when the additional bit 4 of channel #5 is multiplexed at 3.

FIGS. 18 to 20 show the second transmission method of the present invention.
FIG. 2 is an explanatory diagram of the basic configuration of the embodiment and its operation. Figure 18
In the basic configuration block diagram, in addition to the components shown in FIG. 15, a required band determining section 33 is provided which determines the transmission band required by the audio information of each audio channel input from the telephone side.

The information used by the multiplexer 32 for audio multiplexing is the encoded data output by the audio encoder 30 (the band may be fixed or variable) and the information output by the silence detector 31. These are the voice/silence state and the audio encoded data length necessary for the required quality transmission output by the required band determining section 33.

Here, the speech/silence state of speech can be detected by the speech encoder 30 or the multiplexer 32, and the speech encoded data length required for the required quality transmission can be determined by the speech encoder. 30 or can be determined by the multiplexing unit 32.

The multiplexing method in the second embodiment is based on the first
It is determined in exactly the same way as the equations (1) to (3) for the embodiment. However, unlike in the first embodiment, the audio encoded data length Vi is not a value determined as a fixed value in advance, but is determined based on the input audio information, for example, by the required band determining unit 3.
This is the data length as a result of determining the required bandwidth in step 3. Further, the conditions for the speech encoder 30 are also exactly the same as in the first embodiment.

FIG. 19 shows the voice band and quality in relation to the number of voice calls in the second embodiment. In the first embodiment described above, the voice transmission request level has two levels depending on the sound/silence state, whereas in the second embodiment, the sound state is further divided into levels. Since fine-grained control is performed, the sound quality may be further improved compared to the first embodiment.

FIG. 20 is an example of multiplexing processing in the second embodiment. In the figure, channels #2 and #4 are silent, and only side information is transmitted. For example, the required bandwidth for channel #0 is up to 3 additional bits, and #
The required bandwidth for #1 is up to additional bit 4, and the required bandwidth for #3 is up to additional bit 5. When congestion is discarded, the uppermost additional bits in the required bandwidth of each channel are sequentially discarded, for example 10 bits at a time, according to the channel number. All will be multiplexed.

FIGS. 21 to 23 are explanatory diagrams of a third embodiment of the transmission system and its operation. In addition to the components shown in FIGS. 18 to 20 for the second embodiment, the figure also includes the components shown in FIGS. A call detection unit 3 that detects whether a call is placed, that is, whether a call is in progress or a call is disconnected.
4 is provided, and a switch 35 is shown between the telephone side and the voice encoder 30.

The information used by the multiplexing section 32 for voice multiplexing includes, in addition to the information regarding the second embodiment, the state of whether or not a call detected by the call detection section 34 is established. However, this state can be notified from the exchange 35 or determined by either the audio encoder 30 or the multiplexer 32.

As the audio multiplexing method, the following equation (4) is used instead of equation (1) in the first and second embodiments. W′≧Σ(VDFi ×(Vi′−Vs))
+Σ(Vs×CDFi)
・・・・・・(4) (2)
, (3) are used in exactly the same way in the third embodiment. In equation (4), CDFi is a call connection flag for channel i, which is 1 when a call is connected and 0 when a call is disconnected. Further, the conditions for the speech encoder 30 are exactly the same as described above.

In the present invention, as a method of detecting the call connection state in the encoder or multiplexer without notifying the call connection state from the exchange, for example, signal send (SS) and A method is used in which a signal receive (SR) signal is monitored to detect whether a call is being made to a voice channel between multiplexers.

FIGS. 22 and 23 show examples of call detection. FIG. 23 shows that call detection is performed by monitoring the SS signal sent from the transmitting side and the SR signal sent from the receiving side, and this detection method eliminates the need for a status notification path from the exchange.

FIG. 23 shows the call detection logic. In the same figure, for example, 1 indicates a state where the call is completely disconnected for both SS and SR (idle), and 1 indicates a state where the call is completely connected (busy).
When is set to 0, the specified time τ1 is the logical sum of SS and SR.
For example, if it is '0' for 1 second or more, it is determined that the call is in a connected state, and the logical product of SS and SR is the specified time τ2.
For example, if it is '1' for one second or more, a voice call disconnection state is detected.

FIG. 24 is a block diagram of a fourth embodiment of the transmission system. In this embodiment, the multiplexed frame created by the multiplexer 32 is assembled by the packet assembler 36 into a packet format for the label switching network, such as a fixed length packet or ATM cell format, and is transmitted within the label switching network. Ru.

Although FIG. 24 has a configuration in which a packet assembling section 36 is added to the first embodiment shown in FIGS. 15 and 18, a packet assembling section 36 can be similarly added to the second and third embodiments. Of course it is also possible. Further, the operations of the speech encoder 30, silence detection section 31, multiplexing section 32, etc. are the same as in the first to third embodiments.

FIG. 25 is a basic configuration block diagram of a fifth embodiment of the transmission system. In the figure, from the telephone side to the exchange 35
multiplexes audio information for a plurality of audio channels that is input via the audio encoder 30 and encoded by the audio encoder 30, and packets and data that are input from the packet switch, data terminal, etc. through the exchange 37 and the accommodation interface 38. This is an embodiment of a multiplexing device that multiplexes the data by a section 32 and outputs it to the line side. Note that the exchanges 35 and 37 are not necessary in the case of a fixed route connection that does not require a switching function.

The valid information detection unit 39 detects the necessity of transmitting packets and data input to the accommodation interface 38.
For example, the detection is performed based on meaningless portions of the transmission such as all 1's, and whether or not the terminal is in use, and this detection can also be performed by the exchange 37 or the multiplexing unit 32. Information used by the multiplexer 32 to multiplex media other than voice channels, such as data terminals and packet switches, is transmission information generated by the accommodation interface 38 (bandwidth may be fixed or variable).
, is valid information for the transmitted information.

As for the multiplexing method, first, the bandwidth required for multiplexing the packets and data input via the exchange 37 and the accommodation interface 38 is subtracted from the number of bits W' available on the line side, and the bandwidth is divided into multiple voice channels. The bandwidth W″ that can be allocated to is determined by the following equation.

W″=W′−(Σ(DAFi ×(Di −Ds−
i))) -ΣDs-i
・・・
(5) In addition, in order to guarantee the minimum sound quality required at the line design stage, the number of lines that can be accommodated is as follows (6).
) is determined by the formula.

W′≧N×Vmin +ΣDi
(6) Audio information of each audio channel is multiplexed using any of the methods in the first to third embodiments described above. Furthermore, as will be described later, the disposal method may be determined based on past history.

In equations (5) and (6), W″ is the number of bits that can be used in the audio channel, N is the number of audio channels to accommodate, Di is the transmission data length of the packet or each data, and DAFi is the data length. Valid/invalid flag, Ds
-i is control information corresponding to each data.

FIGS. 26 and 27 are explanatory diagrams of a sixth embodiment of the transmission method and its operation. In the configuration block diagram of FIG. 26, an A/D converter 391 constituting the audio encoder,
Only the band division filter 392, ADPCM embedded encoders 393 to 396, and multiplexer (MUX) 397 are shown, and the silence detector and the like are omitted.

In the same figure, input audio information is converted into a digital signal by an A/D converter 391, and is divided into, for example, a band B1 of 0 to 1 KHz by a band division filter 392.
B2 at 1-2KHz, B3 at 2-3KHz, and 3-
4KHz B4 and encoded by ADPCM embedded encoders 393 to 396, respectively,
multiplexed by a multiplexer (MUX) 397,
Output to the packet network side.

FIG. 27 shows an example of the discard level in a multiplexing apparatus that performs band division coding. In the figure, when the band is divided into four, the core information section of each band is assumed to be 2 bits and the additional information section is 3 bits for simplicity. As shown in the figure, for example, at level 1, all core bits and additional bits are transmitted without being discarded. Further, at level 5, two core bits are transmitted for each of the bands B1 to B3, but the core bit for the highest frequency band B4 is discarded and not transmitted.

In the present invention, as explained in FIG. 17, for example, the audio transmission band of each audio channel is transmitted in stages, but in the multiplexed frame on the line, the transmission band of each channel is used as a code. Send. This code has different meanings depending on the type of audio encoder and the operating mode settings.

FIG. 28 shows an example of a multiplexed frame when the transmission band is transmitted as a code. In the same figure,
Following the synchronization flag F of the frame, Ri encoded transmission band information for each voice channel is multiplexed, and then the actual transmission information for each channel is multiplexed. Regarding this transmission band code Ri, an error correction code (
ECC) will be added to provide error tolerance.

As described above, it is possible to efficiently transmit transmission band information necessary for accommodating a voice channel in a variable band to the transmission destination. For media other than audio, similar codes are used to notify the transmission band, mainly whether or not transmission is necessary. For media with few states, such as a data channel, it is possible to reduce the number of bits used, and in this case, efficiency can be further improved by filling the empty space with data from a packet switch.

In FIG. 25 showing the fifth embodiment described above, when multiplexing a medium that requires waiting, such as a packet switch among other media other than voice channels, first the minimum throughput required for that medium is multiplexed. It is also possible to secure a bandwidth for that amount, then perform audio multiplexing arbitration, and then allocate the remaining bandwidth to that media, thereby making effective use of the line. In this case, it is also possible to make the above-mentioned minimum throughput allocation to the media variable depending on the required voice band and activity (speech/silence, call status).

Next, in the first to sixth embodiments, when it becomes necessary to discard audio encoded information, the past history or discard weighting information according to the priority of each channel is taken into consideration. This section explains how to dispose of it. First, as a method of changing the weighting between channels, the number of accommodated lines is determined by the following equation so that the minimum required sound quality, which differs depending on the channel, can be guaranteed at the line setting stage.

W′≧Σ(Vmin−i)
......(7) And in actual multiplexing, the audio encoded data length Vi' to be transmitted satisfies the following equations (8) and (9).
is determined.

W′≧Σ(VDFi×(Vi′−Vs))
+ΣVs ×CDFi
・・・・・・(8)
Vi ≧Vi ′≧Vmin −i
・・・・・・(9) (
7) The meanings of the symbols used in equations (9) are almost the same as in the second and third embodiments, but the encoded data length Vmin required for transmission with minimum sound quality
is constant in the first to third embodiments, whereas in this method, the only difference is that it becomes a value Vmin −i for each channel.

FIG. 29 is an example of a method for determining the discard method for the current transmission frame based on the past discard history. As shown in (1) in the figure, when congestion discard occurs, recovery to the requested bandwidth is performed in stages. Also, as shown in (2), during congestion, the bandwidth allocation when changing from silence to sound is lowered. Note that in FIG. 29, arrows indicate the timing of determining band allocation.

FIG. 30 shows an embodiment of an audio multiplexing system to which the present invention is applied. This figure is almost similar to a part of the multimedia multiplex transmission system shown in FIG. 2, and the same parts are given the same numbers. Analog or digital signals are used in the telephones and extension systems from the telephone 1 to the multiplexer 9, and digital codes are used in the relay trunk system from the multiplexer 9 to the multiplexer 11 via the relay transmission network 10. transmission is performed.

FIGS. 31 and 32 show a basic configuration of a multiplexing device and a basic example of processing of a multiplexing unit according to the present invention. FIG. 31 is a configuration diagram of a multiplex transmission system including multiplexing devices, and multiplexing devices 40 and 41 are connected to both ends of a transmission line 42. Each multiplexing device is composed of a codec (CODEC) 43 and a multiplexing unit (MUX/DMUX) 44 corresponding to each audio channel.

FIG. 32 shows a basic example of multiplexing unit processing. In the present invention, an embedded multiplexing method is used in which audio encoded information is separated into core bits and additional bits and transmitted as described above. Core bits are basic information that cannot be discarded, and additional bits are redundant bits that can be discarded when multiple pieces of audio information become congested in the multiplexing unit. Naturally, sound quality will deteriorate if additional bits are discarded, but even in that case, the core bits are still transmitted, so the sound quality of the core bits is guaranteed at a minimum.

Further, since the additional bits are also discarded in stages at the time of congestion, the deterioration in sound quality can be minimized when the degree of congestion is low. Further, when a call is not in progress or during a call, silent portions are subjected to silence compression and are not sent to the transmission path.

In FIG. 32, no-voice compression is performed on channel 3, and its core bits are also not transmitted. Further, the suppression level of the additional bits is adjusted depending on the congestion situation, and the additional bits for channel 2 are not transmitted at all, but the additional bits for channels 1 and 4 are transmitted.

Next, the necessary functions in the statistical multiplexing method according to the present invention, that is, the statistical multiplexing method using activity fluctuations between audio channels transmitted on the same transmission path, will be explained using FIGS. 31 and 32. In the frame format shown in FIG. 32, information such as whether or not each channel is to be transmitted and the transmission band of transmission information for each channel is transmitted in the header information.

First, the activity identification method for each audio channel and its notification route will be explained. The coder (CODER) of the codec 43 in the multiplexing device 40 in FIG. 31 identifies the call connection state, the presence/absence of sound, and the weighting information of the transmission request as activity identification information.
The result is sent to the multiplexer (MU) of the multiplexing unit 44.
Notify X). Here, the call connection state is identified by monitoring signaling information, the voice/silence state is identified by a voice detector (VDET), which will be described later, and the weighting information of the transmission request has a maximum of 8 levels using the SN ratio as a measure. /Create channel-level information.

The multiplexer in the multiplexing unit 44 arbitrates line multiplexing based on the above information. As a result of arbitration, the bandwidth may be reduced more than the requested bandwidth. In addition, all encoder information is sent to the multiplexer, making it possible to transmit more than the required bandwidth in some cases. After line multiplexing arbitration, the multiplexer transmits the call identification result, voice/silence state, and transmission level (code) of the arbitration result passed from the coder to the destination multiplexer via the transmission path 42, for example. The demultiplexer (DMUX) of the multiplexing unit 44 in the multiplexing device 41 is notified.

The demultiplexer notifies the received information to the decoder of the codec 43 in the multiplexer 41 or the next-stage relay multiplexer, and the decoder receives the encoded data passed by the transmission level notification from the demultiplexer. The size of the waste is recognized and the occurrence of waste is detected using this information.

Other necessary functions of the statistical multiplexing method of the present invention include a relay switching function as a system, measures against transmission errors on transmission paths, and monitoring of transmission path utilization efficiency and fluctuations in transmission quality for each channel. functionality is required.

From the above, the first function required when introducing embedded multiplexing is a function to detect redundancy in voice transmission.
This function identifies whether there is a sound or no sound and notifies the multiplexing unit of the required transmission band.

[0075] The second function is an arbitration processing function for embedded multiplexing, which is a function that deletes unnecessary encoded information in the multiplexer in the multiplexing unit and uses the surplus band to accommodate other audio channels. .

The third function is a discard information notification function, which is a function in which the multiplexer notifies the transmission band of each channel between multiplexing units. You will be notified of the content. An error correction code is added to the header as a countermeasure against line errors.

[0077] The fourth function is a logging function, which is realized by a multiplexer and an external console, and it is possible to check the usage status of the line and the fluctuation status of the audio quality of each channel from the external console.

FIG. 33 is a block diagram of an embodiment of a speech encoder. In the same figure, in the voice encoder, in addition to the four wires used for transmitting and receiving voice signals, there are two wires (SS/SR) for transmitting and receiving signaling information such as dial pulses, for example, a six-wire interface 45, SS, and SR. Call detector (CDET) 4 that detects calls by monitoring signals
6. Multiplexing processing unit (M
UX) 47, internal bus interface circuit for transmission 48
, an analog/digital converter 49 that converts audio codes into digital signals, an encoder (CODER) 50 that outputs audio codes separated into core bits that do not allow discarding and additional bits that allow gradual discarding, and A/D conversion. A voice detector (V
DET) 51 is used for transmission.

On the other hand, for reception, there is an internal bus interface circuit 52 for reception, a multiplexing/demultiplexing processing unit (DMUX) 53 for internal bus interface reception, an audio signal decoder (DECODER) 54 for decoding encoded data, and an audio signal decoder 54 for decoding encoded data. A noise mixer (NMIX) 55, a digital/analog converter 56, and an OD line interface 57, which are noise insertion circuits, are used to reduce the sense of discomfort in the silent section.

FIGS. 34 and 35 are explanatory diagrams outlining the sound/silence determination method. FIG. 34 shows a voice detector (VDET) that detects silent sections of speech in the speech encoder of FIG.
) 51 is shown. Note that the A/D converter 570 corresponds to a combination of converters 49 and 56 in FIG. 33.

FIG. 35 is an explanatory diagram of a method for determining voice/silence. The voice detector 51 analyzes the voice data encoded by the A/D converter 570 using a fixed time unit as a break, and determines whether the interval is sound or silent by analyzing the voice data string. Determine whether

FIG. 36 is an explanatory diagram of an embodiment of the sound/silence determination process. In the figure, a high pass filter (HPF) 5 is used to remove DC offset from the transmitted audio signal.
An audio signal is input to 80, and two processes are performed on the high frequency signal.

In the first process, the power for the input signal is calculated in 581, and it is determined in 582 whether or not the power is equal to or greater than the threshold, and if the power is equal to or greater than the threshold, the state determination unit 585 determines whether or not there is a sound. It is determined that

In the second process, the number of zero crossings, that is, the number of times the signal passes through the 0 level, is counted by 583 for a certain period of time, that is, for each subframe in which one frame of data is divided into several subframes. 584, and if the number of zero crossings is greater than or equal to the threshold, the state determination unit 585
It is determined that there is a sound.

In other words, if it is determined that there is a sound in either the first process or the second process, the determination result is that there is a sound, but the first process determines that a loud sound is heard. The second process has the meaning of detecting the beginning of a word that starts with an unvoiced sound such as "ka" or "sa", that is, a word that has a small sound but a high frequency component.

In the above description of FIGS. 33 to 36, it is assumed that the voice detector in the speech encoder performs silence detection, but the multiplexing unit 44 in FIGS. 31 and 32 performs the voice/silence determination. It is also possible. In this case, the multiplexing unit will understand the encoded information and identify the silence of the audio. The simplest implementation method is to incorporate the decoding section of the speech encoder and the aforementioned voice detector in the multiplexing unit, decode the encoded information once, and perform silence detection in the same way as is done in the speech encoder. This is what we do.

Next, a method of implementing the required bandwidth determining section 33 in the second embodiment of the transmission method described with reference to FIGS. 18 to 20 will be described. The voice encoder 30 can also serve as the required band determining section 33, as described above. The speech encoder used in the present invention uses predictive encoding in which the encoding section and the decoding section operate the predictor in synchronization, and the encoding section uses the difference between the input signal and the prediction result of the predictor (
The prediction error) is quantized.

In an encoder based on this principle, the encoder can easily generate a signal equivalent to the signal reproduced by the decoder, and a device that does this is generally called a local decoder. Therefore, on the encoder side, the error power between the input signal and the signal reproduced by the local decoder is calculated for each fixed sample, and a value called segmental SNR, which is divided by the input signal power, is calculated for all the transmitted band levels. The lowest band level that satisfies a certain SNR value is determined to be the transmission requested band. The relationship between this segmental SNR and the transmission band level will be described later.

On the other hand, if the required band determining section is provided independently of the speech encoder, its processing will be similar to that of a silence detector. That is, although the voice detector described above performs a simple 1/0 determination, it is realized by dividing this determination into levels and determining the corresponding band as the required band. For example, it is possible to make a judgment that divides the sound into four levels: complete sound, gray sound, gray silence, and complete silence. In addition, when determining the required bandwidth by the multiplexing unit, the multiplexing unit performs audio decoding processing in the same way as the silence detection described above, and then
This is achieved by performing the same processing as the determination by the speech encoder.

37 and 38 are embedded ADPC
FIG. 2 is an explanatory diagram of the M encoding method. Figure 37 is a normal ADPC
FIG. 2 is a configuration block diagram of an M encoding method. This figure is an example of a conventional high-compression encoding method, in which a subtracter 590 takes the difference between the input audio information and the output of a predictor 593, and the difference is encoded by a quantizer 591. , is output as encoded information. The encoded information is also input by an inverse quantizer 592 to a predictor 593 for adaptive prediction. This method is based on the premise of synchronous operation on the transmitting and receiving sides, and requires silencing processing when data is discarded.

FIG. 38 is a block diagram of the embedded ADPCM encoding method. The figure is compared to the normal ADP in Figure 37.
Compared to the CM method, the bit deletion unit 594 deletes additional bits from this encoded information as the sum of core information and additional information output from the quantizer 591, and only the core bits as a result of deletion are dequantized. The difference is that it is input to the predictor 596 via the predictor 595. That is, the inner loop that generates the prediction signal passing through the predictor 596 runs only on core bits, and the only bits that are actually ADPCM encoded are the core bits. The above “
"Compression encoding of the core information part using difference information" means this inner loop, and the predicted value with the core information is
96 outputs.

In this embedded ADPCM encoding system, the quantizer 591 on the transmitting side has a resolution corresponding to the number of core bits plus additional bits. Further, the inverse quantizer 595 has a resolution equal to the number of core bits.

On the other hand, the receiving side has an inverse quantizer whose resolution improves according to the number of additional bits received.
Decryption is thereby performed. This equivalently means that the additional bits are simply PCM encoded. In other words, the core bits are differentially encoded using the same prediction loop in the encoder and decoder, whereas the additional bits are simply PCM encoded because they are not used in the prediction loop. It turns out that.

FIG. 39 is an explanatory diagram of the audio/data transfer path within the multiplexer. A voice channel activity identification method, its notification route, etc. will be explained with reference to this diagram. First, from ■ to ■, the PBX 60 gives audio information as an analog signal and signaling information as a 10PPS dial pulse from the PBX 60 to the audio LS (line set, terminal accommodating part) 61, and based on these information, the height of the audio signal is determined. Binary encoding of the signaling information by compression encoding, eg sampling at 0.4 Kbps, and voice activity detection are performed.

As mentioned above, voice activities include the identification of calls by monitoring signaling information, the identification of voice/silence by a voice detector, and the SN detection by a coder.
There are three types of weighting information for transmission requests created using the ratio as a scale. The method of identifying a telephone call by monitoring the signaling information is as described with reference to FIGS. 21 to 23.

Next, from ■ to ■, the audio LS61 to MUX
The activity information detected in 64 is notified via the relay exchange units (EXUs) 62 and 63 from ■ to ■ using a code format (Ri code) to be described later.

[0097] The MUX 64 arbitrates line multiplexing based on the activity information (Ri code) passed corresponding to each channel, and sends the result in 3-bit code format to D.
Notify MUX. As will be described later, the transmission band allocation code has a total of 4 bits, and the remaining 1 bit is used for transmission compression of signaling information. In this case, the information passed from the voice LS 61 is used to identify the voice call and the voice/silence state, and the transmission level is determined by the arbitration result at the MUX, via lines 65 and 66 from ■ to outside 1.

[
0098

[Outside 1]

[0099] Output 2 is sent to the receiving side MUX'64.

[0100]

[Outside 2]

[0101] The DMUX in the MUX' on the receiving side transmits the transmission information based on the arbitration result from outside 3.

[0102]

[Outer 3]

0103 Outside 4, Outside 5 ~ Outside 6
Audio LS61 via EXU62,63, and

[0104]

[Outside 4]

[0105]

[Outer 5]

[0106]

[Outside 6]

[0107] is sent to the next stage relay MUX. Audio LS61
Based on the activity information sent from the DMUX, recognizes the size of the audio encoding information passed from the source audio LS, and performs audio decoding processing based on the recognition result. At this time, the presence or absence of discard by the relay stage is also detected based on this information. And the decryption result from outside 7

[0108]

[Outside 7]

[0109] Send to PBX 60 with Out 8.

[0110]

[Outside 8]

[0111] In addition, the CODEC in FIGS. 31 and 32
43 is located within the audio LS61 in FIG. 39, and the audio encoder in FIG. 33 corresponds to the contents of the audio LS61.

[0112] Within the multiplexing device, there is a node control section 67 that controls a plurality of MUXs 64, and this node control section has a control section 67 that controls the nature of sound quality deterioration caused by discarding additional bits, which depends on the nature of the encoder for each audio channel. The considered multiplexing parameters are stored in advance, and the parameters are set in the MUX 64 when the system is started up. This parameter is based on the characteristics of the encoder, such as the fact that there is little deterioration in sound quality even if only core bits are sent when restoring speech from silence. Furthermore, within the multiplexer, a data LS 68 for a data channel and an interface 69 for a packet switch 70 are provided.

[0113] Here, a method of implementing the effective information detecting section 39 in the fifth embodiment of the transmission system shown in Fig. 25 will be explained. Here, it is simply determined that the data in the multiplexing processing frame period is all '1's as invalid, and only a flag indicating that the data in that frame will not be transmitted (an idle flag to be described later) is transmitted to the receiving side, and the receiving side Next, an example will be described in which transmission of unnecessary information is suppressed by checking the presence or absence of transmission using an idle flag and generating all '1' information for frames that are not transmitted.

In FIG. 39, the valid information detection section is incorporated into either the data LS 68 or the multiplexing unit 64. When detecting with data LS68, data for multiplexed processing frame 1 and cycle is written once to memory or shift register, and all written values are '1'.
This can be realized by providing a comparator that determines whether or not '. Alternatively, a processor is installed in the data LS68, and all the data for one cycle of the multiplexed processing frame is '1'.
This can also be realized by a method of determining whether or not.

Next, when detecting with the multiplexing unit,
As in the data LS, it can be detected by providing memories or shift registers and comparators for the number of data channels, or it can be determined by software using a processor 77 in FIG. 45, which will be described later. If the latter method is used, the hardware scale will be smaller.

FIG. 40 shows an example of a transmission band allocation code. In the figure, the transmission band allocation code is 4 bits, but the lower 3 bits are used to transmit coded transmission band information (RI2 to RI0) for the voice channel, and the most significant bit b3 is the signaling transmission flag. Used for transmission. This signaling transmission flag is set to 1 when signaling information is transmitted, and set to 0 when compressed. Signaling information is transmitted only for frames in which the signaling information has changed, and if the signaling information does not change, the receiving side
The MUX outputs the last received signaling information to the voice channel.

For the data channel, a remote signaling (RS) signal transmission flag is transmitted in the most significant bit as 1 during transmission and 0 during compression. Its meaning is exactly the same as the signaling transmission flag. An idle flag is transmitted to the least significant bit b0 to indicate whether or not it is necessary to transmit the data area, and its value is set to 1 during transmission and 0 during compression. Furthermore, the intermediate bits b1 and b2 are filled with data from the packet switch.

FIG. 41 shows an example of a code table for transmission band information. In the figure, transmission band information RI2 for each audio channel that is multiplexed and transmitted as a result of arbitration.
~RI0 is converted into a 3-bit code and transmitted depending on whether the voice call is disconnected, when there is no voice, and when there is a voice call (6 levels). The transmission level corresponds to levels 1 to 7 every 2 kbps, for example.

Next, as mentioned above, for example, in a speech encoder,
Figure 4 shows the determination algorithm when determining the transmission request level using segmental SNR by a local decoder.
This will be explained in relation to level 1.

In this determination algorithm, first, the segmental SNR corresponding to each transmission band level 2 to 7 for each multiplex frame period, that is, SNR(2) to
SNR (7) is calculated, and then SNR judgment threshold SNR (TH),
For example, when a level higher than the threshold value is found by comparison with 25 dB, that level is set as the transmission band request level. If all segmental SNRs are smaller than the threshold SNR (TH) at the time of comparison up to level 7, level 7 is set as the transmission band request level.

FIG. 42 is an example of serial data multiplexed onto a 64 kbps line. In the same figure, 1
A frame consists of 640 bits and its transmission time is 10ms. And those 640 bits are 160
It is divided into four phases 0 to 3, each consisting of a bit, and the transmission time for one bit is 15.6 μs.

FIGS. 43 and 44 show examples of frame configurations. FIG. 43 shows the overall structure of a frame. One frame consists of four phases, 0 to 3, and the first 11 bits of each phase,
A total of 44 bits are used as a header section, and a total of 596 bits from the 12th bit to the final bit of each phase, that is, the 160th bit, are used as a data multiplexing section.

FIG. 44 shows the structure of the header section. In the figure, the header section has 11 bits for each phase of 0 to 3, and the first bits F1 to F3 are used as a frame synchronization pattern, and SEND is used as a device alarm bit. 2nd to 9th bit of each phase
When multiplexing 8 channels, a total of 32 bits up to bit 4 are transmission band allocation codes Ri of 4 bits each.
(#k) or packet PK from a packet switch
Used for T transmission. The 10th and 11th bits of each phase are used as a storage area for the error correction code ECC for the transmission band allocation code Ri (#k).

FIG. 45 is a block diagram of the configuration of an embodiment of the multiplexing unit. In the figure, the multiplexing unit includes a bus interface circuit 71 as an interface with each channel, a bus interface circuit 72 as a line interface, and a two-port RAM into which transmission request information from each channel is input and multiplexed frame information is output.
73, write control unit 74 for two-port RAM 73;
RAM 75 for storing programs and data, input interface circuit 76 for packet channels, arithmetic processor 77 for multiplexing and demultiplexing processing, output interface circuit 78 for packet channels, two-port RAM 79 for received data, writing to two-port RAM 79 Establishment of frame synchronization between the control unit 80, the multiplex frame aligner unit 82 as a buffer circuit for synchronizing the transmission request information from each channel with the multiplexing cycle, the MSYNC unit 83 as a cycle timing detection circuit, and the MUX. It is composed of a frame aligner 81 as a circuit and a SYNC section 84 as a frame timing detection circuit.

In FIG. 39, the node control section 67 sets the multiplexing parameters for each multiplexing unit 64 as described above, and this setting is performed through the control bus for the package of each multiplexing unit 64.

Although such a control bus is omitted in FIG. 45, specifically, for example, a register in which setting information from the node control section 67 is written is connected to the memory bus of the processor 77, and the multiplexed program A conceivable configuration method is that the processor 77 executing the program reads the contents of the register as necessary.

FIG. 46 is a flowchart of a schematic processing embodiment of the multiplexing unit. In the figure, multiplexing (
MUX) processing is performed. In the multiplexing process, the transmission request information for each channel is received in S86, that is, read out from the two-port RAM 73 in FIG. 45, the audio channel band arbitration process is performed in S87, and a transmission band allocation code (Ri code) is created in S88. An error correction code is created in S89, a transmission path frame is created in S90, and the transmission path frame is transmitted, that is, written into the two-port RAM 73 in S91.

[0128] In the received data separation (DMUX) process, the transmission path frame is received in S92, that is, as shown in FIG.
The data is read from the two-port RAM 79, and correction processing using an error correction code is performed in S93. In S94, the transmission line frame is decomposed into data for each channel, and in S95, the received information and transmission band allocation code for each channel are (Ri
A notification frame for the code) is created, a reception notification frame for each channel is sent out in S96, that is, written into the two-port RAM 79, and a timer is set until the next processing cycle in S97.

[0129] The bandwidth arbitration process in S87 is shown in FIG.
5 processor 77. The arbitration processing program is stored in the RAM 75 or built into the processor 77 as a mask ROM.

FIG. 47 is an explanatory diagram of the processing cycle. The processing cycle T is set to a constant value, such as 5 ms or 10 ms, and the timer times out at every fixed cycle. During this cycle, multiplexing (MUX) processing and demultiplexing (DM) processing are performed.
UX) processing is performed, and the remaining time is idle, that is, waiting for the timer to time out.

FIGS. 47 and 48 are flowcharts of the first embodiment of band arbitration processing during multiplexing. 47 and 48 show the first embodiment of the transmission method shown in FIGS. 16 and 17, and FIG.
This is a band arbitration process corresponding to the second embodiment shown in FIGS. 8 to 20, the third embodiment shown in FIGS. 21 to 23, and the like. Here, as explained in FIGS. 13 and 14, the side information during the processing cycle T is 10 bits, the core information is 20 bits, the total core information part is 30 bits, and the additional information part is 10 bits each for a total of 50 bits. shall consist of.

[0132] The transmission path speed is 64 kbps, the multiplex frame generation period T is 5 ms, the number of accommodated channels is 8 channels, the frame format is 4 flag bits, transmission band allocation code 4 x 8 bits = 32 bits, and error correction code. ECC 8 bits, signaling information 2x
It is assumed that the number of bits BW that can be allocated to the audio channel is 260 bits, excluding 8 bits = 16 bits. Furthermore, the encoded data length Vmin required to transmit audio with the minimum quality is 30 bits.

That is, the initial value (required band) of the audio encoded data length Vi (corresponding to Vi in equation (1)) of each channel actually transmitted is as follows in the first embodiment shown in FIGS. 15 to 17. There are only two levels: 10 bits when there is no sound and 80 bits when there is sound. In the second embodiment shown in FIGS. In the embodiment 3, when the call is disconnected, the bit is 0, when the call is connected and there is no sound, the bit is 10, and when there is a call, the bit is 30.
It is variable in ~80 bits.

When the process starts in FIG. 48, first, in step S100, the initial value of the audio encoded data length of each channel, for example, the total sum of the required bandwidth Vi from each audio channel, is taken for eight channels, In S101, the total sum is subtracted from the number of bits BW that can be allocated to the audio channel, which is 260 bits in this case, to obtain LBW. Then, in S102, it is determined whether the value of LBW is positive or 0, and if it is positive or 0, all the requested bandwidths are satisfied, so the process ends without performing arbitration.

[0135] If LBW is negative in S102, it means that the band is insufficient, so first, in S103, channel j, for example, the channel next to the channel containing the last bit discarded by arbitration during the previous frame transmission, is It is determined whether the audio encoded data length for the channel is smaller than or equal to the encoded data length Vmin for transmission with minimum audio quality. If it is smaller than or equal to Vmin, no more data for that channel can be discarded, so the value of j is incremented in S105, and the process from S100 is repeated.

If Vj is greater than Vmin in S103, 10 bits of the audio data for that channel are discarded in S104, and the value of j is incremented in S105, and then the processing from S100 onwards is repeated. And S102
All processing ends when LBW becomes 0 or positive.

FIG. 49 shows an example of arbitration using the flowchart of FIG. 48. In FIG. 49, T=1 to 15
indicates the arbitration timing when creating a multiplexed frame. First, when T=1, the requested bandwidth from all channels 1 to 8 is all 10 bits, which corresponds to a silent state, and the requested 10 bits are allocated to each channel.

[0138] At T=2, the required bandwidth from each channel is all 50 bits, and the total of 400 bits is larger than the above-mentioned BW of 260 bits, so data must be discarded. For example, if discarding starts from channel 1, first 10 bits are discarded for each of channels 1 to 8. Additionally, 10 bits are discarded once again from channel 1 to channel 6. As a result, a total of 140 bits are discarded and arbitration ends.

[0139] After T=3, arbitration is performed in the same way, but at T=3, audio data is discarded on channel 7,
That is, channel 6 where the last data was discarded at T=2
Channels 7 and 8 and channels 1-
A total of 40 bits are discarded for channel 4, and a total of 30 bits are discarded for channels 5 and 6, and the arbitration ends.

[0140] Hereafter, arbitration is performed in the same manner, but T=
The reason why discarding is stopped at 30 bits for channel 8 in T=9 is that 30 bits are core bits, and the reason why discarding is not done for channel 8 at T=9 is when there is no sound. This is to avoid throwing away side information.

FIGS. 50 and 51 are flowcharts of the second embodiment of bandwidth arbitration processing. Figures 50 and 51 are Figure 2
This is an example of processing for determining the discard method for the current transmission frame from the past discard history explained in Section 9. Figure 50
, in FIG. 51, the core information section, additional information section, number of accommodated channels, number of bits that can be allocated to audio channels, etc. are all the same as in the first embodiment shown in FIG. 48.

When the process starts in FIGS. 51 and 51, first, in S106, i indicating the channel number is set to 1, and in S1
In step 07, a subroutine DISCMD (discard mode renewal subroutine, discard mode update subroutine), which will be described later, is executed as a discard mode update process.
After that, the value of i is incremented in S116, and it is determined in S117 whether the value of i is smaller than 8, and if it is smaller, S1
The process from 07 is repeated.

FIG. 52 shows S107 in FIGS. 50 and 51.
, that is, a flowchart of an embodiment of the discard mode update process. 53 to 55 are examples of arbitration using the flowcharts in FIGS. 50 and 51, and FIG.
, is an example of the relationship between the values of each variable in the bandwidth arbitration process according to the flowchart of FIG. The flowcharts in FIGS. 50, 51, and 52 will be explained using the embodiments in FIGS. 53 to 55 and 56.

FIG. 53 is a diagram 49 for the first arbitration embodiment.
This is a diagram similar to , and shows the required bandwidth for each channel and the bandwidth allocated as a result of the arbitration process at each arbitration timing of T=1 to 15. In the first line of the display for each channel at each arbitration timing, the left side is the requested band and the right side is the allocated band. Further, the leftmost line of the second line indicates a discard mode flag DF (discard mode flag), which will be described later, the center indicates a discard damage variable DMG (damage factor), and the rightmost line indicates a margin GAP (gap factor) for the minimum allocated band. These will be detailed later.

[0145] On the right side of these arbitration results for channels 1 to 8, the overall required bandwidth and allocated bandwidth are shown. The right side of the requested bandwidth is the total request from the encoder for each channel, and the left side is the total requested band considering the discard mode to be described later.

For example, at T=1, the total 80 bits requested from the encoder for each channel matches the total required bandwidth considering the discard mode, but at T=2, (2) in FIG. As explained above, in order to recover from a silent state to a sound state, the bandwidth is allocated to a lower level. Here, in order to allocate only 30 bits of the core information part to each channel immediately after voice restoration, that is, at T=2, the requested bandwidth from each channel is set to 30 bits of the core information itself.
The total is 240 bits.

[0147] The discard mode flag DF, the discard damage variable DMG, and the margin GAP for the lowest allocated band in FIG. 53 indicate the values after the arbitration process has been performed through the processes of FIGS. 50, 51, and 52. . In addition, GAP=
9 indicates that GAP is a negative value. Determination of these values will be explained using FIG. 54 together with the processing in the flowchart.

FIG. 54 shows how much bandwidth is actually allocated in response to a request from a certain channel encoder for arbitration timings T=1 to 11. Note that the channel here corresponds to one channel in FIG. 53.

At arbitration timing T=1 in FIG.
In S107 of FIGS. 50 and 51, the discard mode update process is performed. That is, in S108 of FIG.
= GAP is determined using 80 bits. Here, it is assumed that the encoded data length Vi min necessary to transmit audio with the minimum quality is 30 bits as in FIG. 48. As a result, GAP becomes 5, and it is then determined in S109 whether or not the requested bit Vi is smaller than Vi min .

[0150] In this case, since Vi is larger, S
At 110, it is determined whether the discard mode flag is 0 or not. Here, the discard mode flag DF is a flag that becomes 1 when discarding, and becomes 0 when not discarding. In this case, it is the first arbitration timing, and if it is assumed that the discard mode has not yet entered, the value of the flag remains 0, and the discard mode update process ends and the process moves to S116 in FIGS. 50 and 51. . If it is determined in S117 that the discard mode update process has been performed for eight channels, the process moves to S118 and subsequent steps.

The processes from S118 to S124 in FIGS. 50 and 51 are essentially the same as those in FIG. 48 showing the first arbitration process embodiment. In other words, in response to requests from each channel, if the total is greater than the number of bits BW that can be allocated to the voice channel, the bandwidth of each channel is discarded by 10 bits by arbitration processing, and the value of LBW is 0 or positive. At the point in time, the process moves to S125. However, unlike FIG. 48, for a channel in which 10 bits have been discarded in S122, the value of the discard damage variable DMG is incremented in S123.

After the channel number i is set to 1 in S125 of FIGS. 50 and 51, it is determined in S126 whether or not the value of DMG is positive. At the time of T=1 in FIG. 54, the arbitration process has just started, and at this time the value of DMG is 0.
If so, S128 is executed without going through S127.
It is determined whether the value of DMG is 0 or not. Here, next in S129, the discard mode flag DF is set to 0, and in S1
The value of channel number i is incremented in step S131, and i becomes 8 in step S132.
It is determined whether it is smaller than 8, and if it is smaller than 8,
The process from 126 is repeated, and when the value becomes equal to 8, the process ends. Here, it is assumed that the requested bits for this channel are not reduced at all in the processing from S118 to S124, and as a result, the bits Vi' allocated to this channel are 80 bits.

At arbitration timing T=2 in FIG. 54, there is a silent section, and the requested bits are 10 bits. the result,
In S108 of FIG. 52, the value of GAP becomes -2, in S109 it is determined that the requested bit is smaller than Vi min, and in S114 it is determined whether or not to refer to the change history from silence to sound. Here, (2) in Figure 29
As described above, in order to lower the bandwidth allocation when changing from a silent state to a sound state, the value of DMG is changed in S115. Here, if the value of Vi max is 80 bits, the value of DMG is 6, and S1 in FIGS. 50 and 51
The process moves to step 16.

In the processing from S118 to S124 in FIGS. 50 and 51, the allocated bits for this channel are not reduced from the requested bits, and in S126 the DM
It is determined whether G is positive. At this point, DMG
is 6, that value is subtracted by 1 in S127, it is determined in S128 that DMG is not 0, and S130
Then, the discard mode flag DF is set to 1. Note that the DMG value determined in S127 is the next DMG value in FIG.

[0155] The processing at arbitration timing T=3 is T=
The process is similar to that in 2. T=4 and requested bit is 8
When the bit becomes 0, the value of GAP in S108 of FIG. 52 becomes 5, and the GAP and DMG are compared in S111 after the processes of S109 and 110. In this case, since GAP and the previous DMG are equal, the process of S112 is performed. After S1
After Vi is set to 30 bits in step 13, Figures 50 and 5
The process moves to step S116 of step 1.

In the processing from S118 to S124 in FIGS. 50 and 51, since Vi is already 30 bits, no further discard occurs, and S125 and S12
After the process of 6, DMG is set to 4 in S127, and S130
DF is set to 1, and the processing for this channel ends.

[0157] Even after T=5, all required bits are 8.
0 bit, but the allocated bit is 10 until T=8.
The requested bits and allocated bits finally match at T=9 by only increasing bit by bit. That is, T
T=4 to T=8 are periods in which the discard history is dragged for the silent sections of T=2 and T=3. During this period, the value of DMG is 1
The value gradually decreases, and when the value reaches 0, the period for dragging the discard history ends. That is, the variable DMG is used to convey the degree of past discard occurrence to the next frame and subsequent frames.

FIG. 55 is a diagram illustrating how the history is dragged when congestion abandonment occurs. That is, at T=1, the requested bits are assigned as is, similar to T=1 in (b), but at T=2, after the GAP value is set to 5 in S108 of FIG. 52 for a request of 80 bits, ,
In FIG. 52, it is assumed that no bits are discarded, but 40 bits are discarded due to congestion in the processing from S118 to S124 in FIGS. 50 and 51, and the allocated bits become 40 bits. At this time, the value of DMG is increased by 4 for the discard of 40 bits in S123 compared to the previous DMG=0, and DMG becomes 4. This is the final DMG. Then, DMG=3 after being subtracted by 1 in S127 becomes the next DMG value.

[0159] Even after T=3, the required bit is always 8.
However, at T = 3, 4, and 5, the next DMG is subtracted by 1 and transmitted to the next timing, so the allocated bit is 1 at T = 3 to 5.
It is incremented by 0, and at T=6, the requested bit becomes the assigned bit as is.

FIG. 56 is an example showing how each variable in the flowcharts of FIGS. 50 and 51 is determined according to the request bit and the previous DMG value. Figure 50
, in FIG. 51, only the vertical columns have meaning, and the horizontal arrangement is completely irrelevant.

First, in the leftmost case, the requested bit is 40.
bit, when the previous DMG is 0, and GAP is as shown in Figure 52.
It becomes 1 in S108. In FIG. 56, the discard mode flag D
Assuming that all F are 1, GAP and DMG are compared in S111, but here GAP is compared to DMG.
Since it is larger than , the value of DMG does not change and remains 0. This value is D shown below GAP in FIG.
This is the value of MG. Then, the value of Vi remains unchanged and the process proceeds to S116 in FIGS. 50 and 51. The value of Vi at this time is the new Vi.

[0162] After that, S118 to S124 in FIGS. 50 and 51
Congestion discard is performed in , but in this congestion discard, it is assumed here that the number of bits that can be allocated to this channel is 40 bits, but this value is different from the actual Vi
It is shown as. As a result, the requested bits are not discarded and become the final Vi, that is, the allocated bits, and the final DMG, next DMG, and DF are all 0.
The process ends.

The second column from the left shows the case where the requested bits are 40 bits but the previous DMG was 2. At this time, GAP is set to 1 in S108 of FIG.
DM in S112 because DMG is larger than GAP in 11
G is set to 1, and in S113 Vi, that is, the new Vi is set to 3.
The bit is set to 0, and the process moves to S116 in FIGS. 50 and 51. As mentioned above, in the processing from S118 to S124, it is assumed that the allocatable bits for congestion are 40 bits for this channel, but since Vi is already 30 bits, that value is set as the final Vi.
Therefore, the process moves to S125 and subsequent steps. and S1
In step 27, the value of DMG is set to 0, which becomes the next DMG, and the value of DF also becomes 0.

In the fifth column from the left, the previous DMG is 2 for the requested bit of 70 bits, so after the new Vi value is set to 50 bits in S113 of FIG. Congestion discard is performed in the processing after S118 of 51. As mentioned above, this channel has 4
Since only 0 bits are allocated, 10 bits are discarded in S122, and the DMG, final DM
G is assumed to be 3. Thereafter, in S127, the value of DMG is set to 2, and as a result, DF becomes 1, and the process ends.

[0165] In the bandwidth arbitration processing explained with reference to FIGS. 48 to 56, the problem is how to follow the past discard history. The mode setting for how to drag the discard history is part of the multiplexing parameters set in each multiplexing unit 64 by the node control section 67 in FIG.

[0166] The modes for dragging the discard history are: first, a mode in which the discard history is not dragged, that is, a mode in which arbitration is performed as shown in FIG. 48, and second, a mode in which the discard history is dragged,
In addition to the mode in which recovery is performed one step at a time for each arbitration, that is, the mode in which arbitration is performed in FIGS. 50 and 51, it is possible to consider, for example, a third mode in which recovery is performed one step at a time for every five arbitrations. can. In this mode, Figure 51
The subtraction process of DMGi in S127 is performed once every five times.

FIG. 57 is a basic configuration block diagram of an embodiment of a packet switching network using the high efficiency digital multiplex transmission system of the present invention. In FIG. 57, the multiplexing device shown in FIG.
The configuration is exactly the same as that of , and there is a packet interface unit (PAD) 140 that assembles and disassembles packets between the multiplexing device 135 and the packet network 137a, and between the packet network 137a and the multiplexing device 136. Conversion between frames and packets is performed.

FIG. 58 is a block diagram of the configuration of an embodiment of the packet interface section 140 in FIG. 57. Figure 5
At 8, a packet header is added to the packet from the multiplexer side by the header adding section 141, and after being assembled into a packet, the packet is outputted to the packet network side via the speed absorption buffer 142. Also, for packets input from the packet network side, fluctuations due to packet exchange are absorbed by the arrival fluctuation absorption buffer 143, packet headers are removed by the header removal section 144, and the packets are output as multiplexed frames to the multiplexing section side. .

FIG. 59 shows the speed absorption buffer 142 of FIG.
FIG. In FIG. 58, since a header is added by the header addition unit 141 to the multiplexed frame input from the MUX side at a speed V1, the transmission speed V2 to the packet network side is not the same as V1, but is uniform. When transferring data at high speed, V2 is larger than V1. Speed absorption buffer 1 absorbs this speed difference.
It is 42. Note that there is also a method of transferring packets to the network side in bursts when packets have accumulated and can be transferred, but in such a case, the speed absorption buffer 142
is used as a transfer waiting buffer. Note that FIG. 59 shows a transfer state at a uniform speed.

FIG. 60 is an explanatory diagram of the operation of the arrival fluctuation absorbing buffer 143 similarly shown in FIG. 58. Even if packets are sent from the sending side to the packet network at a constant rate, the arrival delay time of the packets arriving at the receiving side is not constant and varies. The arrival fluctuation absorbing buffer 143 absorbs variations in the arrival delay time, and the output from this buffer to the header removing unit 144 is at a constant speed.

FIGS. 61(a) and 61(b) are examples of packet formats. In FIG. 61(a), only a header section is added to each multiplexed frame, and the frame is converted into a packet as it is and output to the packet network side. In FIG. 61(b), each multiplexed frame is divided into multiple packets, a packet header is added to each packet, and an identifier for reassembling the multiplexed frame is inserted into the header. output to the side.

In FIG. 61(a), for example, the multiplex frame period is 5 ms, the multiplex frame length is 48 bytes, and the execution throughput to the packet network side is 76.8 k.
bps, and the packet format is the ATM cell format, with a header section of 5 bytes and a data section of 4 bytes.
It is 8 bytes.

In addition, in FIG. 61(b), for example, the multiplex frame period is 5 ms, the multiplex frame length is 96 bytes, and the execution throughput to the packet network side is 153 bytes.
．． It becomes 6 kbps. The format of the packet is the same as in FIG. 61(a), with a header section of 5 bytes and a data section of 48 bytes. Here, the multiplexed frame is divided and transmitted, but the multiplexing unit has a function to synchronize the multiplexed frame, and the packet interface section on the receiving side does not need to assemble the frame; There is no need to add extra control bits.

[0174] Fig. 62 shows the case of band division encoding (Fig. 26, Fig. 2
7) is an example of a code table of transmission band information. This transmission band code is, for example, 4 bits for each channel, and the lower 3 bits RI2 to 0 are used by the multiplexer (MUX) on the transmitting side and the demultiplexer (MUX) on the receiving side.
DMUX). The upper 1 bit is used, for example, to notify a discard mode, which will be described later.

As shown in FIGS. 26 and 27, if the side information is separated from the information of each divided band, each band B1
The number of bits transmitted at each discard level for ~B4 is indicated by a number in parentheses. For example, for discard level 1, the number of transmission bits for B1 to B4 is all 5, the transmission level at this time is 40 kbps, and the transmission band information code is 110, and for discard level 5, B1 to B3 The number of transmitted bits is 2 for B4, 0 for B4,
The transmission level is 12 kbps and the code is 010.

FIG. 63 is a flowchart of an embodiment of multiplexing processing in a multiplexing apparatus that performs band division encoding. In the same figure, in S (step) 170, the minimum band, ie, Vmin, for each audio channel is set. This minimum band differs depending on the number j of each channel, and in order to reflect the past discard history, the minimum band for a channel that changed from a silent state to a sound state and a channel whose previous discard level was 5. is set to discard level 5, and channels whose previous discard levels were 1 to 4 are set to discard level 4.

[0177] Next, in S171, the required voice bandwidth is calculated; for example, in FIGS. 13 and 14, the sum of the required transmission bandwidths of each channel excluding the portion of the 50 additional bits that does not need to be transmitted depending on the voice channel is calculated. is,
In S172, the remaining bandwidth within the multiplexed frame is calculated. Then, in S173, it is determined whether or not there is a surplus band, and if there is a surplus band (including 0), the requested bandwidth of each channel can be multiplexed, and the process ends.

[0178] If there is no remaining bandwidth in S173, S1
At 74, the presence or absence of discardable channels is determined. If there is no channel that can be discarded, the lowest band of each channel is lowered to discard level 5 in S175, and if there is a channel that can be discarded, the lowest bandwidth of each channel is lowered to discard level 5, and if there is a channel that can be discarded, the process in S1 is performed without going through the process of S75.
The process moves to step 76. In S176, the discard level is increased by one level for channels that can be discarded, and the process returns to S171. Thereafter, the process continues until it is determined in S173 that there is a surplus bandwidth, and the process ends at the time when it is determined that there is a surplus bandwidth.

FIGS. 64 and 65 show examples of discard modes corresponding to the frequency characteristics of the speaker's voice. For example, Figure 26,
In FIG. 27, encoders 393 to 396 for each of the bands B1 to B4 measure the power of the audio for a certain period of time, and calculate the average value. FIG. 64 shows, for example, the spectral distribution of voice power by band for a male speaker. This distribution is normalized using the power in band B1. FIG. 65 shows the spectral distribution of speech for a female speaker. The importance of each band is determined using these spectral distributions with a certain threshold value (dashed line) as the boundary, and in FIG.
5, we will use discard mode 2, which emphasizes bands B1 and B2.

FIG. 66 is an example of a code table for transmission band information when the discard mode is specified. In the figure, in addition to the above-mentioned side information, 1 bit is additionally used as side information for identifying the discard mode and 1 kbps is used as the transmission level when there is sound.

In FIG. 66, for example, at discard level 5, mode 1, that is, each band B1 to B4 for a male speaker
The number of transmission bits for B1 is the same as mode 2, but in discard level 4, unlike mode 2, the number of transmission bits for B1 is 4 in mode 1, whereas the number of transmission bits for B4 is 0 bits. It becomes.

FIGS. 67 and 68 are flowcharts of an embodiment of the process of determining the discard mode and notifying the receiving side of the mode. FIG. 67 is a flowchart in the case where the sending side notifies the discard mode, and after the receiving side returns acceptance of the change, discarding using the new mode is performed. In the same figure, S1
Before measuring the audio power of each band in step 78, the audio encoder notifies the multiplexer of the predetermined default discard mode, and in step S179, the multiplexer notifies the demultiplexer of the receiving side of the default discard mode. Be notified.

[0183] In S180, the transmitting side coder measures the power of each band in the sound section, determines the discard mode based on the speaker type in S181, and notifies the multiplexer of the discard mode in S182.

[0184] The multiplexer notifies the receiving side demultiplexer of the discard mode in S183, receives a change acceptance notification from the receiving side demultiplexer in S184, and notifies the receiving side demultiplexer of the start of the change in S185. Discard the new mode.

FIG. 68 is a flowchart when the discard mode is notified as audio side information. In this case, the discard mode is changed without receiving a change acceptance notification from the receiving side demultiplexer. First, the default mode is notified in S186, and S
In step 187, the power of each band in the sound section is measured, and in step S18
The discard mode is determined in step S189, and the discard mode is notified to the receiving side as voice side information in step S189.

The method of determining the discard mode explained in FIGS. 64 to 68 will now be further explained. Figure 67, Figure 68
The power measurement of each band of the sound section as the process of S180 or S187 is performed as follows.

First, the power of each band is measured for n samples at a speech/non-speech determination period, for example, every several ms to several tens of ms. The power for each determination period in each band is given by the following equation as an input signal component of the band targeted at Sk.

[0188]

[Math 1]

Here, i is increased by n. Next, in order to calculate the average power only when there is a sound for each band, the power calculation result using the above formula is calculated only when there is a sound within the judgment period, for example, 1000 cycles (replaced as the sound/silence judgment period). Accumulate the
Divide the result by the number of accumulations. That is, the average power during voice presence is determined by the following equation.

[0190]

[Math 2]

[0191] Here, VDFi is a sound presence flag that becomes '1' when a sound presence is determined. Next, the discard mode determination in S181 or S188 is performed based on the calculation of the frequency characteristic distribution as explained in FIGS. 64 and 65. be exposed. This spectral distribution is determined by setting the average power during active periods for band B1 as 1, and dividing the average power during active periods for band B2 by the average power during active periods for band B1, for example. Although FIGS. 64 and 65 explain a method of determining the spectral distribution using a constant threshold value for all bands, in reality, the threshold value may be changed for each band.

FIG. 69 is a block diagram of an embodiment of a system for compressing transmission by making a channel in a reserved state silent. In the figure, when the telephone is on hold, the transmission band is compressed without transmitting hold music as a silent state. The on-hold status is inserted, for example, into the aforementioned Ri code and notified to the receiving side.

In FIG. 69, on the sending side, when the hold button is pressed on the telephone, the hold button is pressed directly, or when the PBX detects the hold tone pattern issued by the telephone, the hold is placed on ■, or on the input side of the voice encoder (coder). When a sound pattern is detected, the result is detected by the hold state detector (R-DET) 191 at ■, and on the receiving side, the hold sound superimposer (R-MIX
) 192, the hold tone is superimposed on the output side of the voice encoder (■), the PBX (■), or the telephone set (■). To release the hold state, change the hook state on the phone,
■ by the hold release button, or by monitoring the signaling signal on the input side of the PBX or coder, the hook status changes, or by a mismatch with the hold music pattern.
or ■, the R-DET 191 is notified, and thereafter the audio is transmitted to the receiving side using the method described above.

In FIG. 69, the pending state detector (R-D
ET) can also be embedded within a PBX, COD, or MUX block. FIGS. 70 and 71 show an embodiment in which this pending state detector is built into the exchange. In FIG. 70, a pending state detector R-DET installed in the exchange
outputs a hold music detection signal when detecting a hold state.

FIG. 71 is an explanatory diagram of a pending state determination method. The method for determining this is that if the called party remains on-hook for a specified time while the call is being connected, it is determined to be on hold, and if the called party remains on-hook for a specified time while on hold, it is determined to be released from hold. . An example of this stipulated time is:
For example, a period between 50 ms and 1 second is considered. If the calling party goes on-hook during a call connection, the call will be disconnected on the spot, but there are many regulations that require the called party to be on-hook for about 30 seconds before the call can be disconnected, which is why this type of hold operation is required. is possible.

FIG. 72 shows an embodiment in which the pending state detector R-DET is built into the multiplexer. In the figure, a signaling signal is input from the exchange to the hold state detector,
The pending state detector (R-DET) detects the pending state in a manner similar to that in FIG.

FIG. 73 is a block diagram of the overall configuration of a system in which a multiplexer accommodates various transmission media. In the figure, in addition to a telephone, a modem to which a terminal is connected, and a FAX are connected to a PBX in addition to a telephone. The output of the multiplexer (MUX) within the multiplexing device is then output to the transmission line via the network port (NP). In this case, for example, F
It is not possible to specify which coder AX is connected to via the PBX.

In other words, in FIG. 73, all the information corresponds to core information depending on the transmission medium, and there is no additional information, so the encoder that encodes the information sent from such a medium is It is necessary to encode the requested bandwidth without discarding it as is. For this purpose, it is necessary to detect, for example, a FAX protocol, and a coder to which the detection signal is input must perform encoding without discarding the input information.

FIGS. 74 to 76 are block diagrams of an embodiment of an encoder that performs encoding according to the FAX protocol detection result. FIG. 74 is an example in which a coder receives a detection signal from a FAX protocol detector (not shown) and sends an encoded signal to a multiplexer without discarding the requested band.

FIG. 75 shows that the coder itself detects the FAX protocol, and when the input signal is a signal from a FAX, it encodes without discarding the requested band, and when the input signal is voice, it encodes according to the above-mentioned discarding method. This is an example of converting. In addition, FIG. 76 shows that the input signal branched by the branching section is input to both the normal voice coder and the FAX protocol termination section, and when the FAX protocol termination section detects the FAX protocol, it is sent to the FAX protocol termination section via the selector. If not detected, the output of the normal audio coder is sent to the multiplexer.

FIG. 75 shows a method in which the coder detects the FAX protocol, and the detection method in this case will be explained using the outline of the G3-FAX transmission procedure shown in FIG. 77. In FIG. 75, when the coder detects transmission of a CNG tone or CED tone in phase A, it determines that the data is from a FAX.

In FIG. 73, a system in which various transmission media are accommodated in a multiplexing device has been described.
In such a system, it is conceivable to change the encoding method by the encoder depending on the nature of the media. For example, in FIG. 75, the encoder can detect the FAX protocol as described above and select a band according to the FAX signal, and in this case, the FAX protocol as shown in FIG.
There is no need to use an encoder that also uses a protocol termination part.

[0203] Furthermore, in the case of a push-button telephone, if encoding is performed such that only the tone code is sent when a tone signal is detected, transmission at a very low rate, for example, 1 kbps or less, can be performed when a tone is detected. Can be done. Furthermore, signals in personal computer communication can be encoded in a band different from that of normal conversational speech, and the necessary encoding speed differs depending on the modulation method depending on the MODEM.

In a system such as that shown in FIG. 73, it is necessary to limit the number of channels that do not cause all of the requested bandwidth to be discarded, and to prevent discarding at least those channels that cannot be discarded. As a method of limiting the number of channels, for example, there is a method using a numbering plan. Figure 7
3, if you want to limit the input port on the multiplexer side that accommodates FAX to only #1, for example, set the FAX dial number to 7xxx and the normal telephone number to 8xxx
Plan your numbers accordingly. PB this number system in advance
If the dial number is set to
Since the first port is busy, the second call is blocked, and the number of channels that can be in a non-discardable state is limited.

[0205] Furthermore, if such a channel that cannot be discarded exists, it is necessary to further increase the number of bits that can be discarded for the channel that can be discarded. For example, in FIGS. 26 and 27, core bits are transmitted even at level 5 for bands B1 to B3, but it is necessary to enable discarding to level 6, in which all of these core bits are not transmitted, that is, to a silent state.

FIG. 78 is an example of processing added to the flowchart of FIG. 63 in order to enable discarding up to the silent state. The process in the same figure is S175 and S in FIG.
176, and after the lowest band of each channel is set to level 5, it is determined in S200 whether there is a channel that can be discarded. If there is no channel that can be discarded, the lowest band of channels that can be discarded to a silent state, that is, discarded in a DSI manner, is set to level 6 in S201, and the process moves to S176. Note that if there is a channel that can be discarded in S174 of FIG.
The process moves to step 76.

FIG. 79 shows a case where the number of voice channels in which a call is connected is small and there is no need to discard the additional information section, or when the sound quality deterioration due to discarding the additional information section is less than the sound quality deterioration due to the operation of the voice detector. FIG. 2 is a block diagram illustrating the configuration of a system that sometimes stops the operation of a voice detector or ignores its detection results. In the figure, for example, a multiplexer (
The MUX) counts the number of calls and determines whether or not it is necessary to discard them, and when there is no need to discard them, it outputs a voice detector ignore/stop signal to the coder, thereby preventing the deterioration of sound quality due to the use of voice detectors. can be prevented.

[0208] In this way, when the number of calls is small and there is no possibility of information being discarded, deterioration of sound quality can be prevented by stopping the operation of the voice detector. Voice detectors are used to detect the presence or absence of voice and cut unnecessary voice information to improve transmission efficiency within the network, but of course the performance is lower than when no voice detector is used. Deterioration of sound quality occurs. For example, the microphone of a telephone handset receives both the speaker's voice and ambient noise, but if the speaker is silent when the speaker is not speaking and the audio information is cut, the ambient noise will not be transmitted and the atmosphere will not be conveyed. become. Additionally, it would be fine to detect silence by determining the content of the conversation, but with ordinary silence detectors, parts that you do not want to be cut off may be determined to be silent, resulting in deterioration of sound quality. Since such deterioration in sound quality occurs even when the number of calls in use is small and no discards occur, the sound quality can be improved by stopping the operation of the voice detector.

[0209] In addition, if a relay exchange is provided between the information source multiplexer or exchange and the receiving side multiplexer or exchange, the number of channels to which calls are connected in the exchange at each relay stage. The data is collected by the multiplexer or switch at the information source via a communication path opposite to the information transmission direction, and when there is no need to discard the additional information, the voice detector at the information source operates. By stopping or ignoring the detection results, the quality of the transmitted information can be improved. In normal line design, the number of connectable voice channels is selected assuming the most congested state, so in many cases there is no need to discard them, and it is possible to improve the sound quality even when relaying in multiple stages. Become. In this case, instead of the information source multiplexer or switch collecting data on the number of channels in each relay stage switch, the switch in the call relay stage with a small number of connected channels is in the opposite direction to the information transmission direction. It is also possible to transmit an operation stop signal to the source audio detector using the communication path.

Next, referring to FIG. 79, a method for determining whether or not the sound quality deterioration due to discarding encoded information is less than the sound quality deterioration due to the operation of the voice detector will be described. This determination can be made, for example, simply from the relationship between line capacity and multiplicity, and the number of channels on which a call is actually established. For example, when multiplexing is performed on a 64 kbps line, a preliminary evaluation shows that 5
Assume that an evaluation result shows that the sound quality is better if the silence detector is stopped up to the channel. This means that all channels are active and are discarded in stages, so even if up to a certain band, such as 11 kbps, is discarded, there is less deterioration due to encoding compared to silent transmission in which no encoded information is sent. This means that the result is less.

[0211] In this case, due to the overhead such as the header of the multiplexed frame, the additional information part is not discarded for at least three channels, and for four to five channels, the additional information part is not discarded.
This corresponds to a case where channel multiplexing causes less deterioration in sound quality due to discarding encoded information than deterioration in sound quality due to operation of the voice detector.

FIG. 80 is a basic configuration block diagram of a division multiplexing method in which data to be multiplexed in a unit of multiplexing in a line set is further divided into several parts and multiplexed and transmitted. If the multiplexing unit (MUX) performs multiplexing processing after taking in all the information of one multiplexing unit from the line set, the delay time until this multiplexed information is sent out to the transmission path increases. Also, in the multiplexing and demultiplexing unit (DMUX), the delay time increases because demultiplexing is performed for each multiplexing unit.

[0213] Normally, it is appropriate for one multiplexing unit to be about 10 ms, but when multi-stage relaying is performed, for example, the influence of the above-mentioned delay becomes even greater. For this reason, a division multiplexing method is shown in FIG. 80 in which the delay time is reduced by further dividing the information of one multiplexing unit.

[0214] In FIG. 80, line set 210a,
The division multiplexing method will be explained assuming that the number of divisions into which one multiplexing unit of information in 210b, . . . is further divided is 'n'. In FIG. 80, first, line sets 210a, 21
0b, . . . output data of one multiplex unit in a form that can be divided into n according to the number of divisions n. In addition, for this one multiplexing unit of information, information indicating the required transmission bandwidth (or information indicating whether or not transmission is necessary) is multiplexed before or at the same time as the beginning of the target multiplexed information. It is assumed that the MUX section 211 in the unit can take in the information.

[0215] The MUX section 211 in the multiplexing unit includes a bandwidth arbitration section 213 that arbitrates the bandwidth at the time of congestion discard;
and a division/multiplexer 214 that divides and multiplexes the information of one multiplexing unit from the line set according to the arbitration result, and this multiplexing and sending processing to the transmission path is performed for one multiplexing unit from the line set. It is assumed that processing for each divided data is completed within 1/n of the processing time. Also, when the first division of multiplexed information is imported, bandwidth arbitration is performed according to the transmission request bandwidth information that has been imported at the same time or prior to that, and the entire process is completed within 1/n of the time including division multiplexing processing. It shall be.

Next, the DMUX section 212 in the multiplexing unit
is a batch dividing unit 215 that separates each channel information in one multiplexing unit and sends it to the line set, and sequentially separates and separates the multiplexed information in the order in which it was divided and multiplexed, that is, the multiplexed information of the first division. It is comprised of a division/separation unit 216 that immediately sends out the data to the line set. In the division multiplexing method described in FIG. 80, first, of the multiplexed information arriving from the transmission path, as much information as is possible to restore the information representing the transmission band of each channel is taken in, the information is restored, and the restored information is Accordingly, the divided and multiplexed data is divided and separated and sent to the line set.

That is, in FIG. 80, when the MUX section 211 takes in 1/n of the information of one multiplexing unit, it multiplexes the information one after another and sends it out to the transmission path. For this reason, multiplexing delay is reduced compared to the case where all the information of one multiplexing unit is taken in and then multiplexed. Also, DMUX section 2
In step 12, the information of one multiplexing unit is sequentially divided and separated by the division and demultiplexing, so that the demultiplexing delay is reduced compared to the case of demultiplexing all at once.

FIGS. 81 to 83 are explanatory diagrams of an embodiment of the division multiplex system. In FIGS. 81 to 83, an example will be described in which information for one multiplexing unit of two channels is divided and multiplexed by n=2, that is, by two.

[0219] The information ■ and ■ of one multiplexing unit (10 ms) for both channels are each 40 bits, and the information ■ and ■ indicating the transmission request band are each 2 bits, and the transmission capacity of the transmission path, that is, the fixed length. Assume that the length of frame ■ is 64 bits.

[0220] The MUX unit 211 first takes in information (1) and (2) indicating the requested bandwidth for transmission, and performs bandwidth arbitration. The total of the data from (1) to (2) is 84 bits, and since the bandwidth of 20 bits is insufficient, bandwidth arbitration is performed, and the arbitration result is as shown in FIG. 82, for example.

[0221] The MUX unit 211 then sends the first divided information ■-
1, -1 and multiplexes the first division information of channel 2 according to the arbitration results a and b. In order to inform the receiving side of the presence or absence of discard during multiplexing, transmission band information c, d is used.
Divide and add. As a result, the first division multiplexing information becomes as shown in (1)-1 whose details are shown in FIG. 76, and this is sent to the transmission path. Subsequently, the MUX section 211 takes in the second divided information (1)-2 and (2), performs similar multiplexing according to the arbitration result, and sends the multiplexed information to the transmission path. The information sent out is (c
) The details are shown as ■-2.

Next, division and separation by the DMUX section 212 will be explained with reference to FIGS. 81 to 83. DMUX section 2
12 first takes in the two-division multiplexing information over a period of time corresponding to one multiplexing unit, and restores information indicating the transmission band from this information. This makes it possible to identify how much of the captured multiplexed information should be distributed to which channels.

[0223] Then, according to the transmission band information c and d, the first
Division multiplexing information■'-1 to channel 1 has 0 bits,
For channel 2, 20 bits are separated, and the restored transmission band information is combined and sent to the line set of each channel. Next, 0 bits for channel 1 and 20 bits for channel 2 are similarly separated from the second division multiplexed information 1'-2 and sent to the line set.

FIGS. 84 and 85 are flowcharts of an embodiment of division multiplexing processing for one multiplexing unit. FIG. 84 shows multiplexing information Xi of one multiplexing unit for a channel and information Ri indicating a transmission requested band for the multiplexing information Xi. One unit of multiplexed information is divided into 1/n pieces, and the divided data are divided into Xi 1 , Xi 2 , . . . in the order of arrival from the line set.
Xin, and the information Ri indicating the required transmission band is also divided into n equal parts, Ri 1 , Ri 2 , . . . Ri n
It is said that Part of these divided data is discarded as a result of bandwidth arbitration, and the information actually transmitted is Xi 1
′, Xi 2 ′, . . . Xi n ′.

FIG. 85 is a flowchart of division multiplexing. In the figure, band arbitration is first performed in S218. For this band arbitration, one of the above-mentioned arbitration methods is used. Then, for each channel, the number of transmission allocated bits Xi' to be actually transmitted for each divided data is determined.

[0226] After that, n is set to 1 in S219, and S220
First, multiplexing is performed on the first divided data of n-division multiplexing. Here, the first piece of divided data for all channels is first taken in, and the portion of the divided data to be transmitted is cut out. Information Rin indicating the transmission band in order from channel 1, and allocation data Xi
n' is multiplexed into an area corresponding to n of the transmission capacity C of the transmission line, and sent out to the transmission line. Then, in S221, the value of n is incremented, and the process of S220 is repeated until the nth piece of divided data.

FIG. 86 shows the DMUX section 21 of FIGS. 81 to 83.
2 is a flowchart of batch separation performed in step 2. First S
In S222, n indicating the order of divided data is set to 1, and in S22
The first divided data is stored in step 3, the value of n is incremented in step S224, and the process continues in step S2 until n pieces of divided multiplexed data are stored.
23 is repeated.

[0228] When n pieces of divided data have been stored,
From the data for one multiplexing unit stored in S225,
Information Ri indicating the transmission band for each channel is restored, i indicating the channel number is set to 1 in S226, and S22
At step 7, the divided data transmitted for channel i are cut out according to the value of transmission band information Ri, and combined in order, and data Xi is sent to the line set for channel i at step S228. The processes of S227 and S228 are repeated for the number of channels, and the batch separation is completed.

FIG. 87 is a flowchart of an embodiment of division and separation processing. In the figure, the processing up to S225 is shown in FIG.
This is exactly the same as the case of batch separation in step 6. Then S229
n indicating the order of divided data is set to 1 in S230, i indicating the channel number is set to 1, and in S231, according to the value of information Ri indicating the transmission band, the stored n-th multiplexing information is assigned to channel i. Transmission data Xi n′
is extracted, and the data is sent to the line set for channel i in S232. After the processes of S231 and 232 are repeated for the number of channels, the process from S230 to 2
The processes up to 32 are repeated n times, the number of pieces of divided data, and the division and separation process is completed.

FIGS. 88 to 90 are explanatory diagrams of delay reduction by the division multiplexing method. 88 to 90 compare delay times in the case of n-division multiplexing, where t is the time corresponding to one multiplexing unit in a line set (basic multiplexing unit). By performing n-division multiplexing, in the MUX section, the delay in capturing line set information, the delay in multiplexing processing, and the delay in sending multiplexed information to the transmission path are each reduced to 1/n, and the time corresponding to one multiplexing unit is reduced to 1/n. has the effect of becoming substantially t/n, reducing multiplexing delay. Also DMUX
In the section, the delay time is increased from 2t to (n+1
)t/n.

[0231] If the resulting reduction in the overall delay time is T, then T is

[0232]

[Math 3]

If n is large, the delay time will be reduced by about three times the unit of multiplexing.

FIGS. 91 and 92 are explanatory diagrams of an embodiment of a voice quality protection system in which voice detection information is simultaneously notified to an encoder and a multiplexer. FIG. 91 shows a conventional method. A voice detector outputs voice detection information that indicates whether there is a sound or no sound in a certain unit of time, and naturally the judgment is made with a delay of that certain period of time compared to the audio data. The result will be output. For this purpose, the voice detection information is transmitted to the multiplexing device in phase with the encoded information processed by the voice encoder.

[0235] The problem in voice detection is the point at which the voice changes from silence to voice, such as at the beginning of a speech or at the beginning of a word.There are various methods that can be used to confirm this point of change, but there are several methods. In order to produce natural reproduced sound, a guard time method is required that substantially advances the sound detection information in time and outputs it to protect the sound state. However, with this guard time method, as mentioned above, it is necessary to temporarily store the audio information in memory in order to match the phase of the audio detection information and the encoded information and send it to the audio encoder, which requires extra storage space. In addition to this, the delay in audio transmission becomes large.

FIG. 92 shows a method according to the present invention, in which the detection result by the voice detector is notified to the multiplex processing section in the multiplexer at the same time as to the voice encoder. For this reason, the detection result is notified to the multiplexing processing unit in advance of the audio information by the time required for processing by the audio encoder, that is, encoding, and the detection result is notified to the multiplexing processing unit in advance of the audio information by the time required for processing by the audio encoder, that is, the time required for encoding. protection, that is, protection of voice quality, can be realized without any additional delay in voice information.

FIGS. 93 and 94 show an embodiment of a method for generating noise on the receiving side according to the noise level on the transmitting side during a silent period. Generally, when there is silence, no audio data is sent from the transmitting side to the receiving side, and the receiving side inserts noise into the silent period to reduce the user's sense of being disconnected from the call and the discomfort when switching between silence and sound. The level of noise inserted on the receiving side is often set to a constant level regardless of the noise level on the transmitting side, and if there is a large difference in the noise level between the transmitting side and the receiving side, it may feel strange. There was a point.

Therefore, in the present invention, as shown in FIG. 93, when the sound detector on the transmitting side detects silence, it simultaneously measures the noise level and sends information indicating the level to the receiving side. As shown at 94, the receiving side generates noise for the silent section according to the noise level information.

[0239] As described above, when transmitting the signaling signal for the voice channel using the signal send (SS) signal and the signal receive (SR) signal, the S
The call connection status is determined when the logical sum of both S and SR signals is '0' for a certain period of time, and the logical product of both signals is '0' for a certain period of time.
When the value is 1', it is determined that the call is disconnected.

[0240] This call detection is performed using signaling signals (SS,
Since the SR) passes through the audio encoder and the multiplexing unit, it can be executed in both the encoder and the multiplexing unit. When performing this in software, the algorithm shown in FIG. 23 may be used, and is executed by the processor 77 in the multiplexing unit shown in FIG. 45, for example. In the case of using hardware, a call detection circuit, which will be described later, may be provided in each voice encoder, or it may be provided in the multiplexing unit for the number of voice channels.

FIG. 95 is a block diagram of an embodiment of the call detection circuit. In the figure, the call detection circuit includes an OR gate 235 to which both SS and SR signals are input, a D flip-flop 236 to which the output of the OR gate is input to the D input, and a counter to which the Q output of the D flip-flop 236 is input to the reset terminal. 237, outside 9 of D flip-flop 236
a counter 238 whose output is input to a reset terminal;
two counters 237

[0242]

[Outer 9]

, 238 are input to the set terminal and reset terminal, respectively.

In FIG. 95, when both the SS and SR signals are 0, the Q output of the D flip-flop 236 is 0;
Output 10 becomes 1, counter 238 is reset,

[0245]

[Outside 10]

[0246] The counter 237 counts up each time a clock is input, and continues for a certain period of time, here τ1 = 640 ms.
When this happens, a carry signal is output, and the RS flip-flop 239 outputs Q=1 as a detection result of the call connection state.

On the other hand, when the SS and SR signals are both 1, the counter 237 is reset, and the counter 237 is reset.
38 counts clock pulses until τ2 = 640 ms and outputs a carry signal at that point, and the RS flip-flop 239 outputs 0 as a Q output indicating call disconnection.

[0248] Furthermore, in a system that transmits signaling information as SS and SR signals, by utilizing the property that the signaling signal hardly changes except during call processing, it is possible to reduce the bandwidth required when transmitting signaling using an out-of-band method. becomes possible. FIG. 96 is an example of a signaling signal. In the figure, on the transmitting side, after activation, the signaling signal maintains a constant value of 0 until the selection number signal transmission starts, takes a value of 0 or 1 when the selection number signal is transmitted, and then becomes constant at 0 again. Also, on the receiving side, it is 1 until the response.
However, after the response, it becomes constant at 0.

[0249] In the present invention, the signaling bits of the frame transmitted immediately before, for example, 4 bits, are all 0,
Alternatively, transmission compression is performed only when all 1s are set and the signaling bits of the currently transmitted frame do not change at all. The reason for adding the condition of all 0s or all 1s is to provide resistance against line random errors.

[0250] That is, on the receiving side, when entering transmission compression, decompression processing is performed during compression based on the rule of all 0s or all 1s, and in the unlikely event that a transmission path error occurs, an error is detected using majority logic. I have decided to make a correction. Further, the same transmission compression method can be used for other media other than voice, ie, remote signaling (RS) signals of data channels, ie, level signals for ensuring status between terminals.

FIGS. 97 and 98 are flowcharts of an embodiment of signaling compression processing. FIG. 97 shows an example of processing on the transmitting side, and when the processing starts, first S240 is performed.
It is determined whether the signaling information is all 0's or all 1's. If neither of these is the case, 1 indicating that signaling information is to be transmitted is set in the signaling transmission flag in S241, and the process ends.

[0252] If it is either all 0s or all 1s in S240, it is determined in S242 whether or not the signaling information has changed from that for the previous frame. If it has changed, it is necessary to send signaling information, so the signaling transmission flag is set to 1 in S241 and the process ends. If there is no change from the previous time, the compression mode is set, the signaling transmission flag is set to 0 in S243, and the process ends.

FIG. 98 shows an embodiment of the signaling compression process on the receiving side. When the process starts, first S244
It is determined whether the received signaling transmission flag is either 0 or 1. If the transmission flag is 1, it means that signaling information is being transmitted, so S24
In step S246, the input signaling information is held, and in step S246, it is output, and the process ends.

[0254] If the signaling transmission flag is 0 in S244, it is determined in S247 as the decompression mode whether the currently held signaling information, that is, the last received signaling information is all 0s or all 1s. If the stored value is either all 0 or all 1, the held value is output in S248 and the process ends.

[0255] If it is neither all 0s nor all 1s, it means that a transmission error has occurred, so in order to correct it, it is determined in S249 whether the number of 1s is 1, 2, or 3. Ru. S25 if 1 or 3
If it is 0, correction is performed using majority logic, and if it is 2, it is assumed that the line is disconnected and correction is made to all 1s in S251, and then the corrected held value is output in S248.
Finish the process.

[0256] In the above embodiment, an example was shown in which the present invention was applied to data transfer, but the present invention is not only applicable to audio transmission, but can also be applied to, for example, image data transmission. . FIG. 99 shows a configuration diagram when the present invention is applied to image processing. This embodiment is similar to the embodiment shown in FIG. 7, and the same parts are given the same numbers and the explanation thereof will be omitted. The difference in this embodiment is that the image encoding means 20 is used instead of the audio encoding means 20.
′ is used.

[0257] FIG. 100 is a diagram showing an example of required bandwidth determination, and at a transmission rate of 0 having a transmission speed of 8 kbps, only side information is transmitted. For transmission rates 1, 2, and 3, the transmission speed is 256, respectively.
kbps, 1024 kbps, and 2048 kbps, and band arbitration is performed at high transmission rates.

[0258]

[Effects of the Invention] As explained in detail above, according to the present invention, for example, for an audio channel, the core information part to guarantee the minimum sound quality is always multiplexed, and the core information part necessary to obtain the desired sound quality is always multiplexed. By sequentially multiplexing the additional information parts starting from the parts with the highest priority, it is possible to minimize the deterioration in communication quality even when congestion is discarded. In addition, for media that requires waiting, such as a packet switch, first secure the packet data band for the minimum throughput of the packet switch, then multiplex the audio of the voice channel into the remaining band, and then packets can be multiplexed to form an efficient multiplexed frame. As described above, according to the present invention, it is possible to pursue both improvement in communication quality and improvement in multiplexing efficiency in statistical multiplexing.

[Brief explanation of the drawing]

FIG. 1 is a diagram showing an example of a voice multiplexing device or switching system.

FIG. 2 is a diagram illustrating an example of a multimedia multiplexing device or switching system.

FIG. 3 is a diagram illustrating a conventional TDM multiplexing method.

FIG. 4 is a diagram showing an example of a conventional statistical multiplexing method.

FIG. 5 is a diagram showing a band allocation method in the DSI method.

FIG. 6 is a principle block diagram of the high-efficiency digital multiplex transmission system of the present invention.

FIG. 7 is a principle block diagram of a multiplexing method based on a transmission required band for a voice channel according to the present invention.

FIG. 8 is a principle block diagram of a multiplexing method according to call detection results of the present invention.

FIG. 9 is a principle block diagram of a multiplexing system when the multiplexing device or switch of the present invention is connected to a label multiplex network such as a packet network or an ATM network.

FIG. 10 is a block diagram of the principle of the present invention.

FIG. 11 is a block diagram of the principle of the present invention.

FIG. 12 is a block diagram of the principle of the present invention.

FIG. 13 is an explanatory diagram of division of audio transmission information in the present invention, and is a diagram showing separation of a core information section and an additional information section.

FIG. 14 is an explanatory diagram of classification of voice transmission information in the present invention, and is a diagram showing a data transmission request range.

FIG. 15 is a basic configuration block diagram of a first embodiment of the transmission method of the present invention.

FIG. 16 is a diagram showing states of band and voice quality in the first embodiment of the transmission system of the present invention.

FIG. 17 is a diagram showing an example of multiplexing processing at the time of congestion discard in the first embodiment of the transmission system of the present invention.

FIG. 18 is a basic configuration block diagram of a second embodiment of the transmission system of the present invention.

FIG. 19 is a diagram showing the voice band and quality with respect to the number of voice calls in a second embodiment of the transmission system of the present invention.

FIG. 20 is a diagram showing an example of multiplexing processing in a second embodiment of the transmission system of the present invention.

FIG. 21 is a basic configuration block diagram of a third embodiment of the transmission system of the present invention.

FIG. 22 is a diagram showing call detection in a third embodiment of the transmission system of the present invention.

FIG. 23 is a diagram showing call detection in a third embodiment of the transmission system of the present invention.

FIG. 24 is a basic configuration block diagram of a fourth embodiment of the transmission system of the present invention.

FIG. 25 is a basic configuration block diagram of a fifth embodiment of the transmission system of the present invention.

FIG. 26 is a basic configuration block diagram of a sixth embodiment of the transmission system of the present invention.

FIG. 27 is a diagram showing a discard level in a multiplexing device that performs band division coding in a sixth embodiment of the transmission system of the present invention.

FIG. 28 is a diagram showing an example of a multiplexed frame when transmitting a transmission band as a code.

FIG. 29 is a diagram illustrating an example of a method for determining a discard method for a current transmission frame based on past discard history.

FIG. 30 is a diagram showing an embodiment of an audio multiplex transmission system to which the present invention is applied.

FIG. 31 is a configuration diagram of a multiplex transmission system including a multiplexer according to the present invention.

FIG. 32 is a diagram showing a basic example of processing of a multiplexing unit in the present invention.

FIG. 33 is a configuration block diagram of an embodiment of a speech encoder.

FIG. 34 is a diagram illustrating an overview of a sound/non-sound determination method.

FIG. 35 is an explanatory diagram of a sound/non-sound determination method.

FIG. 36 is a diagram illustrating an example of sound/silence determination processing.

FIG. 37 is a configuration block diagram of a normal ADPCM encoding method.

FIG. 38 is a block diagram of an embedded ADPCM encoding method.

FIG. 39 is a diagram illustrating a voice/data transfer route within the multiplexing device.

FIG. 40 is a diagram showing an example of a transmission band allocation code.

FIG. 41 is a diagram showing an example of a code table for transmission band information.

FIG. 42 is a diagram showing an example of serial data on a line.

FIG. 43 is a diagram showing the overall structure of a frame.

FIG. 44 is a diagram showing the configuration of a header section.

FIG. 45 is a block diagram showing the configuration of an embodiment of a multiplexing unit.

FIG. 46 is a flowchart of a schematic processing example of a multiplexing unit.

FIG. 47 is an explanatory diagram of a processing cycle in a multiplexing unit.

FIG. 48 is a flowchart of the first embodiment of bandwidth arbitration processing.

FIG. 49 is a diagram showing an example of arbitration using the flowchart of FIG. 48;

FIG. 50 is a flowchart of a second embodiment of bandwidth arbitration processing.

FIG. 51 is a flowchart of a second embodiment of bandwidth arbitration processing.

FIG. 52 is a flowchart of an example of a discard mode update process.

FIG. 53 is a diagram showing an example of arbitration using the flowcharts of FIGS. 50 and 51;

FIG. 54 is a diagram showing an example of arbitration using the flowcharts of FIGS. 50 and 51;

FIG. 55 is a diagram showing an example of arbitration using the flowcharts of FIGS. 50 and 51;

56 is a diagram showing an example of the relationship between the values of each variable in the bandwidth arbitration process according to the flowcharts of FIGS. 50 and 51; FIG.

FIG. 57 is a block diagram showing the configuration of an embodiment of a packet switching network to which the present invention is applied.

58 is a block diagram showing the configuration of the packet interface section of FIG. 57. FIG.

FIG. 59 is a diagram illustrating the action of a speed absorption buffer.

FIG. 60 is a diagram illustrating the operation of the arrival fluctuation absorption buffer.

FIG. 61 is a diagram showing an example of a packet format.

FIG. 62 is a diagram showing an example of a code table for transmission band information during band division encoding.

FIG. 63 is a flowchart of an embodiment of a multiplexing process using a band division encoder.

FIG. 64 is a diagram showing an example of a discard mode corresponding to the frequency characteristics of voice (in the case of a male).

FIG. 65 is a diagram showing an example of a discard mode corresponding to the frequency characteristics of voice (in the case of a woman).

FIG. 66 is a diagram showing an example of a code table of transmission band information for each discard mode.

FIG. 67 is a flowchart of an embodiment of discard mode determination and mode notification processing to the receiving side.

FIG. 68 is a flowchart of an embodiment of discard mode determination and mode notification processing to the receiving side.

FIG. 69 is a block diagram showing the configuration of an embodiment of a system for compressing transmission by making a channel in a reserved state silent.

FIG. 70 is a diagram illustrating an embodiment in which a hold state detector is built into an exchange.

FIG. 71 is a diagram illustrating an embodiment in which a hold state detector is built into an exchange.

FIG. 72 is a block diagram showing the configuration of an embodiment in which a pending state detector is built into a multiplexing device.

FIG. 73 is a block diagram showing the overall configuration of a system in which a multiplexer accommodates various transmission media.

FIG. 74 is a block diagram showing the configuration of an embodiment of an encoder that performs encoding according to a FAX protocol detection result.

FIG. 75 is a block diagram showing the configuration of an embodiment of an encoder that performs encoding according to a FAX protocol detection result.

FIG. 76 is a block diagram showing the configuration of an embodiment of an encoder that performs encoding according to a FAX protocol detection result.

FIG. 77 is a diagram showing an outline of a G3-FAX transmission procedure.

[Figure 78] Figure 6 for enabling discard to silent state
3 is a diagram showing an example of addition processing to the flowchart of FIG. 3; FIG.

FIG. 79 is a block diagram showing the configuration of a system that can stop the operation of the silence detector or ignore the detection result.

FIG. 80 is a block diagram showing the basic configuration of a division multiplexing method.

FIG. 81 is a diagram illustrating an example of a division multiplexing method.

FIG. 82 is a diagram illustrating an example of a division multiplexing method.

FIG. 83 is a diagram illustrating an example of a division multiplexing method.

FIG. 84 is a diagram illustrating an example of division multiplexing processing for one multiplexing unit.

FIG. 85 is a flowchart of an embodiment of division multiplexing processing for one multiplexing unit.

FIG. 86 is a flowchart of an embodiment of batch separation processing.

FIG. 87 is a flowchart of an example of division and separation processing.

FIG. 88 is a diagram illustrating delay reduction by division multiplexing.

FIG. 89 is a diagram illustrating delay reduction by division multiplexing.

FIG. 90 is a diagram illustrating delay reduction by division multiplexing.

FIG. 91 is a diagram illustrating an embodiment of a voice quality protection system that simultaneously reports voice detection information to an encoder and a multiplexing processing unit.

FIG. 92 is a diagram illustrating an embodiment of a voice quality protection system that simultaneously reports voice detection information to an encoder and a multiplexing processing unit.

FIG. 93 is a diagram illustrating an example of a method for generating noise on the receiving side according to the noise level on the transmitting side during a silent period.

FIG. 94 is a diagram illustrating an example of a method for generating noise on the receiving side according to the noise level on the transmitting side during a silent period.

FIG. 95 is a block diagram showing the configuration of an embodiment of a call detection circuit.

FIG. 96 is a diagram showing an example of a signaling signal.

FIG. 97 is a flowchart of an embodiment of signaling transmission compression processing.

FIG. 98 is a flowchart of an embodiment of signaling transmission compression processing.

FIG. 99 is a block diagram of the principle of the present invention.

FIG. 100 is a diagram showing a request bandwidth determination result.

[Explanation of symbols]

20 Voice encoding means 21 Silent section detection means 22 Multiplexing processing means 23 Requested bandwidth determination means 24 Call detection means 25 Packet assembly means 30 Voice encoder 31 Silence detection section 32 Multiplexing section 33 Requested bandwidth determination section 34 Call detection section 36 Packet assembly section 39 Valid information detection section 40, 41 Multiplexing device 43 Codec (CODEC: audio encoder) 44 Multiplexing unit (MUX/DMU
X) 61 Audio LS (line set) 67
Node control unit 68 Data LS (line set) 77
Processor 140 Packet interface unit (PAD
) 191 Hold state detector (R-DET) 2
10a, 210b line set 214
Division/multiplex section 216 Division/separation section

Claims (25)

[Claims] Claim 1: A multiplexing device for transmitting multiplexed signals over a digital line, which includes audio input information, a core information section for ensuring minimum sound quality, and a core information section for obtaining desired sound quality. Audio encoding means that performs separate encoding of additional information that needs to be transmitted and can be discarded in stages according to the priority of transmission, and whether or not it is necessary to transmit audio information. a silent section detecting means for detecting a silent section of the voice input information in order to determine the voice input information; For channels where silence is not detected, only the information necessary for synchronization with the encoder of the communication partner is transmitted, and for channels where silence is not detected, the core information section is transmitted to fixed length digital slots that are repeated at a constant period. and multiplexing the additional information part of each audio channel in which the silence has not been detected in the remaining part of the fixed length digital slot as a result of the multiplexing, starting from the part with the highest priority to be transmitted. 1. A high-efficiency digital multiplex transmission system, characterized in that it has a multiplexing processing means for discarding additional information parts that cannot be multiplexed due to insufficient bandwidth. 2. Requested band determining means for determining each transmission required band of audio input information for each of the plurality of audio channels to be multiplexed on the digital line, wherein the multiplexing processing means does not detect the silence. 2. High efficiency according to claim 1, characterized in that, when discarding the additional information section of each audio channel, the additional information section within the required band determined by the required band determining means is discarded in stages starting from the part with the lowest priority. Digital multiplex transmission method. 3. Call detection means for detecting whether or not each of the plurality of voice channels to be multiplexed is in a call state, wherein the multiplexing processing means detects that the call detection means detects a call disconnected state. Claim 1 or 2, characterized in that no encoded information is transmitted to the audio channel, including information necessary for synchronization with the encoder of the communication partner.
High efficiency digital multiplex transmission method described. 4. Packet assembling means for packetizing the multiplexed frame output by the multiplexing processing means as it is or by dividing it into packets, the multiplexing device label-multiplexing the packet to which the multiplexing device is connected. 4. The high-efficiency digital multiplex transmission system according to claim 1, wherein the high-efficiency digital multiplex transmission system is output to a network. 5. The multiplexing device is configured to perform multiplexing in consideration of the nature of occurrence of sound quality deterioration due to discarding of the additional information section, which depends on the nature of each encoder for each of the plurality of audio channels to be multiplexed on the digital line. a node control unit that holds a multiplexing parameter and sets the multiplexing parameter in the multiplexing processing means, and the multiplexing processing means stores the multiplexing parameter and a history of discarding the additional information section in the immediately preceding transmission frame. Claims 1 and 2, characterized in that the method for discarding the additional information section for the current transmission frame is determined based on
3. The high efficiency digital multiplex transmission system according to 3 or 4. 6. A multiplexing device in which the multiplexing device multiplexes transmission information of a data channel, which is another medium including a data terminal, in addition to a plurality of voice channels, and transmits the multiplexed signal on a digital line. a switching device or a switching device, comprising a valid information detection section for detecting whether or not there is a need for information transmission for each data channel, wherein the multiplexing processing means first detects the transmission of the detected valid information; The high-efficiency digital multiplex transmission system according to claim 1, 2, 3, 4, or 5, characterized in that the multiplexing for the plurality of voice channels is performed after multiplexing the transmission information of the data channels that require . . 7. The multiplexing device multiplexes, in addition to a plurality of voice channels, transmission information of data channels, which are other media including data terminals, and packet data from a packet switch, A multiplexing device or a switching device that transmits the multiplexed signal on a line, the multiplexing device comprising a valid information detection unit that detects whether information transmission is necessary for each data channel, and wherein the multiplexing processing means A guaranteed minimum throughput value of the exchange is stored in advance, and the transmission information of the data channel that needs to be transmitted detected by the effective information detection unit is multiplexed with the packet data corresponding to the minimum throughput of the packet exchange,
Claims 1, 2, and 3, wherein the plurality of voice channels are multiplexed, and packet data is further multiplexed into the remaining portion of the fixed length digital slot.
, 4 or 5, the high efficiency digital multiplex transmission system. 8. The audio encoding means comprises a band division type encoder, and the multiplexing section constituting the multiplexing processing means transmits all core information parts for each audio channel in which silence is not detected. 8. The high-efficiency digital multiplex transmission system according to claim 1, wherein the core information section for a specific frequency component is selectively discarded when the core information section cannot be used. 9. The audio encoding means is configured as a band division type encoder, and the band division type encoder or the multiplexing unit constituting the audio multiplexing unit determines the frequency characteristics of the audio of each of the audio channels. 9. The high frequency information according to claim 8, wherein the multiplexer performs multiplexing processing including selectively discarding the core information section based on a discard mode corresponding to the detected frequency characteristic. Efficient digital multiplex transmission system. 10. The multiplexing processing means maintains a code table for encoding a transmission band, and encodes a transmission information band for each channel actually transmitted on the digital line on a transmission frame. 9. The high-efficiency digital multiplex transmission system according to claim 1, 2, 3, 4, 5, 6, 7, or 8, characterized in that the transmission is carried out using the same method. 11. When the multiplexing device detects a channel in a pending state among the plurality of audio channels,
Notify the communication partner's multiplexer or switch of the hold state of the channel, output the hold tone pattern built into the transmitter's multiplexer or switch, and relay transmission between the multiplexers or switches. 5. The high-efficiency digital multiplex transmission system according to claim 1, 2, 3, or 4, wherein transmission compression is performed in a silent state on the road. 12. The multiplexing device accommodates various transmission media including audio, and the encoder constituting the audio encoding means encodes each media in order to obtain the information quality necessary for the various transmission media. Encoding is performed by changing the encoding method and the number of bits of the core information part that does not allow discarding depending on the characteristics, and the multiplexing section that constitutes the multiplexing processing section selects the transmission media according to the requests of each encoder. 5. The high-efficiency digital multiplex transmission system according to claim 1, 2, 3, or 4, characterized in that multiplexing processing corresponding to the above is performed. 13. When the multiplexing device detects the number of channels to which a call is connected among a plurality of voice channels, and there is no need to discard the additional information section, or the voice quality is improved by discarding the additional information section. When the deterioration is less than the audio deterioration caused by the operation of the audio detector constituting the silent section detection means,
High efficiency digital multiplex transmission according to claim 1, 2, 3 or 4, characterized in that unnecessary deterioration of sound quality is avoided by stopping the operation of the voice detector or ignoring the detection result of the voice detector. method. 14. The multiplexing device that is an information source transmits information on the number of channels to which the call is connected in a switching device that is a relay stage for the information in the label multiplexing network such as the packet network or ATM network. when there is no need to collect the additional information through a communication path opposite to the direction and discard the additional information, stop the operation of the voice detector constituting the silent section detection means;
or by ignoring the detection results of the voice detector,
Claim 1, characterized in that unnecessary sound quality deterioration is avoided.
4. High efficiency digital multiplex transmission system according to 2, 3 or 4. 15. The line set constituting the voice encoding means and the silent section detection means is configured to determine whether or not the data to be multiplexed in one unit of multiplexing in the line set and whether transmission of the data is necessary or not. The multiplexing processing means first takes in information indicating whether or not the transmission is necessary or the requested transmission band, and determines the transmission band of each channel based on the information. and a band arbitration unit, according to the result of the band arbitration, divides the multiplexed data of one multiplexing unit from the line set into a plurality of divided data and sequentially captures the divided data, and each time the data is captured,
5. The high-efficiency digital multiplex transmission according to claim 1, further comprising a division multiplexing section that multiplexes the divided data and sends the divided data to a transmission path, thereby reducing delay time associated with multiplexing. method. 16. The multiplexing processing means first restores information representing a transmission band of each channel from one multiplex unit of multiplexed data in the line set received from a transmission path, and Accordingly, the present invention is characterized by comprising a division and demultiplexing section that separates the received multiplexed data into divided data units according to the order in which they were divided and multiplexed and outputs the divided data to the line set, thereby simultaneously reducing the delay time associated with demultiplexing. 16. The high efficiency data multiplex transmission system according to claim 15. 17. The high-efficiency device according to claim 15, wherein the line set includes a line set for a data channel in addition to a line set constituting the speech encoding means and the silent period detection means. Digital multiplex transmission method. 18. The presence/absence/silence determination result of the voice detector constituting the silence detecting means is simultaneously sent to an encoder constituting the voice encoding means and a multiplexing processing section constituting the multiplexing processing means. Claims 1, 2 and 3 characterized in that the quality of the audio information is protected without increasing the delay of the audio information over the time of transmission and encoding of the audio by the encoder.
or the high-efficiency digital multiplex transmission method described in 4. 19. A voice detector constituting the silence detection means measures a noise level on the transmitting side for the voice channel in which silence has been detected, and the multiplexing processing means:
In addition to the information necessary for synchronization with the encoder of the communication partner for the channel, information indicating the noise level is multiplexed into the fixed length digital slot, and the receiving side calculates the noise level. 5. The high-efficiency digital multiplex transmission system according to claim 1, wherein noise is generated for the silent section according to the information indicated. 20. An audio multiplexing device, a multimedia multiplexing device, or a switching device that transmits signaling information for a voice channel as a signal send (SS) signal and a signal receive (SR) signal,
The SS signal and the SR signal are monitored, and the logical sum and logical product of the values of both the SS and SR signals are calculated, and based on the time when the value of the logical sum or logical product is constant, the voice channel is determined. A call detection method for detecting a call-in-call or call-dropped state. 21. In an audio multiplexing device, multimedia multiplexing device, or switching device that transmits signaling information for a voice channel as a signal send (SS) signal and a signal receive (SR) signal,
A flag indicating whether or not to transmit the SS signal for a fixed transmission cycle unit is added to the transmission frame and transmitted.
A transmission information compression method characterized by compressing transmission of signaling information for an interval in which an S signal does not change. 22. In a multimedia multiplexing device or switching device that transmits a level signal for status confirmation between a communication partner and a communication partner for an accommodated data channel as a remote signaling (RS) signal, Transmission information characterized in that a flag indicating whether or not to transmit the RS signal for each unit is added to a transmission frame and transmitted, and transmission compression of remote signaling information is performed for an interval in which the RS signal does not change. Compression method. 23. A multiplexing device for transmitting multiplexed signals over a digital line, which includes audio input information, a core information section for ensuring minimum sound quality, and a core information section for obtaining desired sound quality. A voice encoding means that separates and encodes the additional information part that needs to be transmitted and can be discarded in stages according to the priority to be transmitted; The additional information portions of each voice channel for which silence has not been detected are gradually multiplexed into the remaining portion of the fixed-length digital slot, starting from the portion with the highest priority to be transmitted, and additional information that cannot be multiplexed due to insufficient bandwidth is multiplexed. A high-efficiency digital multiplex transmission system, characterized in that it has a multiplexing processing means for discarding multiplexed parts. 24. In an exchange that transmits multiplexed signals over a digital line, voice input information, a core information section to ensure minimum sound quality, and a core information section that is transmitted together with the core information section to obtain desired sound quality. Audio encoding means that performs encoding separately from an additional information part that is necessary and can be discarded in stages according to the priority to be transmitted;
a silent section detecting means for detecting a silent section of the audio input information in order to determine whether or not transmission of the audio information is necessary; and a silent section detecting means for detecting the silent section among the plurality of audio channels to be multiplexed on the digital line. For audio channels where silence has been detected, only the information necessary for synchronization with the encoder of the communication partner is transmitted, and for channels where silence has not been detected, the core information section is transmitted at regular intervals. the priority for multiplexing into repeated fixed-length digital slots and transmitting the additional information part of each audio channel in which the silence is not detected in the remaining part of the fixed-length digital slot as a result of the multiplexing; Multiplexing is performed in stages starting from the highest part of
1. A high-efficiency digital multiplex transmission system, comprising a multiplexing processing means for discarding additional information parts that cannot be multiplexed due to insufficient bandwidth. 25. In a multiplexing device or exchange that transmits multiplexed signals on a digital line, image input information is divided into a core information section for ensuring the lowest image quality and a core information section for obtaining a desired image quality. An image encoding means that encodes the additional information section separately from the additional information section that needs to be transmitted together with the additional information section and can be discarded in stages according to the priority to be transmitted, and the additional information section that is necessary for transmitting image information. a required bandwidth determining means for determining a transmission band of the image input information in order to determine the appropriate bandwidth; and a required bandwidth determining means determines that transmission is unnecessary among the plurality of image channels to be multiplexed on the digital line. For image channels that have been transmitted, only the information necessary for synchronization with the encoder of the communication partner is transmitted, and for channels that are determined to require transmission, the core information part is transmitted in a fixed manner that is repeated at regular intervals. The additional information part of each image channel that is multiplexed into a long digital slot, and which is determined to need to be transmitted in the remaining part of the fixed length digital slot as a result of the multiplexing, is the high priority part to be transmitted. A high-efficiency digital multiplexing transmission system characterized by comprising a multiplexing processing means for step-by-step multiplexing and discarding additional information parts that cannot be multiplexed due to insufficient bandwidth.