JP2019050455A - Network quality estimation method and network quality estimation device - Google Patents

Abstract

To provide a network quality estimation method and a network quality estimation device applicable to a streaming application capable of measuring easily the quality of an IP network without giving any influences to other traffic at low cost.SOLUTION: Disclosed network quality estimation method and the network quality estimation device are configured to calculate the packet loss rate and the packet order exchange rate by using a sequence number s(k) of the k-th received packet and a value with the sequence number s(k-1) of a packet, which is received one before that, on the basis of the "deviation amount G(k)" which is the differential value as the value that would be obtained as the sequence number s(k). With this, the network quality is calculated by addition and subtraction only without transmitting any dummy packet for network quality measurement.SELECTED DRAWING: Figure 1

Description

  The present disclosure relates to a network quality estimation method and network quality estimation apparatus for evaluating network quality.

  In a streaming application that packetizes video and audio and delivers it in real time via an IP network, the quality of the IP network has a significant impact on the quality of video / audio to be transmitted. "Packet loss rate", which is the occurrence rate of events where IP packets sent from a sending device are lost in the network, or the occurrence rate of events where the arrival order of IP packets sent sequentially from the sending device is reversed One "packet reordering rate" is an example of a well-known parameter for evaluating the quality of an IP network. Packet loss in IP networks can cause interruptions in received video / audio. Packet reordering in IP networks can cause disturbances in received video / audio. Therefore, properly evaluating or measuring the packet loss rate and packet reordering rate of the IP network is an essential technology for providing streaming applications with less quality degradation.

  An IP network quality measurement device that sends dummy packets to an IP network to estimate the quality of the IP network is already commercially available. However, such devices are generally expensive. In addition, since the dummy packet imposes a load on the IP network, the IP network quality measurement apparatus is not suitable for quality measurement of a commercially available IP network. Therefore, it is desirable to provide a technology that can measure the quality of an IP network without affecting other traffic.

  Patent Document 1 discloses a technique for measuring the quality of an IP network without using a dummy packet by capturing (capturing) and evaluating an IP packet flowing in the IP network at one point of the IP network. It is done. Further, Patent Document 2 discloses a technique in which a receiver for an IP packet is provided with a circular buffer, and the packet loss rate of the received IP packet is measured based on the detection latency determined from the chain length of the circular buffer.

JP, 2005-210515, A Japanese Patent Application Publication No. 2006-510255

  The techniques disclosed in Patent Documents 1 and 2 can accurately measure the quality of an IP network such as a packet loss rate without affecting other traffic. However, in these techniques, it is necessary to prepare a special device (a device for capturing an IP packet in Patent Document 1 and a circulating buffer in Patent Document 2), and there are problems in operation and cost.

  Therefore, the present invention provides a network quality estimation method and a network quality estimation apparatus which can be applied to streaming applications and can measure the quality of an IP network more inexpensively and easily without affecting other traffic. The purpose is to

  In order to achieve the above object, a network quality estimation method and a network quality estimation apparatus according to the present invention grasp the order and omission of sequence numbers assigned to packets, and generate the packet loss and packet order change ratio of the target network. It was decided to estimate

Specifically, the network quality estimation method according to the present invention is
A sequence number obtaining procedure for obtaining a sequence number indicating an order of transmission from a currently received current packet;
A sequence number prediction procedure for predicting the sequence number of the current packet based on the sequence number of the immediately preceding packet received immediately before the current packet;
A network quality estimation procedure for estimating network quality based on a deviation amount which is a difference value between the sequence number of the current packet acquired in the sequence number acquisition procedure and the sequence number of the current packet predicted in the sequence number prediction procedure; ,
I do.

Further, a network quality estimation apparatus according to the present invention is
A sequence number acquisition unit for acquiring a sequence number indicating an order of transmission from a currently received current packet;
A holding unit that holds the sequence number of the packet acquired by the sequence number acquisition unit;
The difference value between the sequence number of the current packet acquired by the sequence number acquisition unit and the sequence number of the current packet predicted based on the sequence number of the immediately preceding packet received immediately before the current packet held by the holding unit. A shift amount calculation unit that calculates a certain shift amount;
A network quality calculation unit that estimates network quality based on the deviation amount calculated by the deviation amount calculation unit;
Equipped with

  The present network quality estimation method and the present network quality estimation apparatus are based on the assumption that the packet to be measured has a "sequence number" indicating the transmission order. Usually, a "sequence number" which is a counter value incremented by one according to the transmission order is given to the header of an RTP (Real-Time Protocol) packet used for a streaming application.

  The present network quality estimation method and the present network quality estimation apparatus sequentially extract the “sequence number s (k)” of the packet from the received packet. s (k) means that it is the sequence number of the kth packet received by the measuring device. The present network quality estimation method and the present network quality estimation apparatus are characterized in that the sequence number s (k) of the kth received packet and the sequence number s (k−1) of the immediately preceding k−1th received packet Each time a packet is received, “deviation amount G (k)” which is a difference between the sequence number s (k) and the wax value (predicted current packet sequence number) obtained from the sequence number s) is calculated. Then, for the network quality evaluation, a packet loss rate and a packet order change rate, which will be described later, are simply calculated using this “amount of shift”.

  It is possible to estimate the occurrence rate of packet loss and packet reordering in the network simply by mounting the present network quality estimation apparatus in the receiving apparatus or connecting it to a part of the network. Therefore, the present invention provides a network quality estimation method and a network quality estimation apparatus that can be applied to streaming applications and can measure IP network quality cheaper and easier without affecting other traffic. can do.

  The present invention provides a network quality estimation method and a network quality estimation apparatus which can be applied to streaming applications and can measure the quality of an IP network more inexpensively and easily without affecting other traffic. Can.

It is a figure explaining a packet structure. It is a figure explaining a packet loss rate and a packet order change rate. It is a figure explaining a receiver provided with a network quality estimating device concerning the present invention. It is a figure explaining the network quality calculation part of the network quality estimating device concerning the present invention. It is a figure explaining the network quality estimation method concerning the present invention. It is a figure explaining the special case in the network quality estimation method concerning the present invention. It is a figure explaining the special case in the network quality estimation method concerning the present invention. It is a figure explaining the special case in the network quality estimation method concerning the present invention. It is a figure explaining the special case in the network quality estimation method concerning the present invention.

  Embodiments of the invention will be described with reference to the accompanying drawings. The embodiments described below are examples of the present invention, and the present invention is not limited to the following embodiments. In the present specification and drawings, components having the same reference numerals denote the same components.

[Packet Configuration]
The configuration of a packet used in the present embodiment will be described. The packet transmitted on the network (this is a packet transmitted by the transmitting device and received by the receiving device) is a counter value that is incremented by one according to the order in which the transmitting device transmits the packet. The sequence number is assigned. The sequence number is a numerical value representing the order in which packets are transmitted. The sequence number is stored in a predetermined area of the header (including the extension header) of the packet. Alternatively, the sequence number may be stored in a predetermined area provided in the payload of the packet.

  A more specific example of the packet configuration that can be used in the present invention will be described using FIG. Encoded data obtained by encoding video, audio and the like is stored in the payload of a Real-Time Protocol (RTP) packet. An RTP header is attached to the RTP payload to form an RTP packet. The RTP header is, as shown in FIG. 1, version (V), padding (P), extension header (X), CSRC count (CC), marker (M), payload type, sequence number, time stamp, synchronization Each area of a transmission source (SSRC) identifier is provided. The sequence number field of the RTP header composed of 16 bits stores a sequence number starting from 1 and incremented by 1 according to the transmission order of RTP packets. The RTP packet is stored in a UDP (User Datagram Protocol) payload. A UDP header is attached to the UDP payload to form a UDP packet. UDP packets are stored in an IP (Internet Protocol) payload. An IP header is attached to the IP payload to form an IP packet. The IP packet is transmitted from the transmitting device to the receiving device via the IP network.

  The packet that can be used in the present invention is not limited to the above-described RTP packet. The present invention can be applied to packets of any format as long as a sequence number indicating the transmission order of the packets is added in the header or payload of the packet.

[Outline of processing]
The outline of the network quality estimation process will be described with reference to FIG. FIG. 2A is a diagram for explaining an outline of a method of estimating a “packet loss rate”, and FIG. 2B is a diagram for explaining an outline of a method of estimating a “packet order change rate”.

  It is assumed that the IP packet transmitter has transmitted 10 IP packets each having sequence numbers 1, 2, 3,..., 10, which are incremented by 1 from 1 to 10. These ten IP packets are received by the receiving device via the IP network. As shown in FIGS. 2 (a) and 2 (b), the receiving apparatus has nine sequence numbers s (k) = 1, 3, 4, 5, 6, 8, 9, 9, 7, and 10 in this order. Suppose that an IP packet is received.

ここで、差分値Ｇ（ｋ）を「ずれ量」と呼ぶこととする。 Here, the sequence number of the IP packet received kth from the start of measurement by the receiving apparatus is represented as s (k). The receiving device obtains the value (s (k-1) +1) which was obtained as the sequence number s (k) of the latest IP packet each time it receives the k-th (ie, the latest) IP packet. Predict. Then, the receiving apparatus calculates a difference value G (k) between the latest sequence number s (k) and the predicted sequence number s (k-1) +1.

Here, the difference value G (k) will be referred to as the “displacement amount”.

  If the kth received IP packet is transmitted after one of the k−1th received IP packet, the sequence number s (k) is larger than the sequence number s (k−1) by one. The deviation amount G (k) is 0 because it is supposed to be. On the other hand, when a packet loss or a change in packet arrival order occurs, the sequence number s (k) does not become a value larger by one than the sequence number s (k-1). For example, when one packet loss occurs, the sequence number s (k) should be a value larger by 2 than the sequence number s (k−1), so the deviation amount G (k) is 1. In the example of FIG. 2A and FIG. 2B, the sequence number of the IP packet received at the second (k = 2) is s (k) = 3, and one before that (k = 1) Since the sequence number of the received IP packet is s (k) = 1, the shift amount calculated at this time is G (k) = 3− (1 + 1) = + 1. The receiving apparatus calculates the amount of deviation G (k) each time an IP packet is received, and uses this “amount of deviation G (k)” to estimate the packet loss rate and the packet order change rate.

(Packet loss rate)
An outline of the packet loss rate estimation process will be described with reference to FIG. When the receiving device receives the last IP packet (k = 9 in FIG. 2A) of the predetermined period, it adds all deviation amounts G (k) obtained so far, and adds up the accumulated deviation amount ΣG ( k) (1 in the example of FIG. 2A) is calculated. The receiving apparatus derives, as the packet loss rate P _Loss , the ratio of the maximum sequence number S _max (10 in the example of FIG. 2A) and the cumulative deviation amount 所 定 G (k) in the IP packet received in a predetermined period.

累積ずれ量フラグΣＮＧ（ｋ）は、パケット到着順序の入れ替えが発生したパケット数を表している。なお、「これまでに得られたすべてのずれ量フラグＮＧ（ｋ）を加算」することは、「負数のずれ量Ｇ（ｋ）をカウントする」ことと等価である。 (Packet order change rate)
Next, with reference to FIG. 2B, an outline of the estimation process of the packet order change rate will be described. The receiving device evaluates whether the calculated deviation amount G (k) is a negative value each time an IP packet is received. When the calculated shift amount G (k) is a negative value, the shift amount flag NG (k) = 1 in the k-th received IP packet is set. Otherwise, the deviation amount flag NG (k) is set to 0 (see Formula C3). When the receiving device receives the last IP packet (k = 9 in FIG. 2B) for a predetermined period, it adds all the shift amount flags NG (k) obtained so far, and the cumulative shift amount flag ΣNG (k) (1 in the example of FIG. 2B) is calculated. The receiving apparatus derives the ratio of the maximum sequence number S _max (10 in the example of FIG. 2B) in the IP packet received in a predetermined period and the cumulative deviation amount flag NGNG (k) as the packet order change rate P _Reorder. Do.

The cumulative deviation amount flag NGNG (k) represents the number of packets in which the packet arrival order has been switched. Note that “adding all the shift amount flags NG (k) obtained so far” is equivalent to “count the shift amount G (k) of a negative number”.

In the above, the maximum sequence number S _max in the received IP packet is used as the denominator for calculating the packet loss rate and the packet order change rate, but the present invention is not limited to this. The point is that the number of transmit packets corresponding to the receive packet to be measured (in the example of FIGS. 2A and 2B, the number of receive packets is 9, but one packet loss has occurred in the network) Therefore, the number of transmitted packets may be estimated as 10) and used as a denominator for calculating the packet loss rate and the packet order change rate. For example, S _max −S _min −1 may be used as a denominator by using the largest sequence number S _max and the smallest sequence number S _min among the received IP packets (Equation C2 and Equation C4). In this case, even if the sequence number of the IP packet transmitted first among the measurement targets is other than 1, the packet loss rate and the packet order change rate can be calculated.

でパケット損失率Ｐ_Ｌｏｓｓを、

でパケット順序入替率Ｐ_{Ｒｅｏｒｄｅｒ}を導出する。また、あらかじめ測定対象のパケット数が決まっている場合は、送信装置がパケットのヘッダ等に送信パケット数に関する情報を格納することで、送信装置が受信装置に伝達するようにしてもよい。 That is, when the minimum value S _min of the sequence number of the IP packet for a predetermined period is not 1,

Packet loss rate P _Loss ,

The packet order change rate P _Reorder is derived by In addition, when the number of packets to be measured is determined in advance, the transmitting device may transmit information to the number of transmission packets in the header of the packet or the like to transmit the information to the receiving device.

[Network quality estimation device]
FIG. 3 is a diagram for explaining the configuration of a receiving apparatus 301 provided with the network quality estimating apparatus 101 according to the present embodiment. The receiving device 301 has a receiving buffer 11 for receiving a packet from the network, a packet processing unit 12 for receiving and processing the packet, and a network quality estimation device 101. The network quality estimation device 101 is incorporated in the receiving device 301 that receives an IP packet, or is configured integrally with the receiving device 301. The network quality estimation apparatus 101 is notified of the sequence number from the packet processing unit 12, and as the network quality, "packet loss rate" which is the occurrence rate of an event that an IP packet transmitted from the transmitting apparatus is lost in the network. The “packet order change rate”, which is the occurrence rate of an event in which the arrival order of IP packets sequentially transmitted from the transmission apparatus is switched, is estimated according to the principle described with reference to FIG.

The network quality estimation device 101
A sequence number obtaining unit 13 for obtaining a sequence number indicating the order of transmission from the current packet currently being received;
A holding unit 14 that holds the sequence number of the packet acquired by the sequence number acquiring unit 13;
A difference value between the sequence number of the current packet acquired by the sequence number acquisition unit 13 and the sequence number of the current packet predicted based on the sequence number of the immediately preceding packet received immediately before the current packet held by the holding unit 14 A shift amount calculation unit 15 that calculates a shift amount which is
A network quality calculation unit 16 that estimates network quality based on the deviation amount calculated by the deviation amount calculation unit 15;
Equipped with

  The sequence number acquisition unit 13 acquires the sequence number s (k) included in the IP packet notified each time the packet processing unit 12 receives the IP packet. The sequence number acquisition unit 13 outputs the sequence number s (k) of the latest (k-th) IP packet to the holding unit 14 and causes the holding unit 14 to hold it.

The holding unit 14 is the sequence number s (k) of the latest IP packet, the sequence number s (k-1) of the IP packet received immediately before that, and the sequence number held by the holding unit 14 for a predetermined period. The small sequence number S _min and the largest sequence number S _max are held.

  The shift amount calculation unit 15 calculates the sequence number s of the latest IP packet from the value of the sequence number s (k−1) of the IP packet received immediately before (k−1) which the holding unit 14 holds. Predict the wax value (s (k-1) +1) that was obtained as (k). Then, the deviation amount calculation unit 15 calculates the deviation amount G (k) that is the difference value between the latest sequence number s (k) acquired by the sequence number acquisition unit 13 and the predicted sequence number s (k-1) +1. Are calculated and output to the network quality calculation unit 16.

  FIG. 4 is a diagram for explaining a specific configuration of the network quality calculation unit 16. The network quality calculation unit 16 includes a packet loss rate calculation unit 20 that calculates a packet loss rate, and a packet order change rate calculation unit 30 that calculates a packet order change rate. The packet loss rate calculating unit 20 includes an adding unit 21, a cumulative deviation amount holding unit 22, and a dividing unit 23. The packet order change rate calculation unit 30 includes a determination unit 31, an addition unit 32, a cumulative deviation amount flag holding unit 33, and a division unit 34.

The packet loss rate calculating unit 20 estimates the packet loss rate according to the above-described number C2. In the adding unit 21, the latest (k-th) deviation amount G (k) output from the deviation amount calculating unit 15 and the preceding (k−1th) accumulation held by the accumulated deviation amount holding unit 22. The shift amount GG (k−1) is added to calculate the latest cumulative shift amount GG (k). The addition unit 21 outputs the accumulated deviation amount GG (k) to the accumulated deviation amount holding unit 22 so as to be held and outputs the accumulated deviation amount GG (k) to the dividing unit 23. The division unit 23 obtains the maximum sequence number S _max and the minimum sequence number S _min held by the holding unit 14 and calculates the number C2. That is, the cumulative deviation amount GG (k) is divided by (S _max −S _min −1). The division unit 23 outputs the value obtained by the division as an estimated value of the packet loss rate P _Loss .

The packet order change rate calculating unit 30 estimates a packet order change rate according to the above-described number C4. The determination unit 31 evaluates whether the latest (k-th) deviation amount G (k) output from the deviation amount calculation unit 15 is a negative value. The determination unit 31 sets the deviation amount flag NG (k) = 1 when the deviation amount G (k) is a negative value according to the number C3, and otherwise the deviation amount flag NG (k) = 0 Set as The addition unit 32 adds the value of the latest (k-th) deviation amount flag NG (k) output by the determination unit 31 and the value one before (k-1) which the accumulated deviation amount flag holding unit 33 holds. The cumulative deviation amount flag NGNG (k−1) is added to calculate the latest cumulative deviation amount flag NGNG (k). The addition unit 32 outputs the accumulated deviation amount flag NGNG (k) to the accumulated deviation amount flag holding unit 33 so as to be held, and outputs the accumulated deviation amount flag NGNG (k) to the dividing unit 34. The dividing unit 34 obtains the maximum sequence number S _max and the minimum sequence number S _min held by the holding unit 14 and performs calculation of the number C4. That is, the cumulative deviation amount flag NGNG (k) is divided by (S _max −S _min −1). The division unit 34 outputs the value obtained by the division as an estimated value of the packet order change rate _Preorder .

  Note that the reset signal is reset (initialized) by inputting a reset signal to the holding unit 14, the accumulated deviation amount holding unit 22, and the accumulated deviation amount flag holding unit 33 after a predetermined period has elapsed or at an arbitrary time. be able to. For example, the predetermined period may be a fixed period required to estimate the network quality. Alternatively, the time may be received from the packet with the lowest value of the sequence number to the packet with the highest value of the sequence number.

FIG. 5 is a flowchart illustrating a network quality estimation method performed by the network quality estimation apparatus 101 according to the present embodiment. This network quality estimation method is
Sequence number acquisition procedure (step S01) for acquiring a sequence number indicating the order of transmission from the current packet currently being received;
A sequence number prediction procedure for predicting the sequence number of the current packet based on the sequence number of the immediately preceding packet received immediately before the current packet;
Network quality estimation procedure for estimating network quality based on the amount of deviation (step S02) between the sequence number of the current packet acquired in the sequence number acquisition procedure and the sequence number of the current packet predicted in the sequence number prediction procedure (step S02) Step S03)
I do.
Here, the amount of deviation in step S02 may be a value calculated using the above-described Equation 1.

In the network quality estimation procedure (step S03), it is determined in step S31 whether or not the deviation amount G (k) is a negative value, and the deviation amount flag NG (k) is determined when the deviation amount G (k) is a negative value. The value is set to 1 (step S33), and in the other cases, the deviation amount flag NG (k) is set to 0 (step S32). Then, the difference between the minimum sequence number S _min and the maximum sequence number S _max among the packets received in the predetermined period is calculated (step S34). Then, in accordance with the number C2 to calculate the packet loss rate _{P LOSS} (step S35), and calculates the packet order change rate _{P Reorder} in accordance with the number C4 (step S36).

[Special case]
Here, attention is focused on the fact that the sequence number stored in the received packet is a finite value. For example, in the case of an RTP packet, the area for storing the sequence number is only 16 bits as shown in FIG. That is, an RTP packet can store only a value from 0 to 65535 in decimal as a sequence number. When the sequence number reaches 65535, it is incremented back to 0 again. Hereinafter, on the premise that a 16-bit sequence number storage area is allocated to the packet, the shift amount G (k) at the timing when the sequence number of the received packet returns from 65535 to 0 will be considered.

  When a packet with sequence number s (k) = 0 is received, the sequence number of the immediately preceding received packet is s (k−1) = 65535, so as shown in FIG. The quantity G (k) = − 65536. Therefore, even if there is no packet loss during the evaluation period of the network quality, an accumulated deviation amount (that is, an estimated value of the packet loss amount) is calculated to be an unreasonable value of kG (k) = − 65536.

(Action 1)
Therefore, as shown in FIG. 6 (b), the timing when the maximum value of the sequence number (65535 in decimal in the example of the 16-bit value) is reached, or when the sequence number returns to the initial value (usually 0) The accumulated deviation amount GG (k) and the accumulated deviation amount flag NGNG (k) may be reset (initialized). That is, when the sequence number s (k) acquired from the packet processing unit 12 is 65535 (or 0), the sequence number acquisition unit 13 outputs a reset signal and holds the reset signal in the holding unit 14. Reset or erase the values stored in the accumulated shift amount holding unit 22 and the accumulated shift amount flag holding unit 33 in the network quality calculation unit 16, and the sequence number changes again from s (k-1) = 1. G (k) may be calculated.

That is, the sequence number acquisition unit 13 resets the holding unit 14 and the network quality calculation unit 16 when acquiring the largest sequence number that can be assigned to the packet, or when acquiring the initial value of the sequence number assigned to the packet. Perform reset operation. Specifically, the sequence number acquisition unit 13 resets s (k), s (k-1), S _max and S _min held by the holding unit 14 in the reset operation, and the network quality calculation unit 16 The cumulative deviation amount _{k k = 2} [G (k)] and the cumulative deviation amount flag _{k k = 2} [NG (k)] which are calculated are reset.

(Action 2)
Even if the estimation of the packet loss rate or the packet order change rate is reset when the sequence number receives the maximum value, it is assumed that the received packet is replaced before and after the sequence number returns to the initial value from the maximum value. May occur. For example, when the packet with the largest sequence number (s (k) = 65535) and the packet with the sequence number smaller by one smaller than the largest value (s (k) = 65534) are received in reverse order, as shown in FIG. As shown in), even after the sequence number s (k) = 65535 is received and reset, the deviation amount G (k) =-65535 is calculated at the timing when s (k) = 0 is received. It will be done. As a result, even if no packet loss occurs thereafter, an invalid value of ΣG (k) = − 65535 is calculated as the cumulative deviation amount. Also, if the packet with the initial sequence number (s (k) = 0) and the packet with the sequence number one larger than the initial value (s (k) = 1) are received in reverse order, as shown in FIG. As shown in), even after the sequence number s (k) = 65535 is received and reset, the cumulative shift amount GG (k) = − 1 is calculated, and as the estimated value of the packet loss amount An invalid value of -1 is calculated.

ただし、Ｍａｘ Ｓはパケットに付与可能な最大のシーケンス番号である。 Therefore, when a packet with a maximum sequence number is received, the deviation amount G (k) for the next received packet may be a value that matches the sequence number of the received packet. In other words, when the sequence number of the immediately preceding received packet is the maximum value (s (k) = 65535), the amount of deviation may be a value that matches the sequence number of the received packet. If this is written as an expression, it becomes as follows.

However, Max S is the largest sequence number that can be assigned to a packet.

  If the number C5 is used, the network quality is set when the maximum value of the sequence number (65535 in decimal in the example of the 16-bit value) is reached or when the sequence number returns to the initial value (usually 0). There is no need to reset the estimate. Even if the same received packet replacement as in FIG. 7A occurs, using the number C5, as shown in FIG. 8A, the cumulative deviation amount can obtain a reasonable value. Similarly, even if the same received packet replacement as in FIG. 7B occurs, as shown in FIG. 8B, the accumulated deviation amount can obtain a reasonable value.

(Action plan 3)
Even if the equation C5 is used to calculate the amount of deviation, the amount of deviation and the amount of deviation if a packet with a sequence number s (k) = 65535, which is the key to solving the problem, is lost and not received in the network An invalid value is calculated as a flag. As shown in FIG. 9A, when a packet with sequence number s (k) = 65535 is not received, when a packet with sequence number s (k) = 0 is received, the amount of deviation G (k) = − A difference between 65535 and the deviation amount flag NG (k) = 1 is calculated.

  Therefore, if the packet with the largest sequence number is lost, the deviation amount G (k) calculated from the received packets before and after this packet should not be used for calculating the packet loss rate or the packet order change rate. Just do it. Specifically, the network quality calculation unit 16 sets a negative threshold value Gth for the deviation amount G (k), and the deviation amount calculation unit 15 sets a negative deviation amount G (k) less than Gth. The negative shift amount is not calculated if the negative shift amount for compensating the negative shift amount G (k) is not calculated within a fixed period shorter than the predetermined period before and after the negative shift amount G (k) is calculated. The displacement amount G (k) is replaced with G (k) = 1.

  In other words, if a shift amount G (k) smaller than a predetermined negative threshold Gth is calculated, a positive shift amount for compensating this shift amount can not be calculated within a certain period before and after the shift amount is calculated. The shift amount may be changed to G (k) = 1, assuming that the packet having the maximum sequence number is lost.

  For example, an example will be described in which the predetermined negative threshold Gth = −60000, and the fixed period is a period corresponding to the number of received packets. It is assumed that a packet (s (k) = 0) whose sequence number is the initial value is received after receiving a packet (s (k) = 65534) whose sequence number is smaller than the maximum value by 1. At this time, the amount of deviation is calculated as G (k) =-65535 (<Gth), and the amount of deviation flag is calculated as NG (k) = 1. The packet with the largest sequence number (s (k) = 65535) if the deviation of G (k) =-Gth (this value becomes positive) is not calculated over the three packets before and after (total 6 packets) Is regarded as a loss, and as shown in FIG. 9B, the shift amount calculated when the packet with the initial sequence number (s (k) = 0) is received is G (k) = 1. Change it and change it to NG (k) = 0 based on it. This process means that the shift amount is set to G (k) = 1 in response to the loss of the packet with the largest sequence number. As a result, the cumulative deviation amount becomes ΣG (k) = 1 (that is, the packet loss amount is 1), and the number of occurrences of packet replacement is corrected to 0. In this example, the fixed period is set to a period equivalent to the number of received packets 6, but may be changed as appropriate depending on the network configuration, the status, and the like.

  Here, the minimum value that can be used as the predetermined negative threshold Gth is a value larger by one than the value where the maximum value of the sequence number is a negative number. For example, as shown in FIG. 1, when the sequence number is represented by 16 bits (the maximum value of the sequence number is 65535), the minimum value that can be used as the predetermined negative threshold Gth is −65534. The deviation G (k) is less than -65534 when only the packet with the maximum sequence number (s (k) = 65535) is lost as shown in FIG. 9A. . On the other hand, use of the predetermined negative threshold Gth = -65534 can not compensate for the case where a plurality of consecutive packets including the packet with the largest sequence number are lost. For example, consider a case where three packets with sequence numbers s (k) 65533 to 65535 are continuously lost. When a packet with sequence number s (k) = 0 is received following reception of a packet with sequence number s (k) = 65532, the amount of deviation is calculated as G (k) = 0− (65532 + 1) = − 65533 Ru. If Gth = -65534 is used as the predetermined negative threshold, this shift amount is not corrected, and therefore the cumulative shift amount is also calculated to be an invalid value of GG (k) =-65555. Therefore, as the predetermined negative threshold Gth, a negative number value larger than "one larger than the value with the maximum value of the sequence number as the negative number" is taken into consideration, taking into consideration the number of packets that may be lost continuously. It may be used. As in the above example, when there is a possibility that at most three packets may be lost consecutively, the predetermined negative threshold may be Gth = -65532.

(Effect of the invention)
A network quality estimation method and a network quality estimation apparatus according to the present invention are values of a sequence number s (k) of a k-th received packet and a sequence number s (k-1) of a packet received immediately before that. From this, the packet loss rate and the packet order change rate are calculated based on the “displacement amount G (k)” that is the difference from the value that would be obtained as the sequence number s (k). According to the present invention, the quality of the network can be measured more inexpensively and easily without affecting other traffic since it is not necessary to transmit a dummy packet for network quality measurement and the network quality can be calculated only by addition and subtraction. You will be able to

  Patent Documents 1 and 2 are devices invented to measure network quality with high accuracy. On the other hand, the present invention is an apparatus invented for measuring network quality inexpensively and simply without obtaining high accuracy as in Patent Documents 1 and 2, and the object of the invention of Patent Documents 1 and 2 I would like to add that it is different.

11: Reception buffer 12: packet processing unit 13: sequence number obtaining unit 14: holding unit 15: shift amount calculation unit 16: network quality calculation unit 20: packet loss rate calculation unit 21: cumulative shift amount holding unit 22: division unit 30 : Packet order change rate calculation unit 31: Determination unit 32: Cumulative deviation amount flag holding unit 33: Division unit 101: Network quality estimation device 301: Reception device

Claims (12)

A sequence number obtaining procedure for obtaining a sequence number indicating an order of transmission from a currently received current packet;
A sequence number prediction procedure for predicting the sequence number of the current packet based on the sequence number of the immediately preceding packet received immediately before the current packet;
A network quality estimation procedure for estimating network quality based on a deviation amount which is a difference value between the sequence number of the current packet acquired in the sequence number acquisition procedure and the sequence number of the current packet predicted in the sequence number prediction procedure; ,
Network quality estimation method to do.   In the network quality estimation procedure, the network quality may be a value obtained by dividing an accumulated shift amount obtained by adding the shift amount over a predetermined period, by the number of packets to be received over the predetermined period calculated from the sequence number. The network quality estimation method according to claim 1.   In the network quality estimation procedure, accumulated deviation obtained by adding the deviation amount flag over a predetermined period using a deviation amount flag which sets “1” when the deviation amount is a negative value and “0” otherwise. The network quality estimation method according to claim 1, wherein a value obtained by dividing an amount flag by the number of packets to be received over the predetermined period calculated from the sequence number is used as the network quality.   If the largest sequence number that can be assigned to the packet is acquired in the sequence number acquisition procedure, or if the initial value of the sequence number to be assigned to the packet is acquired, the sequence number acquisition procedure, the sequence number prediction procedure, The network quality estimation method according to any one of claims 1 to 3, wherein a reset procedure for starting the network quality estimation procedure is performed.   The network quality estimation procedure according to any one of claims 1 to 3, wherein, when the sequence number of the immediately preceding packet is the largest sequence number, the sequence number of the current packet is the deviation amount. The network quality estimation method described in.   The network quality estimation procedure is a negative shift amount in which the shift amount is smaller than a predetermined negative threshold, and the negative shift is within a fixed period shorter than the predetermined period before and after the negative shift amount is calculated. The network quality estimation method according to claim 5, characterized in that when the amount of positive offset that compensates the amount is not calculated, the amount of negative offset is replaced with one. A sequence number acquisition unit for acquiring a sequence number indicating an order of transmission from a currently received current packet;
A holding unit that holds the sequence number of the packet acquired by the sequence number acquisition unit;
The difference value between the sequence number of the current packet acquired by the sequence number acquisition unit and the sequence number of the current packet predicted based on the sequence number of the immediately preceding packet received immediately before the current packet held by the holding unit. A shift amount calculation unit that calculates a certain shift amount;
A network quality calculation unit that estimates network quality based on the deviation amount calculated by the deviation amount calculation unit;
A network quality estimation device comprising:   The network quality calculation unit is characterized in that a value obtained by dividing an accumulated deviation amount obtained by adding the deviation amount over a predetermined period by a number of packets to be received over the predetermined period calculated from the sequence number is the network quality. The network quality estimation apparatus according to claim 7, wherein   The network quality calculation unit is a cumulative deviation obtained by adding the deviation amount flag over a predetermined period using a deviation amount flag that sets “1” when the deviation amount is a negative value, and “0” otherwise. The network quality estimation apparatus according to claim 7, wherein a value obtained by dividing an amount flag by the number of packets to be received over the predetermined period calculated from the sequence number is used as the network quality.   The sequence number acquisition unit resets the holding unit and the network quality calculation unit when acquiring the largest sequence number that can be assigned to the packet or when acquiring the initial value of the sequence number assigned to the packet The network quality estimation apparatus according to any one of claims 7 to 9, wherein:   10. The shift amount calculation unit according to any one of claims 7 to 9, wherein, when the sequence number of the immediately preceding packet is the maximum sequence number, the sequence number of the current packet is the shift amount. Network quality estimation device according to claim 1.   The network quality calculation unit is a negative shift amount in which the shift amount is smaller than a predetermined negative threshold, and the negative shift is within a fixed period shorter than the predetermined period before and after the negative shift amount is calculated. 12. The network quality estimation apparatus according to claim 11, wherein the negative offset amount is replaced with 1 when a positive offset amount compensating the amount is not calculated.

Images