CN108496346A

CN108496346A - The method for predicting speech quality and the speech quality for realizing the above method predict service unit

Info

Publication number: CN108496346A
Application number: CN201780008163.6A
Authority: CN
Inventors: 文炳轸
Original assignee: Individual
Current assignee: Individual
Priority date: 2016-01-25
Filing date: 2017-01-24
Publication date: 2018-09-04
Also published as: CN116319709A; KR101787348B1; KR20170088774A; WO2017131418A1

Abstract

An a kind of side in first and second network communication device executes, the method for predicting the speech quality of other side, including：(a) it is optionally included with the address information (IP address, No. Port) of the one side of the speech quality prediction for one side, the step of SIP (the Session Initiation Protocol) transactions requests required to other side's transmission comprising speech quality prediction；(b) the step of SIP affairs responses of the address information (IP address, No. Port) comprising above-mentioned other side are received from above-mentioned other side in response to above-mentioned SIP transactions requests；(c) it by the address information of the address information of one side and above-mentioned other side, is grouped by using the RTP unrelated whether including 1) with physical medium；2) RTCP is grouped；Or 3) the disguise oneself as camouflage UDP of RTP of above-mentioned other side is grouped by the communication of (calling in the following text " transmission test grouping ") is executed and the step of the transmission test of above-mentioned other side is to predict speech quality；And (d) or after (a) step, after (c) step is first carried out to obtain speech quality prediction result using the address of one side in above-mentioned other side, result is contained in the step of being transferred to one side among above-mentioned SIP affairs response.

Description

It predicts the method for speech quality and realizes the speech quality prediction service of the above method Device

Technical field

The present invention relates to a kind of technologies of prediction speech quality, more particularly to one kind includes that network telephone terminal exists constituting The network communication device of interior networking telephone net is mutual, is not necessarily to actual call generating process, can be asked by using SIP affairs The method of the prediction speech quality of the transmission test prediction speech quality of summation SIP affairs responses and the call for realizing the above method Prediction of quality service unit.

Background technology

Recent old analog and the circuit switched (Circuit based on digital forms such as SS7 (Signaling System 7) Switching) the phone continuous downturn of mode, and based on IP (Internet Protocol) net including Session Initiation Protocol The phone of packet switch (Packet Switching) mode is expanded.

The networking telephone using Session Initiation Protocol of the prior art, the circuit switched being physically protected compared with actual bandwidth The phone of mode, because of the system of the intermediate nodes such as L2, L3, L4 switch and network congestion in the excess load or the packet switching network of terminal State there is a possibility that speech quality reduction.

Ebrean Registered Patent the 10-1011344th is related to a kind of the speech quality measurement and introduction of telephone call service net System, disclosure measure the speech quality of networking person and carry out voice introduction, lead to according to the request of telephone relation person, the in a call heart The call stand-by period shortened with call center consultant is spent, customer satisfaction is improved, reduces the consulting telephone of consultant, So as to save money and improve call center consultant efficiency of operation technology.

In order to overcome the problems referred above, one embodiment of the invention, which provides, a kind of is being made of multiple networking telephone called terminals In the case of, it will be supplied to the Internet telephone calls terminal about the speech quality predictive index of multiple networking telephone called terminals, from And the technology of the good telephone called terminal of particular network of speech quality may be selected.

Invention content

Technical problem

One embodiment of the invention provides one kind before carrying out practical call, predicts speech quality in advance, to when estimated When speech quality drops to certain a reference value or less, it is possible to provide seek the prediction speech quality of the chance of other schemes method and Realize that the speech quality of the above method predicts service unit.

One embodiment of the invention provides one kind in communication process, on the anticipation path of media, converses by interval prediction Quality is simultaneously managed by cache memory within the defined end time, in network telephone terminal speech quality It can avoid method and the realization of the prediction speech quality of the unnecessary prediction operation in the section to almost repeating simultaneously in prediction The speech quality of the above method predicts service unit.

One embodiment of the invention provides a kind of executing media relays in there is identical with actual call process environment When intermediate node, by stages predicts speech quality and preserves structure in the cache, in other speech qualities When section is overlapped during prediction, prevent the transmission test predicted for speech quality from repeating, so as to Effective Operation network The method for predicting the prediction speech quality of operation and the speech quality for realizing the above method predict service unit.

One embodiment of the invention provides a kind of executing media relays in there is identical with actual call process environment When intermediate node, by stages predicts speech quality and preserves structure in the cache, in other speech qualities When section is overlapped during prediction, prevent the transmission test predicted for speech quality from repeating, so as to Effective Operation network The method for predicting speech quality and the speech quality for realizing the above method predict service unit.

One embodiment of the invention provides one kind between network telephone terminal and voip server or practical call connects Between the terminal connect, by the section of media transmission, the RTP/ unrelated whether including with the physical medium for call is generated After RTCP media flow or camouflage UDP flow amount, RTT (Round Trip Time), the packet loss rate of each grouping are calculated, There are when wireless section, for the probability of damage of reply grouping, the side of the prediction speech quality of the integrality of initial data is checked Method and the speech quality prediction service unit for realizing the above method.

One embodiment of the invention is provided in a kind of presence information may be included in networking person and is notified using SIP Event System, or can be provided by the SIP AS with the linkages such as HTTP server the prediction speech quality of speech quality state method and Realize that the speech quality of the above method predicts service unit.

The means solved the problems, such as

In embodiment, it is executed in a side of first and second network communication device, predicts the side of the speech quality of other side Method, including：(a) it is optionally included with address information (the IP of the one side of the speech quality prediction for one side Location, No. Port), to other side transmission comprising speech quality prediction require SIP (Session Initiation Protocol) thing The step of business request；(b) in response to above-mentioned SIP transactions requests the address information (IP for including above-mentioned other side is received from above-mentioned other side Address, No. Port) SIP affairs responses the step of；(c) believed by the address of the address information of one side and above-mentioned other side Breath is grouped by using the RTP unrelated whether including 1) with physical medium；2) RTCP is grouped；Or 3) above-mentioned other side is pretended The communication that (calling in the following text " transmission test grouping ") is grouped at the camouflage UDP of RTP executes the transmission test with above-mentioned other side to predict to lead to The step of talking about quality；And (d) or after (a) step, above-mentioned other side using the address of one side be first carried out (c) step with After obtaining speech quality prediction result, result is contained in the step of being transferred to one side among above-mentioned SIP affairs response.

In one embodiment, above-mentioned (c) step may include when one side is connected with above-mentioned other side by NAT device, By using bipartite address above mentioned information via the bipartite communication of above-mentioned NAT device, be generated in advance for The step of stating NAT pin holes (Pinhole) of address information.

In one embodiment, above-mentioned (c) step may include to one side or other side send successively it is N number of (N be more than 2 Natural number) grouping of above-mentioned transmission test, and confirm respective RTT (Round Trip Time) and whether to respective loss with The step of predicting above-mentioned speech quality.

In one embodiment, the method for predicting speech quality may also include (e) and preserve based on to above-mentioned N number of transmission test The packet loss rate of grouping and the prediction speech quality of propagation delay time prediction, to determine the selection of conversing to above-mentioned other side Step.

In one embodiment, above-mentioned (e) step may include, when above-mentioned other side is made of multiple network telephone terminals, being based on To above-mentioned multiple respective prediction speech qualities of network telephone terminal, the step of selecting particular network telephone terminal to be conversed.

In one embodiment, above-mentioned (e) step may also include when the communication between one side and above-mentioned other side belongs to language When sound communicates, by can currently use and the network telephone terminal with minimum propagation time delay is selected as the above-mentioned networking telephone The step of terminal.

In one embodiment, above-mentioned (e) step may also include when the communication between one side and above-mentioned other side belongs to several When according to communication, by can currently use and the network telephone terminal with minimum packet loss rate is selected as above-mentioned networking telephone end The step of end.

In one embodiment, may also include in the relay of (e) between first and second above-mentioned network communication device In the presence of execute media relays at least one media relays device when, first and second above-mentioned network communication device and it is above-mentioned extremely In a few media relays device (calling in the following text " speech quality prediction meanss "), to be formed in adjacent speech quality prediction meanss it Between relaying section, above-mentioned (a) to (d) step or above-mentioned (c) step is individually performed by executing, is sequentially completed above-mentioned call matter The step of amount prediction.

In one embodiment, above-mentioned (e) step may include to be predicted after the speech quality in section among the above to be stored in height Fast buffer storage and the step of update above-mentioned cache memory when more than specific time.

In one embodiment, above-mentioned (e) step may also include when to the speech quality prediction of failure in given trunk section, The step of reception includes the SIP affairs responses with above-mentioned the reason of failing relevant information about speech quality prediction meanss.

In embodiment, it is executed in voip server or SBC, without calling generating process prediction to multiple networks The method of the speech quality of telephone terminal, including：(a) area between network telephone server or SBC and network telephone terminal Between, it is grouped by using the RTP unrelated whether including 1) with physical medium；2) RTCP is grouped；Or 3) above-mentioned other side is pretended The communication execution that (calling in the following text " transmission test grouping ") is grouped at the camouflage UDP of RTP is respective with above-mentioned multiple network telephone terminals The step of transmission test is to predict speech quality；And it (b) is carried to partner based on the prediction to above-mentioned respective speech quality Prediction result or execution for above-mentioned speech quality is to the network telephone terminal that can be used with classic speech quality Information provide or execute connection attempt the step of.

In embodiment, the side executed in first and second network communication device executes, and predicts the speech quality of other side Method speech quality predict service unit, including：SIP transactions requests portion, is optionally included with for one side The address information (IP address, No. Port) of the one side of speech quality prediction is wanted to other side's transmission comprising speech quality prediction SIP (the Session Initiation Protocol) transactions requests asked；SIP affairs response receiving parts, in response to above-mentioned SIP Transactions requests receive the SIP affairs responses of the address information (IP address, No. Port) comprising above-mentioned other side from above-mentioned other side；

Speech quality prediction section, by the address information of the address information and above-mentioned other side of one side, by using 1) It is grouped with the RTP unrelated whether including of physical medium；2) RTCP is grouped；Or 3) the camouflage UDP for the RTP that disguises oneself as to above-mentioned other side The communication for being grouped (calling in the following text " transmission test grouping ") executes the transmission test with above-mentioned other side to predict speech quality；SIP affairs Acknowledgement transmissions portion, or after transmitting above-mentioned SIP affairs to other side, above-mentioned other side is first carried out pre- using the address of one side After above-mentioned speech quality is surveyed to obtain speech quality prediction result, result is contained among above-mentioned SIP affairs response and is transmitted To one side；And call selection determination section, it preserves and is prolonged based on the packet loss rate and transmission being grouped to above-mentioned N number of transmission test The prediction speech quality of slow time prediction, to determine that the call to above-mentioned other side selects.

Invention effect

Technology disclosed above has the following effects that.It is not intended that specific embodiment need to include all following effects or Following effect is only included, therefore, the interest field of disclosed technology is not limited.

The method of the prediction speech quality of one embodiment of the invention and the speech quality prediction service dress for realizing the above method It sets, before carrying out practical call, predicts speech quality in advance, to drop to certain a reference value or less when estimated speech quality When, it is possible to provide seek the prediction speech quality of the chance of other schemes

The method of the prediction speech quality of one embodiment of the invention and the speech quality prediction service dress for realizing the above method It sets, in communication process, on the anticipation path of media, is being advised by interval prediction speech quality and by cache memory It is managed in the fixed end time, to be can avoid in the prediction of network telephone terminal speech quality to almost repetition simultaneously The unnecessary prediction operation in section

The method of the prediction speech quality of one embodiment of the invention and the speech quality prediction service dress for realizing the above method It sets, when executing the intermediate node of media relays in there is environment identical with actual call process, by stages prediction call matter It measures and preserves structure in the cache, when section is overlapped during other speech qualities are predicted, to prevent from using It is repeated in the transmission test of speech quality prediction, so as to the prediction operation of Effective Operation network.

The method of the prediction speech quality of one embodiment of the invention and the speech quality prediction service dress for realizing the above method It sets, when executing the intermediate node of media relays in there is environment identical with actual call process, by stages prediction call matter It measures and preserves structure in the cache, when section is overlapped during other speech qualities are predicted, to prevent from using It is repeated in the transmission test of speech quality prediction, so as to the prediction speech quality of Effective Operation network.

The method of the prediction speech quality of one embodiment of the invention and the speech quality prediction service dress for realizing the above method It sets, between network telephone terminal and voip server or between the terminal of practical call connection, by the area of media transmission Between, after generating the RTP/RTCP media flow or camouflage UDP flow amount unrelated whether including with the physical medium for call, RTT (Round Trip Time), the packet loss rate for calculating each grouping, when there are wireless section, for the damage of reply grouping Possibility checks the integrality of initial data.

The method of the prediction speech quality of one embodiment of the invention and the speech quality prediction service dress for realizing the above method It sets, may be included in and utilize SIP Event notification architectures in the presence information of networking person, or can be by with HTTP server etc. The SIP AS of linkage provide speech quality state.

Description of the drawings

The schematic diagram of Fig. 1 speech quality forecasting systems of an embodiment to illustrate the invention；

Fig. 2 is the schematic diagram for the structure for illustrating speech quality prediction service unit；

Fig. 3 is to illustrate to generate voip server request and be used for speech quality predictive server and arbitrary network phone The schematic diagram of the process of the transmission test session of speech quality prediction between terminal；

Fig. 4 is to illustrate to generate voip server request and be used for voip server and arbitrary network telephone terminal Between speech quality prediction transmission test session process schematic diagram；

Fig. 5 is to illustrate to generate network telephone terminal request and for speech quality predictive server and arbitrary network phone end The schematic diagram of the process of the transmission test session of speech quality prediction between end；

Fig. 6 be illustrate to generate network telephone terminal request and for voip server and arbitrary network telephone terminal it Between speech quality prediction transmission test session process schematic diagram；

Fig. 7 to Figure 12 is the schematic diagram for illustrating to predict the process of the speech quality between arbitrary network telephone terminal；

Figure 13 is the schematic diagram for the UDP groupings for comparing one embodiment of the invention；

Figure 14 is the schematic diagram for the relationship being described in detail between network telephone terminal and voip server or SBC；

Figure 15 to illustrate the invention an embodiment prediction other side speech quality process precedence diagram；

Figure 16 to illustrate the invention another embodiment without call generating process predict to multiple network telephone terminals The precedence diagram of the process of speech quality.

Specific implementation mode

Only it is used for illustrating the embodiment of structure or function, therefore disclosed skill about disclosed technology contents The interest field of art should not be limited by embodiment described herein.I.e. embodiment can carry out various modifications, have various shapes Formula, therefore, the interest field of disclosed technology should include the equipollent that technological thought can be achieved.The particular embodiment of the present invention It is not that need to include whole above-mentioned purpose active effects or only include following effect, therefore, the interest field of disclosed technology is not It is so limited.

In addition, the term for this present invention need to understand as follows.

The terms such as " first ", " second " are used to distinguish another inscape, but right from an inscape It is not limited by above-mentioned term.For example, the first inscape can be named as the second inscape, and similarly, the second structure It can also be named as the first inscape at element.

One structure " connection " or " access " another structure refer to being directly connected to or accessing another structure or pass through it His structure connects or access.In contrast, a structure and another structure " being directly connected to " refer to that there is no other knots for centre Structure.Illustrate other descriptions of relationship between structure, for example, " ... between " and " between just existing ... " or " adjacent to ... " and " being connected on ... " etc. is also similarly to look like.

It is not distinguished significantly in context, then singular record includes the meaning of plural number, the arts such as " comprising " or " possessing " Language is indicated there are the feature recorded on specification, number, step, action, structure, component or combination thereof, and non-predetermined row Presence except other features of one or more, number, step, action, structure, component or combination thereof or additional possibility Property.

In each step, distinguished symbol (for example, a, b, c etc.) is for used in the facility that illustrates, distinguished symbol is not Illustrate the sequence of each step, unless particular order is expressly recited on context, each step also can be by different with recorded sequence Sequence implement.That is, each step can be implemented by recorded sequence, also can actually implement simultaneously, it can also be in the opposite order Implement.

The present invention can be realized on computer-readable recording medium with computer-readable coding, and computer-readable Recording medium includes preservation can be by the recording device of all kinds for the data that computer system is read.Computer-readable record Medium, such as have ROM, RAM, CD-ROM, disk, floppy disk, optical data storage devices etc..

Unless expressly stated otherwise, the meaning of all terms as used herein including technology or scientific terminology and sheet As the meaning that utility model person of ordinary skill in the field is generally understood.It is generally using with art defined in dictionary The identical term of language have with the meaning equivalent in meaning in the context of the relevant technologies, unless there are specific definition, in the application In do not have ideal or excessive meaning.

The schematic diagram of Fig. 1 speech quality forecasting systems of an embodiment to illustrate the invention.

As shown in Figure 1, speech quality forecasting system 10 includes speech quality prediction service unit 20 and speech quality prediction Server 300, and speech quality prediction service unit 20 includes network telephone terminal 100 and voip server 200.These Network connection can be passed through.

Speech quality predicts that a side of the service unit 20 in network telephone terminal 100 or voip server 200 holds The speech quality of other side can be predicted in row.More specifically, network telephone terminal can be predicted in speech quality prediction service unit 20 100 and voip server 200, network telephone terminal 100 and speech quality predictive server 300 or network telephone terminal Speech quality between 100.

Network telephone terminal 100 can be possessed by user, and can be and voip server 200 and speech quality The computing device that predictive server 300 connects.For example, network telephone terminal 100 can be but unrestricted by realizations such as smart mobile phones.

Arbitrary the Internet telephone calls terminal or calling terminal 102,106,110,114,116,120, the networking telephone it is called Terminal or called terminal 104,108,112,118,122 belong to network telephone terminal.

Speech quality predictive server 300 is for reducing the voip server 200 from executable SIP call treatments Transmission test burden.Here, the generation that call treatment, which refers to control and monitoring, to be called is stateful to ensure to the institute terminated Complete the process of normal talking.It, can be in addition, speech quality predictive server 300 can be not directly dependent upon with SIP network Network telephone terminal 100 carries out IP (Internet Protocol) and connects, and is carried out by linking with voip server 200 The common server of information exchange.

First, to predicting that the transmission test of different relaying section speech qualities illustrates.

The transmission test of speech quality forecast interval can also be used comprising the media that can play as practical converse Practical RTP/RTC groupings carry out.In specific sections, both sides are grouped and using RTP/RCTP is received and dispatched simultaneously by RTCP grouping meters The method for calculating RTT (Round Trip Time) and packet loss predicts that speech quality, both sides share result to calculate two-way call The prediction result of quality can also utilize special body to generate the other side that RTP groupings are transferred in section, then other side receives and exists side by side It transmits again, the method that corresponding main body calculates the RTT and packet loss of each RTP groupings.

It includes the RTP groupings or even of the transmission time information of transmission equipment side not comprising physical medium that transmission, which can also be utilized, Without header, has and be grouped identical IP packet sizes and transmission intercal with RTP, include the UDP of the transmission time information of transmission equipment side Grouping, and be instantly available to transmit again from other side and speech quality is predicted with the RTT of calculating and the method for packet loss.

It the case where being divided into speech quality forecast interval, mutually implementing transmission test simultaneously and is dominated first by side The case where transmitting transmission test grouping, when dominating transmission transmission test grouping first by side, both sides are without knowing other side's Address is transmitted first by side, then receiving side carries out negative direction transmission with reference to the originating address of IP groupings.

Therefore, because can be mutually easy simultaneously by dominating the case where acquisition partner address implements transmission test later by side Implement transmission test, is mainly illustrated based on situation about being dominated by side below.

Fig. 2 is the schematic diagram for the structure for illustrating speech quality prediction service unit.

As shown in Fig. 2, speech quality prediction service unit 20 includes SIP transactions requests portion 21, SIP affairs response receiving parts 22, speech quality prediction section 23, SIP affairs acknowledgement transmissions portion 24, call selection determination section 25 and control unit 26.

SIP transactions requests portion 21 is optionally included with the address letter of the one side of the speech quality prediction for a side Breath (IP address, No. Port), SIP (the Session Initiation required to other side's transmission comprising speech quality prediction Protocol).For example, the SIP transactions requests comprising speech quality predictions request information can be transferred to by SIP transactions requests portion 21 Network telephone terminal 100.In another example SIP transactions requests portion 21 can ask the SIP affairs comprising speech quality predictions request information It asks and is transferred to voip server 200.

In one embodiment, SIP transactions requests portion 21 is in network telephone terminal 100 and multiple servers (networking telephone clothes Be engaged in device 200 or speech quality predictive server 300) between there are when NAT device, can be for generating transmission test session Address information comprising voip server 200 or speech quality predictive server 300 in SIP transactions requests is transferred to network Telephone terminal 100, and network telephone terminal 100 can predict clothes by the voip server 200 or speech quality received NAP pin holes grouping for generating NAT pin holes is transferred to voip server 200 or call by the address information of business device 300 Prediction of quality server 300.

SIP affairs responses receiving part 22 can receive SIP affairs responses.More specifically, 22 sound of SIP affairs response receiving part The SIP affairs responses of the address information (IP address, No. Port) comprising other side should be received from other side in SIP transactions requests.One In embodiment, as the response to SIP transactions requests, SIP affairs responses receiving part 22 can be received from network telephone terminal 100 SIP affairs responses.For example, as the response to SIP transactions requests, it is included as transmission that SIP affairs responses receiving part 22, which can receive, Test the address information of newly assigned Port information.

Speech quality prediction section 23 by the address information of a side and the address information of other side, by using 1) with practical matchmaker The RTP groupings unrelated whether including of body；2) RTCP is grouped；Or 3) the disguise oneself as camouflage UDP grouping of RTP of other side (is called in the following text and " passed Defeated experiment grouping ") communication execute with the transmission test of other side to predict speech quality.

Speech quality prediction section 23, can by using bipartite when a side is connected with other side by NAT device The NAT pin holes (Pinhole) for address information are generated in advance via the bipartite communication of NAT device in location information.

For example, voip server 200 or speech quality predictive server 300 by SIP affairs responses using being connect Object encoding and decoding are predicted in the address for the network telephone terminal 100 that receipts portion 22 obtains to network telephone terminal 100 and speech quality Device implements transmission test by the UDP groupings of RTP or the RTP that disguises oneself as of the generation not comprising practical RTP/RTCP or physical medium To predict speech quality.

In another example network telephone terminal 100 is using the network telephone service obtained by SIP affairs responses receiving part 22 The address of device 200 predicts object codec to speech quality, utilizes the RTP not comprising practical RTP/RTCP or physical medium Or the UDP groupings for the RTP that disguises oneself as implement transmission test to predict speech quality.

I.e. in network telephone terminal 100 and each server (voip server 200 or speech quality predictive server 300) there are when NAT Traversal between, in the SIP affairs that voip server 200 is transmitted to network telephone terminal 100 Include the ground of voip server 200 or speech quality predictive server 300 for generating transmission test session in request Location information is transferred to network telephone terminal 100, and network telephone terminal 100 is to the network telephone service obtained from SIP transactions requests After grouping of the address of device 200 or speech quality predictive server 300 transmission for generating NAT pin holes (Pinhole), network Telephony server 200 or speech quality server 300 can utilize NAT pin holes (Pinhole), to network telephone terminal 100 and lead to Prediction of quality object codec is talked about, is grouped by the UDP of RTP or disguise oneself as RTP of the generation not comprising practical RTP/RTCP real Transmission test is applied to predict speech quality.

At this point, in network telephone terminal 100 and voip server 200 or network telephone terminal 100 and speech quality Before being executed between predictive server 300, SIP certifications to network telephone terminal 100 are can perform, when authentification failure abandons conversing Prediction of quality, and execute transmission test when certification success.

In one embodiment, the address information and port letter that speech quality prediction section 23 can be based on network telephone terminal 100 Breath generates the UDP groupings for the RTP that disguises oneself as, and transmission test is executed by the UDP groupings generated, to predict to network telephone terminal 100 speech quality.Here, multiple camouflage UDP groupings may include size identical with RTP, transmission intercal and transmission time number According to.

Speech quality prediction section 23 N number of (N be more than 2 natural number) above-mentioned transmission test can be sent packets to a side or Other side, and confirm respective RTT (Round Trip Time) and whether to respective loss to predict speech quality.

Speech quality prediction section 23 can be by confirming that generated camouflage UDP is grouped respective RTT (Round TripTime to predict the speech quality of other side whether) and to respective reciprocal loss.Speech quality prediction section 23 can be based on puppet The dress UDP message respective time of reception and the transmission time for being recorded in camouflage UDP message respectively calculate RTT values, and by calculating The quantity of the RTP data of omission and the RTT of whole RT groupings predict that other side's is logical to average and RTT values deviation (i.e. irrelevance) Talk about quality.

In one embodiment, speech quality prediction section 23 produces N number of multiple camouflage UDP groupings successively and is transferred to network Telephone terminal 100, and network telephone terminal 100 is transmitted further to speech quality prediction while receiving multiple camouflage UDP groupings Portion 23.Speech quality prediction section 23 by transmission process again confirm multiple camouflage UDP be grouped respective RTT values and back and forth lose and It is no, to predict the other side i.e. speech quality of network telephone terminal 100.

In addition, speech quality prediction section 23 is to execute in voip server 200 or SBC and generate without calling It, can be whole to network telephone server or SBC and the networking telephone when course prediction is to call and the quality of multiple network telephone terminals Section between end is grouped by using the RTP unrelated whether including 1) with physical medium；2) RTCP is grouped；Or 3) to right Side disguise oneself as RTP camouflage UDP grouping (calling in the following text " transmission test grouping ") communication execution with multiple network telephone terminals respectively Transmission test to predict speech quality.

SIP affairs acknowledgement transmissions portion 24 or after the execution (i.e. (a) step) in SIP transactions requests portion 21, other side utilize After speech quality prediction section (i.e. (c) step) is first carried out to obtain speech quality prediction result in the address of one side, by result It is contained among SIP affairs responses and is transferred to a side.

Call selection determination section can save pre- based on the packet loss rate and propagation delay time being grouped to N number of transmission test The prediction speech quality of survey, to determine that the call to other side selects.

Call selection determination section 25 other side by multiple network telephone terminal structures 100 at when, based on to multiple networking telephones The respective prediction speech quality of terminal selects particular network telephone terminal to converse.

In one embodiment, call selection determination section 25, will when the communication between a side and other side belongs to voice communication It can currently use and the network telephone terminal with minimum propagation time delay is selected as network telephone terminal.In another implementation In example, call selection determination section 24 can will be used currently and be had when the communication between a side and other side belongs to data communication There is the network telephone terminal of minimum packet loss rate to be selected as network telephone terminal.

Call selection determination section 25 exists in the relay between first and second network communication device executes media When at least one media relays device of relaying, in first and second network communication device and at least one media relays device In (calling in the following text " speech quality prediction meanss "), to the relaying section being formed between adjacent speech quality prediction meanss, pass through SIP transactions requests are executed to the response of SIP affairs or speech quality prediction is individually performed, are sequentially completed above-mentioned speech quality prediction Step.

For example, first, first network communication device (or calling terminal) is automatically made last speech quality Target area Between boundary, include the second network communication device in the SIP transactions requests transmitted to the voip server 200 belonging to oneself (or speech quality prediction object called terminal) information transmits speech quality predictions request information.

The second, during SIP transactions requests are transferred to the second network communication device, if SIP network component oneself is not Execute media relays, then SIP transactions requests be transferred to next destination as former state, and if execute media relays, executing packet It is same contained in the last speech quality forecast interval boundary in SIP transactions requests and the transmission test for speech quality prediction When, it will be transmitted comprising the SIP transactions requests for the information that the address of oneself is set as to last speech quality forecast interval boundary To next destination, or will include speech quality prediction result after executing the transmission test for predicting speech quality It is transferred to the SIP transactions requests for the information that the address of oneself is set as to last speech quality forecast interval boundary next Destination.

If third, SIP transactions requests reach callee side voip server, callee side voip server executes matchmaker Body relays, then can execute the last speech quality forecast interval boundary being contained in SIP transactions requests and for matter of conversing The transmission test of prediction is measured, and is also executed between callee side voip server and the second network communication device for predicting After the transmission test of speech quality, including two prediction results are to first network communication device transfers SIP affairs responses, or Execute the last speech quality forecast interval boundary being contained in SIP transactions requests and the transmission examination predicted for speech quality While testing, the SIP affairs comprising the information that the address of oneself is set as to last speech quality forecast interval boundary are asked It asks and is transferred to the second network communication device, or executing the last speech quality forecast interval being contained in SIP transactions requests After boundary and the transmission test predicted for speech quality, it will be set comprising speech quality prediction result and by the address of oneself SIP transactions requests for the information on last speech quality forecast interval boundary are transferred to the second network communication device.

Four, the second network communication devices are set to speech quality forecast interval boundary automatically, and work as SIP transactions requests The second network communication device is reached, then is executing the last speech quality forecast interval boundary being recorded in SIP transactions requests After the transmission test for predicting speech quality, including result is by SIP affairs acknowledgement transmissions to first network communication device, And speech quality prediction result is not contained in SIP affairs in the SIP network component as speech quality forecast interval boundary and is asked When asking middle, it may be included in SIP affairs responses.

Call selection determination section 25 can will be stored in cache memory and simultaneously to the speech quality prediction for relaying section More new cache when more than specific time.For example, the SIP network component as speech quality forecast interval boundary Speech quality prediction result or the SIP affairs responses that can refer to oneself execution, by oneself and as speech quality forecast interval side The speech quality prediction result of other SIP network components on boundary, which is stored in cache memory, to be managed, in high speed When there is the speech quality predictions request repeated within the end time of buffer storage, the transmission test repeated is avoided.

Call selection determination section 25 receives when to the speech quality prediction of failure in given trunk section and includes and failure The SIP affairs responses of the relevant information about speech quality prediction meanss of reason.For example, call selection determination section 25 is in SIP During transactions requests are transferred to the second network communication device, if SIP network component can not continue because of strategy or technical reason It is pre- comprising the speech quality by different sections so far in SIP affairs responses when necessary when carrying out speech quality prediction It surveys result and includes that the information or failure cause that oneself will be set as the failure place that speech quality is predicted are transmitted to first network Communication device.

In addition, call selection determination section 25 provides call matter based on the prediction to respective speech quality, to partner The prediction result of amount or execute the information of the network telephone terminal that can be used with classic speech quality is provided or Connection is executed to attempt.

For example, for the speech quality between network telephone terminal 100 and voip server 200, networking telephone clothes Business device 200 is utilized directly to be used for using the registered address of network telephone terminal 100 or generation are other without calling generating process The transmission test session of speech quality prediction by RTP of the generation not comprising practical RTP/RTCP or physical medium or disguises oneself as After the UDP groupings of RTP calculate packet loss rate, the method for transmission delay value predicts speech quality and preserve information, by network The speech quality predictive information of telephone terminal 100 passes to other side, so that partner is joined before actually call connection It examines or makes partner when there is the multiple network telephone terminals that can reach identical purpose, select the network electricity that speech quality is good Telephone terminal.

Control unit 26 can control speech quality prediction service unit overall operation, can control SIP transactions requests portion 21, SIP affairs responses receiving part 22, speech quality prediction section 23, SIP affairs acknowledgement transmissions portion 24 and call selection determination section 25 it Between control flow or data flow.

Fig. 3 is to illustrate to generate voip server request and be used for speech quality predictive server and arbitrary network phone The schematic diagram of the process of the transmission test session of speech quality prediction between terminal.

More specifically, the arbitrary network telephone terminal 100 in Fig. 3 can execute SIP notes to voip server 200 Volume (or SIP REGISTER) becomes the state that can be conversed, and voip server 200 can be predicted to take to speech quality Business device 300 and network telephone terminal 100 request to generate the transmission test session predicted for speech quality and predict speech quality.

Fig. 3 1. during, network telephone terminal 100 can to voip server 200 ask SIP registration.More For body, network telephone terminal 100 can be that request SIP registration generates SIP REGISTER message, and the SIP by being generated REGISTER message executes SIP registration.At this point, the SIP registration to network telephone terminal 100 passes through such as shared key (Shared ) etc. Secret modes execute, and Shared Secret include account (ID) and password (Password).

It does not include the initial of authentication information that network telephone terminal 100 can transmit to voip server 200 SIPREGISTER message request SIP registrations, and voip server 200 can will include authentication information 401Unauthorized acknowledgement transmissions are to network telephone terminal 100.Here, authentication information may include that authentication method, certification are calculated Method, the region of defining account (ID) and password (Password), nonce values etc..

Network telephone terminal 100 can be in transmitting obtained 401Unauthorized responses from voip server 200 Authentication information is confirmed, according to the calculating eap-message digest (Message Digest) required by voip server 200.Here, Eap-message digest (Message Digest) is directed to the message repeated application one-way Hash function of random length, solid to be reduced into The Bit String of measured length.

It includes calculated eap-message digest (Message that network telephone terminal 100 can be transmitted to voip server 200 Digest second SIP REGISTER message).Available the disappearing with network telephone terminal 100 of voip server 200 It ceases the identical method of digest calculations method and calculates eap-message digest.Voip server 200 is comparable after calculating eap-message digest The eap-message digest result calculated compared with oneself and the eap-message digest result being contained in second SIP REGISTER message.

In one embodiment, voip server 200 in the eap-message digest result oneself calculated and is contained in second When eap-message digest result in SIP REGISTER message is identical, terminate the SIP certifications of network telephone terminal 100.The networking telephone Server 200 transmits 200Ok message after terminating to the SIP certifications of network telephone terminal 100, to network telephone terminal 100, Successful execution utilizes the SIP registration of the network telephone terminal 100 of REGISTER message.

In another embodiment, voip server 200 in the eap-message digest result oneself calculated and is contained in second When eap-message digest result in a SIP REGISTER message differs, the SIP certifications of network telephone terminal 100 cannot be terminated, To the SIP authentification failures of network telephone terminal 100.At this point, voip server 200 can be transmitted to network telephone terminal 100 Such as 403Forbidden (Bad auth) wrong responses prompt SIP registration failure.

Fig. 3 2. during, voip server 200 can be requested to generate to network telephone terminal 100 for conversing The SIP of the representative Dialog-Out message used as SIP is assumed below in the transmission test session of prediction of quality OPTIONS, but it is unrestricted.

Here, SIP OPTIONS message belongs to for asking to user agent (UA, User Agent) or Proxy The inquiry of the current ability (Capability) of the component of the SIP nets such as Server, Registrar, Redirect Server Sip message, and the component of SIP nets can provide oneself accessible sip message kind in the response to the SIP OPTIONS message The information such as class or interpretable language or accessible sip message Body types.But at present except the energy to the SIP components netted Except the inquiry of power, response, IP network whether being also commonly used for the current normal operation of the component for confirming simple SIP nets The purposes such as Ping orders.

Sip message is in addition to the header of usable basic agreement Plays, as the side for adding new function or characteristic Method, the non-standard report that the X- with (eXtension) meaning of (eXperimental) or extension with experiment can be used to start Head.

The characteristics of voip server 200 is using the usable nonstandard pseudo header started with X-, is added into X- The headers such as Quality-Test (header) transmit SIP OPTIONS message to network telephone terminal 100, request to generate for leading to Talk about the transmission test session of prediction of quality.

The header of X-Quality-Test can following form configuration.

X-Quality-Test:Request aservaddr=" 20.20.20.20:1000 ", vservaddr=" 20.20.20.20:1001 ", method=digest, nonce=" 458e3bf6 ", algorithm=MD5, realm=" Lecture ", acodec=ulaw, vcodec=h263

Here, Request is speech quality predictions request, aservaddr is the speech quality prediction service for voice The address of device 300, vservaddr are the address of the speech quality predictive server 300 for image, and method is arranged to Digest means that nonce is to be included in what cryptographic Hash calculated using the authentication method by message digest modes Random value, algorithm are hash algorithm, and the region that realm is definition ID and Password (is mostly used server name in convention Claim), acodec is that use ulaw codecs and vcodec as the codec for voice be as the volume for image Decoder uses h263 codecs.

In upper example (form of the header of X-Quality-Test), codec names have been only used to briefly express. By taking h.264 codec as an example, in order to accurately understand data transmission rate by different codecs, also include such as Profile and The details such as Level are transmitted, with calculate the quantity of frames per second (Frame) or the size of frame (Frame).

The address (aservaddr, vservaddr) of speech quality predictive server 300 for voice or image is to use It, can be pre- by voip server 200 and speech quality in the address for generating the transmission test session for speech quality prediction Server 300 is surveyed to consult to decide by other process.

Voip server 200, can also be in sip message in addition to the method for adding new header in a sip message Solicited message of the middle addition body transmission for speech quality prediction.

For example, addition such as Content-Type in SIP OPTIONS message above:application/call- The header of quality-test, and in content body SIP OPTIONS message is transmitted comprising following content.

Request

aservaddr:20.20.20.20:1000

vservaddr:20.20.20.20:1001

method:digest

nonce:"458e3bf6"

algorithm:MD5

realm:"lecture"

acodec:"ulaw"

vcodec:"h263"

Fig. 3 2. during, the header of the X-Quality-Test of SIP 200Ok message is as follows：

X-Quality-Test:Request aservaddr=" 20.20.20.20:1000 ", vservaddr=" 20.20.20.20:1001 ", method=digest, nonce=" 458e3bf6 ", algorithm=MD5, realm=" Lecture ", response=" dfe56131d1958046689d83306477ecc ", username=" abc ", uri=" sip:Abc@def.com ", acodec=ulaw, vcodec=h263clientaddr=1.1.1.1:10,11

Message digest computation result can be contained in the response of X-Quality-Test headers by network telephone terminal 100 Voip server is transferred in field.The message digest computation of network telephone terminal 100 can respectively calculate A1, nonce and A2, and it is worthwhile to apply hash algorithm to carry out again to A1, nonce and A2.

Here, A1 is to calculate being recorded in ID, Password, realm application hash algorithm of username fields Message digest's as a result, A2 is to the method " OPTIONS " of SIP OPTIONS message and SIP OPTIONS message Request uri application hash algorithms calculate the result of message digest.

Voip server 200 can be passed by comparing the message digest value oneself calculated and from network telephone terminal 100 Defeated obtained message digest value executes the certification to network telephone terminal 100.

Fig. 3 3. during, when to 100 certification of network telephone terminal success, then voip server 200 can be to logical Talk about 300 transmission of authentication information of prediction of quality server.Here, authentication information may include being transferred to each of network telephone terminal 100 For the voice of speech quality predictive server 300 and the IP address (20.20.20.20) of image and No. Port (1000, 1001), the message digest value that voip server 200 oneself calculates during 2..If in Fig. 3 2. in the process to net The authentification failure of network telephone terminal 100 can then be immediately finished speech quality prediction.

Fig. 3 4. during, network telephone terminal 100 can be by will include that the grouping of authentication information is transferred to call matter Predictive server 300 is measured, is generated for the transmission test between network telephone terminal 100 and speech quality predictive server 300 NAT pin holes (Pinhole).

More specifically, network telephone terminal 100 can distribute it is non-for sip message transmit and receive Port (for example, Arbitrary two Port of oneself 5060) are distributed in sip message when Default uses, transmit to obtain to by the 2. process of Fig. 3 Speech quality predictive server 300 address transmission packe, and by Fig. 3 2. during calculate message digest value as former state wrap Contained in being transferred to speech quality predictive server 300 in grouping.

The comparable eap-message digest received from voip server 200 of speech quality predictive server 300 and from network The eap-message digest that telephone terminal 100 receives.In one embodiment, when eap-message digest comparison result is identical, speech quality prediction SIP certification success of the server 300 to network telephone terminal 100, can continue to execute 5. process later.In another embodiment, When eap-message digest comparison result differs, speech quality predictive server 300 loses the SIP certifications of network telephone terminal 100 It loses, and entire speech quality predicts procedure failure.

Fig. 3 5. during, speech quality predictive server 300, which can be recorded in, to be received by the 4. process of Fig. 3 Source IP and Source No. Port number in the IP headers of grouping as a purpose, executes the transmission for speech quality prediction Experiment.The transmission test session generated in above process can cope with NAT (the Network Address of all four kinds of modes Translation), the NAT of four kinds of modes is Full Cone, Restricted Cone, Port Restricted herein Cone、Symmetric NAT。

In fig. 2, if in network telephone terminal 100 and voip server 200 and speech quality predictive server 300 Between be not present NAT Traversal, then voip server 200 can Fig. 3 2. during check network telephone terminal 100 certification success or not.

In one embodiment, in certification success, voip server 200 can be transmitted by the 3. procedure attachment of Fig. 3 Address information (the clientaddr words of X-Quality-Test headers of the network telephone terminal 100 obtained during 2. Section), and process of the speech quality predictive server 300 without the generation NAT pin holes (Pinhole) of the 4. process of Fig. 3, directly hold Capable 5. course prediction speech quality.

If for example, between network telephone terminal 100 and speech quality predictive server 300 be not present NAT Traversal, Then be not necessarily to generate the process of NAT pin holes (Pinhole), therefore, Fig. 3 2. during, be not necessarily in SIP OPTIONS message The address of speech quality predictive server 300 must be included.

In another example if there are NAT Traversal between network telephone terminal 100 and speech quality predictive server 300, Then because speech quality predictive server 300 be based on NAT pin holes (Pinhole) execute transmission test, Fig. 3 2. during, It is not necessarily to that the address of network telephone terminal 100 must be included in SIP 200Ok message.

I.e. network telephone terminal 100 becomes can to converse after completing SIP registration to voip server 200 State, and voip server 200 can request to generate use to speech quality predictive server 300 and network telephone terminal 100 Speech quality is predicted in the transmission test session of speech quality prediction.

Fig. 4 is to illustrate to generate voip server request and be used for voip server and arbitrary network telephone terminal Between speech quality prediction transmission test session process schematic diagram.More specifically, Fig. 4 is to illustrate without using call Quality server 300 is directly executed the schematic diagram of the process of speech quality prediction by voip server 200.

Fig. 4 1. during, network telephone terminal 100 can to voip server 200 ask SIP registration.More For body, network telephone terminal 100 can be that request SIP registration generates SIP REGISTER message, and the SIP by being generated REGISTER message executes SIP registration.At this point, the SIP registration to network telephone terminal 100 passes through such as shared key (Shared ) etc. Secret modes execute.

Fig. 4 2. during, voip server 200 can be requested to generate to network telephone terminal 100 for conversing The transmission test session of prediction of quality.At this point, the 2. process of Fig. 4 remove Fig. 3 2. during, X-Quality-Test headers Aservaddr, vservaddr are arranged to indicate the voip server 200 of the address of non-call prediction of quality server 300 All it is identical except oneself.

Fig. 4 does not execute the 3. process of Fig. 3 because not using speech quality predictive server 300.Fig. 4 3. during, As the 4. process of Fig. 3, network telephone terminal 100 can be by will include that the grouping of authentication information is transferred to network telephone service Device 200 generates NAT pin holes (Pinhole) the execution certification for being used for transmission experiment.

Fig. 4 4. during, voip server 200 can be contained in the grouping received by the 3. process of Fig. 4 IP headers in Source IP and Source No. Port number as a purpose, execute for speech quality prediction transmission examination It tests.The transmission test session generated in above process can cope with NAT (the Network Address of all four kinds of modes Translation), the NAT of four kinds of modes is Full Cone, Restricted Cone, Port Restricted herein Cone、Symmetric NAT。

If NAT Traversal, network are not present between network telephone terminal 100 and voip server 200 Telephony server 200 can be in the certification success or not for 2. checking network telephone terminal 100 in the process of Fig. 4.In one embodiment, If certification, at sky, voip server 200 can be not necessarily to generate NAT pin holes (Pinhole) in network telephone terminal 100 Process directly executes next course prediction speech quality.For example, being contained in the network telephone terminal in SIP200Ok message 100, the Port information that transmission test distribution is predicted for speech quality can be used, the 3. process without Fig. 4 directly executes transmission examination It tests.

I.e. network telephone terminal 100 becomes can to converse after completing SIP registration to voip server 200 State, and voip server 200 can request to generate the transmission test predicted for speech quality to network telephone terminal 100 Speech quality is predicted in session.

Fig. 5 is to illustrate to generate network telephone terminal request and for speech quality predictive server and arbitrary network phone end The schematic diagram of the process of the transmission test session of speech quality prediction between end.More specifically, Fig. 5 is to illustrate by network electricity The schematic diagram of the process of speech quality prediction is asked and directly executed to telephone terminal 100.

Fig. 5 1. during, network telephone terminal 100 can to voip server 200 ask SIP registration.More For body, network telephone terminal 100 can be that request SIP registration generates SIP REGISTER message, and the SIP by being generated REGISTER message executes SIP registration.Here, the 1. process phase of the SIP registration process of network telephone terminal 100 and Fig. 3 and Fig. 4 Together.

Fig. 5 2. during, network telephone terminal 100 can be requested to generate to voip server 200 for conversing The transmission test session of prediction of quality.More specifically, as shown in Figures 3 and 4, SIP can be used in telephone terminal 100 OPTIONS。

The header of X-Quality-Test can following form configuration.

X-Quality-Test:Request acodec=" ulaw ", vcodec=" h263 "

Here, Request is speech quality predictions request, acodec and vcodec are respectively voice and image codec.

It is used for transmission the other information of experiment, acknowledgement transmissions that can be in voip server 200 as SIP OPTIONS SIP 200Ok response messages X-Quality-Test headers in be included in the form of following.

X-Quality-Test:Response aservaddr=" 20.20.20.20:1000 ", vservaddr=" 20.20.20.20:1001 ", method=digest, nonce=" 458e3bf6 ", algorithm=MD5, realm=" Lecture ", acodec=" ulaw ", vcodec=" h263 "

Here, Request is the response to speech quality predictions request, aservaddr is the speech quality for voice The address of predictive server 300, vservaddr are the address of the speech quality predictive server 300 for image, and method is set Being set to digest means that, using the authentication method by message digest modes, nonce is to be included in cryptographic Hash meter The random value of calculation, algorithm are hash algorithm, and realm is the region for defining ID and Password, and acodec is to be used as to be used for It is to use h263 encoding and decoding as the codec for image that the codec of voice, which uses ulaw codecs and vcodec, Device.

Voip server 200 Fig. 5 2. during, can be noted such as the SIP using SIP REGISTER of 1. process Volume request process is such, including the SIP certifications that 401Unauthorized completes network telephone terminal 100 execute later.Only because Speech quality prediction is executed by speech quality predictive server 300, so being omitted in Figure 5.

Fig. 5 3. during, voip server 200 can be performed as the 3. process of Fig. 3, to speech quality 300 transmission of authentication information of predictive server.It is respectively used to leading to here, authentication information may include being transferred to network telephone terminal 100 Talk about the voice of prediction of quality server 300 and IP address (20.20.20.20) and No. Port (1000,1001) of image.One In embodiment, with Fig. 3 2. during calculate the method for eap-message digest it is identical, voip server 200 can calculate message It makes a summary and to 300 transmission of authentication information of speech quality predictive server.

Fig. 5 4. during, with Fig. 3 2. during the mode that is calculated it is identical, network telephone terminal 100 exists After calculating eap-message digest, calculated value is contained in grouping and is first transmitted to speech quality predictive server 300.

Speech quality predictive server 300 is by comparing the eap-message digest received from network telephone terminal 100 and passes through Fig. 5 The eap-message digest that is received from voip server 200 of 3. process, execute the SIP certifications to network telephone terminal 100.

In one embodiment, it when the SIP certifications success to network telephone terminal 100, then surveys server 300 and can receive network The transmission test grouping for speech quality prediction that telephone terminal 100 transmits, and reported with the IP for being contained in transmission test grouping Source IP and Source Port are transmitted again as a purpose in head.The transmission test session generated in above process The NAT (Network Address Translation) of all four kinds of modes can be coped with, the NAT of four kinds of modes is Full herein Cone、Restricted Cone、Port Restricted Cone、Symmetric NAT。

In another embodiment, when to the SIP authentification failures of network telephone terminal 100, then speech quality predictive server 300 refusals transmit or are separately notified to the information about SIP authentification failures to network telephone terminal 100 again, to terminate speech quality Prediction process.

I.e. network telephone terminal 100 becomes the state that can be conversed to the execution SIP registration of voip server 200, And network telephone terminal 100 can request to generate the transmission test session predicted for speech quality to voip server 200. Voip server 200 can will not be that the address information of the speech quality predictive server 300 of oneself is transferred to the networking telephone Terminal 100, and network telephone terminal 100 using the speech quality predictive server 300 of the address execute SIP verification process it Afterwards, it generates transmission test session and predicts speech quality.

Fig. 6 be illustrate to generate network telephone terminal request and for voip server and arbitrary network telephone terminal it Between speech quality prediction transmission test session process schematic diagram.More specifically, Fig. 6 is to illustrate by networking telephone end Ask and directly execute the schematic diagram of the process of speech quality prediction in end 100.Only Fig. 6 does not pass through speech quality predictive server 300, it is logical to directly generate transmission test session prediction with the voip server 200 of the execution SIP registration of network telephone terminal 100 Talk about quality.

Fig. 6 1. during, network telephone terminal 100 can to voip server 200 ask SIP registration.More For body, network telephone terminal 100 can be that request SIP registration generates SIP REGISTER message, and the SIP by being generated REGISTER message executes SIP registration.Here, the 1. mistake of the SIP registration process of network telephone terminal 100 and Fig. 3, Fig. 4 and Fig. 5 Cheng Xiangtong.

Fig. 6 2. during, if Fig. 3, Fig. 4 and Fig. 5 are identical, network telephone terminal 100 can be used SIPOPTIONS to Voip server 200 requests to generate the transmission test session predicted for speech quality.Only, network telephone service at this time In the 200Ok responses that device 200 transmits, aservaddr, vservaddr field of X-Quality-Test headers include oneself IP address and be newly assigned No. Port two of transmission test.

Fig. 6 3. during, when SIP certifications success, network telephone terminal 100 produces transmission test session, can be pre- Survey speech quality.It is identical as mode (Fig. 3, Fig. 4 and Fig. 5) before, network telephone terminal 100 and voip server 200 Message digest value can respectively be calculated.In one embodiment, what 200 comparing cell telephone terminal 100 of voip server transmitted disappears Eap-message digest breath abstract and oneself calculated, the certification success when being worth identical, acceptable transmission tests session and generates request, with net Network telephone terminal 100 executes speech quality prediction.Here, except the object for executing speech quality prediction is voip server It is identical as the 4. process of Fig. 5 except 200.

In addition, voip server 200 Fig. 6 2. during, can as 1. process SIP REGISTER processes that Sample executes speech quality prediction including after the SIP certifications of 401Unauthorized completion network telephone terminals 100.At this point, Fig. 6 3. during, network telephone terminal 100 using Fig. 6 2. during obtain voip server 200 ground Location information is immediately performed transmission test.

I.e. network telephone terminal 100 becomes the state that can be conversed to the execution SIP registration of voip server 200, And network telephone terminal 100 can request to generate the transmission test session predicted for speech quality to voip server 200. Voip server 200 can be included as the ground of the newly assigned Port of oneself of transmission test to the notification of network telephone terminal 100 Location and SIP certification request information, and network telephone terminal 100 executes SIP using the address of the voip server 200 and recognizes After card process, generates transmission test session and predict speech quality.

Speech quality until Fig. 3 to Fig. 6 predicts process, illustrates by executing network telephone terminal 100 and the networking telephone Between server (SIP Proxy, IP-PBX, CSCF etc.) 200 or network telephone terminal 100 and speech quality predictive server Transmission test between 300 measures the process of the prediction speech qualities such as packet loss rate, transmission delay, the initial data property completed. At this point, to solve the problems, such as NAT, when having SBC between network telephone terminal 100 and voip server 200, apply as former state Speech quality until Fig. 3 to Fig. 6 predicts process, and the speech quality between network telephone terminal 100 is predicted by SBC, moreover, SBC can also only serve the work of the prediction process of the speech quality between meta network telephone terminal 100 and voip server 200 With.

It can be given birth to the speech quality predictive server 300 of the voip server 200 or outside that execute SIP registration process Transmission test is executed at the other transmission test session for predicting speech quality.In addition, during SIP registration, network The address for the network telephone terminal 100 that telephony server 200 identifies, also because solving knot by the NATTraversal of SIP itself Structure solves the problems, such as NAT and is stored in voip server 200, the address can be used simply to execute transmission test as former state.

Only, because being the address transmitted for sip message, although sip message and transmission test grouping are possible to be mixed Confuse, but because the pattern of sip message is fixed, when RTP groupings or camouflage (identical timid and transmission intercal) RTP are grouped and record transmission Between UDP grouping pattern be also fixed, differentiation be not impossible, can use as former state SIP registration address execution speech quality it is pre- Survey transmission test.

If networking telephone operator solves the NAT of SIP using SBC (Session Border Controller) Traversal problems and when executing the Topology Hiding of voip server side, generally execute Media by SBC Relay is used at this point, NAT Pinhole can be usually generated in advance in networking person's terminal and SBC when actual call generates.This When, to make speech quality prediction be executed in identical environment when being generated with actual call to the maximum extent, SBC is used in prediction The Media of speech quality similarly carries out Relay.

In Fig. 3 and Fig. 4, when voip server 200 records oneself or speech quality in SIP OPTIONS message The address information of predictive server 300 is transferred to network telephone terminal 100, then by SBC (Session Border Controller it is) that the address of oneself is transferred to network telephone terminal 100 by the address mapping, it is pre- for speech quality to execute The transmission test session of survey.

In Fig. 5 and Fig. 6, the voip server that can be included in by SBC in the response message to SIP OPTIONS 200 or 300 address mapping of speech quality predictive server be the address prediction speech quality of oneself.Only, because being for giving birth to At the address of actual call, it is contemplated that, can be by SBC and network telephone terminal 100 when making troubles because being clashed with actual call Negotiate that the NAT pin holes (Pinhole) for being used for transmission experiment session are generated in advance, this address can also be used.

During speech quality until such as Fig. 3 to Fig. 6 is predicted, transmission test executes after completing SIP affairs, but Also it can be executed among SIP affairs.For example, Fig. 3 2. during, network telephone terminal 100 can be disappeared by SIP OPTIONS The address for receiving speech quality predictive server 300 and authentication information are ceased, eap-message digest is calculated based on the authentication information and will be counted The value of calculating is transferred to speech quality predictive server 300.Speech quality predictive server 300 is worked as from network telephone terminal 100 Calculated message digest value is received, the 3. process of Fig. 3 is executed among the 2. process of Fig. 3, is based on from network telephone service Device 200 transmits obtained authentication information and is compared completion SIP certifications with the eap-message digest that oneself is calculated.In one embodiment, When SIP certifications are successful, then speech quality predictive server 300 may be used at from network telephone terminal 100 and receives eap-message digest The NAT pin holes (Pinhole) generated in the process execute transmission test, and after executing transmission test, as necessary by network Speech quality prediction result is contained in SIP 200Ok message and is transmitted to network telephone terminal 100 by telephony server 200.Separately Outside, it can also be realized in the case of figure 4 with identical method.

In the case of fig. 5, to receive the address of speech quality predictive server 300, network telephone terminal 100 can be first Receive SIP 200OK message.For example, voip server 200 will in the response as initial SIP OPTIONS message While 401Unauthorized responses are first transmitted to network telephone terminal 100, original transmission of authentication information and speech quality The address of predictive server 300, at this point, transmission test can be executed among second SIP OPTIONS affairs.

If in addition, omitting verification process and not deposited among network telephone terminal 100 and speech quality predictive server 300 In NAT device, then voip server 200 Fig. 5 2. during, pass through SIP OPTIONS message and obtain the networking telephone After the address of terminal 100, the 3. process of Fig. 5 while 2. execution in the process, is being predicted to service in Fig. 5 to speech quality The address of 300 transmission network telephone terminal 100 of device, and speech quality predictive server 300 can be executed with network telephone terminal 100 For the transmission test of speech quality prediction, and prediction result is contained in the 2. process of Fig. 5 by voip server 200 SIP 200Ok message in be transmitted to network telephone terminal 100.It can also be realized in the case of fig. 6 with identical method.

In mobile communication wireless net, in the CSCF for belonging to network telephone terminal 100 and voip server 200 There is such as GPRS (General Packet Radio Service) between (Call Session Control Function) When network, it is also necessary to GPRS attach processes, and in network telephone terminal 100 and SGSN (Serving GPRSSupport Node composition is constituted by PDP (Packet Data Protocol) Context Activate between), in SGSN and GGSN It is made of PDP Context Create processes between (Gateway GPRS Support Node).

When the PDP Context are generated activation, then network telephone terminal 100 is endowed the IP address for being used for transmission experiment With No. Port and can utilize its generate transmission test session predict speech quality.Such process be for the maximum extent with reality Identical environmental forecasting speech quality is required when border calling generates, and because can be passed through according to the strategy of mobile operator GGSN obtains Private Ip, can be identical as the case where wired network uses Private Ip, and NATPinhole, which is generated, to be possible to It is needed.

The 2. process of Fig. 3 will send and receive the codec information for being used for transmission experiment, and utilize the information by difference Codec asks the QOS (Quality Of Service) of required bandwidth in PDP generations and activation, utilizes RSVP The QOS correlation skills such as (Resource Reservation Protocol) or DiffServ (Differentiated Service) Art makes QOS be protected, to make to be held in the identical environment of call of the transmission test between real network telephone terminal 100 Row.Such QOS implementation procedures are applicable in the full content of the present invention.

It ought be in actual call connection procedure, generate and activate by first, PDP Context；Or second, utilization The resource reservation to particular path session of RSVP；Or third, utilize the differentiation of the traffic classification by DiffServ；Or the Four, the various methods such as MPLS (Multi-Protocol Label Switching), IntServ (Integrated Service) QOS is provided, then the speech quality for being used for the different sections between the SIP network component including network telephone terminal 100 is pre- The transmission test of survey is also implemented after providing identical QOS.

Fig. 7 to Figure 12 is the schematic diagram for illustrating to predict the process of the speech quality between arbitrary network telephone terminal.Fig. 7 is extremely Network telephone terminal 100 in Figure 12 is illustrated for assuming to send and receive sip message in IMS nets or wired network.

More specifically, Fig. 7 is in IMS nets, for executing the call matter between arbitrary calling terminal and called terminal The message of amount prediction sends and receives process, when calling terminal passes through such as the message requests such as SIP OPTIONS and called terminal Speech quality predict, then IMS components by SIP OPTIONS message in path identical with the transmission path of SIP INVITE messages On transmitted, in the case of actual call between junction call terminal and called terminal, because of communication common carrier strategy, When a variety of causes such as NAT, Topology Hiding, Media Transcoding execute Media Relay, SIP OPTIONS Message pass through IMS net each component be set to speech quality transmission section boundary (terminal the default setting is transmission section Boundary), execute transmission test by variant section, and be averaged or take merely minimum by each speech quality predictive index The method of value, the speech quality predictive index between the worthwhile whole district.It is indicated belonging to calling terminal and called terminal in the case where Fig. 7 Both sides P-CSCF executes the process of Media Relay because of reasons such as NAT.

Fig. 8 is most of identical as Fig. 7, is GPRS (General Packet Radio in calling terminal and called terminal Service) when net, belong to public IP is directly set by GGSN (Gateway GPRS Support Node) or in terminal and There are SBC etc. between P-CSCF, the case where IP public to indirect gain, because the IMS screen components in addition to terminal and SBC are complete Not executing Media Relay, (SBC general execution Media Relay, but ignored herein will be added in fig. 12 It is bright), transmission test is directly executed between calling terminal and called terminal.

Fig. 7 and Fig. 8 shows execute arbitrary network call terminal 102,106 and specific using SIP OPTIONS message The process of speech quality prediction between networking telephone called terminal 104,108.Here, arbitrary network call terminal 102, 106 can be at least two in multiple network telephone terminals 100 in Fig. 1, and the telephone called terminal of particular network 104,108 can Think at least two in the multiple network telephone terminals 100 being not belonging in Fig. 1 of arbitrary network call terminal 102,106 It is a.

In the following, illustrate for convenience, below by the Internet telephone calls terminal 102,106 be recorded as calling terminal 102, 106, and the telephone called terminal of particular network 104,108 is recorded as called terminal 104,108.

SIP transactions requests portion 21 can record the information of calling terminal 102,106 in From headers, be recorded in To headers After the information of called terminal 104,108, using the method similar with Fig. 3 to Fig. 6, such as X-Quality-Test headers are added Ask speech quality prediction.

The case where with Fig. 3 to Fig. 6, is identical, and the header of X-Quality-Test can also be recorded in the body of sip message, Meaning is meant identical herein, therefore will omit what addition body in a sip message was recorded in following content Process.

The header of X-Quality-Test can following form configuration.

X-Quality-Test:Request acodec=ulaw, vcodec=h263, addr1=1.1.1.1:10,11

Here, Request is speech quality predictions request, addr1 is oneself that will be used in speech quality predictions request Oneself IP address, voice port information and video port information.That is addr1 includes the IP address (1.1.1.1) of oneself, voice Port (10) and image Port (11), acodec are that voice uses ulaw codecs, vcodec to compile solution using h263 for image Code device.

The IP address of oneself can be that personal IP or public IP, SBC (Session Border Controller) are present in Between 102,106 and P-CSCF of calling terminal (Proxy CSCF) 400,404, when calling terminal 102,106 possesses personal IP, Calling terminal 102,106 can record personal IP as former state, and being transformed to public IP by SBC later is transmitted to P-CSCF400,404.This When, because P-CSCF400,404 think that calling terminal 102,106 is used for public IP without executing Media Relay, so as to apply The process of Fig. 8.

The case where with Fig. 3 to Fig. 6, is identical, and SIP certifications can be first carried out in calling terminal 102.Fig. 7 1. during, P- CSCF400 can be by being used as the response to SIP OPTIONS message by 401Unauthorized acknowledgement transmissions to calling terminal 102 request certifications, and calling terminal 102 may include authentication information again by SIP OPTIONS message SIP certification requests It is transferred to P-CSCF400.

The value for the SIP OPTIONS message that P-CSCF400 is received by inspection from calling terminal 102, if certification success Next process can be continued to execute, and such as 403Forbidden (Bad auth) error message can be transmitted if authentification failure To calling terminal 102, refuse speech quality predictions request.In following content, the SIP verification process after Fig. 8 can also lead to It crosses identical process to complete, therefore by omission to the explanation of SIP certifications.

As long as P-CSCF400 is receiving the SIP transactions requests for including speech quality predictions request, self-administered calling is eventually End 102 executes telephone connection by the 1. process of Fig. 7 with the called terminal 104 for being recorded in SIP transactions requests, executes Media It, can be by predicting speech quality with the transmission test of calling terminal 102 when Relay.At this point, transmission test and be recorded in Fig. 3 and Method in Fig. 4 is identical.

For example, P-CSCF or IP-PBX passes through GGSN (Gateway GPRS Support Node) quilt in calling terminal 102 Assign individual IP, it is intermediate when being not carried out the SBC of NAT Traversal, the SIP environment being made of IP-PBX the case where or deposit Can be to solve NAT Traversal to execute Media Relay in the case of the calling terminal for being NAT device behind.At this point, If executing the speech quality prediction technique of Fig. 4, by the P-CSCF or IP-PBX management only special service for executing Media Relay Device then executes Fig. 3 speech quality prediction techniques.In addition, Fig. 7 1. during, because completely without X-Quality- The addr1 fields of Test headers, are ignored herein.

When completion and the speech quality of calling terminal 102 predict that then result can be stored in the high speed of oneself by P-CSCF400 In buffer storage (cache memory), it can be preserved in the form of following in X-Quality-Test headers and execute Media The IP address (30.30.30.30) and voice of the P-CSCF400 of Relay and each port (1000,1001) of image.

X-Quality-Test:Request acodec=ulaw, vcodec=h263, addr1=1.1.1.1:10, 11, CQTY=8.9:8.2, addr2=30.30.30.30:1000,1001

P-CSCF400, can be with the transmission path phase of SIP transaction request messages after preserving the packet header X-Quality-Test SIP OPTIONS message is transmitted to next destination in same path.At this point, P-CSCF400 is the rapidity of speech quality prediction, CQTY (Call Quality) field is removed, after transmitting SIP OPTIONS message first, then in transmission 200Ok responses CQTY fields are inserted into the process to be transmitted.Here, CQTY fields respectively indicate voice quality (8.9) and the quality of image (8.2), Addr2 fields indicate IP address (30.30.30.30), voice port (1000) and the video port of call-side P-CSCF400 (1001)。

The 1. process of Fig. 8 belongs to because calling terminal 106 possesses public IP or in calling terminal 106 and P-CSCF404 in itself Between NAT Traversal are executed there are SBS, be considered as calling terminal 106 and possess public IP without P-CSCF404's The case where Media Relay.Therefore, the P-CSCF404 of Fig. 8 can not be set to the side in the section predicted for speech quality Boundary, and the speech quality prediction between calling terminal 106 and P-CSCF404 can not be performed.Even if calling terminal 106 possesses a People IP, if but configured in identical SUBNET environment, Media Relay can not needed.

For example, when SBC executes NAT Traversal, it can preset and be used between calling terminal 106 and SBC The session of Media transmission generates and maintains NAT pin holes, at this point, can be recorded in the addresses Media Relay of oneself by SBC Addr1 fields (for example, can be the address of oneself by the address change of existing call terminal) are transferred to P-CSCF404.Only, because It is the address for generating actual call for the addresses SBC, it is contemplated that when rushing, can negotiate to generate by SBC and calling terminal 106 and use In the NAT pin holes (Pinhole) of transmission test session, the address of SBC can be set in the addr1 of X-Quality-Test headers Field.

Fig. 8 1. during, if calling terminal 106 is sent to the X- of the SIP OPTIONS message of P-CSCF404 Quality-Test headers are as follows, then can be sent to S-CSCF504 as former state.

X-Quality-Test:Request acodec=ulaw, vcodec=h263, addr1=1.1.1.1:10,11

Such as in the case where IMS is netted, in call-side P-CSCF400,404, next mesh of SIP OPTIONS message Ground be the S-CSCF500 for the Home Network (Home Network) for belonging to calling terminal 102,106,504, and next destination is The I-CSCH600 in the domain belonging to called terminal 104,108,602 can be inquired by HSS by the S- belonging to called terminal 104,108 CSCF502,506, the P-CSC402 belonging to called terminal 104,108,406 and called terminal 104,108 sequential delivery.

If there is no the nodes for executing Media Relay for centre, as the case where Fig. 7, X-Quality-Test reports Head is transmitted as P-CSCF400 records until callee side 402,406 as belonging to calling terminal 102.

Here, addr1 is the address of calling terminal 102, addr2 is the ground of the P-CSCF400 of affiliated calling terminal 102 Location, CQTY be Fig. 7 1. during calculated speech quality prediction result.

In the case of fig. 8, it can be transmitted as being recorded the P-CSCF400 belonging to following calling terminal 102.

X-Quality-Test:Request acodec=ulaw, vcodec=h263, addr1=1.1.1.1:10,11

Here, addr1 is the address of calling terminal 106.

Fig. 7 3. during, because called terminal 104 possesses the reasons such as personal IP, executed by callee side P-CSCF402 Media Relay can then predict the speech quality between called terminal 104 and P-CSCF402 by the method for Fig. 3 and Fig. 4, will As a result (voice 8.1, image 8.2) is stored in cache memory.

In addition, Fig. 7 2. during, callee side P-CSCF402 is to the X- for being recorded in SIP OPTIONS message The call-side P-CSCF400 of the addr2 of Quality-Test headers executes the transmission test for speech quality prediction, by result (voice 8.1, image 8.2) is stored in cache memory.At this point, because being generally not present NATTraversal, it is not necessarily to NAT pin holes (Pinhole) generating process, and execute transmission test with the address for being recorded in addr2.

Speech quality prediction can be executed with call-side P-CSCF400 by the 2. process of Fig. 7 by callee side P-CSCF402, It also replaces and transmission test is directly executed with called terminal 104 by the 3. process of Fig. 7,2. Fig. 7 is only carried out by P-CSCF402 Result and the address of oneself are recorded in after SIP OPTIONS message and are transferred to called terminal 104, by called terminal by process 104 predict speech quality with P-CSCF402.

If Fig. 7 2. during need SIP certifications, such as Fig. 3 to Fig. 6 the method using Shared Secret not It is suitble to, and can be in the shape for exchanging the certificate of CA distribution derived from the CA (Certificate Authority) that can trust each other Under state, certification is executed by using the method for certificate.For example, can be by generating challenge and being transmitted, then receiving side is used The key of oneself is encrypted and transmits again, and the public-key cryptography for decrypting certificate for receiving side confirms and transmitted originally The process of challenge executes other side's certification.

After completing speech quality prediction in this way, P-CSCF402 belonging to called terminal 104 (address= 40.40.40.40:2000,2001) can by following form X-Quality-Test header be configured.

X-Quality-Test:Request acodec=ulaw, vcodec=h263, addr1=1.1.1.1:10, 11, CQTY=8.9:8.2, addr2=30.30.30.30:1000,1001CQTY=8.4:8.5addr3=40.40.40.40: 2000,2001CQTY=8.1:8.3addr4=50.50.50.50:3000,3001

Here, addr1 is the address of calling terminal 102, addr2 is the address of call-side P-CSCF400, and addr3 is quilt It is the address of called terminal 104 to cry the address of side P-CSCF402, addr4.In addition, three CQTY fields respectively indicate to exhale in order Cry between terminal 102 and call-side P-CSCF400, between call-side P-CSCF400 and callee side P-CSCF402, callee side P- Speech quality prediction result between CSCF402 and called terminal 104.

Can in the 200Ok responses to SIP OPTIONS message comprising above-mentioned X-Quality-Test transmission of preamble to exhaling It is terminal 102.It can confirm that the prediction result by different sections between calling terminal 102 and called terminal 104 is all contained in In above-mentioned X-Quality-Test headers.

The speech quality predictive index in different sections is respectively pressed, the main body predicted by execution speech quality is stored in speed buffering In memory, when receiving the speech quality predictions request to same object within a certain period of time, actual transmissions examination is not executed It tests, and utilizes the value being stored in cache memory, so as to reduce unnecessary flow.For example, being exhaled when belonging to identical It, can when another calling terminal 102 of side P-CSCF400 being made to ask speech quality prediction to called terminal 104 same as described above Speech quality prediction is only carried out between call-side P-CSCF400 and calling terminal 102, and the both sides of the sages and men of virtue can use guarantor as former state The value being stored in cache memory.

After transmission test by the 2. process of Fig. 7, callee side P-CSCF402 is because be the master for predicting speech quality Body, by result (CQTY=8.4:8.5) it is preserved as the speech quality prediction result with call-side P-CSCF400, and Also preserve the speech quality prediction result with called terminal 104.Call-side P-CSCF400 is in SIP 200Ok response messages, ginseng It examines and is stored in cache memory with the speech quality prediction result of callee side P-CSCF402.

In addition, in call-side P-CSCF400, the worthwhile 2. process by Fig. 7 of SIP 200Ok message and 3. also can refer to The speech quality prediction result of process, as the speech quality prediction result between call-side P-CSCF400 and called terminal 104 It is stored in cache memory.At this point, when belonging to another calling terminal of call-side P-CSCF400 to identical called end 104 request speech quality prediction of end, then only carry out the 1. process of Fig. 7, utilized by callee side P-CSCF400 and be stored in speed buffering Speech quality prediction result in memory transmits SIP 200Ok responses to calling terminal 102 immediately, to terminate matter of conversing Amount prediction.Such cache memory is applicable to using structure in the full content of the present invention.

The case where Fig. 8, because called terminal 108 possesses public IP itself or by SBC, therefore intermediate be not present executes The node of Media Relay, X-Quality-Test headers below can be transferred to called terminal 108 as former state.

X-Quality-Test:Request acodec=ulaw, vcodec=h263, addr1=1.1.1.1:10,11

Addr1 can be can refer to by the 2. process of Fig. 8, called terminal 108 to execute for speech quality with calling terminal 106 The transmission test of prediction.If there are the SBC of callee side SBC and callee side also like Fig. 8 1. during illustrate as with quilt It makes terminal 108 negotiate to generate the NAT pin holes for being used for transmission experiment session, the address may be used and execute transmission test.Work as quilt It is to participate in transmission test the addr1 fields of above-mentioned X-Quality-Test headers are changed to the address information of oneself to be side SBC It is transferred to called terminal 108, executes transmission test using the address of the SBC by called terminal 108, then from SBC to calling terminal Media Relay are executed as former state participates in transmission test.Speech quality prediction result by the 3. process of Fig. 8 is voice 8.8, shadow As 8.9, the personal IP address for being used for transmission experiment of called terminal 108 is 192.168.1.1, and voice port information is 10, shadow Picture port information is 11, and the public IP of callee side SBC transformation is 50.50.50.50, and voice port information is 20, and video port is believed Breath is 21, including the 200Ok message of following X-Quality-Test headers can be transmitted to callee side SBC from called terminal 108.

X-Quality-Test:Request acodec=ulaw, vcodec=h263, addr1=50.50.50.50: 20,21, CQTY=8.8:8.9, addr2=192.168.1.1:10,11

Callee side SBC can change the add1 fields and addr2 fields of X-Quality-Test headers as follows, from callee side SBC is transmitted to callee side P-CSCF406.

X-Quality-Test:Request acodec=ulaw, vcodec=h263, addr1=1.1.1.1:10, 11, CQTY=8.8:8.9, addr2=50.50.50.50:20,21

200Ok message is transferred to calling terminal 106 as former state later, terminates speech quality and predicts process.

Fig. 9 indicates that execution speech quality is pre- between arbitrary the Internet telephone calls terminal and the telephone called terminal of particular network The process of survey.

More specifically, it is that each terminal possesses public IP in Fig. 9 as the case where Fig. 8, because of peace between communication common carrier Entirely, the reasons such as NAT Traversal, Topology Hiding have between IMS nets and play BorderController works The case where IBCF (Interconnection Border Control Function), the IBCF of each communication common carrier is general Execute Media Relay, Fig. 9 indicate because both sides IBCF by with IBGF (Interconnection Border Gateway Function) or the linkage of TrGW (Transition Gateway) executes Media Relay, the biography for speech quality prediction Defeated experiment section is set as calling terminal --- call-side IBGF1 --- callee side IBGF2 --- called terminal, and by not same district Between execute transmission test process.

In addition, Fig. 9 belongs to because arbitrary network call terminal 110 and the telephone called terminal 112 of particular network are all direct Or the reasons such as public IP are indirectly for, between calling terminal 110 and call-side CSCF 1 or called terminal 112 and called Between side CSCF 2 702 be not present Media Relay the case where, indicate centre because between mobile communication carrier safety or There is the IBCF as a purpose with signal processing in the purpose of NAT Traversal, Topology Hiding etc. (Interconnection Border Control Function) and IBGF with media handling as a purpose When (Interconnection Border Gateway Function), how to be held between calling terminal 110 and called terminal 112 The process of row speech quality prediction.

In addition, Fig. 9 not only in IBCF, is still generated when calling transmission path intermediate demand media transcoding etc. MediaRelay, to IBCF the case where it is similar, execute speech quality predict process.Here, IBGF generally by with IBCF Linkage executes Media Relay, and the X- of the SIP OPTIONS message of call-side IBCF 1 is reached by the 1. process of Fig. 9 Quality-Test headers are as follows.

X-Quality-Test:Request acodec=ulaw, vcodec=h263, addr1=1.1.1.1:10,11

Call-side IBCF 1 by linking with call-side IBGF 1 (address=2.2.2.2) because being executed MediaRelay is executed before SIP OPTIONS message is transmitted to callee side IBCF 2 802 or for speech quality prediction Rapidity, while sending out order to call-side IBGF 1 and executing transmission test.

In the case of before being transmitted, X-Quality-Test headers can be configured from mono- 900 reception results of IBGF and as follows It is transmitted to callee side IBCF 2 802.

X-Quality-Test:Request acodec=ulaw, vcodec=h263, addr1=1.1.1.1:10, 11CQTY=8.8:8.9, addr2=2.2.2.2:20,21

In the case where IBCF 1 plays the role of Topology Hiding, to keep consistency, addr1 is deleted as follows It is transmitted to IBCF 2 802 after information.

X-Quality-Test:Request acodec=ulaw, vcodec=h263, CQTY=8.8:8.9,addr1 =2.2.2.2:20,21

Here, existing addr2 can become addr1.

As described above, when Topology Hiding are performed, SIP OPTIONS message reaches IBCF 2 802, then IBCF 2 802 equally need execute Media Relay when, send out order to callee side IBGF 2 902 (address=3.3.3.3), lead to The 2. process for crossing Fig. 9 executes transmission test between IBGF 1 and IBGF 2 902, remembers as follows from 2 902 reception results of IBGF After recording in X-Quality-Test headers, it is transmitted to callee side CSCF 2 702.

X-Quality-Test:Request acodec=ulaw, vcodec=h263, CQTY=8.8:8.9,addr1 =2.2.2.2:20,21, CQTY=9.0:9.1, addr2=3.3.3.3:30,31

If assuming, IBCF 1 executes media transcoding and (in the case of audio, ulaw is converted to iLBC, in video In the case of, will h.263 be converted to h.261), by the SIP OPTIONS message transmitted to IBCF 2 802 of IBCF 1 Acodec and vcodec fields replace with iLBC and h261 is transmitted, thus with being that call is identical, IBCF 1 and IBCF bis- 802, which can utilize iLBC and h261 to execute speech quality, predicts.Such field change has to keep firmly in mind, and is answered in transmission 200Ok When answering message, transmitted after restoring again.

In addition, audio codecs and video codecs predict calling terminal or AS by request speech quality (Application Server) is set arbitrarily, but such as the case where actual call, in transmission for speech quality prediction It is supported all using other SIP OPTIONS Transaction record calling terminals before SIP OPTIONS message Codec list is transmitted, by called terminal therefrom by the intersection with oneself supported codec list, by right The response of SIP OPTIONS message is transferred to calling terminal, then can therefrom select arbitrary codec to execute by calling terminal logical Talk about prediction of quality.

Speech quality predicts process, for rapidity, can be carried out at the same time transmission test and SIP OPTIONS message always It transmits.At this point, because even if CQTY fields are omitted, execute the server of Media Relay address be centainly included in by It transmits, to be set as the boundary in the section predicted for speech quality.Because the acquiescence timer f of SIP Transaction are 32 seconds, transmission test was advisable for 10 seconds or so, and the transmission test in only different sections is performed simultaneously, could no more than Under conditions of timer f, final 200Ok responses are transferred to calling terminal.Transmission test executive agent remember as a result, and The appropriate location of 200Ok response messages is inserted into CQTY fields and is transmitted.

In the case of 2 702 CSCF, because as assumed above, do not executed Media Relay, above-mentioned X- Quality-Test headers are transferred to called terminal 112 as former state, and called terminal 112 (address=4.4.4.4) is by being contained in X- The boundary as the section predicted for last speech quality in Quality-Test headers addr2 (address= 3.3.3.3, IBGF2) and Fig. 9 3. process execute speech quality prediction, and by result (voice 9.2, image 9.3) as follows record CSCF 2 702 is transferred in X-Quality-Test headers.

X-Quality-Test:Request acodec=ulaw, vcodec=h263, CQTY=8.8:8.9,addr1 =2.2.2.2:20,21, CQTY=9.0:9.1, addr2=3.3.3.3:30,31CQTY=9.2:9.3, addr3= 4.4.4.4:40,41

2 702 original samples of CSCF are transferred to IBCF 2 802, in the case of 2 802 IBCF, if the case where with 1 IBCF Equally play the role of Topology Hiding, then the following is hiding intranet information and is transferred to IBCF mono- after deleting addr3 800。

X-Quality-Test:Request acodec=ulaw, vcodec=h263, CQTY=8.8:8.9,addr1 =2.2.2.2:20,21, CQTY=9.0:9.1, addr2=3.3.3.3:30,31CQTY=9.2:9.3

IBCF 1 is Topology Hiding, and the address of the calling terminal 110 of above-mentioned deletion is restored as follows It is transferred to CSCF 1, and CSCF 1 can carry out original sample transmission to calling terminal 110.

X-Quality-Test:Request acodec=ulaw, vcodec=h263, addr1=1.1.1.1:10, 11, CQTY=8.8:8.9, addr2=2.2.2.2:20,21, CQTY=9.0:9.1, addr3=3.3.3.3:30,31CQTY =9.2:9.3

Calling terminal 110 can refer to the speech quality prediction in variant section, is averaged according to strategy use or minimum is pre- Survey speech quality.

Figure 10 indicates that, when there is such as PSTN analog lines switching network, it is pre- that networking telephone called terminal executes speech quality The process of survey.

More specifically, Figure 10 is the case where called terminal belongs to CS (Circuit Switching) domain, in IMS In the case of net, calling in the case is transferred to BGCF (Breatout Gateway Control Function), calling SIP signal passes through SGW (Signaling from BGCF via MGCF (Media Gateway Control Function) Gateway the media for) being transmitted to CS Domain, and calling are transmitted to CS Domain by MGW (Media Gateway), at this point, Because MGW is the last via ground of PS (Packet Switching) Domain, the section that setting is predicted for speech quality herein Boundary, and the transmission test predicted for speech quality is executed between calling terminal.

When destination is PSTN1400, passed to BGCF (Breakout Gateway Control Function) 1000 SIP OPTIONS message is passed, then MGCF (Media Gateway Control Function) 1100 is selected to be transmitted.Hair The number of delivering letters is responsible for by SGW (Signaling Gateway) 1200, and media are responsible for by MGW (Media Gateway), thus with Circuit-switched network connects.In the case of circuit-switched network, because of the channel for using bandwidth to be physically protected, beyond the present invention Range, need not carry out again speech quality prediction operation.Speech quality is predicted until MGW1300.

Figure 10 belongs to because calling terminal 114 is used directly or indirectly in the reasons such as public IP, in calling terminal 114 and call-side The case where Media Relay are not present between P-CSCF 1, including the OPTIONS of following X-Quality-Test headers disappears Breath is transferred to P-CSCF408 by the 1. process of Figure 10, is transferred to MGCF1100 as former state without alterations later.

X-Quality-Test:Request acodec=ulaw, vcodec=h263, addr1=1.1.1.1:10,11

Execute Media Relay because MGCF110 is connected to MGW1300, so as to by the 2. process of Figure 10 to MGW ( Location=2.2.2.2) 1300 send out order execute speech quality prediction.MGCF803 is as the structure for executing speech quality prediction Obtain voice 8.0, image 9.0 as a result, then extremely by the 200Ok message transmissions comprising following X-Quality-Test headers BGCF1000 is transferred to calling terminal 114 successively as former state, terminates the process for executing speech quality prediction.

X-Quality-Test:Request acodec=ulaw, vcodec=h263, addr1=1.1.1.1:10, 11CQTY=8.0:9.0, addr2=2.2.2.2:20,21

At this point, because not being to complete speech quality prediction until final called terminal, 200Ok message is substituted, at other Information comprising failure place (MGCF) and failure cause (being converted to circuit-switched network) etc. in SIP Error message is passed It is defeated.Only, the content for including above-mentioned X-Quality-Test headers as former state transmits speech quality prediction knot until MGW1300 Fruit.The technical reasons such as circuit switched are not converted to only, and are the over load for the communication network predicted by speech quality, it is specific Region can also be prohibited speech quality prediction because of the policy reason of communication common carrier.

Figure 11 indicates that the SIP AS by external execute the process of speech quality prediction.More specifically, Figure 11 is indicated non- The case where speech quality between such as sip application server request call terminal and called terminal of calling terminal is predicted.

It, can also be by SIP AS in addition to calling terminal 116 is by transmitting SIP OPTIONS message direct requests (Application Server) 1500 etc. ask specific call terminal 116 to arrive specific called terminal by random communication protocol 118 speech quality prediction.If at this point, by taking IMS nets 1600 as an example, the P- belonging to calling terminal 116 can be obtained first The information of CSCF410.In this domain belonging to calling terminal 116, pass through SLF (Subscriber Location Function) It obtains the address of HSS (Home Subscriber Server) and is inquired after the address for the S-CSCF for obtaining call-side to HSS, Obtain the information of P-CSCF410.Equally in the case of wired network, obtain by similar methods belonging to calling networking person SIP Proxy information, SIP Proxy or P-CSCF410 requests of the SIP AS1500 to call-side later arbitrarily exhale The speech quality between terminal 116 and called terminal 118 is made to predict.

When the 1. process by Figure 11, include the speech quality prediction of 118 information of 116 information of calling terminal and called terminal Request reaches call-side P-CSCF410, then by the 2. process of Figure 11, call matter is asked from P-CSCF410 to calling terminal 116 Amount prediction.The same address for using SIP OPTIONS message, To headers and Request URI as calling terminal 116, and X- Quality-Test headers can be configured to following form.

X-Quality-Test:Request acodec=ulaw, vcodec=h263, peer=" sip:abc@ def.com"

Here, Request is speech quality predictions request, acodec is audio coder ＆ decoder (codec), and vcodec is that image compiles solution Code device, peer are the information of called terminal.

When calling terminal 116 obtains speech quality predictions request by the 2. process of Figure 11, then it can pass through the 3. mistake of Figure 11 Journey itself executes in called terminal 118 and between 116 speech quality prediction.

Can the 3. process of according to circumstances Figure 11 be replaced by the process of Fig. 7, Fig. 8, Fig. 9, Figure 10, Figure 12 etc..Pass through Figure 11's 4. process, calling terminal 116 can be transmitted to P-CSCF410 by the 2. process of Figure 11 obtain to speech quality predictions request As a result it is transmitted to P-CSCF410.The X-Quality-Test headers of 200Ok message can following form configuration.

X-Quality-Test:Request acodec=ulaw, vcodec=h263, peer=" sip:abc@ Def.com ", CQTY=8.0:9.0

In upper example, it can confirm that the final structure of only worthwhile different junction call prediction of quality results is passed.In addition, The speech quality prediction result for respectively pressing different sections, may include that the address information on the boundary in section is transmitted together.

Speech quality prediction result between call-side P-CSCF410 reception calling terminals 116 and called terminal 118, Then speech quality prediction result can be transmitted to SIP AS1500 by the 5. process of Figure 11.

Figure 12 is the schematic diagram for indicating to reduce the process of speech quality predicted flow rate, and Figure 13 is more of the invention one real Apply the schematic diagram of the UDP groupings of example.More specifically, Figure 12 illustrates the SBC for executing Media Relay being set as conversing The boundary in the section of prediction of quality carries out the process of speech quality prediction.

So far, real in actual call connection procedure by SBC while being responsible for the NAT Traversal of terminal Border executes Media Relay, and the also practical execution Media Relay in the transmission test predicted for speech quality, but does not have As the interval border of speech quality prediction.If there are multiple SBC and multiple terminals and SBC linkages, SBC is set as conversing Record is simultaneously predicted by the speech quality between cache management SBC in the boundary in prediction of quality section, so that it may to subtract Few speech quality predicted flow rate.

In fig. 12, it is assumed that the personal IP address of calling terminal 120 is 192.168.1.10, passes through SBC changes later Public ip address is 20.20.20.20.

1. Figure 12's is predicted by the request of calling terminal 120 and the speech quality of called terminal 1222 in the process, called terminal 122 information setting is in the To headers of SIP OPTIONS message, and the X-Quality-Test of SIP OPTIONS message Header can following form configuration.

X-Quality-Test:Request acodec=ulaw, vcodec=h263, addr1=192.168.1.10: 10,11

Include the SIP OPTIONS message of above-mentioned X-Quality-Test when call-side SBC 1 is received, then ignores Addr1 fields simultaneously execute calling terminal 120 and itself speech quality between 1700 predicts (voice by the method for Fig. 3 and Fig. 4 8.9, image 8.2), and it is transferred to call-side Proxy 1 to configure X-Quality-Test headers as follows.

X-Quality-Test:Request acodec=ulaw, vcodec=h263, addr1=192.168.1.10: 10,11, CQTY=8.9:8.2, addr2=20.20.20.20:10,11

At this point, call-side Proxy 1 and callee side Proxy 2 1802 arrives callee side because not executing Media Relay Above-mentioned X-Quality-Test headers are transmitted as former state until SBC bis- (address=30.30.30.30) 1702.Because of callee side SBC 2 1702 executes Media Relay and is set as the boundary in the section predicted for speech quality for these reasons, to Execute speech quality prediction, by result (voice 8.3, image 8.4) be recorded as it is following and be transferred to called terminal (address= 192.168.2.10)122。

X-Quality-Test:Request acodec=ulaw, vcodec=h263, addr1=192.168.1.10: 10,11, CQTY=8.9:8.2, addr2=20.20.20.20:10,11CQTY=8.3:8.4addr3=30.30.30.30: 10,11

Callee side SBC 2 1702 is executed with the prediction of the speech quality of called terminal 122 and immediately to the transmission of Proxy 2 1802 200OK message.Called terminal 122 executes the transmission test with callee side SBC 2 1702 by the 3. process of Figure 11, by result (voice 8.5, image 8.6) is recorded as following and by SIP 200Ok messagings to callee side SBC 2 1702.

X-Quality-Test:Request acodec=ulaw, vcodec=h263, addr1=192.168.1.10: 10,11, CQTY=8.9:8.2, addr2=20.20.20.20:10,11CQTY=8.3:8.4addr3=30.30.30.30: 10,11CQTY=8.5:8.6addr4=192.168.2.10:10,11

The 200Ok message is transferred to calling terminal 120 as former state, terminates speech quality and predicts process.

The present invention, according to the strategy of each communication common carrier, can use in the case of actually call connection to improve safety IPSEC (Internet Protocol Security) or SRTP (Secure Real-Time Transport Protocol) Etc. technologies, during speech quality is predicted, also for by different sections with actual call connection procedure phase to the maximum extent It is tested in same environment, IPSEC SA (Security Association) associations can be first carried out between the boundary in section After quotient, the ESP (Encapsulating Security Payload) or AH (Authentication of IPSEC are used Header) mode executes the encryption and certification being grouped to transmission test.

In addition, when using the SRTP specially developed for RTP, to RTB points not comprising practical RTP groupings or physical medium Group can execute the encryption and certification to respective packets after executing key exchange between the boundary in section in a manner of SRTP.

Using the transmission test of UDP groupings, the process generation for generating UDP groupings has and is actually subjected to test The RTP of object codec be grouped the UDP groupings of identical size and transmission intercal.Such as in case of ULAW, because every The interval that the data of 8000 samples are divided into 20ms by the second is transmitted (data transmission of 160 sample * 50 1 second), generates packet Size (acquiescence 12 bytes+basis of arbitrary data and the RTP headers for being contained in RTP groupings containing 160 samples (=160 byte) The header or growth data that the type of codec includes) size dummy data UDP grouping send and receive, one Determine to generate to be grouped in the standard time to be tested.

It even uses ULAW and executed speech quality prediction experiment within 10 seconds time, then the present invention generates 500 points Group, and it is contained in the transmission time information for including current transmission side in the dummy data of each grouping.When other side is transmitted again, then pass Defeated side is calculated flat by the time received to 500 each grouped comparisons and the sending time being contained in the dummy data of grouping Equal RTT (Round Trip Time) acquires the deviation of the quantity of the grouping of omission and the RTT values of each grouping and average RTT values After (irrelevance) value, speech quality is predicted using above-mentioned value.When comprising wireless network, whether checking the integrality of initial data It can also be contained in during speech quality prediction.

In the audio coder ＆ decoder (codec)s such as such as AMR (Adaptive Multi Rate), because being passed using variable according to Network status Defeated rate, UDP groupings are also similar to actual conditions, can be according to Network status application variable bit rate VBR.

In the case of image, most of to improve compression ratio, image information is used by believing comprising the independent of image entirety The I-frame (Intra Frame) of breath, with reference to before I or the P-frame (Predictive that are encoded of P-frame Frame), the B-frameBi-directionally predictive that bi-directional predicted front and back I or P-frame are encoded Frame) GOP (Group Of Pictures) concept constituted, it is of different sizes because of frame, and gop structure itself can also Other forms are configured to, statistical method etc. can be used, UDP groupings can also be adjusted to transmission intercal and packet size and reality The similar situation of situation is transmitted with the state similar with image frame when practical converse to the maximum extent.Practical RTP groupings With the light reference chart of comparison 13 of the UDP groupings for the present invention.

Figure 14 is the schematic diagram for the relationship being described in detail between network telephone terminal and voip server or SBC.More Specifically, Figure 14 is the signal for indicating the relationship between network telephone terminal 124,126 and voip server 200 in detail Figure.

Network telephone terminal 1 belongs to the case where network telephone terminal that such as smart mobile phone is connected by wireless network. Network telephone terminal 1 and BTS (Base Transceiver Station) 1900 or Node B1900 are wirelessly connected, via The BSC (Base Station Controller) 2000 of wireless network is managed by BTS1900 or nothing is managed by Node B1900 The RNC (Radio Network Controller) 2000 of gauze is connected to wired network.

Later, network telephone terminal 1 is to be connected to circuit-switched network, can be connected to MSC (Mobile Switching Center (is omitted) in Figure 14, or to be connected to the packet switching networks such as internet, is endowed by SGSN2100 and GGSN2200 IP address, and the internet by being made of L3Switch and L2Switch, L4Switch etc. for being used as router 2600 is direct Or according to circumstances voip server 200 is connected to via SBC1704.

Voip server 200 plays the SIP Proxy and individual enterprise IP- of the P-CSCF and wired network of IMS nets The server of the effects that PBX, and from being passed through with network telephone terminal 1 in SIP nets from the viewpoint of Session Initiation Protocol First SIP object that SIP engages in the dialogue.

Network telephone terminal 1 by GGSN2200 can be endowed personal IP because of reasons such as the strategies of communication common carrier, it Public IP is transformed to by NAT device afterwards and is connected to voip server 200 via internet, at this point, to solve NAT Traversal problems can be connected to voip server 200 via SBC1704.In addition, in calling terminal and called terminal Between when generating calling, media are not using SBC1704's generally via SBC there is NAT Traversal yet yet In the case of, it, can be by 200 relay media of voip server when generating calling.

Network telephone terminal 2 126 is connect via having wireless router 2300 and DSL modem 2400 by xDSL etc. Networking (Access Network) is connected to wired network.DSLAM(Digital Subscriber Line Access Multiplexer it) 2500 is set to communication common carrier and is connect with DSL modem 2400, providing internet to networking person connects Enter service.Later as network telephone terminal 1 the case where it is identical, by by be used as router 2600 L3Switch and The internet of the compositions such as L2Switch, L4Switch is connected to voip server directly or according to situation via SBC1704 200。

According to the strategy of communication common carrier, communication common carrier can by DSL modem 2400 to networking person provide it is personal or Public IP, even if in the case where obtaining public IP, if using having wireless router, network telephone terminal 2 126 that will be assigned Give personal IP.At this point, as the case where network telephone terminal 1, to solve the problems, such as NAT Traversal, via SBC1704 is connected to voip server 200 or by 200 relay media of voip server.

In the present invention, arbitrary network telephone terminal 124,126 is predicted before calling generates for reaction the above situation Speech quality state, between voip server 200 and network telephone terminal 124,126 or SBC1704 and the networking telephone Speech quality is predicted in section between terminal 124,126 by transmission test, using whole as the networking telephone in Idle states The benchmark of the speech quality state at end 124,126.In the state of i.e. before determining conversation object, even if connecting with any other side It connects, speech quality prediction is carried out to the media transmission path of common longest distance.Because of the reasons such as such as IPv6 environment, wireless Public IP is obtained by network telephone terminal 1 by GGSN2200 in the case of net, or passes through DSL in the case of wired network Modem2400 obtains public IP by network telephone terminal 2 126, when actually conversing, because media are not via SBC1704 or network Telephony server 200 is transferred directly to distant terminal, lead to the problem of with the present invention speech quality forecast interval it is Chong Die, when compared with The case where being measured by the other speech quality measuring apparatus of non-network telephony server 200, have can be preferably The advantages of state of reaction network telephony server 200.

Figure 15 to illustrate the invention an embodiment prediction other side speech quality process precedence diagram.

In fig.15, the method for predicting speech quality can be executed in a side of first and second network communication device, prediction The speech quality of other side.

SIP transactions requests portion 21 is optionally included with the address letter of the one side of the speech quality prediction for a side Breath (IP address, No. Port), SIP (the Session Initiation required to other side's transmission comprising speech quality prediction Protocol) (S1501 steps).

SIP affairs responses receiving part 22 receives the address information (IP comprising other side from other side in response to SIP transactions requests Location, No. Port) SIP affairs response (S1502 steps).

Speech quality prediction section 23 by the address information of a side and the address information of other side, by using 1) with practical matchmaker The RTP groupings unrelated whether including of body；2) RTCP is grouped；Or 3) the disguise oneself as camouflage UDP grouping of RTP of other side (is called in the following text and " passed Defeated experiment grouping ") communication execute with the transmission test of other side to predict speech quality (S1503 steps).

SIP affairs acknowledgement transmissions portion 24 after (a) step, other side using the address of a side be first carried out (c) step with After obtaining speech quality prediction result, result is contained among SIP affairs responses and is transferred to a side (S1504 steps).

Call selection determination section can save pre- based on the packet loss rate and propagation delay time being grouped to N number of transmission test The prediction speech quality of survey, to determine that the call to other side selects (S1505 steps).

In figure 16, the method for predicting speech quality can execute in voip server or SBC, without calling life At course prediction to the speech quality of multiple network telephone terminals.

Section of the speech quality prediction section 23 between network telephone server or SBC and network telephone terminal, passes through profit It is grouped with the RTP unrelated whether including 1) with physical medium；2) RTCP is grouped；Or 3) the puppet for the RTP that disguises oneself as to above-mentioned other side Fill UDP grouping (calling in the following text " transmission test grouping ") communication execute and above-mentioned multiple respective transmission tests of network telephone terminal with Predict speech quality (S1601 steps).

Call selection determination section 25 provides the pre- of speech quality based on the prediction to respective speech quality, to partner It surveys result or executes to the information offer of the network telephone terminal that can be used with classic speech quality or the company of execution Connect trial (S1602 steps).

Above-described embodiment is only to illustrate the present invention and unrestricted, it will be understood by those of ordinary skill in the art that, it can be with It modifies, deform or equivalent replacement should all cover in this hair without departing from the spirit and scope of the present invention to the present invention In bright right.

Reference sign

10：Speech quality forecasting system 20：Speech quality predicts service unit

21：SIP transactions requests portion 22：SIP affairs response receiving parts

23：Speech quality prediction section 24：SIP affairs acknowledgement transmissions portion

25：Call selection determination section 26：Control unit

30、40：IP Header 32、42：UDP Header

34：RTP Header 36：RTP Data

44：The transmission time 46 being contained in UDP groupings：Dummy Data

100、102、104、106、108、110、112、114、116、118、120、122、124、126：Network telephone terminal

200：Voip server 300：Speech quality predictive server

400、402、404、406、408、410：P-CSCF(Proxy CSCF)

500、502、504、506、508：S-CSCF(Serving CSCF)

600、602：I-CSCF(Interrogating CSCF)

700、702：CSCF(Call Session Control Function)

800、802：IBCF(Interconnection Border Control Function)

900、902：IBGF(Interconnection Border Gateway Function)

1000：BGCF(Border Gateway Control Function)

1100：MGCF(Media Gateway Control Function)

1200：SGW(Signaling Gateway) 1300：MGW(Media Gateway)

1400：PSTN(Public Switched Telephone Network)

1500：SIP AS(Application Server)

1600：IMS (IP Multimedia Subsystem) net

1700、1702、1704：SBC(Session Border Controller)

1800、1802：SIP Proxy

1900：BTS (Base Transceiver Station) or Node B

2000：BSC (Base Station Controller) or RNC (Radio Network Controller)

2100：SGSN(Serving GPRS Support Node)

2200：GGSN(Gateway GPRS Support Node)

2300：Wireless Router (have wireless router)

2400：DSL(Digital Subscriber Line)Modem

2500：DSLAM(Digital Subscriber Line Access Multiplexer)

2600：Router 2700：Backbone network

Claims

1. a kind of method of prediction speech quality, executes in a side of first and second network communication device, predict that other side's is logical In the method for talking about quality, including：

(a) be optionally included with for one side speech quality prediction one side address information (IP address, No. Port), SIP (the Session Initiation Protocol) affairs required to other side's transmission comprising speech quality prediction The step of request；

(b) address information (IP address, the Port that include above-mentioned other side are received from above-mentioned other side in response to above-mentioned SIP transactions requests Number) SIP affairs responses the step of；

(c) by the address information of the address information of one side and above-mentioned other side, by using 1) including with physical medium Whether unrelated RTP groupings；2) RTCP is grouped；Or 3) (in the following text " transmission tries is called to the disguise oneself as camouflage UDP grouping of RTP of above-mentioned other side Test grouping ") communication the step of executing with the transmission test of above-mentioned other side to predict speech quality；And

(d) or after (a) step, (c) step is first carried out to obtain call matter using the address of one side in above-mentioned other side After measuring prediction result, result is contained in the step of being transferred to one side among above-mentioned SIP affairs response.

2. the method for prediction speech quality according to claim 1, it is characterised in that：Above-mentioned (c) step is included in above-mentioned When one side is connected with above-mentioned other side by NAT device, by using bipartite address above mentioned information via above-mentioned NAT device Bipartite communication, the step of NAT pin holes (Pinhole) for address above mentioned information is generated in advance.

3. the method for prediction speech quality according to claim 1, it is characterised in that：Above-mentioned (c) step includes to above-mentioned One side or other side send N number of (N is more than 2 natural number) above-mentioned transmission test grouping successively, and confirm respective RTT (Round Trip Time) and whether to respective loss to predict above-mentioned speech quality the step of.

4. the method for prediction speech quality according to claim 1, it is characterised in that：Further include that (e) is preserved based on to upper The packet loss rate of N number of transmission test grouping and the prediction speech quality of propagation delay time prediction are stated, to determine to above-mentioned right The step of call selection of side.

5. the method for prediction speech quality according to claim 4, it is characterised in that：Above-mentioned (e) step is included in above-mentioned When other side is made of multiple network telephone terminals, based on to above-mentioned multiple respective prediction speech qualities of network telephone terminal, choosing Select the step of particular network telephone terminal is conversed.

6. the method for prediction speech quality according to claim 5, it is characterised in that：Above-mentioned (e) step further includes when upper When stating the communication between a side and above-mentioned other side and belonging to voice communication, it can will currently use and there is minimum propagation time delay Network telephone terminal the step of being selected as above-mentioned network telephone terminal.

7. the method for prediction speech quality according to claim 6, it is characterised in that：Above-mentioned (e) step further includes when upper When stating the communication between a side and above-mentioned other side and belonging to data communication, can will currently it use and with minimum packet loss rate Network telephone terminal is selected as the step of above-mentioned network telephone terminal.

8. the method for prediction speech quality according to claim 1, it is characterised in that：Further include (e) above-mentioned first and When there is at least one media relays device for executing media relays in the relay between the second network communication device, upper It states in first and second network communication device and above-mentioned at least one media relays device (calling in the following text " speech quality prediction meanss "), To the relaying section being formed between adjacent speech quality prediction meanss, by executing above-mentioned (a) to (d) step or individually holding Row above-mentioned (c) step, is sequentially completed the step of above-mentioned speech quality is predicted.

9. the method for prediction speech quality according to claim 8, it is characterised in that：Above-mentioned (e) step includes will be to upper The speech quality prediction for stating relaying section, which is stored in cache memory and updates above-mentioned high speed when more than specific time, delays The step of rushing memory.

10. the method for prediction speech quality according to claim 9, it is characterised in that：Above-mentioned (e) step further include when pair When the speech quality prediction of failure in given trunk section, receive comprising relevant pre- about speech quality with above-mentioned the reason of failing The step of surveying the SIP affairs responses of the information of device.

11. a kind of method of prediction speech quality, executes in voip server or SBC, pre- without calling generating process It surveys in the method for the speech quality of multiple network telephone terminals, including：

(a) section between network telephone server or SBC and network telephone terminal, by using 1) with the packet of physical medium The unrelated RTP groupings containing whether；2) RTCP is grouped；Or 3) (in the following text " transmission is called to the disguise oneself as camouflage UDP grouping of RTP of above-mentioned other side Experiment grouping ") communication execute with above-mentioned multiple respective transmission tests of network telephone terminal to predict speech quality the step of； And

(b) based on the prediction to above-mentioned respective speech quality, to partner provide above-mentioned speech quality prediction result or It executes and the information of the network telephone terminal that can be used with classic speech quality is provided or executed what connection was attempted Step.

12. a kind of speech quality predicts service unit, executed executing the side in first and second network communication device, prediction In the speech quality prediction service unit of the method for the speech quality of other side, including：

SIP transactions requests portion is optionally included with the address letter of the one side of the speech quality prediction for one side Breath (IP address, No. Port), SIP (the Session Initiation required to other side's transmission comprising speech quality prediction Protocol) transactions requests；

SIP affairs response receiving parts receive the address comprising above-mentioned other side from above-mentioned other side in response to above-mentioned SIP transactions requests and believe The SIP affairs responses of breath (IP address, No. Port)；

Speech quality prediction section, by the address information of the address information and above-mentioned other side of one side, by using 1) and real The RTP groupings unrelated whether including of border media；2) RTCP is grouped；Or 3) the camouflage UDP groupings for the RTP that disguises oneself as to above-mentioned other side The communication of (calling in the following text " transmission test grouping ") executes the transmission test with above-mentioned other side to predict speech quality；

SIP affairs acknowledgement transmissions portion, or after transmitting above-mentioned SIP affairs to other side, above-mentioned other side utilizes the ground of one side After the above-mentioned speech quality of prediction is first carried out to obtain speech quality prediction result in location, result is contained in above-mentioned SIP affairs One side is transferred among response；And

Call selection determination section preserves pre- based on the packet loss rate and propagation delay time being grouped to above-mentioned N number of transmission test The prediction speech quality of survey, to determine that the call to above-mentioned other side selects.