JP3668144B2 - Packet communication voice simulation apparatus and method - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a packet communication voice simulation method and apparatus, and more particularly to a packet communication voice simulation method and apparatus for simulating deterioration of voice quality that occurs when voice information is packetized, transmitted, and reproduced.
[0002]
[Prior art]
Voice packet communication such as VoIP (Voice over Internet Protocol), in which voice information is packetized and transmitted on the transmission side and reproduced on the reception side, has recently begun to be put into practical use.
In such voice packet communication, when packet delay or loss occurs due to network congestion or the like, packets carrying voice information do not arrive correctly at the receiving side and are not reproduced correctly, resulting in voice quality degradation. . Therefore, accurately estimating the degree of quality degradation is important in network design and operation.
[0003]
In order to estimate the voice quality degradation in such voice packet communication, it is necessary to first clarify the behavior related to delay and loss in the network at the packet level, that is, to perform a simulation at the network level. .
Then, based on the behavior of the packet obtained by the simulation at the network level, an operation for clarifying how the voice actually deteriorates, that is, a simulation at the voice level may be performed.
[0004]
Conventionally, network simulation apparatuses that perform simulation at the network level already exist. For example, a network simulation device called OPNET manufactured by OPNET Technologies, Inc. simulates the behavior by simulating packet contention occurring in each device in the network for each individual packet. According to this device, finally, regarding the packet output from a predetermined part of the network,
1. Loss information (whether or not lost) that indicates whether the packet is lost in the network
2. Arrival time information indicating the arrival time of the packet if it is not lost
Thus, network performance information at the packet level can be obtained.
[0005]
Conventionally, there is also an emulator device that reproduces the same network quality when the behavior (packet loss rate, etc.) of a packet in a network portion is clarified by simulation or actual measurement.
According to this emulator device, voice deterioration due to the network can be reproduced by connecting an actual voice terminal to the emulator device.
[0006]
[Problems to be solved by the invention]
Here, it is conceivable to estimate voice quality degradation in voice packet communication by operating an emulator device based on network performance information at the packet level obtained by the conventional network simulation device as described above.
On the other hand, in the real-time terminal application used in such voice packet communication, using the fluctuation absorbing buffer provided in the receiving side terminal, the fluctuation of the packet arrival timing caused by the network side factor is reduced, Relieves audio breaks. In addition, a packet that has arrived beyond the allowable time by the fluctuation absorbing buffer has the same effect as a packet loss on the network side.
[0007]
Therefore, simply combining the network simulation device and the emulator device does not consider the influence of the fluctuation absorbing buffer as described above, so only simulation results different from the actual one can be obtained. -There was a problem that simulation could not be performed with high accuracy to the extent that it could be used for evaluation.
An object of the present invention is to provide a packet communication voice simulation apparatus and method capable of accurately simulating quality degradation in voice packet communication.
[0008]
[Means for Solving the Problems]
In order to achieve such an object, the packet communication voice simulation apparatus according to the present invention simulates the behavior of a packet in the network for each packet transmitted from the transmission side terminal, and each packet in the network is simulated. A network simulation unit that outputs network performance information including the presence / absence of loss and arrival time at the receiving terminal, and the fluctuation absorbing function of the receiving terminal based on the network performance information obtained by the network simulation unit Fluctuation absorption simulation unit that generates degradation processing information indicating the contents of degradation processing for each packet, taking into account the impact on each packet by the fluctuation absorption function, and from the transmission side terminal Sound entered as The information is provided with a voice degradation processing unit that performs degradation processing for each packet based on the degradation processing information from the fluctuation absorption simulation unit and outputs as degraded voice information that is reproduced at the receiving terminal. .
[0009]
When generating degradation processing information, the fluctuation absorption simulation unit uses the minimum buffer size that indicates the fluctuation absorption allowable time of the fluctuation absorption buffer used in the fluctuation absorption function of the receiving terminal, and the maximum capacity that can be stored in the fluctuation absorption buffer. Based on the buffer size, the packet loss determination is performed for each network performance information, and degradation processing information is generated and output from the loss information indicating the loss determination result and the arrival time included in the network performance information. You may do it.
[0010]
As a specific example when generating degradation processing information, the fluctuation absorption simulation unit compares the arrival interval of the packet calculated for each network performance information with the minimum buffer size, and if the arrival interval is larger than the minimum buffer size, Alternatively, it is possible to determine that the packet is a late arrival packet delayed in the network and generate loss information instructing to cause a voice interruption immediately before the packet.
[0011]
As another specific example when generating degradation processing information, the fluctuation absorption simulation unit calculates the packet accumulation amount of the fluctuation absorption buffer for accumulating the packet for each network performance information, and the packet accumulation quantity is the maximum buffer. If it exceeds the size, it may be determined that the packet is missing in the fluctuation absorbing buffer of the receiving terminal, and loss information instructing discard of the packet may be generated.
[0012]
When the minimum buffer size is unknown, select the assumed minimum buffer size that is expected as the minimum buffer size, and send it from the transmitting terminal under various network loads between the transmitting terminal and the receiving terminal. Among the received packets, the lost packets that did not arrive at the receiving terminal and the late arrival packets whose arrival interval between the previous arrival packets exceeded the assumed minimum buffer size were determined as invalid packets, and the number of invalid packets was counted. Then, the ratio of the number of invalid packets to the total number of packets sent from the transmitting terminal is calculated as the invalid packet rate, and the subjective quality evaluation value obtained by the actual subjective quality evaluation by the subject and the invalid packet rate at that time are calculated. Deriving a regression curve showing the correspondence between subjective quality assessment and invalid packet rate, and calculating the contribution rate of the regression curve , Different assumptions minimum buffer among the plurality of regression curves obtained by size, it may be used assuming minimum buffer size that most regression curve its contribution ratio is obtained as the minimum buffer size.
[0013]
Furthermore, when performing subjective quality evaluation in parallel, the fluctuation absorption simulation unit determines the arrival interval between a lost packet that has not arrived at the receiving terminal and the immediately preceding arrival packet among the packets corresponding to the network performance information. The late arrival packet exceeding the assumed minimum buffer size shall be determined as an invalid packet, and an invalid packet rate indicating the ratio of the number of invalid packets determined as invalid packets by the fluctuation absorption simulation unit to the total number of packets is calculated. A subjective quality evaluation estimation unit that outputs a subjective quality evaluation value corresponding to the invalid packet rate may be further provided based on a predetermined estimation model indicating the relationship between the invalid packet rate and the subjective quality evaluation value.
[0014]
In addition, the packet communication voice simulation method according to the present invention simulates the behavior of a packet in the network for each packet sent from the transmission side terminal, and determines whether there is a loss in the network and arrives at the reception side terminal for each packet. Network performance information consisting of the arrival time to be output, and based on these network performance information, the operation of the fluctuation absorbing function of the receiving terminal is simulated, and the influence of each fluctuation absorbing function on each packet is considered. Degradation processing information indicating the content of degradation processing for the packet is generated for each packet, and voice information input as from the sending terminal is subjected to degradation processing for each packet based on the degradation processing information and received. Output as degraded audio information that is played back on the terminal. It is intended.
[0015]
When generating degradation processing information, based on the minimum buffer size indicating the fluctuation absorption allowable time of the fluctuation absorbing buffer used in the fluctuation absorbing function of the receiving terminal, and the maximum buffer size indicating the capacity that can be accumulated in the fluctuation absorbing buffer. Alternatively, the loss determination of the packet may be performed for each network performance information, and the deterioration processing information may be generated from the loss information indicating the loss determination result and the arrival time included in the network performance information.
[0016]
As a specific example of generating degradation processing information, the arrival interval of the packet calculated for each network performance information is compared with the minimum buffer size. If the arrival interval is larger than the minimum buffer size, the packet is delayed in the network. It may be determined that the packet is a late arrival packet, and loss information may be generated instructing to cause a sound interruption immediately before the packet.
[0017]
In addition, as another specific example when generating degradation processing information, when calculating the packet accumulation amount of the fluctuation absorbing buffer that accumulates the packet for each network performance information, and the packet accumulation amount exceeds the maximum buffer size, Alternatively, it may be determined that the packet is missing in the fluctuation absorbing buffer of the receiving terminal, and loss information instructing discard of the packet may be generated.
[0018]
When the minimum buffer size is unknown, select the assumed minimum buffer size that is expected as the minimum buffer size, and send it from the transmitting terminal under various network loads between the transmitting terminal and the receiving terminal. Among the received packets, the lost packets that did not arrive at the receiving terminal and the late arrival packets whose arrival interval between the previous arrival packets exceeded the assumed minimum buffer size were determined as invalid packets, and the number of invalid packets was counted. Then, the ratio of the number of invalid packets to the total number of test packets transmitted from the transmission side is calculated as the invalid packet rate, and the subjective quality evaluation value obtained by the actual subjective quality evaluation by the subject and the invalid packet rate at that time From this, a regression curve indicating the correspondence between subjective quality assessment and invalid packet rate is derived, and the contribution rate of the regression curve is calculated. And, different assumptions minimum buffer among the plurality of regression curves obtained by size, it may be used assuming minimum buffer size that most regression curve its contribution ratio is obtained as the minimum buffer size.
[0019]
Furthermore, when performing subjective quality evaluation in parallel, the arrival interval between the lost packet that did not arrive at the receiving terminal and the packet that arrived immediately before the packet that corresponds to the network performance information exceeds the assumed minimum buffer size. The late arrival packet is determined as an invalid packet, an invalid packet rate indicating a ratio of the number of invalid packets determined as invalid packets to the total number of packets is calculated, and a predetermined relationship between the invalid packet rate and the subjective quality evaluation value is calculated. The subjective quality evaluation value corresponding to the invalid packet rate may be output based on the estimated model.
[0020]
DETAILED DESCRIPTION OF THE INVENTION
Next, embodiments of the present invention will be described with reference to the drawings.
FIG. 1 is a block diagram showing a packet communication voice simulation apparatus according to a first embodiment of the present invention.
The packet communication voice simulation apparatus 10 includes a fluctuation absorption simulation unit 1, a network simulation unit 2, and a voice deterioration processing unit 3. In addition, a multilingual DB 4 and a compression encoding unit 5 are provided as a configuration for inputting voice data used for simulation in the packet communication voice simulation apparatus 10. Further, a compression decoding unit 6 and a deteriorated sound DB 7 are provided as a configuration for outputting and storing the sound data subjected to the deterioration processing by the packet communication sound simulation apparatus 10.
[0021]
The multilingual DB 4 stores sample audio data 4A used for simulation in advance. Here, voice degradation due to communication varies depending on the language and text content, so in addition to Japanese, voice samples of various languages such as English and French, and various text contents are used for wide speech quality evaluation. Are stored as sample audio data 4A.
This sample audio data 4A is a linear PCM audio data sampled at 16 kHz, that is, an audio signal obtained by sampling 16000 times per second, each encoded with 16 bits (hereinafter referred to as a frame) in time series. It is stored in the form of combined digital audio data.
[0022]
FIG. 2 shows the relationship between frames and packets.
As described above, the encoded audio data 5A generated by the compression encoding unit 5 is composed of, for example, a 16-bit frame sequence, and in actual audio packet communication, these multiple frames are stored in one packet. Is transmitted. FIG. 2 shows an example in which three frames are stored in one packet. Here, the frames F1 to F3 are stored in the packet P1, and the frames F4 to F6, F7 to F9, F10 to F12,... Are stored in the packets P2, P3, P4,. In the degradation process in units of packets, it is not necessary to perform the process of adding header information as in an actual packet, and the voice degradation processing unit 3 of the present invention simply recognizes a predetermined number of frames as one packet. Just process it.
[0023]
The compression encoding unit 5 compresses the sample audio data 4A read from the multilingual DB 4 based on a predetermined code compression method, and outputs the compressed audio data 5A. In voice packet communication, normally, in order to transmit voice information efficiently, the original voice data is compressed and transmitted by a predetermined code compression method on the transmission side, and the original voice data is reproduced by decoding on the reception side. Yes. As a code compression method, for example, ITU-T recommendation G.I. 729 and G. 711 or the like is used. When simulating quality degradation in voice packet communication, the compression coding unit 5 performs coding processing based on a code compression method used in actual voice packet communication.
[0024]
On the other hand, the network simulation unit 2 simulates the behavior of each packet by simulating packet contention occurring in each device in the network to be simulated.
As the simulation result, the presence / absence of packet loss in the network and the arrival time when the packet arrives at the receiving side without loss are output as network performance information 2A.
For the network simulation unit 2, an existing device such as the above-described OPNET can be used as it is.
[0025]
Based on the network performance information 2A from the network simulation unit 2, the fluctuation absorption simulation unit 1 outputs, for each packet, degradation processing information 1A indicating the content of the degradation processing for the packet in consideration of the influence of the fluctuation absorption buffer. To do.
The fluctuation absorption simulation unit 1 includes an invalid packet determination unit 11 that determines invalid packets for individual packets, and a loss information setting unit that sets loss information of individual packets based on the determination result of the invalid packet determination unit 11 12 are provided.
[0026]
Based on the degradation processing information 1A from the fluctuation absorption simulation unit 1, the speech degradation processing unit 3 performs degradation processing in units of packets on the encoded speech data 5A to be processed, and outputs it as degraded encoded speech data 3A. To do.
The voice degradation processing unit 3 includes a degradation processing information recognition unit 31 that recognizes degradation processing information 1A corresponding to the encoded voice data 5A read in units of packets, and a degradation process captured by the degradation processing information recognition unit 31. Based on the information 1A, there is provided a degradation processing unit 32 that performs degradation processing on the encoded audio data 5A and outputs it as degraded encoded audio data 3A.
[0027]
FIG. 3 shows a schematic process of the packet communication voice simulation apparatus 10.
The network simulation unit 2 simulates packet behavior in the target network, and generates network performance information 2A. 3, for each packet P1 to P4,..., Network performance information N1 to N4,.
The fluctuation absorption simulation unit 1 generates, for each of these network performance information N1 to N4,..., Degradation processing information Y1 to Y4, which includes arrival time and loss information in consideration of the fluctuation absorption buffer of the receiving terminal. Output. The speech degradation processing unit 3 performs degradation processing of the encoded speech data 5A for each packet based on the degradation processing information Y1 to Y4,..., And outputs it as degraded encoded speech data 3A.
[0028]
The compression decoding unit 6 decodes the degraded encoded audio data 3A from the audio degradation processing unit 3 based on the code compression method used by the compression encoding unit 5, and outputs the result as degraded audio data 6A. The deteriorated audio data 6A has the same data format as the sample audio data 4A.
The deteriorated sound DB 7 stores deteriorated sound data from the compression decoding unit 6.
In addition, what is necessary is just to select the encoding system used by the compression encoding part 5 and the compression decoding part 6 according to the environment where simulation is performed.
[0029]
Next, the operation of the packet communication voice simulation apparatus 10 according to the present embodiment will be described.
The network simulation unit 2 simulates packet behavior in the target network, and generates network performance information 2A in units of packets.
The fluctuation absorption simulation unit 1 takes in the network performance information 2A, and generates and outputs the degradation processing information 1A in units of packets.
[0030]
FIG. 4 is a flowchart showing the fluctuation absorption simulation process in the fluctuation absorption simulation unit 1.
In the fluctuation absorption simulation unit 1, first, the invalid packet determination unit 11 takes in the minimum buffer size of the fluctuation absorption buffer as an attribute of the receiving terminal to be simulated (step 100), and sets the maximum buffer size of the fluctuation absorption buffer. Capture (step 101).
The fluctuation absorbing buffer of the VoIP device used as a receiving side terminal has two reference sizes, that is, a minimum buffer size and a maximum buffer size.
[0031]
The VoIP device has a function of temporarily storing a packet in a fluctuation absorbing buffer without waiting for voice processing immediately after receiving the packet, and always waiting for a certain time before starting the processing. A certain time in this function is called a minimum buffer size (standby buffer size) here. Normally, the arrival packet is processed at a constant speed in accordance with the voice reproduction speed in the VoIP device. Therefore, the packet with late arrival, that is, the late arrival packet causes voice interruption. If this function is used, smooth voice processing can be performed even for a packet that is delayed to some extent. The minimum buffer size corresponds to the allowable time of the late arrival packet.
[0032]
On the other hand, depending on the congestion state of the network, packets may arrive in groups in a burst manner. Even in such a case, the VoIP device needs to process these packets at a constant speed in accordance with the audio playback speed, and has a function of temporarily holding packets that have exceeded the processing speed in the fluctuation absorbing buffer. Have. The maximum value of the capacity that can be held in this function is referred to herein as the maximum buffer size (maximum accommodating buffer size). Therefore, the maximum buffer size corresponds to the maximum value that can be held without losing packets that have arrived above the processing speed.
[0033]
The invalid packet determination unit 11 takes in the values of the minimum buffer size and the maximum buffer size given in advance, and then takes in the network performance information 2A one by one corresponding to each packet, and sends the packet to the fluctuation absorbing buffer. The accumulated state is updated as accumulated (step 102).
Next, as lost packet determination processing, it is checked whether or not the captured network performance information is lost, and it is determined whether or not the packet is lost in the network (step 103).
[0034]
If there is a loss (step 103: YES), it is determined that the packet is lost, and the loss information setting unit 12 sets “loss” as the loss information of the degradation processing information 1A corresponding to the packet (step 104). .
If all the network performance information has not been processed (step 112: NO), the process returns to step 102 and shifts to the process for the next network performance information.
[0035]
On the other hand, if there is no loss in step 103 (step 103: NO), the packet arrival is determined from the difference between the arrival time of the network performance information corresponding to the immediately preceding packet and the arrival time of the network performance information as the late arrival packet determination process. The interval is calculated (step 105), and the arrival interval is compared with the minimum buffer size (step 106).
Here, if the arrival interval is larger than the minimum buffer size (step 106: YES), the packet has arrived after the packet processing wait time, that is, the allowable time, and therefore a break occurs between the previous packet and the packet. The loss information setting unit 12 sets “discontinuous” as the loss information of the degradation processing information 1A corresponding to the packet (step 107), and the process proceeds to step 112.
[0036]
If the arrival interval is equal to or smaller than the minimum buffer size in step 106 (step 106: NO), the packet accumulation amount in the fluctuation absorbing buffer at that time is calculated as the missing packet determination process (step 108). Are compared with the maximum buffer size (step 109).
Here, if the packet accumulation amount exceeds the maximum buffer size (step 109: YES), it is determined that the arrival packet overflows from the fluctuation absorbing buffer and is lost, and the subsequent packet is clogged to be processed next. Then, the loss information setting unit 12 sets “clogged” as the loss information of the degradation processing information 1A corresponding to the packet (step 110), and the process proceeds to step 112.
[0037]
In step 109, if the packet accumulation amount is equal to or smaller than the maximum buffer size (step 109: NO), it is determined that the packet is normally transferred and received by the network, and the loss information setting unit 12 “Normal” is set as the loss information of the degradation processing information 1A corresponding to the packet (step 111), and the process proceeds to step 112.
If all the network performance information processing is completed in step 112 (step 112: YES), the series of fluctuation absorption simulation processing is ended.
[0038]
In this way, the fluctuation absorption simulation unit 1 sets the loss information of the degradation processing information 1A for each network performance information 2A in packet units, and further sets the arrival time of each network performance information 2A corresponding to the degradation processing information 1A. And output to the voice deterioration processing unit 3.
The voice degradation processing unit 3 performs degradation processing on a packet basis for the encoded voice data 5A based on the degradation processing information 1A.
[0039]
FIG. 5 is a flowchart showing the deterioration process in the voice deterioration processing unit 3.
In the speech degradation processing unit 3, first, the degradation processing information recognition unit 31 captures degradation processing information 1A corresponding to a new packet (step 120), and the degradation processing unit 32 converts the encoded speech data 5A into a new packet. A predetermined number of corresponding frames are captured (step 121).
Next, when the loss information of the degradation processing information 1A indicates “normal” (step 122: YES), the corresponding frames fetched from the encoded audio data 5A are output as they are without being degraded (step 123). If all the packets have not been processed (step 129: NO), the process returns to step 120 and shifts to the process for the next packet.
[0040]
If the loss information is not “normal” (step 122: NO) and indicates “interruption” (step 124: YES), the arrival time of the deterioration processing information 1A corresponding to the immediately preceding packet and the deterioration processing information of the packet are displayed. The packet arrival interval is calculated from the arrival time of 1A (step 125). Then, a frame corresponding to the interruption corresponding to the arrival interval (for example, silence: bit value 0) is output (step 126), and then the corresponding frames fetched from the encoded audio data 5A are output as they are without deterioration. (Step 127). Then, the process proceeds to step 129.
[0041]
If the loss information is not “discontinuous” (step 124: NO), it is determined that it is “clogged” or “lost”, and each frame is discarded without being output (step 128). Then, the process proceeds to step 129.
In step 129, when the processing for all the packets is completed (step 129: YES), the series of deterioration processing is terminated.
In this way, the speech degradation processing unit 3 performs degradation processing on the encoded speech data 5A in units of packets based on the degradation processing information 1A, and outputs the degradation speech data 3A.
[0042]
FIG. 6 shows a state of deterioration processing in the voice deterioration processing unit 3.
FIG. 6A shows encoded audio data 5A, and each frame is subjected to a degradation process for each packet P1 to P6,. FIG. 6B shows deterioration processing information 1A, and deterioration processing information Y1 to Y6,... Corresponding to each packet P1 to P6,.
FIG. 6C shows the state of deterioration processing in the sound deterioration processing unit 3, and FIG. 6D shows deterioration encoded sound data 3A.
[0043]
In the speech degradation processing unit 3, for example, when the degradation processing information Y1, the loss information indicates “normal”, so that each frame corresponding to the packet P1 is not degraded as it is as the degradation encoded speech data 3A. Is output.
In the degradation processing information Y2, since the loss information indicates “interruption”, a frame corresponding to the interruption is output immediately before the packet P2, and each frame corresponding to the packet P2 is not deteriorated as it is thereafter. Is output as degraded encoded audio data 3A.
[0044]
In the degradation processing information Y3, since the loss information indicates “clogging”, each frame corresponding to the packet P3 is not output but discarded. Since the loss information of the next degradation processing information Y4 indicates “normal”, each frame corresponding to the packet P4 is output following the frame corresponding to the previous packet P2.
Further, since the loss information of the degradation processing information Y5 indicates “loss”, each frame corresponding to the packet P5 is not output but discarded. Then, since the loss information of the next deterioration processing information Y6 indicates “normal”, each frame corresponding to the packet P6 is output following the frame corresponding to the previous packet P4.
[0045]
As described above, based on the network performance information 2A indicating the behavior of the packet in the network simulated by the network simulation unit 2, the fluctuation absorption simulation unit 1 uses the fluctuation absorption buffer provided in the receiving terminal. The degradation processing information 1A in consideration of the influence is generated in units of packets, and the speech degradation processing unit 3 performs degradation processing on the encoded speech data 5A in units of packets on the basis of the degradation processing information 1A. As compared with the conventional combination of the network simulation device and the emulator device, the influence of the fluctuation absorbing buffer is taken into consideration, so the simulation result is almost the same as the actual simulation result. Obtain and study and evaluate network design and operation It can be accurately simulated to a degree that can be used.
[0046]
For example, when the network simulation device and the emulator device are simply combined, in the example of FIG. 6 described above, only the loss with respect to the packet P5 is output from the network simulation device as the network performance information. Therefore, the interruption in the packet P2 and the clogging of the packet P3 are not reproduced, and these packets P2 and P3 are output as normal packets.
On the other hand, according to this embodiment, it can be seen that the interruption in the packet P2 and the clogging of the packet P3 can be reproduced, and the voice packet communication is simulated accurately.
[0047]
In the packet communication speech simulation apparatus 10, the degradation processing for the encoded speech data 5A is performed separately from the compression encoding unit 5 and the compression decoding unit 6 depending on the simulation environment and the VoIP device. , Versatility can be obtained and can be applied flexibly in a wide range.
Further, in the fluctuation absorption simulation unit 1, the invalid packet determination unit 11 determines whether or not degradation processing is necessary for each packet and the content of the degradation processing based on the minimum buffer size and the maximum buffer size of the fluctuation absorption buffer. Therefore, the influence of the fluctuation absorbing buffer can be checked in detail for each packet.
[0048]
Next, a second embodiment of the present invention will be described with reference to FIG. FIG. 7 is a block diagram showing a packet communication voice simulation apparatus according to the second embodiment.
Compared to the packet communication voice simulation apparatus 10 according to the first embodiment described above, the subjective quality estimation unit 8 is added to the packet communication voice simulation apparatus 10A according to the present embodiment. Is the same as in the first embodiment.
[0049]
Subjective quality evaluation is a form in which an application is finally used, and a person who is a user evaluates the quality. The subjective quality estimation unit 8 estimates the subjective quality evaluation according to the opinion evaluation method used for the overall quality evaluation of the telephone system. The opinion evaluation method is a kind of psychometric method called “Method of Successive Categories”. The test subject evaluates the quality of the voice under test in five levels from "very good" to "very bad" in light of the quality of the telephone used every day. Then, a score of 4 to 0 (or 5 to 1) is given to the evaluation, and a MOS value (Mean Opinion Source) is obtained by calculating the weighted average value based on the voter turnout by many subjects. (See, for example, Asaya, “Quality design of communication networks”, The Institute of Electronics, Information and Communication Engineers, 1993, Mar, etc.).
[0050]
In the subjective quality estimation unit 8, based on the determination result in the invalid packet determination unit 11 of the fluctuation absorption simulation unit 1, the invalid packet indicating the proportion of invalid packets that are not reproduced by the VoIP device out of all packets corresponding to the network performance information 2 </ b> A. The rate is calculated, and the MOS value is estimated from the invalid packet rate. At this time, the relationship between the invalid packet rate and the MOS value is modeled in advance, and the MOS value is estimated from the invalid packet rate using the estimation model. In this embodiment, a regression curve is used as this estimation model, but the present invention is not limited to this. The subjective quality evaluation value is not limited to the MOS value, and other evaluation values may be used.
[0051]
Next, the subjective quality estimation operation will be described with reference to FIG. FIG. 7 is a block diagram showing the subjective quality estimation unit. The subjective quality estimation unit 8 includes an invalid packet rate calculation unit 81 that calculates an invalid packet rate based on a determination result in the invalid packet determination unit 11, and an invalid packet rate calculated by the invalid packet rate calculation unit 81. An evaluation estimation unit 82 that outputs a MOS value 84 indicating an estimation result of subjective quality with reference to an estimation model 83 indicating a correspondence relationship between an invalid packet rate and a MOS value.
[0052]
The invalid packet rate calculation unit 81 calculates the rate of occurrence of invalid packets determined to be lost packets or late arrival packets, that is, the invalid packet rate among the determination results of the invalid packet determination unit 11.
Here, a lost packet and a late arrival packet are assumed as invalid packets that do not function normally in a VoIP device and cause a reduction in subjective quality. The lost packet refers to a packet lost on the network for some reason, that is, a packet in which “loss” is set as loss information. The late arrival packet refers to a packet that has arrived later than usual for some reason and has not been normally reproduced by the VoIP device, that is, a packet in which “clogged” is set as loss information.
[0053]
FIG. 8 shows a method for determining lost packets and late arrival packets.
In the lost packet determination (step 103 in FIG. 4) of the invalid packet determination unit 11 described above, out of the network performance information 2A of each packet, a packet indicating that there is a loss is determined to be a lost packet. For example, a packet that has been determined to have not arrived from the packet sequence number assigned in the transmission order of the transmission packet is considered to be lost. In FIG. 8, packet P _Three , P ₇ Are determined to be lost packets.
[0054]
In the late arrival packet determination process (steps 105 and 106 in FIG. 4) of the invalid packet determination unit 11, the arrival interval of each packet is calculated from the arrival time of the network performance information 2A, and the arrival interval and the minimum buffer size (allowable) (Time), a late arrival packet is determined, and a packet corresponding to the late arrival packet is determined as a lost packet. In FIG. 8, the arrival interval t ₁ ~ T ₇ Is calculated and the fluctuation absorption allowable time t _BS (For example, 30 ms). In this case t ₆ > T _BS Packet P ₉ Is determined as a late arrival packet.
It is also possible to include lost packets that have been determined as lost packets by the invalid packet determination unit 11 (steps 108 and 109 in FIG. 4).
[0055]
FIG. 9 is a flowchart showing the subjective quality estimation process in the subjective quality estimation unit 8.
In the invalid packet rate calculation unit 81, the number of lost packets P determined by the lost packet determination of the invalid packet determination unit 11. _L , And late arrival packet number P determined by late arrival packet determination _D Are counted (step 140), and the invalid packet rate R is calculated as follows. _P Is calculated (step 141).
R _P = (P _L + P _D ) / Total number of transmitted packets
The evaluation estimation unit 82 estimates the MOS value 84 by referring to the estimation model 83 based on the invalid packet rate calculated by the invalid packet rate calculation unit 81 (step 142), and ends a series of subjective quality estimation processes.
[0056]
The estimation model 83 here is a model in which the relationship between both is modeled in advance using a regression curve with the invalid packet rate obtained by actual measurement or the like as an explanatory variable and the MOS value at that time as an objective variable.
As the regression curve, any one of various functions such as a quadratic function and an exponential function may be selected as necessary.
[0057]
FIG. 10 is a regression curve showing the relationship between the invalid packet rate and the MOS value, and FIG. 11 is a regression curve showing the relationship between the packet loss rate and the MOS value.
These regression curves are obtained by actually generating various network loads and actually testing them. For the MOS value, 20 subjects evaluated each of the 224 condition audio files recorded by playing audio over the network, and the average value was used by calculating the MOS value for the audio file. . In addition, the invalid packet rate and the packet loss rate at this time were calculated, and a regression curve was obtained from these correspondences. The packet loss rate is the ratio of the number of lost packets to the total number of transmitted packets, and the regression curve in FIG. 11 does not take into account late arrival packets with respect to the fluctuation absorbing buffer.
[0058]
Comparing FIG. 10 and FIG. 11, it can be seen that the variation of FIG. 10 is smaller than that of FIG. For example, when the invalid packet rate of FIG. 10 is 0%, the MOS value is in the range of 2.9 to 4.3, whereas when the packet loss rate of FIG. 11 is 0%, the MOS value is 1.7 to It is in the range of 4.3, and there is a variation of 1.8 times or more compared to FIG. Also, the contribution rate (determination coefficient) of the regression curve is 0.8215 in FIG. 10 and 0.6999 in FIG. 11, and FIG. 10 shows a better fit to the regression curve data. You can see that Therefore, the MOS value can be estimated with higher accuracy by considering the late arrival packet exceeding the fluctuation absorption allowable time.
[0059]
Thus, in the invalid packet determination unit 11, from the network performance information 2A for each packet, the number of lost packets that have not arrived at the receiving side terminal and the arrival delay when the packet arrival interval exceeds the minimum buffer size of the fluctuation absorbing buffer. The invalid packet rate calculation unit 81 calculates the ratio of the number of invalid packets to the total number of transmitted packets, that is, the invalid packet rate, from the determination result, and the evaluation estimation unit 82 calculates the invalid packet rate. Based on the above, the estimation model 83 indicating the relationship between the invalid packet rate and the MOS value prepared in advance is referred to and the corresponding MOS value is estimated and output. Therefore, the accuracy without performing the actual subjective quality. The subjective quality can be estimated well.
[0060]
Therefore, it is possible to obtain subjective quality (for example, MOS value) from the same network performance information 2A in parallel with the degraded speech data 6A obtained from the degradation encoded speech data 3A subjected to degradation processing by the speech degradation processing unit 3. it can. As a result, when the deteriorated voice data 6A is used for the examination and evaluation of the design and operation of the network, each deteriorated voice data 6A can be selected with the subjective quality, or any other obtained when any deteriorated voice data 6A is used. The evaluation result and subjective quality can be compared, and network design and operation can be examined and evaluated in more detail and accurately.
[0061]
Next, a case where the minimum buffer size is unknown will be described.
In the first embodiment and the second embodiment described above, the case has been described in which the minimum buffer size of the fluctuation absorbing buffer of the VoIP device is known, but the minimum buffer size may be unknown.
In such a case, an equivalent minimum buffer size closest to the actual minimum buffer size may be determined and used.
[0062]
FIG. 12 is a flowchart showing the minimum buffer size estimation process.
First, a new minimum buffer size is assumed (step 150).
The relationship between the assumed minimum buffer size and the invalid packet rate is represented in a graph as shown in FIG. The lowest value of the invalid packet rate corresponds to the lost packet rate. An assumed minimum buffer size is selected within the range of this graph.
Then, using the assumed minimum buffer size selected instead of the above-mentioned minimum buffer size, the actual subjective quality evaluation by the subject is performed (step 151), and the number of invalid packets is counted as described above (step 152).
[0063]
Subsequently, the invalid packet rate is calculated (step 153), and the regression curve and the contribution rate of the regression curve are obtained for each assumed minimum buffer size (step 154).
FIG. 14 shows a correspondence example between the assumed minimum buffer size and the contribution rate.
This contribution rate is an evaluation index indicating the goodness of fit to regression curve data (in this case, invalid packet rate and MOS value), and is generally a coefficient of determination R in regression analysis. ² It is also called.
[0064]
In this way, different hypothetical minimum buffer sizes are sequentially selected and steps 150 to 154 are repeatedly executed (step 155), thereby obtaining a regression curve for each hypothetical minimum buffer size.
Then, among the obtained regression curves, the regression curve having the highest contribution rate is used as the estimation model, the assumed minimum buffer size at that time is selected as the estimated minimum buffer size (step 156), and the actual minimum buffer size is selected. It may be used instead of. In FIG. 14, the contribution ratio = 0.8215 is the highest, and the assumed minimum buffer size at this time is used as the minimum buffer size.
[0065]
As a result, even when the minimum buffer size of the VoIP device is unknown, the degradation processing and the subjective quality evaluation are accurately estimated using the minimum buffer size closest to the actual minimum buffer size and considering the influence of the fluctuation absorbing buffer. Can do. As for the method of selecting the assumed minimum buffer size, the assumed minimum buffer size is changed at a somewhat large interval when the contribution rate is low, and the assumed minimum buffer size is changed by decreasing the interval as the contribution rate increases. The effective assumed minimum buffer size can be found efficiently.
[0066]
In each of the above embodiments, the IP network using the VoIP device has been described as an example. However, the present invention is not limited to this, and any communication method can be used as long as it is a data transmission method using packets. Applicable and similar effects can be obtained.
[0067]
【The invention's effect】
As described above, in the present invention, the network simulation unit simulates the packet behavior in the network for each packet transmitted from the transmission side terminal, and the presence or absence of loss in the network and the reception side terminal for each packet. The network performance information consisting of the arrival time of arrival at the network is output, and based on this network performance information, the fluctuation absorption simulation unit simulates the operation of the fluctuation absorption function of the receiving terminal, and each fluctuation absorption function Deterioration processing information indicating the contents of the deterioration processing for the packet is generated for each packet in consideration of the influence on the packet, and the voice deterioration processing unit absorbs fluctuations in the voice information input from the transmitting terminal. Based on the degradation processing information from the simulation unit, The degradation process is performed on a per-unit basis, and output as degraded audio information that is played back on the receiving terminal. Compared to a simple combination of a network simulation device and an emulator device as in the past, Since the influence of the fluctuation absorbing buffer is taken into consideration, a simulation result almost the same as the actual one can be obtained, and the simulation can be performed with a high degree of accuracy so that it can be used for the examination and evaluation of the network design and operation.
[Brief description of the drawings]
FIG. 1 is a block diagram showing a packet communication voice simulation apparatus according to a first embodiment.
FIG. 2 is an explanatory diagram showing a relationship between a frame and a packet.
FIG. 3 is an explanatory diagram showing a schematic process of the packet communication voice simulation apparatus.
FIG. 4 is a flowchart showing fluctuation absorption simulation processing.
FIG. 5 is a flowchart showing deterioration processing.
FIG. 6 is an explanatory diagram showing a state of deterioration processing in a sound deterioration processing unit.
FIG. 7 is a block diagram showing a packet communication voice simulation apparatus according to a second embodiment.
FIG. 8 is an explanatory diagram showing a method for determining lost packets and late arrival packets.
FIG. 9 is a flowchart showing a subjective quality estimation process.
FIG. 10 is a regression curve showing the relationship between the invalid packet rate and the MOS value.
FIG. 11 is a regression curve showing the relationship between the packet loss rate and the MOS value.
FIG. 12 is a flowchart showing a minimum buffer size estimation process.
FIG. 13 is a graph showing a relationship between an assumed allowable time and an invalid packet rate.
FIG. 14 is a correspondence example between an assumed minimum buffer size and a contribution rate.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 10,10A ... Packet communication voice simulation apparatus, 1 ... Fluctuation absorption simulation part, 11 ... Invalid packet determination part, 12 ... Loss information setting part, 1A ... Degradation processing information, 2 ... Network simulation part, 2A ... Network performance information, 3 ... voice deterioration processing unit, 31 ... deterioration processing information recognition unit, 32 ... deterioration processing unit, 3A ... deterioration encoded voice data, 4 ... multilingual DB, 4A ... sample voice data, 5 ... compression encoding part, 5A ... code Voice data, 6 ... compression decoding unit, 6A ... degraded voice data, 7 ... degraded voice DB, 8 ... subjective quality estimation unit, 81 ... invalid packet rate calculation unit, 82 ... evaluation estimation unit, 83 ... estimation model, 84 ... MOS value.

Claims (12)

Simulated voice quality degradation that occurs when voice information is packetized and transmitted between a transmitting side terminal and a receiving side terminal connected via a network, and the degraded voice played back by the receiving side terminal In a packet communication voice simulation apparatus that outputs information,
Simulate packet behavior in the network for each packet sent from the transmitting terminal, and network performance consisting of the presence or absence of loss in the network and arrival time at the receiving terminal for each packet A network simulation unit for outputting information;
Based on the network performance information obtained by this network simulation unit, the operation of the fluctuation absorbing function of the receiving terminal is simulated, and the degradation process for the packet in consideration of the influence on each packet by the fluctuation absorbing function A fluctuation absorption simulation unit that generates degradation processing information indicating the contents of each packet,
The voice information input from the transmission side terminal is subjected to degradation processing for each packet based on the degradation processing information from the fluctuation absorption simulation unit, and the voice information after degradation reproduced at the reception side terminal A packet communication voice simulation apparatus, comprising: The packet communication voice simulation apparatus according to claim 1,
The fluctuation absorption simulation unit is based on the minimum buffer size indicating the fluctuation absorption allowable time of the fluctuation absorption buffer used in the fluctuation absorption function of the receiving terminal and the maximum buffer size indicating the capacity that can be accumulated in the fluctuation absorption buffer. Determining the loss of the packet for each network performance information, and generating and outputting the degradation processing information from the loss information indicating the loss determination result and the arrival time included in the network performance information, Packet communication voice simulation device. The packet communication voice simulation apparatus according to claim 2,
The fluctuation absorption simulation unit compares the arrival interval of the packet calculated for each network performance information and the minimum buffer size, and if the arrival interval is larger than the minimum buffer size, the delay arrival delayed in the network. A packet communication voice simulation apparatus characterized by generating loss information that is judged to be a packet and instructing the voice to be interrupted immediately before the packet. The packet communication voice simulation apparatus according to claim 2,
The fluctuation absorption simulation unit calculates a packet accumulation amount of the fluctuation absorption buffer for accumulating the packet for each network performance information, and when the packet accumulation amount exceeds the maximum buffer size, A packet communication voice simulation apparatus, characterized in that it determines that a packet is missing in the fluctuation absorbing buffer and generates loss information instructing discard of the packet. The packet communication voice simulation apparatus according to claim 1 or 2,
If the minimum buffer size is unknown,
Select an assumed minimum buffer size expected as the minimum buffer size;
Under various network loads between the transmission side terminal and the reception side terminal, among the packets sent from the transmission side terminal, the arrival of the lost packet that did not arrive at the reception side terminal and the previous arrival packet The late arrival packets whose interval exceeds the assumed minimum buffer size are determined as invalid packets, and the number of invalid packets is counted.
Calculating the ratio of the number of invalid packets to the total number of packets sent from the transmission side terminal as an invalid packet rate;
From the subjective quality evaluation value obtained by the actual subjective quality evaluation by the subject and the invalid packet rate at that time, a regression curve indicating the correspondence between the subjective quality evaluation and the invalid packet rate is derived, and the contribution of the regression curve Calculate the rate,
A packet communication voice simulation apparatus characterized by using, as the minimum buffer size, an assumed minimum buffer size from which a regression curve having the highest contribution ratio among a plurality of regression curves obtained for different assumed minimum buffer sizes is used. . The packet communication voice simulation apparatus according to claim 1 or 2,
The fluctuation absorption simulation unit is a late arrival packet in which an arrival interval between a lost packet that has not arrived at a receiving terminal and a previous arrival packet among the packets corresponding to the network performance information exceeds the assumed minimum buffer size. Are considered invalid packets,
Based on a predetermined estimation model indicating the relationship between the invalid packet rate and the subjective quality evaluation value, calculating the invalid packet rate indicating the ratio of the invalid packet number determined to be invalid packets by the fluctuation absorption simulation unit to the total number of packets. A packet communication voice simulation apparatus, further comprising a subjective quality evaluation estimation unit that outputs a subjective quality evaluation value corresponding to the invalid packet rate. Simulated voice quality degradation that occurs when voice information is packetized and transmitted between a transmitting side terminal and a receiving side terminal connected via a network, and the degraded voice played back by the receiving side terminal In a packet communication voice simulation method for outputting information,
Simulate packet behavior in the network for each packet sent from the transmitting terminal, and network performance consisting of the presence or absence of loss in the network and arrival time at the receiving terminal for each packet Output information,
Based on these network performance information, the operation of the fluctuation absorbing function of the receiving terminal is simulated, and the deterioration processing information indicating the contents of the deterioration processing for the packet in consideration of the influence on each packet by the fluctuation absorbing function For each packet,
The audio information input from the transmitting terminal is subjected to deterioration processing for each packet based on the deterioration processing information, and output as deteriorated audio information reproduced by the receiving terminal. Packet communication voice simulation method. The packet communication voice simulation method according to claim 7,
When generating the degradation processing information, the minimum buffer size indicating the allowable fluctuation absorption time of the fluctuation absorbing buffer used in the fluctuation absorbing function of the receiving terminal, and the maximum buffer size indicating the capacity that can be accumulated in the fluctuation absorbing buffer Based on the loss information indicating the loss determination result and the arrival time included in the network performance information, the degradation processing information is generated based on the loss determination of each packet based on the network performance information. Packet communication voice simulation method. The packet communication voice simulation method according to claim 8, wherein
When generating the degradation processing information, the arrival interval of the packet calculated for each network performance information is compared with the minimum buffer size. If the arrival interval is larger than the minimum buffer size, the packet is delayed in the network. A packet communication voice simulation method characterized by generating loss information that determines that a packet is a late arrival packet and instructs to cause a voice interruption immediately before the packet. The packet communication voice simulation method according to claim 8, wherein
When generating the degradation processing information, calculate the packet accumulation amount of the fluctuation absorbing buffer that accumulates the packet for each network performance information, and if the packet accumulation amount exceeds the maximum buffer size, the receiving side A packet communication voice simulation method, characterized in that it is determined as a lost packet that has been lost in a fluctuation absorbing buffer of a terminal, and loss information that instructs to discard the packet is generated. The packet communication voice simulation method according to claim 7 or 8,
If the minimum buffer size is unknown,
Select an assumed minimum buffer size expected as the minimum buffer size;
Under various network loads between the transmission side terminal and the reception side terminal, among the packets sent from the transmission side terminal, the arrival of the lost packet that did not arrive at the reception side terminal and the previous arrival packet The late arrival packets whose interval exceeds the assumed minimum buffer size are determined as invalid packets, and the number of invalid packets is counted.
Calculate the ratio of the invalid packet number to the total number of test packets transmitted from the transmission side as an invalid packet rate,
From the subjective quality evaluation value obtained by the actual subjective quality evaluation by the subject and the invalid packet rate at that time, a regression curve indicating the correspondence between the subjective quality evaluation and the invalid packet rate is derived, and the contribution of the regression curve Calculate the rate,
A packet communication voice simulation method characterized by using, as the minimum buffer size, an assumed minimum buffer size from which a regression curve having the highest contribution ratio among a plurality of regression curves obtained for different assumed minimum buffer sizes is used. . The packet communication voice simulation method according to claim 7 or 8,
Of each packet corresponding to the network performance information, a lost packet that has not arrived at the receiving terminal and a late arrival packet whose arrival interval between the immediately preceding arrival packets exceeds the assumed minimum buffer size are determined as invalid packets. ,
An invalid packet rate indicating a ratio of the number of invalid packets determined to be invalid packets to the total number of packets is calculated, and the invalid packet rate is calculated based on a predetermined estimation model indicating a relationship between the invalid packet rate and the subjective quality evaluation value. A packet communication voice simulation method characterized by outputting a corresponding subjective quality evaluation value.

Images