This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Trend Communications is an international company supplying hand-held test equipment and on-line monitoring systems to the communications market. Trend’s solutions are intended to cover businesses involved with broadband access, voice, datacom, network management, photonic transmission, metropolitan and mobile networks.
Trend has always been at the forefront of the communications test market, and our strength is based on the robustness and high quality of our products. Our solutions combine excellence and high technology with ease of use, covering such technologies as Triple Play, xDSL, 3G/UMTS, ISDN, IP, Carrier Ethernet, NG-SDH/SONET and NGN.
At Trend our mission is to be the preferred supplier of Field-Deployable Testers through innovative design and cost leadership.
Trend Communications is a subsidiary of IDEAL INDUSTRIES, INC.
Triple play is a business concept; a bundle of services rather than a completely new development.
1. Triple play is not a new technology, but a marketing concept for delivering three services: broadband access, television and telephone services over a single access network.
2. If mobile services are included, the bundle is often referred to as Quadruple Play.
3. There are two concepts closely related to triple-play:• • • • Service bundling: all the services are bundled into a commercial product.• • • • Technological convergence: one network supports voice/data/video applications.
4. Triple play can be delivered over various network types - copper, fibre, coaxial and wireless.
5. Inter-operability is not a requirement, but IP is at the heart of every implementation.
Telecom operators are embracing a new strategy to deliver new, thrilling services by means of next generation networks. This packet of services includes line rental and fixed telephony with a combination of Internet access, IP television, video-on-demand, entertainment applications and, eventually, cellular phone services.
Triple Play is not only a set of multiple information flows, but it is a way to make a wide range of devices and terminals manage data, audio and video applications.
Advances in technology and new regulations have made ISPs, Cable and Mobile operators competitors of telcos in voice and data access services. So, companies that were originally in different markets, are now all racing to bundle and offer the same services, using their own version of a converged network.
Cable OperatorTelecom Operator
DOCSISFDM
VoIP
Internet Provider
DSL
TV
DSL (and Fiber)
DSLVoIP
IPTV
IPTV
Mobile Operator
GPRS
3G
GSMPCM
VoIP
VoiceData
Video
VoiceData
AnyIMSHFC
Telcos, Cable, Mobiles and ISPs Become Competitors
More competition leads to:• • • • More segmentation: smaller scale than before.• • • • Less differentiation: to meet the same customer requirement due to convergence.• • • • Low entry barrier: more players, business cycle shorter.• • • • Price reduction: low margin.
Unfortunately for telcos, ISPs, cable and mobile operators are also offering phone services. The revenues of fixed telephony are declining, because mobile phones are so popular, and there is more competition now when cable operators also offer broadband access and voice services.
It is easy to understand that bundling has become a protective strategy for incumbent operators to keep in business by means of wireline access while for cable and competitive operators it is seen as a threat.
Telcos’ StrategiesFocus on urban and high-speed connectionsThose customers are more likely to contract new bundled services and stay loyal.
Cost is a key factorResidential customers are very sensitive to cost when contracting commodities such as telephony, TV and broadband access.
It is Video, can’t you see!Only Video-on-Demand is really new. Television is not, because there are many services based on broadcast, satellite and cable.
The Mobile ConvergenceThe “mobile vs. fixed lines” time is over. Integration of both worlds is strategic.
Keep it simple and reliableOne bill, one provider is probably less important than a reliable service: but it should be simple to manage by the customer and easy to maintain by the operator.
Think TankNetwork convergence makes it possible to provision any type of telecom service.
Subscribers• • • • Yankee Group (Aug. 2006): The US market has been calculated at 32 million annually, with an
average operator spending rate of about $ 4 000 per subscriber• • • • Pyramid research (2006): World market 35 million dollars by 2010
ARPU• • • • Heavy Reading (2006): ARPU can be increased by 100% when bundled services are running• • • • Gartner research (2006): Monthly european ARPU for fixed voice, Internet and TV is €93,70• • • • Fastweb (Italy) obtains an ARPU of €900 a year
Revenues• • • • Forrester (2006): Initial cumulative loss higher than €3 000 per subscriber
Although triple play strategies may start only with service bundling, migrating to an IP-centric converged network needs to be part of the strategy of the operators involved, in order to reduce the delivery costs and simplify the management structure.
3Gmultimedia
in/outdoors cells
Triple Play(voice, data, video)
Mobile bundle
Fixed bundleVoice
Broadband
Voice
SMS
TV
video
GSM+WiFi
VoD
Multiservice(TV, VoD, VoIP, Internet, Mobile...)
Multiaccess(copper, fiber, wireless)
One bill
One vendor
One network
Multiplatform(PC, TV, Mobile, Game Console)
Bundling
Convergence Support Provision
Competition, Target Customer, Segmentation, Timing, Cost, Tariff, Cultural
Latest improvements on DSL• • • • ADSL2: Real-time rate adaptation support and Inverse Multiplexing for ATM (IMA)• • • • ADSL2: Native support (no ATM) of packet-based services (for example Ethernet)• • • • ADSL2+: Higher bit rates than ADSL2, but less reach• • • • VDSL: Faster than ADSL, but it needs remote DSLAMs closer to the customers• • • • VDSL2: Matches Fast Ethernet rate at 100 m and has native support for packet-based services• • • • VDSL2: Asymmetric and Symmetric configurations, compatible with POTS and ISDN• • • • VDSL2: Needs remote DSLAM deployment and FTTN
• • • • Passive Optical Network (PON) is an optical technology for the access network, based only on passive elements like splitters. In a PON, the transmission medium is shared, and traffic from different stations is multiplexed. Due to the use of simple and inexpensive transmission elements and a shared medium, a PON is a cost-effective solution for the optical access network.
• • • • Active Ethernet is an alternative technology based on point-to-point optical links instead of a shared infrastructure such as PON. It can provide higher bandwidth per user than any other access technology, but it is also more expensive.
EFM interfaces provide low and medium speeds when compared to the available LAN or WAN standards. The new interfaces, however, are optimized to be profitable in the existing and newly installed provider access networks.
The access technology used depends on:• • • • Target bit rate or application to be rolled out• • • • Distance to the Central Office (CO) short distances enable xDSL; if longer then FTTx is a must• • • • Installed base, i.e. a lot of copper and good quality facilitate ADSL2+ and VDLS2 rollout• • • • Fiber availability close to the subscriber’s site makes FTTN installation easier• • • • Budget i.e. xDSL is inexpensive (if copper is available), fiber is not expensive, but digging is!
Native Ethernet has important drawbacks:• • • • Connectionless: This is often an advantage, but it limits QoS and requires constant address learning• • • • Privacy/efficiency: Switches and bridges use broadcasting for learning (IEEE 802.1d)• • • • VLAN limitations of 4 094 identifiers cannot be used in a WAN (IEEE 802.1q) • • • • Non-hierarchical MAC addresses are flat, so the switching table does not scale well• • • • It takes seconds to restore the Spanning Tree Protocol (STP). It cannot match 50 ms!• • • • No Ring topologies can be used, because STP allows only tree or star topologies• • • • Limited QoS, because native Ethernet is basically a best-effort technology• • • • Poor Management of nodes, topologies, events, performance• • • • Network Demarcation, the CPE and the Operator network must be separated clearly
MPLS manages traffic streams by separating route selection and packet-forwarding functions.
Pseudowire Edge-to-Edge Emulation PWE3 require a Tunnel label, used for guiding the frame through the MPLS domain, and a VC label, used to identify each customer’s traffic matching an MAC, Port or VLAN tag to a constant label.
Including VPLS (the PWE3 multipoint implementation) the Metro Ethernet network can provide easily:• • • • QoS to support triple play services• • • • Increased scalability overcoming the MAC address explosion issues• • • • Integrated protection architectures• • • • Advanced management
• • • • Provides routing, not QoS• • • • Overcomes many IP scaling problems• • • • Flexible and efficient, increasing the performance of IP networks• • • • Makes traffic isolation per customer or flow possible• • • • Transparent to QoS protocols
Therefore MPLS makes QoS provisioning easier, but using tools like DiffServ, RSVP, or ATM
Deploying Ethernet in MAN / WAN environments makes it necessary to develop new types of SDH
•••• Enhanced ADMs are like a traditional ADM, but they include Ethernet interfaces to enable access to new services, and TDM interfaces for legacy services.
•••• Packet ADMs have a configuration similar to enhanced ADMs: They include TDM and packet interfaces, but packet ADM offers common packet-based management for both new and legacy services.TD
Fundamentally NG SDH is a packet-enabled technology made possible by three elements:• • • • GFP (Generic Framing Protocol) is an encapsulation procedure for packet data, performing bit rate
adaptation, managing features such as rate adaption, priorities, channel selection and submultiplexing.
• • • • VCAT (Virtual Concatenation) is a mechanism that assigns granular bandwidth sizes rather than exponential provisioning of contiguous concatenation. This is why VCAT is flexible and efficient.
• • • • LCAS (Link Capacity Adjustment Scheme) modifies the allocated VCAT bandwidth dynamically by adding/removing members. LCAS is also being used to implement diversity for traffic resilience.
Based on resource reservation using end-to-end signalling• • • • Applications require resource management• • • • Resource reservation is done per flow by means of Reservation Protocol (RSVP) signalling• • • • Guaranteed service and controlled load for QoS-sensitive flows• • • • Source-to-destination packet handling at each hop and per each flow• • • • It’s not very scalable: RSVP is end-to-end and too complex• • • • Large packet processing and resource reservation makes RSVP inappropriate for core routers
Key QoS control is managed at the Ingress router• • • • IP packets are marked and classified into categories or DSCP• • • • In charge of packet access, shaping and policing
Core routers just forward packets• • • • Fast routing to the next hop (rather than end-to-end management like in RSVP)• • • • Packet scheduling per DSCP. No previous signalling, no resource reservation• • • • End-to-end QoS built with PHBs
DiffServ does not guarantee a QoS but manages flows differently• • • • Simple and scalable solution
Ingress Router
Egress Routers
Flow identificationPacket markingAccess control
Core RoutersTraffic schedulingPer hop forwarding
Data Flows
EF
AF1
AF2
AF3
AF4
BE
+ -
EF: Expedited ForwardingAF1-4: Assured Forwarding (x4)BF1: Best Effort (lowest priority)
The right architecture depends on many factors. Each business case must be evaluated independly:• • • • Environmental: competition, culture, affordability, consumer behavior• • • • Segmentation: niche, mass, residential, enterprise• • • • Differentiation: services bundle, content, language, premium or value price• • • • Technology: QoS with new & existing technologies and low network deployment costs• • • • Time to Market: when and where to launch• • • • Service Costs: leadership or premium
Triple Play applications are often a combination of several types of information, and a number of parameters such as bandwidth, source/destination relationship, type of routing, QoS, and symmetry
1. Video contribution, which may include television and video on MPEG demand applications. Contents is coded, formatted and streamed according the addressing scheme and the protocols to transport the signal and distribute the programs across the subscribers.
2. UDP is the most common higher-layer envelope to forward packets across the network.
3. Access network, audio an video packets reach the customer premises thought the access network which may be based on ADSL2, ADSL2+, VDSL2, FTTN, PON, or WiMax
4. Customer Premises, to reach the Set Top Box (STB) where are signals decoded and finally displayed on a TV or a PC. The Internet Group Membership Protocol (IGMP) is used for channel tuning.
Video Contribution Customer PremisesDistribution network
The IP television is a recent achievement thanks to the development of the following technologies
1. Carrier Ethernet, that can guarantee seamless Video streaming over converged networks
2. New architectures for IP network to support differentiated QoS for video applications, and allow a bidirectional or interactive service between the Content Provider and the Subscriber
3. Availability of a new generation of high performance IP routers and Ethernet switches
4. Evolution of First Mile technologies (xDSL, FTTx, Wireless) than can deliver several Mbit/s
5. Rich middleware software that can differentiate each IPTV service implementing options like video on demand, pay-per-view, VCR, multiple definitions, etc.
Video on Demand (VoD)• • • • Unicast Service• • • • Real-time QoS is not a must• • • • Pause, Stop, Backwards, Forward, etc. options controlled with RTSP protocol• • • • Rich middleware like subscription VoD, network video recorder and personal video recorder
IP Television• • • • Multicast Service• • • • Real-time QoS is required • • • • RTP & RTCP protocols for quality control• • • • Channel zap with IGMP
• • • • Digital Video is encoded with the help of MPEG-2, MPEG-4, or WM9.• • • • Content – where, who and how• • • • Cost – competitive• • • • Quality – viewing experience• • • • Convenience – shifted time TV, PVR• • • • Coverage – accessibility (fixed line or mobile)• • • • DRM – business model
Audio-visual Services and MPEGThe Moving Picture Experts Group (MPEG) is a working body within the ISO that is responsible for developing video and audio encoding, compression and standards for digital television delivery, IPTV, commercial advertisements and multimedia digital video applications.
1. MPEG-1 (1993), typical rates up to 1.856 Mbit/s• • • • Coding of audio/video for digital storage media• • • • Used in CD Video• • • • Video resolution, generally 352 x 240/288 • • • • MP3 is the audio draft of MPEG-1
2. MPEG-2 (1995), typical rates from 2 to 9 Mbit/s • • • • Rates generally around 4 Mbit/s with ADSL2+• • • • Video resolution generally 720 x 480, 720 x 576 or 544 x 576• • • • Used in Cable, DBS, DVD, VoD and HDTV• • • • When used with HDTV, MPEG-2 typically runs at 19.3 Mbit/s
3. MPEG-4 (1999), typical rates from 5 kbit/s to 10 Mbit/s• • • • Developed by the ITU to enable wireless single-user video services• • • • Mobile/POTS 5 kbit/s to 64 kbit/s• • • • Internet 64 kbit/s to 364 kbit/s• • • • Broadcast/VoD 364 kbit/s to 10 Mbit/s• • • • High efficiency for IPTV
Digital television requires that pictures be digitized so that they can be processed by computer hardware.
Each pixel is represented by:• • • • one luminance number, that describes the brightness• • • • two crominance numbers that describe the color of the pixel. • • • • 4:2:2 means crominance horizontally subsampled by a factor of 2 relative to the luminance, 4:2:0 the
factor is horizontal and vertically subsampled.
pixels
lines
Standard TV Digitalization (4:2:0 at 25 frames/s)Luminance: 720 lines x 576 pixels x 25 fr x 8 bits = 82,94 Mbit/s
25f/s
Crominance: 720 lines/2 x 576/2 x 25 fr x 16 bits = 41,47 Mbit/s
High Definition TV Digitalization1920 lines x 1080 pixels x 25 fr x 8bits = 1.49 Gbit/s
Compression is necessary to reduce the bandwidth requirements• • • • Lower SDTV: 24bit/pixel x 480x640pixels/frame x 30frames/s = 221,16 Mbit/s!• • • • Market moves toward HDTV therefore high efficient compression is necessary• • • • Pixel characteristics is correlated with neighbors then in some extend its value is predictable• • • • Human eye is less sensitive to detail near edges or around shot changes
A two dimensional DCT is performed on small block of 8x8 pixels. The magnitude of each DCT coefficient indicates the contribution of vertical and horizontal frequencies to the original image. Note that:• • • • DCT: Converts and image block into frequency component,• • • • DCT: does not reduce the size of the image, in fact increases it! from 8 bits/pixel to 11• • • • DCT tends to concentrate the energy into the low frequency coefficients matching eye sensitive• • • • The non uniform coefficient distribution is a result of spatial redundancy in the block• • • • The coefficient weight is done according human perception: high freq are coarsely quantized• • • • DC coefficient: is called when the value is 0• • • • The scanning and compression algorithm produces a variable length code• • • • The final coding results in a I-frame
641 41E 51E 4A2 F5 456 428 52
4A2 603 4A2 E39 23C 136 6 36
410 182 297 149 11 A1 2 1F
A7 49 10 1C 33 12 41 13
7F1 845 7F9 234 4F1 912 41 445
478 645 437 458 306 877 817 B3
282 564 252 52 67E 31C 1AE 11
134 7F1 541 349 156 52 3 21
25 1A 11 8 0 2 1 0
5 5 7 6 2 1 0 0
3 2 A 1 0 0 0 0
1 1 0 0 0 0 0 0
123 58 69 24 F 3 0 2
78 2E 21 6 1 4 3 1
1C 8 5 1 3 2 0 0
2 6 9 3 1 0 0 0
25 1A 11 8 0 2 1 0
5 5 7 6 2 1 0 0
3 2 A 1 0 0 0 0
1 1 0 0 0 0 0 0
123 58 69 24 F 3 0 2
78 2E 21 6 1 4 3 1
1C 8 5 1 3 2 0 0
2 6 9 3 1 0 0 0
Sequence of Images
Image Macroblock Block
Slide
16 pixel16
pix
els
8 pixels
8 pi
xels
DCT
8 bits/pixel
DCT coefficients8 bits/pixel 11 bits/pixel
Human eye is less
Weighting
sensitive to high freq.set 0 below perceptionand minimize high freq.
Compression
Zig/zag scan with detection of zerosHuffman encoding
I-Frames: Intra-frame coded independly to other pictures.• • • • Compression is achieved with DCT reducing the spatial redundancy but not temporal.
P-Frame: Predicted pictures can use previous I or P pictures for motion compensation• • • • Each block can either be predicted or intra coded
B-Frames: Bidirectional predicted pictures from previous or later I or P frames (never B-frame) for motion• • • • Each block can either be forward/backward/bidirectional predicted or intra coded• • • • Forward prediction requires to change the natural frame order causing a reordering delay at reception• • • • B-frames achieve the highest degree of compression, I-frames the lowest
A Group of Pictures (GOP) is described the number of pictures (N) and the spacing of P pictures (M) in our sample GOP N=12 M=3.• • • • In theory, the number of B-frames that may occur between any two I- and P-frames is unlimited• • • • In practice, there are typically up to twelve P- and B-frames occurring between each I-frame. • • • • One I-frame will occur approximately every 0.4 seconds during video showtime.
Two type of streams can be generated with the same source signal:• • • • Program Stream, intended media with low errors probability like CD-ROM• • • • Transport Stream, for noisy medias i.e. Satellites, IPTV uses shorter packets and independent clocks
• • • • PID=000: PAT - Programme Association Table, lists the PIDs of tables describing each programme.• • • • PID=001: CAT - Conditional Access Table, defines the type of scrambling used + management info.• • • • PID=X: PMT - Programme Map Table, defines the set of PIDs associated with audio, video, data...• • • • PID=010: NIT - Network Information Table, contains details of the bearer network used• • • • PID=Y: PES - Packetized Elementary Stream, each independent sequence of voice, video or data
4 bytes
PayloadHeader
184 bytes
Adaption field
Transport Scrambling ControlPID
Transport PriorityPayload Unit Start Indication
Transport error indication Sync byte
Adaption Field Control Continuity Counter
PID8 1 1 1 13 2 2 2
Transport
Transport
TP Header
AudioVideo
DataAudio
PAT table
- Network Info: PID=10- Program H: PID=306- Program X: PID=032- Program Z: PID=510
IGMP snooping is a method for intelligent forwarding of multicast packets within a layer-2 broadcast domain. IGMP registration information is snooped to create a distribution list of workstations to know which end-stations will receive packets with a certain multicast address.
Channel Zapping IGMP is a test used to measure the delay that occurs when a user joins or leaves a specific multicasting group. In other words, it is an IPTV channel zapping measurement.
Examples of TV Strategies• • • • Operator A: Targets high-end residential customers with busy life style for a premium
charge (4-10 times more than cable) for TSTV, VoD and HSI access.• • • • Operator B: Differentiates itself by delivering 25-50 Mbit/s VDSL2 bandwidth with
HDTV content at competitive rates to raise barriers to entry and strengthen its position in the market.
• • • • Operator C: Targets the mass market by using DTT for broadcasting, delivering content via VoD to use less bandwidth and make QoS control easier. This significantly reduces costs and risks in early deployment.
• • • • Operator D: Delivers multicast services at lower prices than cable for the mass market, with attractive content to boost market share and revenue for the time being.
• • • • Operator E: Targets specific sport fans for mobile TV during major sports events. Uses DVB-H technology to differentiate the service.
• • • • Operator F: Targets enterprise customers by delivering educational content for the campus via FTTx and LAN, to deploy the network at very low maintenance costs.
• • • • SIP is used at the control plane and RTP/UDP is used for the voice transport• • • • H.323 was the first, and still is the most used, easy internet-working POTS & ISDN
(but it is getting less popular)• • • • SIP, used for IP phones, is currently popular, as it is very flexible• • • • SIP can be integrated easily with PCs, e-mail, web and corporate platforms
IP is data-oriented, but can also support multiple applications based on voice and video.
Why VoIP?• • • • Subscribers: cheaper calls, integration with PCs and e-mail• • • • Carriers: convergence across a unique network• • • • Service providers: new business opportunity• • • • Manufactures: new market demands
VoIP uses known protocols such as:• • • • IP, TCP, UDP (User Datagram Protocol)• • • • RTP (Real Time Protocol), RTCP (Real-Time Control Protocol)• • • • SIP (Session Initiation Protocol), H.323 (ITU-T)
SIP is the protocol used to establish IP sessions between users, to set up VoIP calls, as well as multimedia conferences, multimedia distributions or multicast sessions. However, this protocol does not transport voice or multimedia contents.
What it covers:• • • • User search• • • • Call Init, Control and Close • • • • IP address, UDP/TCP • • • • Changes during the session• • • • Supplementary services
What it doesn’t cover:• • • • Type of network to be used• • • • Type of codecs to be used• • • • Session details (formats,
codecs...)• • • • Where and how the proxy,
registers, redirections etc. are implemented
Note that the caller does not send the Invite message directly to the recipient, but to an SIP proxy that locates the user and starts negotiating the session parameters.
Voice stream Packets ^ timestamps Packets with jitter
Retimed stream
IP
Clock Clock
RTP Voice stream
Hellooo!
Report ReportRRSR RTCP
Real Time Protocol (RTP) RFC 3550• • • • Used to transport voice and video signals
in real time• • • • Congestion produces jitter at the far end• • • • RTP inserts a timestamp in all voice packets • • • • Timestamps are used to ensure that all voice
packets that are delivered to the far end maintain the time that was originally generated.
Note that RTP does not provide QoS, but just transports timing signals.
Real-Time Control Protocol (RTCP)
RTCP complements the RTP protocol with information on the QoS received: delays, loss, jitter, etc. It provides:• • • • persistent session information• • • • basic session management• • • • performance feedback to communication
The problem of delivering QoS in packet networks can be compared with water distribution.• • • • In the diagram, two pumps supply water for two towns, Town A and Town Z.• • • • Some water is lost in the pipelines between the pumps and the towns.
The water company now needs to deliver water to a third town, Town M.• • • • A new pump could theoretically supply water for Town M.• • • • As more water is pumped in the pipelines, more water is lost before it reaches its destination.• • • • The result is that now Town M can get the water it needs, but there is not enough water for Town Z.
• • • • The water company solves the problem by installing a fourth pump.• • • • Now much more water is lost in the pipelines, but all three towns can receive the water they need.• • • • The ratio between water pumps and serviced towns is now 1.33 pumps / town.• • • • What would happen if it were necessary to deliver water to a fourth town?
Town A Town M
Over-provisioning
An extra pump is added
Much more water is lost
At least, the water supply cannow be guaranteed to town Z
• • • • There is an alternative to over-provisioning: fixing the distribution pipelines.• • • • Fixing the network can be more expensive than over-provisioning.• • • • The ratio between water pumps and serviced towns is now 0.66 pumps / town.
Triple Play is an application that runs over a large stack of telecom/datacom protocols. This means that bad quality of service or loss of service can be caused by many different factors:• • • • CPE faults• • • • Access faults, depending on the technology used• • • • IP networks must support proper QoS and multicast requirements• • • • Service availability and performance
DSL service providers can choose between two possible configurations for the local loop:• • • • Fixed data rate: The transmission bandwidth between the customer premises and the CO is fixed.
The transmission performance (SNR and noise margin) may change. It must be checked that the upstream and downstream rates match the configured values.
• • • • Fixed SNR: The local loop performance (SNR and noise margin) is fixed. Transmission rates may differ for each customer. It must be checked that the noise margin is 6 dB or better.
DSL service providers can choose between two possible configurations for the local loop:• • • • Optical TDR: check the physical conditions of the fiber, including continuity.• • • • Optical characterization: evaluate attenuation and absolute power level during transmission.• • • • Bandwidth: PONs use a multipoint-to-point topology; the more subscribers there are,
the more critical the system is.• • • • Efficiency: The PON is a shared medium; the scheduling performance must be checked.• • • • Security: Downstream signals are encrypted for all subscribers, to guarantee privacy.
• • • • During synchronization and training the modem and the DSLAM agree the upstream/downstream bit rates in the local loop.
• • • • The actual transmission rates can be smaller due to congestion in the network or in the remote system.• • • • The bit rate that the FTP client or server can send or receive depends on the local loop, the
transmission conditions of the whole network and the path between the client and the server.
The typical data applications (generally Internet-based) that need to be to checked are:• • • • Web browsing performance; a basic facility for residential customers• • • • FTP capacity for file uploading and downloading• • • • Traffic statistics compiled during data browsing• • • • PPP authentication
The QoE test aims is to measure how good the service is from the customer’s point of view. In practice it is a combination of packet impairments and video content measurements.
Quality of Service (QoS) Quality of Experience (QoE)
Content Quality- PCR Jitter- Coding distortion- Server overload
Video: Blocking, blurring, visual noise, loss of colour, edge distortion, pixelization, audio/video sync...
Transport Quality- Packet loss- Latency and delay variation- TCP Retransmissions
Voice: Distortion, noise, echo, loss of interactivity, interruptions, accessibility...
Video rendering is important for qualitative video performance assessment.
A single lost packet in an MPEG-2 video stream is displayed as several errored pixels or even lines in• • • • a video frame (spacial error propagation), • • • • several video frames with errored pixels (temporal error propagation).
The video preview test can be performed from different points in the customer premises or in the local exchange:• • • • The WAN connection is used to test the service provider network, but not the subscriber network.• • • • The LAN connection is useful to test the combined performance of the subscriber network and the
service provider network.• • • • The LAN Connection can be used to diagnose problems in the modem/router in the customer
premises due to firewalls, NAT, and unicast or multicast routing.
How to check the video IGMP delay:• • • • Channel zap, checks the delay in receiving the image when the channel is changed.• • • • Transmission: TV channels are transmitted in IP networks by using multicast IP datagram flows.• • • • Joining/Leaving an IP multicast group is managed by the Internet Group Management Protocol
(IGMP). Joining/leaving multicast groups may take time. The user sees this as excessive delays and degraded service.
Problems with NAT arise, because with SIP, there is some addressing information that is carried in the application payload. This information is bypassed by devices that only work at layer 3.• • • • SIP responses may fail to find the way back to the originator of the transaction if the “Via” or “Contact”
fields of the SIP requests cannot be resolved to a public IP address.• • • • The media transport protocol, usually RTP, may fail to find the participants of a session if they are
MOS (Mean Opinion Score): To arrive at an MOS score, a tester assembles a panel of “expert listeners” who rate the quality of speech samples that have been processed by the system under test. • • • • Ideally, a panel would consist of a mix of male and female listeners of various ages• • • • The samples should reflect a range of typical voice conversations• • • • Each panelist rates the quality of the system output from 1 to 5, with 1 indicating the worse and 5 the best• • • • The scores of the panelists are then averaged
E-Model: a computational model that uses transmission parameters (errors, packet loss, delay, echo...) to predict the subjective quality of voice. Good for conversational MOS evaluation using R-factor.
PSQM (Perceptual Speech Quality Measure), defined by ITU-T P.861, uses pre-recorded voice signals that are transmitted at the origin and compared at reception in the 300 - 3 400 Hz frequency range. Created to evaluate codec performance, basically the distortion of the voice signal. PSQM is not designed to reflect the effects of packet loss or jitter.
PAMS (Perceptual Analysis Measurement System) also compares an output signal with the input signal, but using a different algorithm based on factors of human perception to measure voice quality, scoring on a 1 to 5 scale that can be correlated to MOS.
PESQ (Perceptual Evaluation of Speech Quality), developed based upon PAMS and an improved version of PSQM called PSQM+. Uses the best features of both: the robust time-alignment techniques of PAMS with the accurate modelling of PSQM. It targets not only VoIP, but also ISDN, GSM and POTS.
1. In lightly loaded packet networks, delay, delay variation and packet loss ratio can be kept relatively under control. However, under higher traffic loads, carriers must face a dilemma: increase the capacity of their networks further or implement QoS policies.
2. MPLS is perhaps the best QoS solution among the currently available options, including IntServ and DiffServ. It combines the features of Intserv and Diffserv with other features that are very much appreciated by carriers: Connection oriented technology, traffic engineering, carrier-class protection...
3. Providers must offer differential QoS both on a per customer basis and on a per service basis. Every service has its own QoS needs. Providing QoS on a per service basis is not just a problem of prioritization. The network must be prepared to fulfill the needs of every application.
4. QoS tests are performed as end-to-end measurements. For example, a tester can be used to measure delay, delay variation and packet loss, for several services/subscribers, from the customer premises.
5. Residential subscribers experience service deficiency rather than abstract QoS problems; because they buy services rather than transmission facilities. This makes QoE testing an important requirement of test equipment in residential applications.
Acronym ListACELP: Algebraic CELPADPCM: Adaptive Differential PCMADM: Add/Drop MultiplexerADSL: Asynchronous Digital Subscriber LineADSL2+: ultra-high-speed ADSLAPON: ATM PONARPU: Average Revenue Per UserB-Frame: Bidirectional Prediction FrameBPON: Broadband PONCAT: Conditional Access TableCELP: Code Excited Linear PredictionCPE: Customer Premises EquipmentCVSD: Continuously Variable Slope DeltaDiffServ: Differentiated ServicesDOCSIS: Data Over Cable Service Interface SpecificationDSCP: Differentiated Services Code PointDSLAM: Digital Subscriber Line Access MultiplexerDTT: Digital Terrestrial TelevisionDVB: Digital Video BroadcastDWDM: Dense WDMEPON: Ethernet PONES: Elementary StreamESCON: Enterprise Systems ConnectionEVC: Ethernet Virtual Connection EVPL: Ethernet Virtual Private Line EVPLAN: Ethernet Virtual Private LANFR: Frame RelayFTTB: Fibre To The BuildingFTTC: Fibre To The CurbFTTH: Fibre To The HomeFTTN: Fibre To The NetworkFTTP: Fibre To The PremisesGEM: G-PON Encapsulation ModeGFP: Generic Framing ProtocolGPRS: General Packet Radio ServiceGPON: Gigabit PONGSM: Global System for Mobile communicationHDTV: High Definition TVHFC: Hybrid Fibre/Coaxial network HSDPA: High Speed Downlink Packet AccessI-Frames: Intra Frame
IGMP: Internet Group Membership ProtocolIntServ: Integrated ServicesIPTV: IP TelevisionISP: Internet Service ProviderISUP: ISDN User PartLAN: Local Area NetworkLCAS: Link Capacity Adjustment SchemeLLC: Logical Link ControlLPC: Linear Predictive CodingLSP: Label-Switched PathLSR: Label-Switched RouterMAC: Media Access LayerMGC: Media Gateway Controllers MGCP: Media Gateway Control ProtocolMOS: Mean Opinion SquareMPEG: Moving Picture Experts GroupMPLS: Multiprotocol Label SwitchingMSPP: MultiService Provisioning PlatformMSTP: MultiService Transport PlatformMSSP: MultiService Switching PlatformNAT: Network Address TableNG SDH: Next-Generation SDHNGN: Next-Generation NetworkNIT: Network Information TableOAM: Operation Administration and MaintenanceOLT: Optical Line TerminationONU: Optical Network UnitOSPF: Open Shortest Path FirstOTN: Optical Transport NetworkP-Frame: Prediction FramePAMS: Perceptual Analysis Measurement SystemPCM: Pulse Code ModulationPCR: Program Clock ReferencePES: Packetized Elementary StreamPESQ: Perceptual Evaluation of Speech QualityPHB: Per Hop BehaviorPMT: Programme Map TablePPP: Point to Point ProtocolPON: Passive Optical NetworkPON: Private Optical NetworkPSTN: Public Switched Telephone Network
PSQM: Perceptual Speech Quality MeasurePOTS: Plain Old Telephone SystemPWE3: PseudoWire edge-to-edge EmulationQoE: Quality of ExperienceQoS: Quality of ServiceR-Factor: 0 to 100 ratio for voice quality (ITU G.107)RPE-LTP: Regular Pulse Excitation - Long Term PredictionRSVP: Resource Reservation ProtocolRTCP: Real Time Control ProtocolRTP: Real Time ProtocolSAN: Storage Area Network SDH: Synchronous Digital NetworkSDTV: Standard Definition TVSLA: Service Level AgreementSNT: Signal to Noise RatioSIP: Simple Internet ProtocolSMS: Short Message ServiceSTB: Set Top BoxSTUN: Simple Traversal of UDP through NATPAT: Programme Association TableTP: Transport PacketsTS: Transport StreamTSTV: Time Shift TelevisionUTP: Unshielded Twisted Pair cableVC: Virtual ConcatenationVPN: Virtual Private NetworkVDSL: Very High Bit rate DSLVLAN: Virtual LANVPLS: Virtual Private LAN Service VPWS: Virtual Private Wire ServiceVoD: Video On DemandVoIP: Voice over Internet ProtocolVSELP: Vector-Sum Excited Linear PredictionWDM: Wave-Division MultiplexingWiFi: Wireless FidelityWiMAX: World-wide Inter operability for Microwave AccessWIS: WAN Interface SublayerWLAN: Wireless Local Area NetworkWM: Windows Media