Download PDF VoIP Service Quality: Measuring and Evaluating Packet-switched Voice (Professional Telecom)

Free download. Book file PDF easily for everyone and every device. You can download and read online VoIP Service Quality: Measuring and Evaluating Packet-switched Voice (Professional Telecom) file PDF Book only if you are registered here. And also you can download or read online all Book PDF file that related with VoIP Service Quality: Measuring and Evaluating Packet-switched Voice (Professional Telecom) book. Happy reading VoIP Service Quality: Measuring and Evaluating Packet-switched Voice (Professional Telecom) Bookeveryone. Download file Free Book PDF VoIP Service Quality: Measuring and Evaluating Packet-switched Voice (Professional Telecom) at Complete PDF Library. This Book have some digital formats such us :paperbook, ebook, kindle, epub, fb2 and another formats. Here is The CompletePDF Book Library. It's free to register here to get Book file PDF VoIP Service Quality: Measuring and Evaluating Packet-switched Voice (Professional Telecom) Pocket Guide.

Trends in voice quality performance can be monitored continually with data created at each element in a network. Monitoring voice quality over an entire network using the methods of the preferred embodiment allows automatic collection of additional call information to be included in a management call record for post-analysis. The analyst can trace call routes to identify network call paths as well as phone numbers where collected data indicates problems in the network or where customers may comment of having poor quality or connection problems with calls.

The assembly of a selected set of LEC performance indicators can be aggregated to search for, and evaluate patterns in, VOIP network performance indicators across all hierarchical network levels. Voice quality reports include analysis of transient data flowing through the network for real-time or offline analysis. An aggregated data report for a network module, node, group of components, or an entire network division includes all the lower level voice quality indicators e.

For example, each of the nodes could report a voice quality score or data indicator that includes all of the voice quality indicators for the DSPs, channels, and ECUs in each of the hardware devices and packet transfer statistics for all components that comprise the node. To determine an aggregated performance of DSPs in a node, only the node data needs to be queried for performance indicators since a report of the node's voice quality indicator data includes all of the data indicators for all related modules within the node.

Direct raw data and voice quality indicator data sets may selectively be gathered and stored for offline analysis. The isolated components in the network may then be investigated to search for related data sets reporting error flags and the raw data for the individual network components creating the error flag investigated throughout the levels of the network. For example, if a specific phone number is consistently experiencing QoS problems with calls, the network behavior of the entire call path can be traced and evaluated.

A problem with a call may not be caused by a hardware failure but could be a performance problem that is flagged by the reporting of voice data in a specific part of the network. The present invention allows a network administrator to isolate the problem down to an individual module within a channel and take corrective action in the problematic component prior to complete failure of the component or failure of the network. Through data collection and correlation, periodic pro-active offline audits of an targets aspects of network performance can be performed in order to increase quality of the voice network without causing interruptions in service.

In an alternative exemplary embodiment, LEC data on each hierarchical level of the network is reported using fuzzy data sets. Referring to the flowchart in FIG. The LEC voice and non-voice network data may be gathered S and reported to central monitoring server S Data may be generated for individual network elements or aggregated together in any combination possible. To report fuzzy LEC data sets, raw LEC data is gathered from network elements and transmission lines and submitted for fuzzification S All fuzzy LEC data sets could be analyzed independently or aggregated together to provide a single fuzzy LEC quality score for a group of network elements, a leg of a network, or the entire VOIP network.

Thus, fuzzy LEC data determinations are made of the health of hardware, software, and communication links for each hierarchical level, down to modules in each of the nodes. As stated above, to avoid the problem of dealing with an overwhelming mass of LEC network data, the fuzzy LEC data is also analyzed through the rules and thresholds and then fused with other LEC related or non-related data to create quality assessments of one or more reduced and simplified values.

An example of fusing data is to focus on tracking LEC and non-LEC related data, such as packet transmission quality in combination with signal quality, echo cancellation, and voice power levels. The fuzzy LEC data sets reflect network component operation and voice quality status and are based on fuzzy logic. Fuzzy logic has the advantages of the ability to model expert systems comprising inputs with uncertainties that cannot be modeled with pure logic. Fuzzy inference is the process of formulating the mapping from a given input to an output using fuzzy logic.

In other words, fuzzy logic uses a system with inputs that can be true or false to a certain degree, according to membership in a set. Fuzzy systems are based on rules that may be obtained using heuristics e. The flexibility in which additional functionalities may be added for a process control are also advantages of the fuzzy inference system. The fuzzy inference system of the present invention provides an operational reporting technique that results in a superior way over existing methods or systems.

Fuzzy logic may be considered an extension of conventional Boolean logic in that logical values are not restricted to zero FALSE and one TRUE , but rather can take on any value between zero and one inclusive. This provides greater flexibility and precision in the description process. On the other hand, with fuzzy logic, membership is represented by a continuum of values. One individual may receive 0. Applied to voice quality monitoring in a VOIP network, this method be used to track data from one or more network sources over time while the administrator is periodically observing the data for trends in the data that may trend towards optimal performance or trend towards a failure of performance.

A fuzzy inference system FIS is a system that uses fuzzy logic to map one or more inputs to one or more outputs. The FIS employed in the exemplary embodiment is based on Mamdani's fuzzy inference method. However, it is understood that one skilled in the art will recognize that the present invention is not limited merely to Mamdani or any particular fuzzy logic method.

Mamdani's method uses fuzzy inference in which both the inputs and outputs are treated as fuzzy variables. A fuzzy inference system may generally be described functionally in the following five steps:. However, the LEC data may also be assigned flags to indicate whether the behavior is over, under or between thresholds given by an administrator. Thus, an important concept of the present invention is that one or more fuzzy values can be used to reflect a single LEC quality data assessment or many fused assessments for a VOIP network.

File Extensions and File Formats

According to the alternative embodiment, the LEC performance of each associated module in each node , 98 is evaluated using fuzzy reports of operational data S Fuzzy LEC data sets can indicate the channels in the DSP that are performing properly and which are under-performing and which are failing to perform. A final aggregated fuzzy report is then produced that reflects the operations and voice quality of the incorporated elements.

The fuzzy performance indicators of LEC on the VOIP network can be used to search and evaluate patterns of network performance across all levels of the network. A snapshot of all levels of the network may be evaluated for VOIP voice quality and status over time, evaluated in an offline analysis. The fuzzy LEC data reports include analysis of transient voice and system data flowing through the network and the behavior of each network element S Such flags can then be correlated with other flags from the same or different hierarchical levels to indicate behavior of all echo cancellers in a LAN at a certain level of network use, or the performance throughout the entire VOIP.

The fuzzy LEC data report for a VOIP network may be configured to include a single node and its modules, a group of nodes, servers, and gateways, or all network elements including data transmission statistics throughout the network. Fuzzy LEC data can be searched to isolate errors. Fuzzy reported data can be correlated, or fused, together with LEC and non-LEC data from throughout the network to determine an overall behavior of the VOIP network instead of scoring performance of individual network components.

This makes it possible to view trends of operational performance throughout the network in any combination of views. A user may look up and down legs of in the entire VOIP network to analyze where the errors indicated by the fuzzy LEC data set reporting are occurring. Thus, different types of data in the VOIP network may be fused together such as modules, voice channels, groups of software functions, packet transfer and network congestion, time division multiplex data, echo cancellation, and so forth to create different assessments of the factors that affect voice quality within the network.

Fuzzy and direct LEC data is collected from each level of the VOIP network for evaluation of trends of operational and voice quality problems in the network. For example, all errors in a VOIP may be reported in and from other devices connected to a single voice gateway device. The fuzzy LEC data sets from the voice gateway may then be further analyzed to search for errors within a specific processor, voice channel, or module.

Direct raw and fuzzy LEC data sets may selectively be collected and stored for offline analysis. The isolated components in the network may then be investigated to search for related fuzzy data sets reporting error flags and the raw data for the individual network components creating the error flag investigated throughout the levels of the network.

This allows a network administrator to isolate the problem down to an individual module within a channel and take corrective action in the problematic component prior to complete failure of the component or failure of the network. A problem with a call may not be caused by a hardware failure but could be a performance problem that is flagged by the fuzzy reporting of LEC data in a specific part of the network. Through data collection and correlation, periodic pro-active offline audits of an entire network performance, from central servers and media gateways down the hierarchical levels to software modules in individual voice channels, can be performed in order to increase quality of the network without causing interruptions in service.

By fusing fuzzy LEC data sets together, trends in data and network performance can be researched and analyzed. If a specific phone number is consistently experiencing echo problems with calls, the network behavior can be traced and evaluated. To accomplish fuzzy LEC data reporting, each monitored network element can either continuously calculate and transmit the fuzzy data or periodically report the data to monitoring server Time of periodicity for polling a lower level node for data or transmitting the data to a higher level can differ according to configuration by the network manager.

If raw data is not requested or needed from a network component, then only the fuzzy data report needs to be transmitted. The preferred use of fuzzy reporting of LEC data affecting voice quality, instead of merely reporting raw hardware or data transfer statistics, characterizes the behavior of a VOIP network either in correlated groups of network elements or the network as a whole. Fuzzy LEC data from different hierarchical levels, from remote hardware components, or from any combination of nodes and modules can be correlated together to provide an indication of VOIP network performance.

The present invention has an advantage that is a simple way to proactively identify and flag potential problems in a voice network to allow rapid response to major voice quality issues that impact customer's voice services, and allow service providers to monitor network voice performance in order to proactively improve and optimize voice quality in the network.

The present invention provides further advantages of real-time indication to a network administrator of potential network issues that can proactively be addressed prior to customer problem reports. Thus, proactive maintenance of VoIP networks is provided on a comprehensive scale over all hierarchical levels of the networks. Because many varying and different embodiments may be made within the scope of the inventive concept herein taught, and because many modifications may be made in the embodiments herein detailed in accordance with the descriptive requirements of the law, it is to be understood that the details herein are to be interpreted as illustrative and not in a limiting sense.

Year of fee payment : 4. Year of fee payment : 8. Monitoring voice quality passively using line echo cancellation data across a telecommunications network and reporting monitoring data to a central network management system. Network is monitored for potential voice quality issues for pro-active isolation of problems prior to customer complaints about the problems.

Line echo cancellation related and non-related data for IP and other networks is gathered and correlated together to provide voice quality assessments of network performance. Active tests perform calculations on test or simulated calls and thus intrude on normal network usage, while passive tests can perform calculations on active calls in live networks without any interruption of service It is costly to test the quality of voice networks at the component and system level and to measure the performance of active networks, since revenue-producing traffic must be interrupted to perform the tests.

SUMMARY The limitations of the prior art are overcome by the present invention's technique for intelligent real-time monitoring of voice network conditions. A fuzzy inference system may generally be described functionally in the following five steps: 1. Fuzzification of inputs through membership functions; 2. Application of fuzzy operations as defined by the rules; 3. Implication to create fuzzy outputs for each rule; 4. Aggregation of fuzzy rule outputs; and 5.

Defuzzification of aggregated fuzzy output. Step five, defuzzification of aggregated fuzzy output, is implemented in the exemplary embodiment because direct fuzzy outputs are used to report operations of the VOP network and network components. It is understood that one skilled in the art will recognize that defuzzification of aggregated fuzzy output may also be implemented in the embodiments without departing from the scope of the present invention.

The method of claim 1 , wherein said correlating further comprises: tracing routes of a voice call in said VOIP network; and. The method of claim 1 , wherein said generating comprises gathering and comparing said analyzed LEC data to determine quality of a call setup and voice data transmissions along a route in said VOIP network. A method of real-time monitoring a voice over packet network VOIP , comprising: determining, with a fuzzy inference system, line echo cancellation performance of a plurality of network elements from throughout the entire VOIP network;.

The method of claim 4 , wherein said analyzing comprises defining a plurality of rules for scaling an output of said fuzzy inference system; and aggregating a plurality of said scaled outputs into a single fuzzy score, wherein said score determines a quality of LEC for an element of said VOIP network.

The method of claim 5 , wherein said aggregating comprises aggregating fuzzy data sets from a lower hierarchical level in said network into a higher hierarchical level in said VOIP network. The method of claim 4 , wherein said analyzing comprises monitoring for LEC of a voice over Internet Protocol call by analyzing fuzzy LEC data sets from components in said network that are associated with said call.

This makes it possible to view trends of call statistics throughout the network in any logical combination of correlations. Such evaluations can be performed in real-time or off-line. A user may look up and down the hierarchical levels and call routes in the entire VOIP network. Thus, different types of data in the VOIP network may be fused together to create different views of network performance, such as LEC modules, voice channels, groups of software functions, packet transfer and network congestion, time division multiplex data, echo cancellation, and so forth.

Trends in voice quality performance can be monitored continually with data created at each element in a network. Monitoring voice quality over an entire network using the methods of the preferred embodiment allows automatic collection of additional call information to be included in a management call record for post-analysis. The analyst can trace call routes to identify network call paths as well as phone numbers where collected data indicates problems in the network or where customers may comment of having poor quality or connection problems with calls.

The assembly of a selected set of LEC performance indicators can be aggregated to search for, and evaluate patterns in, VOIP network performance indicators across all hierarchical network levels. Voice quality reports include analysis of transient data flowing through the network for real-time or offline analysis. An aggregated data report for a network module, node, group of components, or an entire network division includes all the lower level voice quality indicators e.

For example, each of the nodes could report a voice quality score or data indicator that includes all of the voice quality indicators for the DSPs, channels, and ECUs in each of the hardware devices and packet transfer statistics for all components that comprise the node. To determine an aggregated performance of DSPs in a node, only the node data needs to be queried for performance indicators since a report of the node's voice quality indicator data includes all of the data indicators for all related modules within the node.

Direct raw data and voice quality indicator data sets may selectively be gathered and stored for offline analysis. The isolated components in the network may then be investigated to search for related data sets reporting error flags and the raw data for the individual network components creating the error flag investigated throughout the levels of the network. For example, if a specific phone number is consistently experiencing QoS problems with calls, the network behavior of the entire call path can be traced and evaluated.

A problem with a call may not be caused by a hardware failure but could be a performance problem that is flagged by the reporting of voice data in a specific part of the network. The present invention allows a network administrator to isolate the problem down to an individual module within a channel and take corrective action in the problematic component prior to complete failure of the component or failure of the network.

Through data collection and correlation, periodic pro-active offline audits of an targets aspects of network performance can be performed in order to increase quality of the voice network without causing interruptions in service. In an alternative exemplary embodiment, LEC data on each hierarchical level of the network is reported using fuzzy data sets. Referring to the flowchart in FIG. The LEC voice and non-voice network data may be gathered S and reported to central monitoring server S Data may be generated for individual network elements or aggregated together in any combination possible.

To report fuzzy LEC data sets, raw LEC data is gathered from network elements and transmission lines and submitted for fuzzification S All fuzzy LEC data sets could be analyzed independently or aggregated together to provide a single fuzzy LEC quality score for a group of network elements, a leg of a network, or the entire VOIP network.

Thus, fuzzy LEC data determinations are made of the health of hardware, software, and communication links for each hierarchical level, down to modules in each of the nodes. As stated above, to avoid the problem of dealing with an overwhelming mass of LEC network data, the fuzzy LEC data is also analyzed through the rules and thresholds and then fused with other LEC related or non-related data to create quality assessments of one or more reduced and simplified values.

An example of fusing data is to focus on tracking LEC and non-LEC related data, such as packet transmission quality in combination with signal quality, echo cancellation, and voice power levels. The fuzzy LEC data sets reflect network component operation and voice quality status and are based on fuzzy logic. Fuzzy logic has the advantages of the ability to model expert systems comprising inputs with uncertainties that cannot be modeled with pure logic.

Fuzzy inference is the process of formulating the mapping from a given input to an output using fuzzy logic. In other words, fuzzy logic uses a system with inputs that can be true or false to a certain degree, according to membership in a set. Fuzzy systems are based on rules that may be obtained using heuristics e. The flexibility in which additional functionalities may be added for a process control are also advantages of the fuzzy inference system. The fuzzy inference system of the present invention provides an operational reporting technique that results in a superior way over existing methods or systems.

Fuzzy logic may be considered an extension of conventional Boolean logic in that logical values are not restricted to zero FALSE and one TRUE , but rather can take on any value between zero and one inclusive. This provides greater flexibility and precision in the description process. On the other hand, with fuzzy logic, membership is represented by a continuum of values. One individual may receive 0. Applied to voice quality monitoring in a VOIP network, this method be used to track data from one or more network sources over time while the administrator is periodically observing the data for trends in the data that may trend towards optimal performance or trend towards a failure of performance.

A fuzzy inference system FIS is a system that uses fuzzy logic to map one or more inputs to one or more outputs. The FIS employed in the exemplary embodiment is based on Mamdani's fuzzy inference method.

However, it is understood that one skilled in the art will recognize that the present invention is not limited merely to Mamdani or any particular fuzzy logic method. Mamdani's method uses fuzzy inference in which both the inputs and outputs are treated as fuzzy variables. A fuzzy inference system may generally be described functionally in the following five steps:. However, the LEC data may also be assigned flags to indicate whether the behavior is over, under or between thresholds given by an administrator.

Thus, an important concept of the present invention is that one or more fuzzy values can be used to reflect a single LEC quality data assessment or many fused assessments for a VOIP network. According to the alternative embodiment, the LEC performance of each associated module in each node , 98 is evaluated using fuzzy reports of operational data S Fuzzy LEC data sets can indicate the channels in the DSP that are performing properly and which are under-performing and which are failing to perform.

A final aggregated fuzzy report is then produced that reflects the operations and voice quality of the incorporated elements. The fuzzy performance indicators of LEC on the VOIP network can be used to search and evaluate patterns of network performance across all levels of the network. A snapshot of all levels of the network may be evaluated for VOIP voice quality and status over time, evaluated in an offline analysis. The fuzzy LEC data reports include analysis of transient voice and system data flowing through the network and the behavior of each network element S Such flags can then be correlated with other flags from the same or different hierarchical levels to indicate behavior of all echo cancellers in a LAN at a certain level of network use, or the performance throughout the entire VOIP.

The fuzzy LEC data report for a VOIP network may be configured to include a single node and its modules, a group of nodes, servers, and gateways, or all network elements including data transmission statistics throughout the network. Fuzzy LEC data can be searched to isolate errors. Fuzzy reported data can be correlated, or fused, together with LEC and non-LEC data from throughout the network to determine an overall behavior of the VOIP network instead of scoring performance of individual network components.

This makes it possible to view trends of operational performance throughout the network in any combination of views. A user may look up and down legs of in the entire VOIP network to analyze where the errors indicated by the fuzzy LEC data set reporting are occurring. Thus, different types of data in the VOIP network may be fused together such as modules, voice channels, groups of software functions, packet transfer and network congestion, time division multiplex data, echo cancellation, and so forth to create different assessments of the factors that affect voice quality within the network.

Fuzzy and direct LEC data is collected from each level of the VOIP network for evaluation of trends of operational and voice quality problems in the network. For example, all errors in a VOIP may be reported in and from other devices connected to a single voice gateway device.

Practical Voice Over IP (VoIP) HT16

The fuzzy LEC data sets from the voice gateway may then be further analyzed to search for errors within a specific processor, voice channel, or module. Direct raw and fuzzy LEC data sets may selectively be collected and stored for offline analysis. The isolated components in the network may then be investigated to search for related fuzzy data sets reporting error flags and the raw data for the individual network components creating the error flag investigated throughout the levels of the network.

This allows a network administrator to isolate the problem down to an individual module within a channel and take corrective action in the problematic component prior to complete failure of the component or failure of the network. A problem with a call may not be caused by a hardware failure but could be a performance problem that is flagged by the fuzzy reporting of LEC data in a specific part of the network. Through data collection and correlation, periodic pro-active offline audits of an entire network performance, from central servers and media gateways down the hierarchical levels to software modules in individual voice channels, can be performed in order to increase quality of the network without causing interruptions in service.

By fusing fuzzy LEC data sets together, trends in data and network performance can be researched and analyzed. If a specific phone number is consistently experiencing echo problems with calls, the network behavior can be traced and evaluated. To accomplish fuzzy LEC data reporting, each monitored network element can either continuously calculate and transmit the fuzzy data or periodically report the data to monitoring server Time of periodicity for polling a lower level node for data or transmitting the data to a higher level can differ according to configuration by the network manager.

If raw data is not requested or needed from a network component, then only the fuzzy data report needs to be transmitted. The preferred use of fuzzy reporting of LEC data affecting voice quality, instead of merely reporting raw hardware or data transfer statistics, characterizes the behavior of a VOIP network either in correlated groups of network elements or the network as a whole. Fuzzy LEC data from different hierarchical levels, from remote hardware components, or from any combination of nodes and modules can be correlated together to provide an indication of VOIP network performance.

The present invention has an advantage that is a simple way to proactively identify and flag potential problems in a voice network to allow rapid response to major voice quality issues that impact customer's voice services, and allow service providers to monitor network voice performance in order to proactively improve and optimize voice quality in the network.

The present invention provides further advantages of real-time indication to a network administrator of potential network issues that can proactively be addressed prior to customer problem reports. Thus, proactive maintenance of VoIP networks is provided on a comprehensive scale over all hierarchical levels of the networks.

Because many varying and different embodiments may be made within the scope of the inventive concept herein taught, and because many modifications may be made in the embodiments herein detailed in accordance with the descriptive requirements of the law, it is to be understood that the details herein are to be interpreted as illustrative and not in a limiting sense.

Year of fee payment : 4. Year of fee payment : 8. Monitoring voice quality passively using line echo cancellation data across a telecommunications network and reporting monitoring data to a central network management system. Network is monitored for potential voice quality issues for pro-active isolation of problems prior to customer complaints about the problems. Line echo cancellation related and non-related data for IP and other networks is gathered and correlated together to provide voice quality assessments of network performance.

Active tests perform calculations on test or simulated calls and thus intrude on normal network usage, while passive tests can perform calculations on active calls in live networks without any interruption of service It is costly to test the quality of voice networks at the component and system level and to measure the performance of active networks, since revenue-producing traffic must be interrupted to perform the tests.

SUMMARY The limitations of the prior art are overcome by the present invention's technique for intelligent real-time monitoring of voice network conditions. A fuzzy inference system may generally be described functionally in the following five steps: 1. Fuzzification of inputs through membership functions; 2. Application of fuzzy operations as defined by the rules; 3. Implication to create fuzzy outputs for each rule; 4. Aggregation of fuzzy rule outputs; and 5. Defuzzification of aggregated fuzzy output.

Step five, defuzzification of aggregated fuzzy output, is implemented in the exemplary embodiment because direct fuzzy outputs are used to report operations of the VOP network and network components. It is understood that one skilled in the art will recognize that defuzzification of aggregated fuzzy output may also be implemented in the embodiments without departing from the scope of the present invention.

The method of claim 1 , wherein said correlating further comprises: tracing routes of a voice call in said VOIP network; and. The method of claim 1 , wherein said generating comprises gathering and comparing said analyzed LEC data to determine quality of a call setup and voice data transmissions along a route in said VOIP network. These tests are based on substitution of the subjective tests by appropriate mathematical models or algorithms. The objective tests can be divided on intrusive and non-intrusive. The intrusive one uses two speeches for determination of final speech quality.

The first speech is original non-degraded speech and the second one is the same speech degraded by transmission over the telecommunication chain. It enables to obtain more precise results, however this method cannot be used for real-time speech quality measurement. On the other hand, non-intrusive methods do not require the original source speech, since the evaluation of speech quality is based only on the degraded speech signal processing. Hence, the non-intrusive methods are suitable for real-time monitoring of speech quality. The PESQ is one of the most widely used objective methods developed for end-to-end speech quality assessment in a conversational voice communication.

The signal Y t is result of a transmission of the signal X t through a communication system. The PESQ method generates a prediction of the quality which would be given to the signal Y t in subjective listening test. Principle of PESQ method for speech quality assessment. The range of converted value is from 1 to 4. The recalculation is defined by the next formula: : 0. Frequency characteristics of the speech signal and signal level alignment must be in accordance with recommendation ITU-T P.

Speech processing for speech quality assessment The speeches used for subjective and objective tests are original digital studio recordings, obtained by separation from dialogs of two people. Their content does not imply any emotional response at the listeners. All recordings are in Czech language spoken by natural born Czech speakers without speech aberrations.

The length of utterances is selected to fulfil the requirements of both types of tests. Therefore it is between 8 an 12 s. Tested signals is coded by bit linear PCM Pulse Code Modulation sampled with 8 kHz or 16 kHz sample rate down sampled from the original studio quality recordings sampled with 48 kHz. Three individual cases are analysed: i comparison of the impact of individual and consecutive packet losses of conventional telecommunication frequency bandwidth 3.

Each of the speeches is encoded using PCM. At the beginning, the analyzed speech is split into sections with the same length. The individual sections can be understood in terms of packet that will be transmitted through a network. The length of all packets is chosen, in each round of analysis, to be equal to the following values: 10, 20, 30, or 40 ms. The lengths of packets correspond to 80, , , or samples per packet for the speeches sampled with 8 kHz and , , , or samples for the speeches sampled with 16 kHz.

The number of elements in the vector R corresponds to the number of packets, to which the speech is divided. The random vector Vl contains random numbers in the range from to 1 generated with uniform distribution. The number of vector elements does not depend only on the length of the speech, but depends also on the length of individual packets. The packet losses are randomly determined and the number of lost packets is calculated according to the Packet Loss Ratio PLR and the total number of packets in the speech.

The threshold is determined according to the required PLR. The vector Vpl contains only numbers "one" and "zero", where zeros represent the losses. The final speech is a product of multiplication of the modified vector Vpl and the relevant speech split into R packets. Packets, which are multiplied by zero corresponds to the lost parts, and packets multiplied by the number one remained unchanged.

The total duration of lost packets is the same for all speeches with the same PLR, regardless of the amount of packets contained in a speech R. For example, if the speech is divided into packets with 10 ms length, the number of these packets is twice the number of packets created in case of the packets with 20 ms length.

Random vector vi is generated twenty times for each value of PLR, each sections length, and for each of the speeches. Repetition of generation of the random vector limits negative effects of random drop of losses, which could affect results. The random losses of packets, described above, are characterized by random appearance in time. Beside this, the consecutive losses frequently occur in real networks. For example, during the short outage in communication e. More probable is the loss of several packets.

Therefore, the consecutive packet loss can be expressed as a loss of sequence of subsequent packets. Assessment of Speech Quality in VoIP 33 Speech processing of consecutive packet losses is similar to the generation of the individual losses, with slight modification to follow the effect of losing packets in groups. Randomly generated vector Vl is thus shortened to the length r. The shortest of the analyzed speeches has length of 8 s. For the packets with duration of 40 ms in groups of twenty packets, it gives the overall duration of the ms. Hence, the higher length of consecutively lost packet cannot be accommodated into these speeches.

To eliminate the random factor of position of consecutive packet loss, twenty repetitions for each of the speeches and for each of the length of group of consecutive packet loss are performed, as in the case of the random individual packet losses to suppress the affection of results by effect of placement of losses into different parts of speeches.

For our analysis, the narrowband telecommunication channel - Hz is separated to four sub-bands. Lower count of frequency bands would give less information on individual frequency influence and higher count would distort the speech too little as it would be undistinguishable from the original non-distorted speech.

Another parameter which has to be set is the choice of corner frequencies of these bands. One possibility is to take them linearly according to the telecommunication channel band and the other one is to choose corner frequencies nonlinear ly with unequal bandwidth of each band. An ear perceives each frequency differently, which means that if the frequency of perceiving tone will increase linearly, the listener will subjectively sense only logarithmic increase.

Relating to this fact, the corner frequencies are chosen with geometrical interval.


  • Lectures on fractal geometry and dynamical systems.
  • Golf Resorts: Where to Play in Hawaii.
  • VoIP Service Quality - William C. Hardy - Google книги.

The resulting Mel-frequencies are presented in Table 3. Note that the lowest band band 1 is extended to 50 Hz. Frequency bands used for harmonic distortion. I" Frequency responses of band stop filter for all four levels of suppression and four Mel-bands a band 1; b band 2; c band 3; d band 4. Each band is suppressed in four levels: 10, 20, 30, and 40 dB.

Frequency responses of band stop filters for all four bands and suppression levels are shown in Fig. The algorithm for generation of packet losses with exact ratio is modified to place the lost packet only to the specific location of the particular phonetic elements. It requires the phonetic transcription of the speech and subsequent exact definition of the first and the last samples of each phonetic element. All individual speech sounds of the speech are considered as analyzed phonetic elements. The vowels and diphthongs are always voiced sounds, therefore they carry higher energy.

Also the nasals and liquids are predominantly voiced with higher energy level. The plosives and affricates as well as fricatives contain either voiced or unvoiced consonants, which have the same mechanism of its origination in voice organs. The classification of the speech sounds into groups is presented in Table 4.

Classification of the speech sounds into groups for speech quality assessment purposes. The speech is furthermore processed in the following way. First, the amount of packets contained only in speech sounds of investigated group of phonetic elements is determined denoted R. The coefficient H expresses the probability of loss of each packet in investigated group of phonetic elements. The length of V g is equal to R and each element of V g represent whether the packet belonging to an element of a selected group will be lost or not.

At the end, the packets labeled as "lost" are replaced by zeros silence. Results of speech quality assessment As mentioned in previous section, three types of speech modification are investigated. This section provides the results of all performed tests. While maintaining all speech packets no packets are lost , the speech quality reach the maximum. Of course this maximum is independent on the packet length.

By increasing the PLR the significant speech quality degradation can be observed from Fig. The comparison of speeches with 3. On the other hand, the speeches with bandwidth of 7 kHz achieve higher score than speeches with 3. The maximum gap between results for both bandwidths is approximately up to 0. With further increase of the PLR, both bandwidths perform in closer way. The impact of bandwidth is also decreasing with higher duration of packets. For packet duration of 40 ms, the maximum difference between both bandwidths is up to 0. Only the results of objective test are presented in this section since the results of subjective one are included in Becvaretal.

Thus the separation of the speech into packets with 40 ms length results in a higher impairment than splitting the speech into 10, 20, or 30 ms segments. This discrepancy is roughly 1. The largest difference for speeches with bandwidth 3. This analysis clearly shows that it is profitable to utilize shorter speech packets. The evaluation of the quality of speech influenced by consecutive packet losses is performed under the same conditions as the tests of individual packet losses.

Objective assessment of the different frequency bandwidth and lengths of packets over packet loss ratio. Several lengths of packets are considered for analysis. The longer duration of consecutive losses firstly lower the speech quality from approximately 2. Then, the speech quality gradually increases with length of consecutive packet losses to ms, where the MOS rating is again approximately 2. Further increase of the consecutive packet loss leads to only insignificant speech quality improvement.

This effect is not noticeable for packet length over 40 ms, where the MOS score is continuously rising over the length of consecutive loss. From Fig. This limit is sixty, ten and, three losses for packets with length of 10, 20, and 30 ms respectively for 3. The situation for 7 kHz channel is the same, however the final speech quality is lower between 0. Objective assessment of the impact of consecutive packet losses.

This is achieved at four, two, or one consecutively lost packets with length of 10, 20, and 40 ms respectively. The results also show that the same summarized duration of lost parts of speech are almost identical and independent on the length of individual packets. For example, MOS score for the loss of sections with 40 ms length is the same like for consecutive loss of bursts of two packets with 20 ms length or loss of bursts of four packets of 10 ms length.

This fact is more noticeable in Fig. This figure presents the impact of individual packet lengths over the overall length of consecutive losses. The results presented in Fig. Hence, the division of packets to 40 ms length is not appropriate from the point of view of the individual losses. The converted results of the objective tests of consecutive packet losses. Five different speeches for four bands and four suppression levels are processed for each distortion. Therefore, the overall duration of the subjective listening test is approximately 20 minutes per a listener.

Overall, 26 listeners participates on the subjective listening test. Software Tester Brada, developed at Czech technical University in Prague is used for the listening test. The listeners participated on the subjective testing have been selected from students and employees of the university with respect to above mentioned recommendations. The results are presented in Fig. Subjective assessment of the harmonic distortion. The results show that the biggest influence at the reception quality of speech is obtained by the components contained in the first band 50 - Hz. This phenomenon is caused by higher energy carried by lower frequency components in comparison with lower energy of higher frequency components.

Therefore, the suppression of low frequencies causes Assessment of Speech Quality in VoIP 39 significant decrease in the speech quality. Only slightly lower impact is caused by the frequency components contained in the fourth frequency band - Hz. The lowest degradation of the speech quality is noticeable in the second and in the third bands - Hz and - Hz. In all cases, the higher attenuation of the signal in individual bands leads to the decrease of the speech quality. The suppression of all bands has the similar impact on the speech quality according to PESQ. Also the impact of the level of attenuation is negligible since the drop of the speech quality is only between 0.

Objective assessment of the harmonic distortion by PESQ method. The average subjective score is always lower than objective one.


  • Which QoS implementation model would be most effective for your organization?;
  • Dining With The Doctor: The Unauthorized Whovian Cookbook.
  • Orphan Star (Pip & Flinx, Book 3).
  • Re-reading the Gospel of Mark Amidst Loss and Trauma!
  • McGraw-Hills SAT, 2010 Edition;

The difference between results of both tests is very significant for all bands between 0. Average MOS score of each bands over levels of suppression. The software Tester is also used for the subjective speech quality assessment. As well the listeners participated on the subjective testing have been selected from students and employees of the Czech Technical University in Prague. Four speeches are generated for each pair of parameters group and packet loss ratio to eliminate effect of random drop of losses.

The overall duration of the subjective listening test is approximately 20 minutes per a listener. The results of subjective test are presented in Fig. From this figure can be observed that the most significant group of phonetic elements from the speech quality point of view are groups containing vowels and diphthongs. This fact is caused by two reasons.

QoS (quality of service)

The first one, based on lexical aspect, says that the vowels are basement of nearly all syllables and words; hence its modification or unintelligibility can cause the change of the meaning of the whole word. The second aspect is the signal processing. From this side, the vowels as well as sound voices contains high amount of energy. Therefore, its loss leads to the loss of major part of information. The second most important group consists of nasals and liquids since all speech sounds included in this group are voiced and thus they carry high energy.

The difference between this group and group with vowels and diphthongs is marginal; it is roughly 0. The next most perceptible impact is caused by fricatives. This group contains voiced as well as unvoiced consonants. Therefore the speech quality in comparison to the first and second group is higher roughly from 0. The lowest important group consists of plosives and affricates. This group contains also voiced and unvoiced consonants; however its energy is the lowest of all groups.

VoIP Service Quality

Its impact on the speech quality is less perceptible by users. The average MOS score is by roughly 0. Subjective assessment of an impact of phonetic elements. The results of objective speech quality measurement are depicted in Fig. These results show slightly lower speech quality than subjective test. The slightly higher negative impact less than 0. Objective assessment of an impact of phonetic elements by PESQ method. The average subjective score is always higher than the results of objective evaluation. Average MOS score of each group of phonetic elements.

The difference in speech quality of groups containing only voice sounds and groups containing also unvoiced sound is considerable in results of both subjective as well as objective tests. For example, the speech quality is roughly by 1 MOS higher if the packet losses hit only plosives and affricates than if the losses are in vowels and diphthongs. This fact should influence the design of packet loss concealment mechanisms to put more focus on elimination of losses of vowels, diphthongs, nasals or liquids.

Conclusions This chapter provides an overview on the speech quality assessment in VoIP networks. On the other hand, while the speech is affected by consecutive packet losses or by individual losses with higher packet loss ratio, the narrowband channel reaches better score. The consecutive packet losses can leads to the higher speech quality while the duration of losses is long enough comparing to the individual losses. The exact duration of loss that reaches higher score than individual one depends on the length of packets. The objective method PESQ is not able to handle with the harmonic distortion and its results do not match the subjective one.

The evaluation of the importance of the groups of phonetic elements shows that the most considerable elements are vowels and diphthongs. On the other hand, the speech quality is affected only slightly by losses of plosives or affricates. References Bachu, R. Barriac, V. Discussion on unified objective methodologies for the comparison of voice quality of narrowband and wideband scenarios.

Becvar, Z. Proceeding of Digital Technologies , pp. Benesty, J; Sondhi, M. Springer handbook of speech processing, Springer- Verlag, pp. Brada, M. Proceedings of Research in Telecommunication Technology , pp. Clark, A. Fasti, H. Friedlander, B. Hanzl, V. Hassan, M.

In Proceedings of Wireless Telecommunications Symposium, pp. ITU-T Rec. Terms and definitions related to quality of service and network performance including dependability. August The E-model, a computational model for use in transmission planning. March One-way transmission time. May Pulse Code Modulation of Voice Frequencies. Methods for Subjective Determination of Transmission Quality. February Perceptual evaluation of speech quality PESQ : An objective method for end-to-end speech quality assessment of narrow-band telephone networks and speech codecs.

Mapping function for transforming P.

November Kondo, K. ED, No. Linden, J. Molau, S. Oouchi, H. Robinson, D. Sing, J. EA, No. Proceedings of 2nd IP-Telephony Workshop, pp. Ulseth, T. Telektronikk, Vol. Introduction The Internet and its packet based architecture is becoming an increasingly ubiquitous communications resource, providing the necessary underlying support for many services and applications. It is known that, due to real time requirements, voice over IP VoIP needs tighter delivery guarantees from the networking infrastructure than data transmission. While such requirements put strong bounds on maximum end to end delay, there is some tolerance to errors and packet losses in VoIP services providing that a minimum quality level is experienced by the users.

Therefore, voice signals delivered over IP based networks are likely to be affected by transmission errors and packet losses, leading to perceptually annoying communication impairments. This chapter is concerned with voice signal reconstruction methods and quality evaluation in VoIP communications. An overview of suitable solutions to conceal the impairment effects in order to improve the QoS and consequently the Quality of Experience QoE is presented in section 2.

Among these, simple techniques based on either silence or waveform substitution and others that embed voice parameters of a packet in its predecessor are addressed. Section 3 provides a brief review of relevant algebra concepts in order to build an adequate basis to understand the fundamentals of the signal reconstruction techniques addressed in the remaining sections. Since signal reconstruction leads to linear interpolation problems defined as system of equations, the characterization of the corresponding system matrix is necessary because it provides relevant insight about the problem solution.

In such 46 VoIP Technologies characterisation, it will be shown that eigenvalues, and particularly the spectral radius, have a fundamental role on problem conditioning. This is analysed in detail because existence of a solution for the interpolation problem and its accuracy both depend on the characterisation of the problem conditioning. Section 4 of this chapter describes in detail effective signal reconstruction techniques capable to cope with missing data in voice communication systems.

Two linear interpolation signal reconstruction algorithms, suitable to be used in VoIP technology, are presented along with comparison between their main features and performance. The difference between maximum and minimum dimension problems, as well as the difference between iterative and direct computation for finding the problem solution are also addressed. One of the interpolation algorithms is the discrete version of the Papoulis-Gerchberg algorithm, which is a maximum dimension iterative algorithm based on two linear operations: sampling and band limiting.

A particular emphasis will be given to the iterative algorithms used to obtain a target accuracy subject to appropriate convergence conditions. The importance of the system matrix spectral radius is also explained including its dependence from the error pattern geometry. Evidence is provided to show why interleaved errors are less harmful than random or burst errors. The other interpolation algorithm presented in section 4 is a minimum dimension one which leads to a system matrix whose dimension depends on the number of sample errors.

Therefore the system matrix dimension is lower than that of the Papoulis-Gerchberg algorithm. Besides an iterative computational variant, this type of problem allows direct matrix computation when it is well-conditioned. As a consequence, it demands less computational effort and thus reconstruction time is also smaller. In regard to the interleaved error geometry, it is shown that a judicious choice of conjugated interleaving and redundancy factors permits to place the reconstruction problem into a well conditioned operational point.

By combining these issues with the possibility of having fixed pre- computed system matrices, real-time voice reconstruction is possible for a great deal of error patterns. Simulation results are also presented and discussed showing that the minimum dimension algorithm is faster than its maximum dimension counterpart, while achieving the same reconstruction quality. Finally section 6 presents a case study including experimental results from field testing with voice quality evaluation, recently carried out at the Research Labs of Portugal Telecom Inovacao PT Inovacao.

Voice signal reconstruction and quality evaluation 2. Channel coding can be used to protect transmitted signals from packet loss but it introduces extra redundancy and still does not guarantee error-free delivery. In order to achieve higher quality in VoIP services with low delay, effective error concealment techniques must be used at the receiver. Typically such techniques extract features from the received signal and use them to recover the lost data. Enhanced VoIP by Signal Reconstruction and Voice Quality Assessment 47 The different approaches to deal with voice concealment can be classified in either source- coder independent or source-coder dependent Wah et aL, The former schemes implement loss concealment methods only at the receiver end.

In such receiver-based reconstruction schemes, lost packets may be approximately recovered by using signal reconstruction algorithms. The latter schemes might be more effective but also more complex and in general higher transmission bandwidth is necessary. In such schemes, the sender first processes the input signals, extract the features of speech, and transmit them to the receiver along with the voice signal itself. Among several possible solutions, it is worth to mention those algorithms that try to reconstruct the missing segment of the signal from correctly received samples. For instance, waveform substitution is a method which replaces the missing part of the signal with samples of the same value as its past or future neighbours, while the pattern matching method builds a pattern from the last M known samples and searches over a window of size N the set of M samples which best matches the pattern Goodman et aL, , Tang, In Aoki, the proposed reconstruction technique takes account of pitch variation between the previous and the next known signal frames.

In Erdol et aL, two reconstruction techniques are proposed based on slow-varying parameters of a voice signal: short-time energy and zero-crossing rate or zerocrossing locations. The aim is to ensure amplitude and frequency continuity between the concealment waveform and the lost one. This can be implemented by storing parameters of packet k in packet k Splitting the even and odd samples into different packets is another method which eases interpolation of the missing samples in case of packet loss.

Particularly interesting to this work is an iterative reconstruction method proposed in Ferreira, a , which is the discrete version of the Papoulis Gerchberg interpolation algorithm. A different approach, proposed in Cheetham, , is to provide mechanisms to ease signal error concealment by acting at packet level selective retransmissions to reduce the dependency on concealment techniques.

Another packet level error concealment method base on time-scale modification capable of providing adaptive delay concealment is proposed in Liu et aL, In practical receivers, the performance of voice reconstruction algorithms includes not only the signal quality obtained from reconstruction but also other parameters such as computational complexity which in turn has implications in the processing speed. Furthermore in handheld devices power consumption is also a critical factor to take into account in the implementation of these type of algorithms.

These methods take into account the most significant human voice and audition characteristics along with possible impairments introduced by current voice communication systems, such as noise, delay, distortion due to low bitrate codecs, transmission errors and packet losses. Quality 48 VoIP Technologies evaluation methods for voice can be classified into subjective, objective and parametric methods.

In the first case there must be people involved in the evaluation process to listen to a set of voice samples and provide their opinion, according to some predefined scale which corresponds to a numerical score. The Mean Opinion Score MOS collected from all listeners is then used as the quality metric of the subjective evaluation. The evaluation methods are further classified as reference and non-reference methods, depending on whether a reference signal is used for comparison with the one under evaluation.

If the MOS scores are obtained in a conversational environment, where delays play an important role in the achieved intelligibility, then this is referred to as MOScqs 2 - Even though a significant number of participants should be used in subjective tests ITU-T, , every time a particular set of tests is repeated does not necessarily lead to exactly the same results.

Subjective testing is expensive, time-consuming and obviously not adequate to real- time quality monitoring. ITU-T P. Since the reference methods interfere with the normal operation of the communication system, they are usually known as intrusive methods. The PESQ method transforms both the original and the degraded signal into an intermediate representation which is analogous to the psychophysical representation of audio signals in the human auditory system.

Such representation takes into account the perceptual frequency Bark and loudness Sone. Then, in the Bark domain, some perceptive operations are performed taking into account loudness densities, from which the disturbances are calculated. This is commonly called the raw MOS since the respective values range from -1 to 4.

More Views

It is often necessary to map raw MOS into another scale in order to compare the results with MOS obtained from subjective methods. They are also called single-ended or non-intrusive methods. The E-Model, described in the Rec. ITU-T G. While signal based methods use perceptual features extracted from the speech signal to estimate quality, the parametric E-Model uses a set of parameters that characterize the communication chain such as codecs, packet loss pattern, loss rate, delay and loudness.

Then the impairment factors are computed to estimate speech quality. This model assumes that the transmission voice impairments can be transformed into psychological impairment factors in an additive psychological scale. Based on the value of R, which is comprised between and , Rec. Annex B of Rec. Algebraic fundamentals This section presents the most relevant concepts of linear algebra in regard to the voice reconstruction methods described in detail in the next sections.

The most important mathematical definitions and relationships are explained with particular emphasis on those with applications in signal reconstruction problems. An element of any of these sets is called a vector. In digital signal processing, such vector components are known as signal samples. The solution of many signal processing problems is often found by solving a set of linear equations, i. The conjugate transpose of the mxn matrix A is the nxm matrix A H obtained from A by taking the transpose and the complex conjugate of each element aij.

Matrix B is called the inverse of A, and it is denoted by A" 1. In other words, when A is multiplied by v, the result is the same as a scalar X multiplied by v. Note that it is much easier to multiply a scalar by a vector than a matrix by a vector. The spectrum of A is defined as the set of its eigenvalues, while the spectral radius of A, denoted by p A , is the supremum 6 among the absolute values of its spectrum elements.