JP2004080279A - Quality control method of video stream, its program, and recording medium - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To achieve simple quality control of video streams considering burst characteristics. <P>SOLUTION: The video streams are measured in advance, stream characteristics (P, ρ, δ) (P: maximum traffic rate, ρ: long-time average rate, δ: amount of one-time burst data) are examined, and packet loss rate (ε) to be permitted is determined. Additionally, a link speed C in each node in a network, and buffer time T are examined. Then, a maximum value KO of K to be permitted when allowing K streams to flow to the same link is determined, and a multiplexing parameter A = C/K0 is defined and expressed by the two parameters α, β. Input side conditions and output side are expressed by α, β, respectively. By regarding, for example, stream speed and link speed in each node as ασ and βC, respectively, application can be made to a node independent distribution system such as RSVP-TE. <P>COPYRIGHT: (C)2004,JPO

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to a quality control technique of an IP network, and in particular, as a quality control for a video stream traffic having a periodic burst, parameterizing a statistical multiplexing effect and using the parameter to provide a quality control function to the network. It is about technology.
[0002]
[Prior art]
In general, a video stream has a property that bursts are generated at a constant cycle and the amount of data flowing in one burst is constant. FIG. 17 shows an example of such a video stream. Stream characteristics (traffic) can be expressed by three parameters P, ρ, and σ. Here, ρ represents the long-term average rate, P represents the instantaneous maximum rate, and σ represents the amount of data flowing in one burst. Such a stream has the property of causing significant quality degradation due to packet loss, but it is difficult to control the quality by retransmission control, such as being handled by applications that use real-time properties and multicasting, and it is necessary to perform quality control on the network side. is there.
[0003]
First, as a conventional technique for performing quality control on bursty traffic, a standard technique has already been established as an ATM (Asynchronous Transfer mode) traffic control technique (literature: Paul). Ferguson; Geoff Huston, "Internet QoS", Ohmsha, pp. 123-154, 2000). The properties common to them are as follows.
1. The traffic is processed by UPC (Usage Parameter Controller) according to the convenience of the network. UPC generally prevents congestion by prohibiting excessive traffic network entry when all network resources are used.
2. A VC (virtual channel) is allocated to each flow, and control is performed on a VC basis.
3. Quality control is performed by estimating a statistical multiplexing effect by using a QoS (Quality of Service) parameter. Parameters that affect communication quality include delay time, delay variation, data loss rate, and peak speed.
[0004]
Further, as a quality assurance technology in an IP network, there is RSVP-TE using RSVP (Resource Reservation Setup Protocol) which is a protocol for managing a transmission band of a specific communication channel between routers (Reference: D. Awduch et al., “RSVP-TE: Extension to RSVP for LSP Tunnels”, IETF Networking Group RFC 3209, 2001).
1. The presence of UPC is not mandatory.
2. Since the processing is performed by the path control protocol, the processing is generally performed by autonomous distribution of each node, and the path calculation processing load is smaller than that of the ATM.
3. Many are based on a resource reservation model like RSVP-TE, and the parameters are mainly simple.
[0005]
Thus, the quality control of ATM and IP has different directions. There are other control methods at a lower layer, such as rate shaving on a VLAN basis, but these are static controls that operate independently of IP routing.
[0006]
[Problems to be solved by the invention]
The ATM traffic control technology requires the implementation of UPC, which imposes strict requirements on accommodating nodes, and allocates VCs to each flow. Therefore, when the number of flows increases, the route calculation load reduces the IP routing control. There was a problem such as a rapid increase compared to. Further, in the existing traffic control technology based on IP, only simple parameters such as a link bandwidth and a required bandwidth can be used. Therefore, the burst property is not considered. When the burst property is considered, the maximum peak rate needs to be used as the required bandwidth. There is a problem that the traffic control cannot be performed in consideration of the effect of the statistical multiplexing effect.
[0007]
These problems are problems that assume general traffic as input traffic. As described above, the output characteristics of the video stream have some regularity, and the packet loss target value is very small.
[0008]
An object of the present invention is to use such characteristics of a video stream to enable traffic control in consideration of burst characteristics and the effects of statistical multiplexing effects using an existing IP routing protocol without using UPC. It is to be.
[0009]
[Means for Solving the Problems]
The present invention is roughly classified into two procedures: a parameterization based on a statistical analysis in advance, and a procedure for performing a judgment process in real time when a user requests viewing. The reception determination unit required for the real-time procedure is placed in each node in the network or the QoS server existing in the network.
1. Parameterization by Prior Statistical Analysis (1) In order to model a video stream, a server stream is measured in advance, and stream characteristics (P, ρ, σ) are determined. In addition, it checks how much packet loss the codec type of the stream allows, and sets an allowable packet loss rate (ε).
(2) In order to estimate the multiplexing characteristic at the node j, the link speed (Cj) and the buffer time (Tj) are checked for all the links Lj to be quality controlled in the network.
(3) Statistical analysis is performed based on the results of (1) and (2), and the packet loss rate (Ploss (K)) when K (P, ρ, σ) streams are sent through the same link Is obtained, and the maximum value K0 of K satisfying Ploss (K) <ε is obtained.
(4) By defining A≡C / K0, a multiplexing parameter Aij when the stream-side conditions Si (Pi, ρi, σi, εi) and the node-side conditions Lj (Cj, Tj) are determined is obtained.
(5) Aij is decomposed into α and β parameters such that Aij ≒ αi / βj.
[0010]
The stream multiplexing characteristic of S (P, ρ, σ) defined by modeling has already been formalized (Literature: Kei Ue, "Traffic Control for Guaranteeing QoS in Information Sharing Planet Home", NII Journal No. 3, pp. 23-2001). The packet loss rate (Ploss (K)) when K streams with the same (P, ρ, σ) flow through the same link can have an upper limit as follows.
log (Ploss (K)) = − sup [s (Kμ−C) −Klog (Kρ / C)]: S ≧ 0
Here, μ is called an execution band for lossless multiplexing, and assuming that the maximum value of the delay allowed at the node is the buffer time (T), the following is obtained.
μ = max [ρ, Pσ / {(σ + T (P-ρ)}]
At this time, the condition for Ploss (K) <ε is expressed by the equation (1) or (2).
K> C / μ (1)
Klog (Kρ / C) <log (ε) (2)
Here, equation (2) indicates that if ε is sufficiently small and C / ρ is not so large, there is a case where a natural number solution is not provided. Tend to have.
If equation (2) has a solution for all links, the parameters are determined using equation (2); otherwise, the parameters are determined using (1).
When equation (2) is used, there is almost no difference for each link type.
A = K0ρ / C, α = A, β = 1 (3)
When the equation (1) is used, the following operation is performed using the time Ton of one burst.
A = [mu] = P [sigma] / {[sigma] + T (P- [rho])} = P * / {1 + T / Ton} ~ [sigma] / T where T >>> where Ton is the property of the stream, and T is the node Because of the nature, it can be determined as follows.
α = σ / ρ, β = T (4)
[0011]
2. Procedure for performing determination processing in real time when a user requests viewing (1) A reception determination unit is placed in the network, and for each link Lj in the network, the used bandwidth (ス ト リ ー ム Aijρi or Σαipi) of the stream currently flowing on the link Lj Make a note of it.
(2) When the user terminal issues a new viewing request to the stream source, the stream source issues an acceptance determination request to the acceptance determination unit.
(3) Upon receiving the reception determination request, the reception determination unit determines the link to be subjected to quality control, checks the required bandwidth AIjρI or αIρI of the stream for which the new viewing request has been received, and determines the determination formula ｛for each link Lj. {Aijρi + AIjρI} <Cj} or {αipi + α {ρ} <βjCj}
Returns true / false of the value of to the stream source.
(4) If true is returned, the stream source distributes the stream, and when the distribution ends, notifies the reception determination unit of an end. If false is returned, the stream source refuses to deliver the stream.
(5) Upon receiving the end notification, the reception determining unit corrects the used bandwidth of the stream to a value obtained by subtracting the value of the used bandwidth of the stream.
[0012]
BEST MODE FOR CARRYING OUT THE INVENTION
Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
<Parameters by prior statistical analysis>
FIG. 1 shows a processing flowchart of parameterization by prior statistical analysis. This is common to each embodiment described later.
First, the traffic of the video stream is measured in advance, the characteristics of the video stream are modeled by (P, ρ, σ), and the allowable packet loss rate (ε) is determined (S1). As shown in FIG. 17, P indicates the maximum traffic rate, ρ indicates the long-term average rate, and δ indicates the data amount of one burst. Further, the link speed C and the buffer time (T) of each node of the network (NW) are investigated as node characteristics (output side) (S2).
[0013]
Next, a statistical analysis is performed based on the results of steps S1 and S2, and the packet loss rate (Ploss (K)) when K streams with the same stream characteristics (P, ρ, σ) flow on the same link is calculated. Then, the maximum value K0 of K that satisfies Ploss (K) <ε is determined (S3). Then, the multiplexing parameter A = C / K0 is defined by the K0 (S4).
[0014]
Since both input and output conditions are involved in A, the condition is divided into an input side and an output side, and is expressed by two parameters α and β, such as A = α / β. Here, α represents an input condition and β represents an output condition.
[0015]
As described above, the parameters α and β are determined by determining whether Klog (Kρ / C) <(ε) has a natural number solution, where K> C / μ (S5). Determine. That is, when a solution having a natural number is provided, α = K0ρ / C and β = 1 are set based on the equation (3) (S6). If there is no natural number solution, α = α = δ / P and β = T from the above equation (S7).
[0016]
Hereinafter, each embodiment to which the present invention is applied will be described.
<Example 1>
In a closed network in which the source and destination of a stream can be identified when the source and destination of the stream are known, it is possible to determine whether or not the communication quality can meet the standard by the end-to-end of the source and destination at the time of a viewing request. is there.
[0017]
FIG. 2 shows a network configuration example of the present embodiment. In FIG. 2, 100 is a stream source, 110 is a node, 120 is a user terminal, 130 is a QoS server, and an acceptance determination unit 135 exists in the QoS server 130. The reception determination unit 135 manages the link speed, the used bandwidth, and the like of the interfaces (IFs) of all the nodes.
The user terminal 120 makes a stream viewing request to the stream source 100. The stream source 100 makes a reception determination request to the reception determination unit 135 of the QoS server 130 in response to the viewing request of the user terminal 120, and if the answer is true, distributes the stream. Each node 110 transfers a packet based on the route control. The reception determination unit 135 will be described later.
[0018]
<Example 2>
This is because when the buffer time of the corresponding node is sufficiently longer than the burst time of a single stream in a network where the steps and nodes where the quality degradation due to traffic multiplexing occurs are predetermined, the burst period and the long-term average rate Even if streams differing from each other, the upper limit value of the packet rate at the corresponding node is calculated, and it is possible to determine whether the communication quality can meet the criterion by End-to-End.
[0019]
FIG. 3 shows a network configuration example according to the present embodiment. This is basically the same as FIG. 2 except that only the access section is a bottle net. The reception determination unit 135 in the QoS server 130 manages the link speed, the used bandwidth, and the like of the IF of the target node.
[0020]
FIG. 4 shows a functional block diagram of a configuration example of the stream source 100 and the QoS server 130 in the first and second embodiments. The stream source 100 includes a server communication processing unit 101, a stream distribution application unit 102, a traffic classifier 103, a packet scheduler 104, and the like. The server communication processing unit 101 makes an acceptance determination request to the QOS server 130 in response to the viewing request of the user terminal, receives the response, and if the response is true, distributes the stream to the stream distribution application unit 102 and the packet schedule 104. When the distribution is completed, the QOS server 130 is notified of the completion. Other functions are basically the same as the conventional one.
[0021]
The QoS server 130 includes a communication processor 131, a route calculation unit 132, a reception determination unit 135, and the like. The communication processor 131 manages the communication control of the stream source 100 and each node 110, and the route calculation unit 132 controls the route selection of the packet. Since these are not directly related to the present invention, the details are omitted. The present invention relates to the reception determining unit 135.
[0022]
FIG. 5 shows a configuration example of the reception determining unit 135 in the QoS server 130. The reception determination unit 135 includes an interface function unit 1351, a bandwidth calculation / determination unit 1352, a stream data management DB 1353, a node data management DB 1354, a bandwidth usage management DB 1355, a recording medium 1356, and the like. FIG. 6 shows a specific example of the management contents in each of the management DBs 1353, 1354, and 1355.
[0023]
FIG. 7 shows an example of a determination processing flow in the first and second embodiments. Hereinafter, a real-time determination process at the time of a user viewing request in the first and second embodiments will be described with reference to FIG.
The acceptance determining unit 135 of the QoS server 130 manages the bandwidth ΣAipi of the stream currently flowing in all nodes in the network (S11). When the user terminal 130 issues a new viewing request to the stream source 100, the stream source 100 makes an acceptance determination request (bandwidth ρ) to the QoS server 130 (S12). The QoS server 130 calculates the path of the stream (S13), calculates ΣAipi + AIjρI for all nodes i on the path, and determines ΣAipi + AIjρI <Cj (S14). If the determination is false, NG is returned to the stream source 100 (S15). On the other hand, if the judgment is true, AIjρI is added to the used band ΣAipi for the link on the route, and OK is returned to the stream source 100 (S16). When the OK is returned from the QOS server 130, the stream source 100 performs stream distribution (S17), and when the distribution ends, reports an end to the QOS server 130 (S18). Upon receiving the end report from the stream source 100, the QoS server 130 subtracts AIjρI from the used band for the link on the corresponding path (S19).
[0024]
<Example 3>
This is because in a network that has the function of grasping the link speed and current bandwidth used by each node for its own interface and establishing a session when necessary resources can be secured for a new session, By setting the assumed link speed (βC) and using the assumed long-term average rate (αρ) as the required band, the stream transmission is performed only when the communication quality meets the criterion in End-to-End. It is possible.
[0025]
FIG. 8 shows a network configuration example of the present embodiment. In FIG. 8, reference numeral 100 denotes a stream source, 110 denotes a node, and 120 denotes a user terminal. Each reception determination unit 115 manages the IF band (Σαρ) and the link speed (βC) of the IF of the own node.
[0026]
The user terminal 120 issues a stream viewing request to the stream source 100. The stream source 100 makes a reception determination request to a node 110 (hereinafter, a reception node) in which the stream source 100 is accommodated. The receiving node 110 requests the αIρI of the band to each node on the route by an RSVP-Path message, makes a determination in the reception determining unit 115 of each node, and returns a RSVP-Resv message if the determination is true. , To the stream source 100. If the response is true, the stream source 100 distributes the stream.
[0027]
FIG. 9 shows a functional block diagram of a configuration example of the stream source 100 and each node 110 in the present embodiment. Note that one of the nodes 110 is a receiving node. The stream source 100 is the same as that of FIG. 4 except that the server communication processing unit 101 is replaced with the RSVP processing unit 105. The reception node 110 includes an RSVP processor 111, a routing processor 112, a traffic classification unit 113, a reception control unit 114, a reception determination unit 115, a packet scheduler 116, and the like. Here, the reception determination unit 115 is related to the present invention.
[0028]
FIG. 10 shows a configuration example of the reception determining unit 115 in each node. The reception determination unit 115 includes an interface function unit 1151, a bandwidth calculation / determination unit 1152, a link information management DB 1153, a bandwidth usage management DB 1154, a storage medium 1155, and the like. FIG. 11 shows a specific example of management contents in the management DBs 1153 and 1154.
[0029]
FIG. 12 shows an example of a determination processing flowchart in the third embodiment. Hereinafter, a real-time determination process at the time of a user viewing request in the third embodiment will be described with reference to FIG.
The reception determination unit 115 of each node 110 manages the band used (自 αρ) and the link speed (βC) of the own node (S21). When the user terminal 130 issues a new viewing request to the stream 100, the stream source 100 requests a band (αIρI) from the receiving node 110 by an RSVP-Path message (S22). The receiving node 110 that has received the message determines the output IF according to the routing protocol (S23), calculates Σαρ + αIρI for the output IF, and determines Σαρ + αIρI <βjCj (S24). Here, if the determination is false, return the Path message discard to the stream source 100 (S25). On the other hand, if the determination is true, it is determined whether the user terminal 130 is accommodated immediately below (S26). If not, the bandwidth (αIρI) is requested from the output IF to the next node by an RSVP-Path message. (S27). Hereinafter, when the determination is true, similar processing is performed in each node. When the user terminal 130 is accommodated immediately below, the corresponding node 110 returns the RSVP Resv message one immediately before, adds αIρI to the used band for the corresponding link, and registers it in the reception control unit (S28). Thus, when the determination is true at each node 110 on the route, the receiving node 110 replies an RSVP-Resv message to the stream source 100. Upon receiving the RSVP-Resv message, the stream source 100 distributes the stream, and transmits the RSVP-Conf message to each node 110 on the route via the receiving node during distribution (S29). In each node 110, when RSVP-Conf does not come, αIρI is subtracted from the used band and deleted from the registration of the reception control unit (S30).
[0030]
<Example 4>
The network configuration is the same as that of the third embodiment. However, by storing the stream parameters for the stream sources to which the stream has been streamed in the past, the pre-analysis is automated and performed on demand, and the end-to-end is performed. The stream transmission is enabled only when the communication quality satisfies the criterion in -End. The network configuration is the same as that of FIG. 8, and each node 110 has a reception determination unit 115.
[0031]
FIG. 13 shows a functional block diagram of a configuration example of the stream source 100 and each node 110 in the present embodiment. The configuration of the stream source 100 is the same as that of FIG. The configuration of the node 100 is the same as that of FIG. 9 except that the traffic measurement unit 117 is added. However, the internal configuration of the reception determination unit 115 is slightly different.
[0032]
FIG. 14 illustrates a configuration example of the reception determination unit 115. The reception determination unit 115 includes an interface function unit 115, a bandwidth calculation / determination unit 1152, a stream data management DB 1156, a link information management DB 1153, a bandwidth usage management DB 1154, a storage medium 1155, and the like. This is basically similar to FIG. 5 described above. FIG. 15 shows a specific example of management contents in each management DB 1156, 1153, 1154. Here, the contents of the stream data management DB 1156 are added as needed.
[0033]
FIG. 16 shows an example of a flowchart of the determination process in the fourth embodiment. The difference from FIG. 12 is that the process (S34) of searching for the parameters (P, ρ, Toff) from the IP address of the stream source, and when the stream is distributed, the traffic measurement unit detects the stream (source and destination). That is, a process (S41) for measuring the parameter (P, Toff, ρ) is added to the set of nations.
[0034]
It is to be noted that a part or all of the processing functions of each unit in the apparatus shown in each embodiment are configured by a computer program, and that the program can be executed by a computer to realize the present invention. Needless to say, the processing procedures shown in FIG. 7, FIG. 12, FIG. 16, and the like can be configured by a computer program, and the computer can execute the program. In addition, a program for realizing the processing function by the computer or a program for causing the computer to execute the processing procedure is stored in a computer-readable recording medium such as an FD, an MO, a ROM, a memory card, The program can be recorded on a CD, a DVD, a removable disk, or the like, stored and provided, and the program can be distributed through a network such as the Internet.
[0035]
【The invention's effect】
According to the present invention, it is possible to perform quality control by applying the resource reservation model to a video stream having a strong burst property. Also, by separating the parameters into two, α and β, it is also possible to apply to quality control of a node autonomous distributed type. As an example of the improvement of the accommodation ratio by considering the traffic multiplexing effect, FIG. 18 shows the result of calculating K0 and A based on the data of a commercially available stream source and the data of the layer 3 switch. FIG. 19 shows the calculated data. It can be seen that the accommodation ratio is much higher in the result calculated in conditional expression (2) than in the case calculated in conditional expression (1).
[Brief description of the drawings]
FIG. 1 is an example of a processing flow of parameter determination by prior statistical analysis according to the present invention.
FIG. 2 is an example of a network configuration according to a first embodiment of the present invention.
FIG. 3 is an example of a network configuration according to a second embodiment of the present invention.
FIG. 4 is a functional block diagram of a stream source and a QoS server according to the first and second embodiments.
FIG. 5 is a functional block diagram of a reception determining unit in the QoS server.
FIG. 6 shows management contents of each management DB in the reception determination unit.
FIG. 7 is a flowchart illustrating an example of a determination process according to the first and second embodiments.
FIG. 8 is an example of a network configuration according to a third embodiment of the present invention.
FIG. 9 is a functional block diagram of a stream source and each node according to a third embodiment;
FIG. 10 is a functional block diagram of a reception determining unit in each node.
FIG. 11 shows management contents of each management DB in the reception determination unit.
FIG. 12 is a flowchart illustrating an example of a determination process according to a third embodiment.
FIG. 13 is a functional block diagram of a stream source and each node according to a fourth embodiment of the present invention.
FIG. 14 is a functional block diagram of a reception determining unit in each node.
FIG. 15 shows management contents of each management DB in the reception determination unit.
FIG. 16 is a flowchart illustrating an example of a determination process according to a fourth embodiment.
FIG. 17 shows stream characteristics of a video stream targeted by the present invention.
FIG. 18 shows an example of a statistical multiplexing effect according to the present invention.
FIG. 19 shows an example of calculation data of FIG.
[Explanation of symbols]
100 stream source 110 node 115 reception determination unit 120 user terminal 130 QoS server 135 reception determination unit

Claims (7)

In a network having a stream source that outputs a video stream in which the burst period is constant and the data amount of a single burst is constant, a plurality of streams are input to one node and traffic is multiplexed from one link. When outputting, a method of calculating the upper limit of the packet loss rate that occurs without being absorbed by the node buffer, determines whether the communication quality satisfies the standard, and controls the quality of the video stream,
The stream source is capable of selecting a plurality of long-term average rates, and has a property that the burst period is constant and the data amount of one burst changes even if the long-term rate changes, and the buffer time of each node A video stream quality control method characterized by determining whether communication quality satisfies a standard even if streams having different long-term average rates are mixed in a range sufficiently longer than one burst time of the stream. 2. The video stream quality control method according to claim 1, wherein in a closed network in which a route can be specified when the source and destination of the stream are known, the communication quality can be satisfied by the end-to-end of the source and destination at the time of a viewing request. A video stream quality control method, characterized in that the method 2. The video stream quality control method according to claim 1, wherein each node grasps the link speed and the current bandwidth used for its own interface and establishes a session when necessary resources can be secured for a new session. In a network having the function of performing the communication, a link speed is set by imitating a node, and a long-term average rate is imitated as a required bandwidth, so that a stream can be streamed only when communication quality satisfies an end-to-end standard. A quality control method of a video stream, which enables transmission of a video stream. 2. The video stream quality control method according to claim 1, wherein in a network in which a step and a node in which quality deterioration due to traffic multiplexing occurs are predetermined, a buffer time of the node is sufficiently longer than one burst time of the stream. In the case of, even if streams with different burst periods and long-term average rates coexist, the upper limit of the packet loss rate at the corresponding node is calculated, and it is determined whether the communication quality can meet the standard by End-to-End. A quality control method for a video stream, wherein the quality control method is enabled. 4. The video stream quality control method according to claim 3, wherein the pre-analysis is performed on demand by automating the pre-analysis by storing the stream parameters for the stream sources to which the stream has been streamed in the past. A video stream quality control method characterized in that a stream can be transmitted only when communication quality satisfies a standard. A program for causing a computer to execute the video stream quality control method according to claim 1. A recording medium recording a program for executing the method of controlling the quality of a video stream according to claim 1 on a computer.

Images