WO2003055147A1

WO2003055147A1 - Evaluation of complex communication paths

Info

Publication number: WO2003055147A1
Application number: PCT/IL2001/001142
Authority: WO
Inventors: Yaron Cahana; Hanoch Levy; Joseph Levy
Original assignee: Comgates, Ltd.
Priority date: 2001-12-11
Filing date: 2001-12-11
Publication date: 2003-07-03
Also published as: AU2002222470A1

Abstract

A method of calculating a quality measure of a communication path. The method includes transmitting packets on at least two segments of the communication path, determining a value of at least one quality parameter for the at least two segments of the communication path responsive to the transmitted packets, and calculating a value of the quality measure of the path based on the determined parameter values of the at least two segments.

Description

EVALUATION OF COMPLEX COMMUNICATION PATHS FIELD OF THE INVENTION

The present invention relates to transmission networks and particularly to methods of assessing paths within a network.

BACKGROUND OF THE INVENTION

Packet based networks, used for example for computer communications, generally include many routers connected in a complicated mesh, such that between many source and destination computers there are a plurality of parallel paths in the network. In order to provide high quality of service (QoS), for example for voice over LP (VoIP) applications, it is required to have real time information on the quality of the paths. There are, however, different parameters used for rating connections. These parameters include, for example, cost, delay, and jitter.

The ITU-T G.107 recommendation, "The E-model, a computational model for use in transmission planning", "TIPHON Release 3; Technology Compliance Specification; Part 5: Quality of Service (QoS) measurement methodologies." ETSI TS 101 329-5 vl.1.1 (2000-11), and A. D. Clark, "Modeling the Effects of Burst Packet Loss and Recency on Subjective Voice Quality", Columbia University 2001 IP telephony workshop, http://www.telchemy.com/references/tech_papers/iptel2001.pdf, the disclosures of which documents are incorporated herein by reference, describe a complex parameter (designated R) to be used in evaluating the quality of VoIP sessions over communication paths. Generally, determining the R value of a path includes measuring various parameters of the path, including an average delay (d) of the path, an inter arrival time (iat) for each of the packets transmitted on the path, packet loss statistics and one or more CODEC parameters. A delay impairment factor (Lj) and an equipment impairment factor (I_e) are then determined from the measurements. The ETSI document suggests calculating the impairment factor (I_e) as a function of a packet loss factor I_e(packet loss), a packet delay variation (also known as jitter) factor I_e(pdv) and a CODEC related factor

I_e(CODEC), which are determined from the measured parameters. The packet loss factor

I_e(packet loss) is determined from the packet loss statistics by fitting the statistics in a

Markov model which differentiates between burst and gap conditions. R is then calculated as a function of the delay and impairment factors, i.e., R= f {I_d,I_e(pdv),I_e(packet loss), I_e(CODEC)} .

Measuring the parameters required for calculating R requires a certain amount of resources, including transmission of test packets on the path. When the number of paths in a network, for which the measurements need to be performed, is very large, for example in graph structure networks, the measurements required for the calculation of R for all the paths may require substantial resources which could otherwise be used in handling the transmission of data.

In such cases, a simpler measure is generally used, for example taking into consideration only one parameter (e.g., jitter, delay). Alternatively or additionally, only a limited number of paths in the network are evaluated.

Some routing methods evaluate a path by receiving from each of the routers along the path one or more parameters of a segment of the path adjacent the router. The one or more parameters of the path provided by the routers are generally preconfigured or are determined by viewing the internal load on the router, for example according to the utilization of its buffers.

SUMMARY OF THE INVENTION

An aspect of some embodiments of the present invention relates to a method of evaluating a path of a communication network, or a group of paths passing through common points. The method includes actively measuring one or more quality parameters for at least two partial segments of the evaluated path. A quality measure for the path is then calculated based on the measured values of the one or more quality parameters in the at least two segments forming the path. Each of the segments may include a single link connecting two adjacent routers or a plurality of links that connect end-points of the segment through a plurality of routers. In some embodiments of the invention, a plurality of quality parameters are measured for each of the segments.

Optionally, measuring the one or more quality parameters of the path, segments includes out of band testing, i.e., transmitting test packets along the segments and determining the quality parameters by evaluating the arriving test packets. Alternatively or additionally, the quality parameters are determined using inband testing, i.e., from data packets transmitted along the segments. In some embodiments of the invention, one or more test fields are added to the data packets from which the quality parameters are determined. Optionally, the one or more added fields include a time stamp, for example added the IP header of the packets.

In some embodiments of the invention, the test or data packets used in measuring the quality parameters are transmitted using a network protocol layer, such as the Internet protocol (IP), so that the test packets may be easily routed to specific destination units. For example, when an evaluated segment includes a plurality of routers, using a network protocol layer in the test packets allows simple routing of the test packets to the other end of the segment. In an exemplary embodiment of the invention, the transmitted test packets use a transport protocol layer, such as UDP and/or RTP. In some embodiments of the invention, the test packets include an application layer portion in accordance with a known or proprietary application layer protocol. Optionally, the application layer portion of the test packets is similar to the application layer portions of data packets transmitted in the network, particularly of packets transmitted based on routing decisions which use the measurements of the test packets.

In some embodiments of the invention, in determining the quality measure for the entire path, the effect of the routers connecting the segments is taken into account. Alternatively, the effect of the routers is considered negligible and is ignored.

Optionally, values of the quality measure are determined for a plurality of paths in the network based on a single set of measurements for segments of the network. Measurements of the one or more quality parameters are determined for a plurality of segments of the network. A plurality of paths are constructed from the segments for which measurements were acquired and for each path (or for each group of paths passing through same points), the values of the parameters of the segments forming the path are used in calculating the quality measure for the path.

In some embodiments of the invention, the evaluation of the paths is used for routing, for example for selecting a route with at least a predetermined quality of service (QoS) and/or to find a best route between two points. Alternatively or additionally, the evaluation is used for other purposes, for example for billing and/or quality assurance.

There is therefore provided in accordance with an embodiment of the present invention, a method of calculating a quality measure of a communication path, comprising transmitting packets on at least two segments of the communication path, deteimining a value of at least one quality parameter for the at least two segments of the communication path responsive to the transmitted packets, and calculating a value of the quality measure of the path based on the determined parameter values of the at least two segments.

Optionally, the quality measure comprises a measure for evaluating voice over IP connections, such as the R parameter of the E model. Optionally, determining the value for the at least one parameter comprises providing values for the same parameter for each of the at least two segments. Optionally, determining a value of at least one quality parameter comprises providing values for a packet delay variation factor, a delay impairment factor and/or a CODEC related factor.

Optionally, calculating the value of the quality measure of the path comprises calculating a function of the values of at least one of the quality parameters of the at least two segments. Optionally, calculating the value of the quality measure of the path comprises calculating a linear combination of the values of at least one of the quality parameters of the at least two segments. Optionally, calculating the linear combination of the values comprises calculating a linear combination in which the values of the different segments have non-equal coefficients. Optionally, calculating the linear combination comprises calculating the linear combination with coefficients assigned according to the relative values of the parameters of the different segments.

Optionally, calculating the linear combination comprises calculating the linear combination with coefficients assigned to the minimum and maximum values of the values of the parameters. Optionally, calculating the linear combination comprises calculating using coefficients determined responsive to packet transmission simulations. Optionally, calculating the value of the quality measure of the path comprises deriving, for at least one of the segments, an average delay of the segment responsive to an impairment delay factor of the segment. Optionally, deriving, for at least one of the segments, an average delay of the segment comprises deriving the average delay from an impairment delay factor derived from the average delay through a non-reversible function. Optionally, calculating the value of the quality measure of the path comprises calculating a packet loss factor responsive to packet loss probabilities of the at least two segments. Optionally, calculating the packet loss factor comprises calculating according to a loss bursty model.

Optionally, calculating according to the loss bursty model comprises calculating according to the loss bursty model of the TIPHON document. Optionally, one or more of the at least two segments comprises a multi-link segment.

Optionally, transmitting packets comprises transmitting test packets. Alternatively or additionally, transmitting packets comprises transmitting data packets. Optionally, the method includes altering at least one of the data packets before transmitting the packet.

Optionally, transmitting packets comprises transmitting using a network layer protocol, such as the IP protocol. Optionally, transmitting packets comprises transmitting using a transport layer protocol and/or an application layer protocol. Optionally, transmitting packets comprises transmitting packets which carry time stamps used in determining the value of at least one quality parameter. Optionally, the method includes calculating the value of the quality measure for a plurality of paths in a network. Optionally, the method includes selecting a route for data packets to be transmitted through the network, responsive to calculating the value of the quality measure. Optionally, at least two of the plurality of paths include at least one common segment. Optionally, calculating the value of the quality measure of the path comprises calculating a complex quality measure which depends on values of a plurality of quality parameters.

There is therefore provided in accordance with an embodiment of the present invention, a method of calculating a voice quality measure of a communication path, comprising providing values of a plurality of quality parameters for at least two segments of the communication path, and calculating a value of the voice quality measure of the path based on the provided parameter values of the at least two segments. BRIEF DESCRIPTION OF THE DRAWINGS

Exemplary non-limiting embodiments of the invention will be described with reference to the following description of the embodiments, in conjunction with the figures. Identical structures, elements or parts which appear in more than one figure are preferably labeled with a same or similar number in all the figures in which they appear, and in which:

Fig. 1 is a schematic illustration of a communication network, in accordance with an embodiment of the present invention;

Fig. 2 is a schematic illustration of a network segment for which a complex quality measure is calculated, in accordance with an embodiment of the present invention; and

Fig. 3 is a Markov diagram used in modeling the packet loss on a network segment, as is known in the art, for example in annex E of the ETSI document described above.

DETAILED DESCRIPTION OF EXEMPLARY EMBODIMENTS

Fig. 1 is a schematic illustration of a communication network 100, in accordance with an embodiment of the present invention. A plurality of VoIP telephones 102 (and/or other types of telephones), media gateways 103 and/or computers 104 are connected to each other through communication network 100. Network 100 optionally comprises a mesh of links 106 and routers 108. In Fig. 1, some of the portions of network 100 are shown as segments 112, which may include a single link or a plurality of links and internal routers. In evaluating paths within network 100, the particulars of segments 112 are considered irrelevant and segments 112 are tested as a single link. Segments 112 may include, for example, multi-link segments, which connect only two extreme points, subnetworks which may be evaluated as a single element and/or any other network segment which may be evaluated as a single link at least for the evaluated path.

In some embodiments of the invention, some or all of routers 108 include a data gathering unit (DGU) 114, implemented in software and or hardware, which collects data used to evaluate the quality of the paths in network 100, as described hereiribelow. The routers in segments 112 optionally do not include DGUs 114 and do not participate in evaluating segments of network 100. Alternatively or additionally, one or more routers in segments 112 include DGUs 114 which are used for evaluating some segments they are connected to but are not used for evaluating other segments to which they are connected. Further alternatively or additionally, one or more data gathering units 114 are mounted on other communication elements, such as switches, proxies and/or firewalls. In some embodiments of the invention, one or more data gathering units 114 are stand alone units located near respective routers. The DGUs 114 are optionally preconfigured with the IP addresses of the neighboring DGUs 114 with which they are to communicate. Alternatively or additionally, a distributed method is used to discover the neighboring DGUs 114 and optionally a number of hops to each of the neighboring DGUs 114.

Data gathering units 114 are optionally associated also with at least some of the end apparatus, e.g., telephones 102, media gateways 103 and/or computers 104, so as to allow evaluation of the path between the end apparatus and the neighboring DGUs 114. Alternatively or additionally, DGUs 114 are associated with, or hosted by, edge routers, which are directly connected to the end apparatus.

It is noted that in the prior art data gathering units are generally associated only with end apparatus and are not located within, or near, routers 108, as measurements of quality parameters are generally performed between end apparatus. hi some embodiments of the invention, data gathering units 114 periodically perform one or more tests to determine values for one or more quality parameters for adjacent segments. Data gathering units 114 optionally perform the tests by cooperating with another data gathering unit 114 on an opposite end of a tested link 106 or network segment 112. One or more of the data gathering units 114 performing the tests optionally store the results of the tests. In some embodiments of the invention, the data gathering units 114 also determine one or more complex quality parameters, which depends on a plurality of different quality parameters, based on the test results. Alternatively or additionally, one or more of the data gathering units 114 determines values of one or more quality measures for one or more paths in the network. Further alternatively or additionally, the DGUs 114 transmit the results to one or more central calculation units 116, which calculate quality measures for paths in network 100. The evaluated paths may include end to end paths which connect end apparatus through the network and/or partial paths which connect two points of interest within network 100. In some embodiments of the invention, DGUs 114 and/or central calculation unit 116 also perform end to end measurement of quality parameters of one or more network paths, as is known in the art, in addition to performing measurements for partial segments of one or more same or other paths and calculating quality measures for these paths accordingly. In some embodiments of the invention, the one or more quality measures comprise a voice over IP (VoIP) related quality measure, such as the R factor of the E- model and/or any other quality measure, such as used for routing and/or quality of service purposes.

In some embodiments of the invention, the tests include transmitting test packets between data gathering units at opposite ends of a tested segment 112 or link 106. Optionally, the test packets are transmitted using a network protocol layer, e.g., the Internet protocol (IP), such that the test packets are easily routed through a segment 112 including one or more routers, hi some embodiments of the invention, the test packets use a transport protocol layer, e.g., UDP and/or RTP, to allow simple delivery to data gathering unit 114 at the opposite end of the tested segment 112. In other embodiments of the invention, the test packets comprise ICMP echo request packets or similar control packets, thus using packets of existent protocols and/or packets which are used for tasks other than performing the tests, for performing the tests. Optionally, data gathering units 114 include application layer software which does not require special control of the communication ports of a router on which it is installed. Thus, data gathering units 114 may be installed on various types of routers without requiring adaptations to each router type. Optionally, the test packets carry a time stamp in their application layer portion. The time stamp may be used to follow the transmission time, delay, and/or other parameters of the transmitted packets. Alternatively or additionally, the test packets carry a time stamp in their IP header, as is performed by some existent network elements. Further alternatively, the test packets do not carry a time stamp, so that they may be shorter in size.

In some embodiments of the invention, the tests are performed every few milliseconds, seconds, minutes, hours, or days, as required. The test times are optionally selected as a compromise between the desire to collect up to date data and the need not to overload the network due to the tests. Optionally, the tests are performed in all the segments at substantially the same time. Alternatively, the tests are performed at different times throughout the network, for example at random times.

Alternatively or additionally to transmitting test packets, in some embodiments of the invention, the determination of the one or more quality parameters is performed using data packets transmitted in the network without relation to the tests. Optionally, the DGUs 114 add one or more test fields, such as time stamp fields, to the data packets used in determining the one or more quality parameters. The test fields are added to the LP header, a UDP header, an RTP header, as a tail of the packet and/or in any other convenient point within the data packets. Alternatively or additionally, the DGUs 114 do not alter the data packets so as not to interrupt the transmission of data through the network. In some embodiments of the invention, all the data transmitted through the network is used for evaluating the quality parameters. Alternatively, a predetermined amount of packets or a predetermined percentage of the packets are used. Further alternatively or additionally, only certain types of packets are used by the DGUs 114 to determine the quality measures. In some embodiments of the invention, at some times test packets are used to measure the one or more quality parameters of the segments, while at other times and/or for other segments, altered or unaltered, data packets are used to measure the one or more quality parameters of the segments. For example, a thorough measurement session using test packets may be performed at start up and/or periodically (e.g., every few minutes, hours), and adjustment measurements using data packets are performed at a higher rate to update those measurements.

In some embodiments of the invention, central calculation unit 116 receives the test results and calculates the value of one or more quality parameters for end to end paths of the network, based on the test results. The values of the quality parameters are optionally used to select a communication path for connecting telephones 102, to determine billing amounts and/or to provide quality assurance measures, h some embodiments of the invention, in addition to or instead of central calculation unit 116, local path calculations are performed at a plurality of DGUs 114. hi an exemplary embodiment of the invention, each DGU 114 along a path in the network calculates a partial value of quality parameters of the path up to the current DGU 114. The partial value is then forwarded to an adjacent DGU 114 further in the path, which calculates, based on the partial values, further partial values which take into account one or more additional network segments.

Optionally, values of one or more quality parameters for routers 108 are also determined. Optionally, the values of the quality parameters are determined based on known characteristics of the routers and or based on the internal state of the router (e.g., the load on its buffers). Alternatively or additionally, predetermined values of the quality parameters are configured for the routers. Further alternatively or additionally, tests are performed on routers 108 to determine values for the quality parameters. The tests are optionally performed by transmitting signals through a network segment including the tested router and subtracting the effects of the portions of the segment other than the tested router. Alternatively or additionally, the tests of routers 108 are performed between the external interfaces of the routers.

Using the above described method, a large number of end-to-end paths in network 100 are evaluated using a limited number of measurements performed on segments 112 forming the end-to-end paths.

The number of DGUs 114 in network 100, which participate in measuring the quality parameters, may vary. For example, a first implementation of a network evaluation system may include installing data gathering units 114 at a limited number of points in the network. The measurements will then be performed between each two data gathering units 114In some embodiments of the invention, two neighboring data gathering units 114 test a plurality of different paths between them by fransmitting fixed route packets along different paths connecting the neighboring DGUs 114. The fixed route packets may include, for example, packets which state the route they must pass in their IP header. Alternatively, the test packets transmitted between the two neighboring data gathering units 114 are not limited in the route they may pass and therefore the tests do not differentiate between parallel routes, i.e., non-identical routes which connect the same end points, connecting the two neighboring DGUs 114. Further alternatively or additionally, one or more of DGUs 114 have a plurality of interfaces and the interfaces used for transmitting and/or receiving the test packets define the route on which the test packets are to pass. For example, each interface of a DGU 114 may have a different IP address and the source and/or destination IP address of the test packets defines the IP address to be used.

At any later time, the management of network 100 may install data gathering units 114 in additional routers of the network, to enhance the routing decisions. Installing additional data gathering units 114 may in some cases also reduce the number of test packets passing on the links of the network.

In some embodiments of the invention, some of the data gathering units 114 do not participate in the evaluation of links at all times. For example, during work hours all the data gathering units 114 may be employed, while during the night only a limited group of data gathering units 114 participate in evaluating the paths. Thus, more detailed results, e.g., for more segments, are provided during work hours so as to allow for more accurate routing. Alternatively, fewer DGUs 114 are used when the network is loaded, so as to interfere less with the transmission of data.

Fig. 2 is a schematic illustration of a network segment 200C for which a complex quality measure is calculated, in accordance with an embodiment of the present invention. Routers 108A and 108B are connected through a router 108C over segments 112A and 112B, which connect router 108A to router 108C and router 108C to 108B, respectively. Segments 112A and 112B may include a single link 106 or may include a path formed of a plurality of links.

Following is an exemplary calculation of the complex quality measure R of the E model for segment 200C based on test values received for segments 112A and 112B. The calculation for segment 200C may be performed for itself, i.e., to evaluate a path consisting of segment 200C, or as part of an evaluation of a network path including segment 200C and one or more other segments. It will be obvious for those skilled in the art to calculate the R measure for paths of any number of segments, according to the lines of the following description. Furthermore, the principles of the present invention may be used for complex quality measures other than the R measure.

In some embodiments of the invention, central calculation unit 116 receives, for each of segments 112A and 112B, values for the delay impairment factor Ifi A and 1% B , the packet delay variation factor / (pdv) and I_e (pdv) and the CODEC related factor I_e

(Codec) and I_e (Codec), h addition, a delay d_r of router 108C is received or estimated (d_r may also be neglected). Central calculation unit 116 optionally also receives, for each of segments 112A and 112B, stationary probabilities of states 1 and 3 of the known packet loss Markov model (see Fig. 3 below), denoted herein as ^ f ,^ ,π_\ and πf (in the art they are sometimes denoted using the letter p). hi addition, packet loss probabilities for segments 112A and 112B, i.e., Pff A ,Pι ft , respectively, are received.

Optionally, the average delays of segments 112A and 112B

are calculated from the received delay impairment factors I A and 1^ ft , respectively. For values of I_d greater than 0, the average delay d is optionally calculated based on the following known equation:

I_d = 25{1 + X⁶ l⁶ - 3(1 + (— )⁶)V⁶ + 2}

in which X=log2(d/100). Optionally, the equation is solved for a given Lj value using a numerical inversion method and/or any other numerical method known in the art.

As is known in the art, when the average delay is between 0 and 100 milliseconds, the delay impairment factor is set to 0, and therefore j is a non-reversible function.

Therefore, when the received delay impairment factor ( j) equals zero, d is optionally set to a predetermined value, which is considered to most likely represent the delay d represented by the received value of the delay impairment factor Lj. For example, the average delay value which minimizes the estimation error for I_d=0, e.g., 50, may be used.

Alternatively or additionally, when Ij A and/or 1^ ft is equal zero, measurement of the average delay dC of the entire segment 200C is performed in addition to the measurements received for segments 112A and 112B. Further alternatively or additionally, central calculation unit 116, stores for some or all of the segments of network 100, average delay values d which are used in estimating the delay of the segments when Lj of the segment is zero. Further alternatively or additionally, central calculation unit 116 is provided with the average delays d of some or all of the segments, such that their determination from the delay impairment factor

is not required. In this alternative, central calculation unit 116 optionally does not receive or utilize values of the delay impairment factor Lj, for segments for which average delay values d are received.

Based on the average delay values

of segments 112A and 112B, the average delay dC of segment 200C is calculated, optionally as the sum of the delays

d^B) of the segments 112A and 112B and of the delay d_r of router 108C, i.e., d^c= d^A+ d^B+d_r. The delay impairment factor of segment 200C is then calculated from the average delay dC according to the known equation:

0 d^{C '} ≤ 100ms ή 25{1 + X^βγl^β - 3(1 + (— )⁶)¹/⁶ + 2} d^C > lOOms

where

Based on the received values of Pi ,P , ^-₃ ,^-₃ ,π_\ and of , the values of the combined burst loss density

and the combined gap loss density (DgC) of segment

200C are calculated according to the following known equations (from the above mentioned TIPHON document):

where both D^C and D C _are probabilities expressed as percentages.

An exemplary derivation of these equations is described hereinbelow with reference to Fig. 3. It is noted, however, that the use of these equations does not depend on the accuracy of the assumptions in the below derivation. In addition, other equations may be used, for example depending on different received parameters, e.g.,

^₂ and /or ^ , for example using the known probabilistic equality π + Pχ + ₃ = 1.

The packet loss factor I_eC(packet loss) of segment 200C is then optionally calculated from D_DC and D C. as is known in the art.

In some embodiments of the invention, the packet delay variation factor I_eC(pdv) of segment 200C is estimated based on a linear combination of the packet delay variation factors I_e-A-(pdv) and I_eB(pdv) of segments 112A and 112B. Optionally, the linear combination is of the minimum and maximum values of the packet delay variation factors

I_eA(pdv) and I_e^(pdv). Optionally, the coefficients of the linear combination are determined based on simulations in accordance with various traffic flow scenarios. In an exemplary embodiment of the invention, in accordance with simulations performed by the inventor, the linear combination is of the form:

I_e (pdv) = /b +A

+j0₂ m_aχ{l (pdv),I*(pdv)} in which βo^Φ, β 1=0.7 and β2=0.9. Alternatively, more accurate values may be used based on the below described simulations and/or based on other simulations or tests.

Alternatively, central calculation unit 116 receives the inter arrival times (iat) of the packets transmitted on segments 112A and 112B and adds the respective iat(A) and iat(B) values to generate iat(C) values for segment 200C. The packet delay variation factor I_eC(pdv) of segment 200C is then computed based on the calculated inter arrival times (iat) of the segment.

In some embodiments of the invention, the CODEC factor I_eC(Codec) of network segment 200C is set to be equal to the value of the CODEC factor I_e(Codec) of one of segments 112A and 112B, as for compatibility, the CODEC value is equal for both segments 112A and 112B.

Central calculation unit 116 then calculates the R value of segment 200C according to the known equation:

R^c = f{I_dC I_eC(pdv),I_eC(p_acket loss), I_eC(Codec)}.

The value of RC may be used, for example, to select a communication path for connecting telephones 102, to determine billing amounts and or to provide quality assurance measures.

Fig. 3 is a Markov diagram 300 used in modeling the packet loss on network segments, as is known in the art. The segment is modeled to be either in a burst condition represented by states 2 and 3, or in a gap condition represented by states 1 and 4. For simplicity of the calculations, states 1 and 4 are referred together as state 1.

Segment 200C is in the gap condition only if both of segments 112A and 112B are in the gap condition. Under the assumption that the loss of packets on segments 112A and 112B is independent, the probability for the segment 200C to be in the gap condition is:

P^C(GAP) = P^A (GAP) ■ P^B (GAP) = π(πξ The probability of segment 200C being in the Burst condition is therefore:

P^c (Burst) = 1 - πfπ?

Segment 200C can be in state 3, i.e., loss during a burst condition, if segment 112A or segment 112B are in state 3. Therefore, the probability for segment 200C to be in state 3 is the sum of the respective probabilities of segments 112A and 112B being in state 3 minus the probability of both segments 112A and 112B being in state 3, i.e., π-f = τc£ + πξ - π^Aπ£ .

As is known in the art, the burst loss density Dtø is given by the ratio of the probability of being in state 3 and the probability of being in the burst condition

multiplied by 100, i.e., D = 100 — . Therefore, by substituting for the values of

P(burst)

segment 200

As is known in the art, the gap loss density Dg, is the probability of loss of a packet when the transmission is in the gap condition, i.e., Dg= P(Loss|Gap). According to the known connection: pξ = P^c (Loss\Burst)P^C (Burst) + P^C (Loss] Gap) P^C (Gap)

and since the stationary probability of state 3 is equal to the probability of losing a packet in the burst condition: ^ = P^C (Loss] Burst) P^C (Burst)

it follows that:

The combined packet loss on segment 200C in terms of the loss on segments 112A and 112B is equal to:

Therefore: _n r. l-(l-Pέ)(l-Pξ)-( + - )

Following are simulation results performed to determine whether a linear combination of the packet delay variation factor I_e(pdv) of segments 112A and 112B could be used to evaluate the value of the packet delay variation factor I_eC(pdv) for segment 200C, and what the coefficients of the combination should be.

In each simulation, a series of inter arrival times (iat) for each of segments 112A and 112B were calculated. According to the two series the packet delay variation factor I_e(pdv) was calculated for each of segments 112A, 112B and 200C. A regression method was used to fit the packet delay variation factors to a linear combination of the form:

I_e (pdv) = β + β mm{I^(pdv),I^B(pdv)} + β m3x{lf(pdv),I^B(pdv)} in which β₀,β_l,β₂ axe coefficients of the linear combination. In addition, the squared correlation coefficient (R^) linearity measure, known in the art, was determined.

Two sets of simulations were performed. A first set of simulations, summarized in table 1, was performed using inter arrival time (iat) values in accordance with a uniform Gamma distribution on [0,1]. A second set of simulations, summarized in table 2, was performed using inter arrival time (iat) values in accordance with an exponential Gamma distribution with a mean value of 1. Five simulations including 500 pairs of sequences with 1000 values of inter arrival time (iat) values were performed for a variety of jitter buffer tjb and discard (t_di_scard) values.

As can be seen in tables 1 and 2, a high linear correlation (in terms of R^) between I_eC(pdv) and the linear combination of I_e^ pdv) and I_e^(pdv), was found. Rounding up of βj and β2 coefficients to account for some inaccuracies (e.g., the known Δ parameter), and to compensate for the βo coefficient that is dropped, it is suggested to use the heuristic βτ= 0.7 and β2= .9.

Table 1 Numerical results of first study: Uniform Gamma parameters

Table 2 Numerical results of first study: Exponential Gamma parameters

It will be appreciated that the above described methods may be varied in many ways, mcluding, changing the order of steps, and/or performing a plurality of steps concurrently. It should also be appreciated that the above described description of methods and apparatus are to be interpreted as including apparatus for carrying put the methods and methods of using the apparatus.

The present invention has been described using non-limiting detailed descriptions of embodiments thereof that are provided by way of example and are not intended to limit the scope of the invention. It should be understood that features and/or steps described with respect to one embodiment may be used with other embodiments and that not all embodiments of the invention have all of the features and/or steps shown in a particular figure or described with respect to one of the embodiments. Variations of embodiments described will occur to persons of the art.

It is noted that some of the above described embodiments may describe the best mode contemplated by the inventors and therefore may include structure, acts or details of structures and acts that may not be essential to the invention and which are described as examples. Structure and acts described herein are replaceable by equivalents which perform the same function, even if the structure or acts are different, as known in the art. Therefore, the scope of the invention is limited only by the elements and limitations as used in the claims. When used in the following claims, the terms "comprise", "include", "have" and their conjugates mean "including but not limited to".

Claims

1. A method of calculating a quality measure of a communication path, comprising: transmitting packets on at least two segments of the communication path; determining a value of at least one quality parameter for the at least two segments of the communication path responsive to the transmitted packets; and calculating a value of the quality measure of the path based on the determined parameter values of the at least two segments.

2. A method according to claim 1, wherein the quality measure comprises a measure for evaluating voice over IP connections.

3. A method according to claim 2, wherein the quality measure comprises the R parameter of the E model.

4. A method according to any of the preceding claims, wherein determining the value for the at least one parameter comprises providing values for the same parameter for each of the at least two segments.

5. A method according to any of the preceding claims, wherein determining a value of at least one quality parameter comprises providing values for a packet delay variation factor.

6. A method according to any of the preceding claims, wherein determining a value of at least one quality parameter comprises providing values for a delay impairment factor.

7. A method according to any of the preceding claims, wherein determining a value of at least one quality parameter comprises providing values for a CODEC related factor.

8. A method according to any of the preceding claims, wherein calculating the value ' of the quality measure of the path comprises calculating a function of the values of at least one of the quality parameters of the at least two segments.

9. A method according to any of the preceding claims, wherein calculating the value of the quality measure of the path comprises calculating a linear combination of the values of at least one of the quality parameters of the at least two segments.

10. A method according to claim 9, wherein calculating the linear combination of the values comprises calculating a linear combination in which the values of the different segments have non-equal coefficients.

11. A method according to claim 10, wherein calculating the linear combination comprises calculating the linear combination with coefficients assigned according to the relative values of the parameters of the different segments.

12. A method according to claim 10 or claim 11, wherein calculating the linear combination comprises calculating the linear combination with coefficients assigned to the minimum and maximum values of the values of the parameters.

13. A method according to any of claims 9-12, wherein calculating the linear combination comprises calculating using coefficients determined responsive to packet transmission simulations.

14. A method according to any of the preceding claims, wherein calculating the value of the quality measure of the path comprises deriving, for at least one of the segments, an average delay of the segment responsive to an impairment delay factor of the segment.

15. A method according to claim 14, wherein deriving, for at least one of the segments, an average delay of the segment comprises deriving the average delay from an impairment delay factor derived from the average delay through a non-reversible function.

16. A method according to any of the preceding claims, wherein calculating the value of the quality measure of the path comprises calculating a packet loss factor responsive to packet loss probabilities of the at least two segments.

17. A method according to claim 16, wherein calculating the packet loss factor comprises calculating according to a loss bursty model.

18. A method according to claim 17, wherein calculating according to the loss bursty model comprises calculating according to the TIPHON loss bursty model.

19. A method according to any of the preceding claims, wherein one or more of the at least two segments comprises a multi-link segment.

20. A method according to any of the preceding claims, wherein transmitting packets comprises transmitting test packets.

21. A method according to any of the preceding claims, wherein transmitting packets comprises transmitting data packets.

22. A method according to claim 21, comprising altering at least one of the data packets before transmitting the packet.

23. A method according to any of the preceding claims, wherein transmitting packets comprises transmitting packets according to a network layer protocol.

24. A method according to claim 23, wherein transmitting packets comprises transmitting packets according to the IP protocol.

25. A method according to any of the preceding claims, wherein transmitting packets comprises transmitting using a transport layer protocol.

26. A method according to any of the preceding claims, wherein transmitting packets comprises transmitting using an application layer protocol.

27. A method according to any of the preceding claims, wherein transmitting test packets comprises transmitting packets which carry time stamps used in determining the value of at least one quality parameter.

28. A method according to any of the preceding claims, comprising calculating the value of the quality measure for a plurality of paths in a network.

29. A method according to any of the preceding claims, comprising selecting a route for data packets to be transmitted through the network, responsive to calculating the value of the quality measure.

30. A method according to claim 29, wherein at least two of the plurality of paths include at least one common segment.

31. A method according to any of the preceding claims, wherein calculating the value of the quality measure of the path comprises calculating a complex quality measure which depends on values of a plurality of quality parameters.

32. A method of calculating a voice quality measure of a communication path, comprising: providing values of a plurality of quality parameters for at least two segments of the communication path; and calculating a value of the voice quality measure of the path based on the provided parameter values of the at least two segments.