VOIP Technologies

VolP TechnologiKashiharalished by InTecha Trine 9 51000 Rijeka CroatiaCopyright e 201All chapters are Open Ake Attribution 3

0 license, which permits to copy,work is properly cited Afteryork has been published by InTech, autare the authoo make other pereferencing or pe of thest explicitly identify the original sourceStatements and opinions expressed in the chapters are these oftributessarily those of the editors or publisher No responsibility is accepterfor theation contained in the pub lished artidesassumesury to persons or property arising outof any materials, instructions, methods or ideasthe bookblistanager Jelena marusioCover Designer Martina Siromage Copyright WitR, 2010 Used under licenFirst published Febidditional hard copies can be obtained from orders@ intechweb orgVolP Technologies, Edited by Shigeru KashiharaSBN978-953-307549-5

VolP Quality Assessment TechnologiesMousa al-akhras and iman al momaniTelephone Network(PSTN) carriers for carrying voice traffic Before uscircuit switching network a dedicated channel or circuit is established from these sender ter and that path is selected over the most efficient route using intelligentAccordingly, it is not necessary for a phone call fromDuring callonce the route is determined, that path or circuit stays fixed throughoutthe callted to the phone call from theginning to the end of the call The established circuit cannot be used by other callers until thenins unavailable to other users even when no actual communicationis taking place, therefore, circuit switching is carrying voice with high fidelity from source todestination( Collins, 2003) Circuit switching is like having a dedicated railroad track withonly one train, the call, is permitted on the track at one timeodayercial telephone networks that based ont switching technology hatatures, including: Availability, Capacity, Fast Response and HiglQuality(Collins, 2003)

The quality is the main focus of this chapterOne alternative technologyitching networks such as Internet Protocol (IP)nepacket switching technology, no circuit is built from the sender to the receiver and packetssent over the most effectiveacktime of sending that packet, consequently differetake differfrom thever within the sameice over IP(VolP)networks istelecommunication andactive area of research Networks of the future will use ip as there transport network as IP is seen as the long-termr all types of traffic includingoice and video VolP will become the main standard for third generation wireless networksBos leroy 2001: Heiman, 1998P networksThis provides promises of greater flexibility and advanced services thanthe traditional telephony with greater possibility for cost reduction in phone calls VolP alsoas other advantages, including: number portability, lower equipment cost, lower bandwidthrequirements, lower operating and management expenses, widespread availability of Ip, andother advantages(Collins, 2003; Heiman, 1998; Low, 1996; Moon et al, 2000; Rosenberg et al

2P Technologies(Collins,lP can be used in many applications, including: call centre integration, directoryinferencing, fax over IP, and Radio/ TV BroadcastingIP technologydopted brswitchingtechnology This adoption was motivated by the abovetages and to shareof the high revenue achieved by telecommunication companies However, to be able tompete with the highly reputable PSTN networks, VolP networks should be able to achievecheaper solutions than what PsIN does, but regardless ofw the cost of theservice is, it is the user perception of the quality what matters If the quality of the voice ispoor, the user of the traditional telephony will not be attracted to the VolP service regardlessfrom the fact that customers who are used to thehigh-quality telephony networks, expect to receive a comparable quality from any potentialnd they are doing this task very well, bory non real-time traffic such as email or file transferre characterised by beind the rerefore theyrt real-timePpch as voice traffic in addition to dThe best-effort nature of ipnetworks causes several degradations to the speech signal before it reaches its destinationecause of the time-varying characteristics(eg packetclayof IP networksThese characteristics which are normal to data trafficserious deterioration to theeal-time traffic and prevent IP networks from providing the high quality speech ofterprovided by traditional PSTN networks for voice services

Sharing ofdedicated to theall in contrast to what is happeningin traditional circuit switching telephony such as PSTN where the required resources arellocated to the phone call from the start to the end with the absence of resource dedication,Among the problpacket loss whichderflow in intermediate rdue to the long time taken by packets to reach their destinations( Collins, 2003) Real-timePplications are also sensitive to delay since they require voice packets to arrive at thereceiving end within a certain upper bound to allow interactivity of the voice call (ITU-T,urce to thewhich causes packettime to vary, a phenomenon known as jitter Due to the problem of jitter, it is not easy to playpackets in a steady fashion to the listener(Narbutt Murphy, 2004; Tseng Lin, 2003; Tsengal, 2004) The above challetySPeechsignal before it reaches its destination Many solutions have been proposed to alleviate theseproblems and the quality of the received speech signal as perceived by the end user is greatlyaffected by the effectiveness of these solutionsross the path from the sender to the receiver aif it is possible to allocate the required bandwidth and maintain the giverQos target for all existing calls, or otherwise to reject the call(Mase, 2004) Among theolutions that have been proposed to impt Cacand tothe available bandwidefficiently are: Resource Reservation Protocol(RSVP), Differentiated Service (DiffServ)

VolP Quality AssessmentMultiProtocol Label Switching(MPLS), and End-to-end Measurement Based AdmissioControl(EMBAC) Reservinges is difficult and very expensive proposal as it requireshang the IntimeThereforelive networks andkees from legal, commercialnd technical reality would be a necessitygreement usually requiring the company to providelevel of quality, otherwisnies for poor quality Also,measuring the quality gives the chance to network administrators to overcome temporalproblems that could affect the quality of ongoing voice calls Measurement of the quality alsoallows service providers to evaluate their own and their competitors service using a standarddicator ofsatisfaction of the service provided (Takahashi et al2004: Zurek et al 2002To this end, a specialised mechanism is required for measuring the speech qualityy One of driving forinformation and communication technology As the global focathe private sector in developing telecommunication networks and services, ITU's role isto help the world communicate ITU-Telecommunication Standardisation Sector (ITU-T,toward standardising and regulating international telecomdwide towardis goal, ITU-T study technical, operating and tariff qder the name of Recommendations for the purpose of standardising telecommunicationsITU-Ts Recommendations are divided into categories that are identified byfor example P800(ITU-T, 1996b)

ITU-T has a formal recognition as it is part of ITU which isquality formany of these standards are considered in this chapter Speechquality in ITU-T standards is expressed as Mean Opinion Score(MOS)which ranges betweenI and 5, with 1 corresponds to poor quality and 5 to excellent quSome standards measure the speech quality or the MOS subjectively by setting lab conditionsterms of MOS This method is standardised in ITU-T Recommendation P 800(ITU-T, 1996b)Other methods are objective that depend on comparison of the received signal with there the perceived quality in terms of MOS, these methods are knowntrusive methods as they require the injection of the original signal to analyse the distortionof the received signal The most recent method for measuring the speech qualitf Speech Quality(PESQ) PESQ is standardisedRecommendation P862 (ITU-T, 2001) Yet another objective category depends on either theed signal or the nearameters to estimattrusively withoutP563(ITU-T, 2004)and the E-model as defined in ITU-T RecoITU-T2009),Mosed by othsearchers, and the authors of this chapter independentITU-T, these attempts will bediscussed in detail later in the chapto

P TechnologiesThe selection of a method for VolP qualitshould take the characteristics of ipnetworks and voice calls into consideration Such characteristics that affect the selectionTo able to do this, an objectivthated signal at theside without the need for the original speech signal at the sender side; ie a non-intrusiveThis chapter aims to servin VolP networks The rest of this chapter is organised as foltechniques and discusses the main requirements ofpplicable technique in VolP environment Sections 3 and 4 discusses subjective and objectivedistinguish between different qualityt methodspresentedd possibilites for future work aresection 62 Categories of volP quality assessment technologiesobjective methods Objective methods can be either intrusive or non-intrusive Non-intrusivemethodsbe eithearametric-basedcts differentclassificationsThe primaryality A subjective qualityaffect the quality of service of VolP, among there:packet loss, delay, jitter, loudness, echo, and codec distortionthe subjectiethod is used, the most widely accepted metricthe Mean Opinion Score(MOS)as defined by ITu-T Recommendation P800 (ITU-T, 1996bHowev

er, although subjective quality assessment is the most reliable methtime-consuming and expensive as any other subjective test

Thus other methodautomatically estimate quality objectidered This can be done intrusivelyby comparing the reference signal with the degraded signaely utilisingphysical quality parameters or theed signal without using the reference signalapplicability of any solution for measuring the speech quality in VolP networks shouldake into consideration the nature of ip networks and the characteristics of voice trafficAmong the desired features for a VolP speech quality assessment solution are1 Automatic: It should providespeech quality online while the network isrunning2 Non-intrusive: It should be able to providement of the speech quality dependingspeech signal or network parameters without the need for the originaAccurate: It should provide accurate measurement of speech quality to reflect how thequality is perceived by the end4 With the changing world, it should be applicaband emerging applications andking conditions As such it should avoid the subjectivity in estimating parametershe E-model (section 42) for example depends on subjective tests to estimate packet

VolP Quality Assessment TeclSpeech Quality AssessmentlibrationObiective MethodsParametric-basedFig 1 Overview of VolP measurement methods(Sun, 2004)Based on the above requirements and from the previous ditrusive solutions that will be discussed in sections 3 and 4such task as they are manual and intrusive, respectivelylutions that will be discussed in section 4

2 antes forrusive subjective solutions for measuring(ITU-T, 2004)and the E-model (ITU-T, 20093 Subjective assessment of qualityTU-T Recommendation p 800 in whichtest subjects) perform the subjectiests of voice quality and give their opinions on the quality(ITU-T, 1996b) Subjective testsand thearithmetic mean of these opinions is calculated In listening tests, one subject is listening topre-recorded sentences(ITU-T,1996bTo conduct a subjective experiment according to the ITu-T Recommendation P 800, strict labthan 20 m In case of recording them must be a volume of 30to 120 m with an echo duration lower than 50000-300 ms is preferred) and a background noise lower than 30 decibel(dB) The recordingshort phrases, taken from newspapersth Alsehe sound pressure level should be measured from a vertical position above the subjects seatRecommendation P800 also specifies other conditions regarding the subjects who participatethe test such as they have not been directly involved in work connected with assessment of

echnologiesof telephone circuits, or related work such as speech coding, also they hayn any subjective test whatever for at least the previous six months, andheardsentence lists before(ITU-T, 1996bategory Rating, ACR) or relative to the subjective qualiIn as inry Rating, DCR), or Comparison Category Rating(CCR)(ITU-T, 1996b)on rating is Mean Opinion Score(MOS)which is an ACRmetric with five-point scale: (5)Excellent, (4)Good, (3)Fair, (2)Poor, (1)Bad (ITU-T, 1996b)MOS is internationally accepted metric as it provides direct link to the quality as perceived)for a set of subjects When the subjective test is listening-only, the resultermsstening subjectiveity; ie MOS- Listening Quality Subjectiernational the resultuality; ie MOS-Conversational QualityAlthough the overall quality of VolP must be discussecstening qualityhelpful in analysing the effect of individual qualityfactors such as distortion due to speech coding and packetIn DCR test two samples(A and B)are present: A represents the reference sample with thethe A sample according to the following five-point degradation category scale: degradation is2(4)audible but not annoying

(3)slightly annoying,(2)annoying, and(1)veryannoying The samples must be compothen samplThe results (opinions) are averaged as Degraded MOS(DMOS) Each configuration isjudgements on speech samples from at least four talkers DCR testitivity and used with high-quality voice samples, this is especially uswhen the impairment is small andof therequired asCR is inappropriate to discover quality variations as it tends to lead to low sensitivitydistinguishing among good quality circuits(ITu-T, 1996b; Takahashi et al-, 2004)The CCr method is similar to the dCr method as subjects are presented with a pair of speechample(A)followed by the degradedB) In the DCR method, listeners althe amount by which sample B is degraded relative to sample A In the CCr procedure, therder of the processed and unprocessed samples is chosen at random for each trial, On halftrials, the order is reversed Listeners use the following scale: (3)Much Better, (2)Better,Slightly Better,(O) About the Same, (-1)Slightly Worse, (-2)Worse, and (-3) Much WorseTU-T, 1996b) In this technique listeners provide two judgements with one respoCCR method over the dcrthe possibility to assess speechprocessing that either degrades or improves the quality of the speech The quantity evaluatedrepresented as Coron MOS(CMOSResults of mos scores shouldhardware platform, listening groups and test data and slight variation between differentabjective tests should be expected although the above rigid conditions should guarantee

VolP Quality Assessmentmethodshe most famous subjective quamethodology, but other methods have also been proposed Diagnostic Rhyme Test (RT)here the subject task is to recognise one of two possible wordsa set oDeat) Diagnostic Acceptability Meare based on results of teshods evaluating the quality of aunicationn the acceptability of speech as perceived by a trained normative listener (Spanias, 1994)Liparameterused mospinion rating methodstill the most famousand widely used methodAlthough subjective quality measureris the most accurate and reliable assessmetthere are fted with subjectivefrom the strictrating methods as mentioned above that the inherebjective MOSare that it is: time-consuming, expensir meeting the demandt in communication networksavoid the limitations of the subjective tests

ve speech qualityulates the opinions of human testers algorithmicallyto the rating ol subjects, where the aim is to predict MOS values thatclose as possible to the rating obtained from subjective test and to avoid thebjective assessment methods However, as subjective methods are the most accurate andreliable methods for meg speech quality, they are used to calibrate objective methodsTherefore thecy, effectiveness and performance evaluation of objecd on objective metrics of speech signal ornetwork, Objective qualitytegorised into two groups: Intrusive speech-layer models andIntrusive models(Signal-based and parametric-based ) Figure 2 shows the three main types of objective41 Intrusivctive assessment ofas input-to-outpun-basedtorted or output) speech sity or speech-layer objective models are full-reference methods for meaThey provide an accurate method for measuring speech quality as they require the original orreference speech signal as input and produce moS by comparingthe post-transmitted signal withginal one(double-ended) using a distancethe quality of the original signal However, such methods are inapplicable arison wionitoringtraffic because it is difficult or impossib

NTECOPEN ACCESSUBLISHERNTECH openree online editions of intecBooks and journals can be found atwww

intechoper

ontentsVolP Quality Ass 1nek becvar

lukas novak and michal vondraand Voice Quality Assessment 45Filipe TavaresChapter 4 An IntroductiVoIPnd-to-End Elements and oos Parameters 79Estrada-Vargas and D Torres-Rs of clMechanisms on volesQos Properties 95VolP System for Enterprise Network 12mmbedded3ortoles-ComerasJaume Nin-Guerrero and Josep Mangues-Bafalluy

Chapter 10 VolP Features Oriented Uplink SchedulingScheme in Wireless NetworksChapter 11 Scheduling and Capacity of volP Servicesreless OFDMA Systems 237Reliable sessioiation Protocol 253Chapter 13 Multi-path Transmission, Selectioand Handover Mechanigh-Quality volP 277nd-to-End Handover Management for vCommunications in Ubiquitous Wireless Networks 295Shigeru Kashihara, Muhammad Niswar, YKazuya Tsukaguru Yamaguchi and Yuji Oig New Approach

PrefaceVoice over IP(VolP) is undoubtedly ative communication toolCompared with the public switched telephone network(PSTN), VolP offers the benefitcation and infrastructure cost This makes it possible for everyne to easily keep in touch with family, friends, and clients around the world Ftthermore volp has theating with varplications, such as Web systems, presentation software, andhe near future, we expect VolP to provide a rich multimediaHowever, voice communication over the Internet is inherently less reliable than PSTNntially works as a best-effort network without a Quality of ser-vice(QoS) guarantee, andoice data cannot be retransmitterfer from packet loss, delay, and jitter due to interference from other data packetswireless media, including reduction of signal strength and radio interference

Additionally, we need to pay closer attention to security issues related to VolP communications Thus, VolP technologies are challenging research issuesThis book comprisedee of volP research, fromfuch of thefocused on the kdenhaapproaches that illustrate how VolP aspires to be a powerful and reliablenication tool We hope that you will enjoy reading these diverse studies, and will findlot of useful information about VolP technologies Finally, I would like to thank allthors of the chapters for their great contributionsShKashiharaNara Institute of Science and Technology