DISPATCH WG B. Khasnabish Internet-Draft ZTE USA, Inc. Intended status: Informational G. Fernando Expires: January 4, 2014 Compression Labs, Inc. L. Ya ZTE Corporation July 3, 2013 End-point based Multimedia QoE Management draft-khasnabish-dispatch-qoe-management-01 Abstract This draft describes a method for improving the quality of experience (QoE) for real-time video and other multimedia services using features and functions of the end-point only, that is, without requiring any upgrade to the network transport infrastructure. Any upgrade to the network transport infrastructure not only incurs significant costs, these are also time consuming and technology- dependent. Therefore, these QoE improvement mechanisms are significantly more attractive to both network operators and service providers. Status of this Memo This Internet-Draft is submitted in full conformance with the provisions of BCP 78 and BCP 79. Internet-Drafts are working documents of the Internet Engineering Task Force (IETF). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet- Drafts is at http://datatracker.ietf.org/drafts/current/. Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress." This Internet-Draft will expire on January 4, 2014. Copyright Notice Copyright (c) 2013 IETF Trust and the persons identified as the document authors. All rights reserved. This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents Khasnabish, et al. Expires January 4, 2014 [Page 1] Internet-Draft End-point based Multimedia QoE Management July 2013 (http://trustee.ietf.org/license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Simplified BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Simplified BSD License. Khasnabish, et al. Expires January 4, 2014 [Page 2] Internet-Draft End-point based Multimedia QoE Management July 2013 Table of Contents 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 4 1.1. Method of controlling QoE for real-time video and other multimedia services . . . . . . . . . . . . . . . . 4 2. Conventions used in this document . . . . . . . . . . . . . . 5 3. Preliminary Survey of existing QoE methods . . . . . . . . . . 5 3.1. Measurement Techniques for Network Quality . . . . . . . . 5 3.1.1. Network-based measurements . . . . . . . . . . . . . . 6 3.1.2. Content Inspection . . . . . . . . . . . . . . . . . . 6 3.1.2.1. Audio/Video Content Analysis . . . . . . . . . . . 6 3.1.2.2. Other Analysis Techniques at the Bitstream level . . . . . . . . . . . . . . . . . . . . . . 6 3.2. Metrics Used for QoE Assessment . . . . . . . . . . . . . 6 3.2.1. Direct Metrics . . . . . . . . . . . . . . . . . . . . 7 3.2.2. Indirect Metrics . . . . . . . . . . . . . . . . . . . 7 4. Proposed Toolkit for Managing QoE . . . . . . . . . . . . . . 8 4.1. Multi-layer Elastic Virtualized Buffer Stack . . . . . . . 11 4.2. Jitter Buffer . . . . . . . . . . . . . . . . . . . . . . 11 4.3. Error resilience and concealment methods . . . . . . . . . 11 4.4. Transport rate clinging . . . . . . . . . . . . . . . . . 12 4.5. Transport rate adaption . . . . . . . . . . . . . . . . . 12 4.6. Endpoint Resource Reallocation . . . . . . . . . . . . . . 12 4.7. Adaptation . . . . . . . . . . . . . . . . . . . . . . . . 14 4.8. Virtualized displays . . . . . . . . . . . . . . . . . . . 14 4.9. Use of trick play methods . . . . . . . . . . . . . . . . 14 4.10. Synchronization in multicast . . . . . . . . . . . . . . . 14 5. Security considerations . . . . . . . . . . . . . . . . . . . 15 6. IANA considerations . . . . . . . . . . . . . . . . . . . . . 15 7. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . 15 8. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . 15 9. References . . . . . . . . . . . . . . . . . . . . . . . . . . 15 9.1. Normative References . . . . . . . . . . . . . . . . . . . 15 9.2. Informative References . . . . . . . . . . . . . . . . . . 15 Appendix A. Appendix . . . . . . . . . . . . . . . . . . . . . . 16 A.1. Factors Influencing QoE . . . . . . . . . . . . . . . . . 16 A.1.1. Service . . . . . . . . . . . . . . . . . . . . . . . 16 A.1.2. Content . . . . . . . . . . . . . . . . . . . . . . . 16 A.1.3. Terminal Device Used for Consuming Content . . . . . . 16 A.1.4. View/Listener Characteristics . . . . . . . . . . . . 16 A.2. Range of Network Architectures . . . . . . . . . . . . . . 17 A.2.1. Managed vs. Unmanaged Networks . . . . . . . . . . . . 17 A.2.2. Unicast vs. Multicast Delivery . . . . . . . . . . . . 17 A.2.3. Range of Service Offerings . . . . . . . . . . . . . . 17 A.2.3.1. Traditional Closed Network operators . . . . . . . 17 A.2.3.2. Over-the-Top Service Offerings . . . . . . . . . . 17 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 18 Khasnabish, et al. Expires January 4, 2014 [Page 3] Internet-Draft End-point based Multimedia QoE Management July 2013 1. Introduction This Internet Draft describes a method for improving the quality of experience (QoE) for real-time video and other multimedia services without requiring any upgrade to the network transport infrastructure. It is recognized that network quality of service (QoS) based management of multimedia services has a very valid role. However, such management techniques are orthogonal to endpoint based QoE management, and thus are outside the scope of this Internet Draft. This Internet Draft defines a common set of QoE parameters that are applicable for HTTP, Websocket or RTP sessions. Hence, these parameters would be used for controlling the QoE for any real-time video and other session-based multimedia services over an Internet- protocol (IP) based network. The multimedia session may be displayed on a mobile communication terminal, a tablet/phablet, a laptop, a television, etc. This method facilitates a high-quality user experience with minimal user input for interactively controlling the QoE. Any user input may be through a touch sensitive device or verbal commands. This method may be applicable for improving QoE for both multicast and OTT (over the top) service scenarios, and may apply equally to transit/intermediate network elements. Depending on the particular protocol (RTP/UDP vs. HTTP/TCP) different tools from the toolkit may be applied. 1.1. Method of controlling QoE for real-time video and other multimedia services Data transport protocols for the Internet have been designed to be agnostic to the data type. There has been minimal attention given to the requirements for time-critical and high priority services, such as real-time communications, etc. This has lead to such services providing unpredictable and inconsistent user experiences, and thereby leading to lost revenue for the service provider. Network architects have often used quality of service (QoS) as the metric for objectively determining the level of service to be provided. From a customer perspective the QoS measure is not satisfactory as it doesnOt take into consideration the subjective experience; instead it is necessary to measure the quality of experience (QoE). QoE is the user experience that can be quantified in terms of the subjective experience, and it depends on various factors, including the type of content, expectations of the consumer of the content, the device on which the content is -consumed- and the environment in which the content is consumed. Admittedly, these are Khasnabish, et al. Expires January 4, 2014 [Page 4] Internet-Draft End-point based Multimedia QoE Management July 2013 indeed subjective measures. However, there are standardization efforts (in particular in ITU-T) to evaluate and quantify QoE as there is a real need for such quantitative metrics by service providers. This Internet Draft proposes a method for transport impairment independent control of QoE for multimedia services. The display of the multimedia session could be in a mobile communication terminal, a tablet, a laptop, a television, etc. This proposal relates to facilitating high-quality user experience with minimal user input (could be through a touch sensitive device or a verbal command) for interactively controlling the QoE. This method is applicable to operator managed services and to unmanaged (i.e., OTT - over the top) service. Additionally, this may apply to intermediate network elements as well. The focus of this proposal is for the QoE management to be transport independent, and hence it is applicable to a range of media transports, including HTTP, WebSocket or RTP sessions. In order to improve quality it is common practice to focus on improving network infrastructure (bandwidth increase, etc.). However, our proposal enables improvements in quality by utilizing transport independent QoE management. 2. Conventions used in this document The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in RFC-2119 [RFC2119]. In this document, these words will appear with that interpretation only when in ALL CAPS. Lower case uses of these words are not to be interpreted as carrying RFC-2119 significance. 3. Preliminary Survey of existing QoE methods 3.1. Measurement Techniques for Network Quality Currently there are two measurement techniques: (a) Network-based Measurement, and (b)Content Inspection. ITU-T and other standards bodies have been developing standards to address QoE measurement in the area of multimedia/IPTV [2], [3]. Below, these measurement techniques have been categorized. These techniques vary according to the performance, complexity and Khasnabish, et al. Expires January 4, 2014 [Page 5] Internet-Draft End-point based Multimedia QoE Management July 2013 feasibility. It is left to service providers and network operators to determine which of these to be used, and how to combine these according to different needs, multimedia content type, network, end user equipment and service type. 3.1.1. Network-based measurements These are objective network metrics which thus do not require accessing and inspecting the content. These metrics (one-way delay, packet loss and jitter) are uses in QoS assessment as well. However, in the context of QoE the intent is to predict multimedia quality based on these metrics. 3.1.2. Content Inspection Content inspection is applied at various levels, and these are described below. 3.1.2.1. Audio/Video Content Analysis There are objective and subjective content analysis techniques that are used to determine the distortion that may be introduced at a given stage in the end-to-end multimedia delivery system. These content analysis techniques may be based on the following approaches: (a) Non-reference (NR)-- the NR approaches assess quality without knowledge of the original source, (b) Reduced-reference (RR) -- the RR approach uses an alternative channel to send parameters corresponding to the delivered content, thus enabling evaluation of content quality, and (c) Full-reference (FR) -- the FR method requires the original content to be available in order to assess quality. Hence, this approach is not practical for most delivery scenarios. 3.1.2.2. Other Analysis Techniques at the Bitstream level By inspection of the bitstream it is possible to extract the following parameters from the headers that can be used to assess quality: (a) Timestamp accuracy/error, (b) PCR jitter/clock accuracy, (c) Media Encapsulation and Packetization (MEP) Header. The above parameters require deep packet inspection, but they do not require decoding of the audio/video sequences 3.2. Metrics Used for QoE Assessment In this section we shall provide details of metrics for QoE assessment that are currently used in the industry. In [4] the metrics used for QoE assessment have been categorized as being either -direct- or -indirect-. These will be described below. Khasnabish, et al. Expires January 4, 2014 [Page 6] Internet-Draft End-point based Multimedia QoE Management July 2013 3.2.1. Direct Metrics These metrics directly affect the user's perception of the audio/ video experience. peak signal to noise ratio (PSNR), structural similarity (SSIM) and video quality metric (VQM) are three commonly used objective metrics for quality analysis. For subjective analysis the commonly used methodology is the ITU-T Recommendation BT500 [5]. ITU-T have defined the mean opinion score (MOS)[6] which is a meta- metric which uses values form several other metrics. Due to the cumbersome nature of subjective analysis techniques there has been much work done to map objective metrics to MOS, for example, there's a mapping of PSNR to MOS. 3.2.2. Indirect Metrics There are indirect metrics which do not directly correlate to the multimedia content quality. Instead they correspond to artifacts that determine the user's satisfaction of the multimedia service. These indirect metrics include the following. o Responsiveness to user requests -- time delay from user request to response by the system. Example -- responsiveness to pausing video. o Start-up time -- time delay from user requesting content to receiving the content. o Delivery synchronization -- when several users access a given content do they all receive the content at the same time? o Live content delay -- time delay from content being available at the encoder to it being received by the user. For live streaming this is an important requirement. This is applicable to progressive downloading as well. o Picture freezing -- due to mismanagement of buffers. One example is buffer underflow. o Multimedia synchronization -- the accuracy at which audio, video and associated data (including subtitles) are synchronized has a significant impact on the QoE. For example, if audio is earlier or later than video by 160ms, then this will lower the QoE. o Content/Channel Switching time -- time delay from the user requesting content/channel switching to the display of the requested content. Khasnabish, et al. Expires January 4, 2014 [Page 7] Internet-Draft End-point based Multimedia QoE Management July 2013 o Blocking artifact -- consistency of the data in one frame. o Mosaic -- if packet is lost during transport, and it can not be recovered by the client, mosaic will appear. 4. Proposed Toolkit for Managing QoE We propose a toolkit for managing QoE for a multimedia system. P.NBAMS [2] is a quality assessment model that uses bitstream information in addition to prior knowledge of the media stream as well as client buffer information. P.NAMS [3] is another quality assessment model that uses only prior knowledge of the media stream as well as the client buffer information. In both of these Recommendations Figure 1e illustrates the "Embedded operation mode" where QoE management is performed at the client/end-point. Our proposal similarly aims to perform QoE management at the client/ end-point and thus it can be described as being performed in a Transport Independent Fashion (XIF). Once a session starts with a specific transport bandwidth, the endpoint's objective is to maintain persistently a consistent QoE even when the session bandwidth fluctuates. The smartness or intelligence resides at the endpoint, and the use of virtualization makes the simplistic implementation of the operation a technically feasible one. This is shown in Figure-1 Khasnabish, et al. Expires January 4, 2014 [Page 8] Internet-Draft End-point based Multimedia QoE Management July 2013 +------------------------------------------------------------------+ |/----------\ Endpoint management agent | || |+---------------------------------------------------+ | || || QoE/MOS ctl | | || +---------------------------/ | | ||Present- || | | | || ation ||QoE/ | | | || agent ||MOS +-----------+ +-----\------+ /----------\ | | || ||ctl | Decoding | | QoE/MOS | | | | | ||(Audio, +------+ and | | | | Storage | | | || Video, || | control +--+ model +---+ | | | || Text, +------+ | | | | | | | || etc.) ||Final+-----+-----+ +-----+------+ \----+-----/ | | || ||media |Refined media | | MRD/ | | || ||stream |streams | | RPP | | || || +------+--------------+---------------+-----+ | | |\----------/| | Physical/Virtual buffer and QoE/MOS | | | | | | management agent (with a multitude of | | | | | | tools in toolkit | | | | | | *Multilayer elastic virtualized buf. stack| | | | | | *Jitter buffer | | | | | | *Error correction and concealment | | | | | | *Transport rate clinging | | | | | | *Transport rate adaptation | | | | | | *Endpoint resource allocation(among audio,| | | | | | video, message, etc.) | | | | | | *Depth adaptation | | | | | | *Virtualized display | | | | | | *Synchronization in multicast | | | | | | *Trick-play methods(Slow/Fast, | | | | | | Look-ahead/Reverse, etc. | | | | | +------------------/------------------------+ | | | +-----------------------|---------------------------+ | +------------------------------------|-----------------------------+ |Raw |media |stream _,..\..,, _-` `', / \ | Network | | | \ / `-, ,-` ``''--'`` Figure 1: QoE management at the client/end-point in a Transport Khasnabish, et al. Expires January 4, 2014 [Page 9] Internet-Draft End-point based Multimedia QoE Management July 2013 Independent Fashion (XIF) Media content is received over an IP based network. However, it is expected that the actual transport would include any of the following -- RTP/UDP, HTTP/TCP, HTTP/UDP, etc. HTTP/UDP is relevant if the roundtrip link condition is poor or RTT (round-trip delay time) is long, etc. A QoE Management Agent in the endpoint will receive the media content. The management agent has a toolkit of "tools" which will be used to maintain the QoE. This toolkit includes the following tools o Multi-layer Elastic Virtualized Buffer Stack o Jitter Buffer o Error resilience methods o Transport Rate Clinging o Transport rate adaption o Endpoint Resource Reallocation (among audio, video, messaging, etc. services) o Adaptation, e.g., depth for image, surroundings of sound, etc. o Virtualized Display o Use of trick play methods o Synchronization in multicast Each of these shall be described in detail further down in this section. However, it is noted that there may be other tools which could also lead towards improving QoE. Any such tools may also be added as appropriate. With the aid of these tools the QoE management agent will maintain the QoE based on policy as set for the given endpoint. Resource pre-positioning policy information will be available in the local database. The QoE model will be applied to determine the QoE, and the metrics will be signaled to the presentation agent. In parallel with QoE evaluation, the media which has been "refined" by the QoE management agent is sent to the decoding and control agent where the content is decoded. The decoded data is then sent to the presentation agent. The decoding and control agent receives QoE signal information from the presentation agent. With this information the decoding and control agent is able to apply any remedial steps in order to maintain QoE. Khasnabish, et al. Expires January 4, 2014 [Page 10] Internet-Draft End-point based Multimedia QoE Management July 2013 4.1. Multi-layer Elastic Virtualized Buffer Stack Virtualization of buffer allows increase of the effective available buffer space (depth) over the amount that is actually physically available. This can be achieved, for example, by assigning temporarily larger than actually required physical buffer/memory chunk so that during process execution/operation a larger memory block can be temporarily used (must make sure that the allocation/ operations do not overlap or interfere with each other, as required by virtualization). This helps improve performance without extra costs. This is equivalent to vertical expansion of the buffer space. When "Elasticity" is added to the virtualized buffer, it allows incorporation of Elastic behavior (i.e. capability to expand and contract seamlessly in order to maintain a certain output rate) in the operation of the buffer module. This is equivalent to horizontal expansion of the buffer space. 4.2. Jitter Buffer Jitter is the variation of packet latency over a network. . The severity of jitter can vary significantly depending on network type and current conditions. Therefore, a device called jitter buffer is commonly introduced in the receiver. The jitter buffer can de-jitter the incoming stream of packets and provide a constant flow of data to the decoder. 4.3. Error resilience and concealment methods Packet loss is due to either random loss and network congestion. In first situation, error resilience methods including Automatic Repeat reQuest (ARQ) and Forward Error Correction (FEC) may also be used to maintain QoE. Unequal Error Correction (UEP) which is based on FEC may be more commonly used. The original data can be classified into several levels, the most important data has highest priority. Different priority has different redundancy. With network congestion, ARQ and FEC are not suitable as they could make the situation worse. Hence, different measures should be adopted according to network characteristics. When receiving multimedia communication payloads, the receiver may evaluate network statistics, including packet loss , consecutive packet loss, delay, jitter etc. from this data it is possible to determine if the network is congested. The receiver sends these parameters to the sender, and the sender can adjust the packet delivery strategy in real time based on these parameters: if the network is congested, sender can modify it's encode rate or send rate to decrease congestion, and make sure the protected data from FEC or ARQ is under the total rate. Khasnabish, et al. Expires January 4, 2014 [Page 11] Internet-Draft End-point based Multimedia QoE Management July 2013 Moreover, the network condition is unknown in the initial state of the communication, then sender may take its strategy supposing the network condition is poor. Packet Loss Concealment (PLC) is another tool that may be used for this purpose. The other method is using two or more encoders. When one frame is lost, the frame from another encoder can be used. It is important to process these two streams switchover. The delay of the encoders should be known for synchronization. Mute frame can be added to align the start time of frames. When switching between two streams, smooth the joins of different frames. 4.4. Transport rate clinging The intent of transport rate clinging is to maintain a steady rate of bit stream (e.g., media). The brute force method is to introduce blank packets in the event the required media data rate goes lower than a given threshold. Instead of using this brute force method, it is possible to reallocate the bandwidth to other sub-processes through appropriate virtualization. 4.5. Transport rate adaption During multimedia transport, the network may not be stable. The available bandwidth may be time-varing. In order to maintain a consistent QoE, transport rate should change according to the available bandwidth. If the transport rate is too high, packet loss will occur. Otherwise, bandwidth resource is waste. 4.6. Endpoint Resource Reallocation Endpoint resource reallocation involves dynamically adjusting virtualized resources among sub-sessions, audio, video, text/ messaging, etc. The aim is to ensure that these different types of sub-sessions maintain a consistent QoE. A flowchart for dynamically adjusting virtualized resources among sub-sessions (audio, video, text/messaging, etc.) is presented in Figure 2. +----------+ | Start | +----/-----+ | | | +---------------------------------\-------------------------------+ | | Khasnabish, et al. Expires January 4, 2014 [Page 12] Internet-Draft End-point based Multimedia QoE Management July 2013 | +--------+ | | |Local | +----------------+ | | |RPP |------Historic trend | | | | | |based session- |---> +---------------------+ | | +--------+ |specific res. | | |-------------------- | | | |allocation | | | Projected resource | | | +----------------+ | |requirements for next| | |+-------------+ |-->| few time intervals/ | | ||Res. utiliz. | | | periods for | | ||monitor | /-----------------\ | | maintaining | | ||(trick-play, | |Current resource | | | consistent QoE | | || ECC, | |utilization | | +---------------------+ | || rate adjust-| |trends & reqs. |--> | ` | || ment,...)| | for managing | | +---'--+ | || |-| QoE | | |MRD/ | | |+-------------+ \-----------------/ | |RPP | | | | +------+ | +-------------------------------------------------|---------------+ /--------------------------------\--/ | | | | \, \, .'` `., .'` ., ,-` ', ,-` ', .` `', ,.` Procure ', .'`Distribution of '`, -' resources from ',. ,-` is sufficient for ', NO ,`MRD/RPP for adjustment', -, the desired QoE ----->, for resource allocation . `', ,-` `', among sessions ,-` `'. ,-` `'. ,-` `'., .-` `'., .-` '., _.'` '., _'` `/' `/' | | YES | DONE | | | +------\-------------------------------------\----+ | Request for additional resources | | for maintaining overall QoE | | is fulfilled | | | +-------------------------/-----------------------+ | | /-----\------\ | END | | | \------------/ Khasnabish, et al. Expires January 4, 2014 [Page 13] Internet-Draft End-point based Multimedia QoE Management July 2013 Figure 2: A flowchart for dynamically adjusting virtualized resources among sub-sessions 4.7. Adaptation Adaptation (of image/picture depth) requirements may depend on the type of scene/picture, e.g., Talking head, sports, action movie, and so on. The number of GOPs/frames required to recover from any losses would depend on the type of scene/picture. 4.8. Virtualized displays The goal here is to manage the real-estate of the screen. Depending on the intended resolution and screen size at which a given piece of media content is generated there is some opportunity to maintain the QoE for, say, a lower sized display. 4.9. Use of trick play methods Various mechanisms from trick play can be utilized for slow/fast and Look-Ahead/Reverse adjustment of the scene. Such methods provide the means to catch up with "real-time" in the event that there has been a major hiccup in delivery. Voice recognition may be used to determine if catch up with Oreal-timeO is required. However, if network condition or the environment of the endpoint is too poor to guarantee the quality of voice data, the sender can generate text message from the actual voice by voice recognition module, and send text message to receiver. Network condition can feedback by packet loss, jitter, delay etc. The environment of the endpoint include voice output device, ambient noise, and so on. The receiver may choose to directly present the text message, or it can use a text-to-speech module to reconstruct the audio signal, and then present it. 4.10. Synchronization in multicast In multicast system, the server usually send I-frames or independently decodable frames periodically, the endpoint system display may be out of sync because of the different access times. A re-encode module can be added to the multicast system where it backs up the non-I frames from multicast server and re-encode them to new I frames, these new I frames use smaller bandwidth than the original I frames . When client join multicast group, it send request to server, the server can choose the latest I frame, send it and the following frames to client. Meanwhile, the server send the information to tell the client how to display, including the started time of the first I frame and the distance to the current frame which is send to other clients that already in the multicast group, then client can play these frames quickly until synchronize with other Khasnabish, et al. Expires January 4, 2014 [Page 14] Internet-Draft End-point based Multimedia QoE Management July 2013 clients. 5. Security considerations TBD 6. IANA considerations This Internet Draft includes no request to IANA. 7. Conclusions We have presented a tool-kit based approach for improving the quality of experience (QoE) for video and other multi-media services in a transport impairment independent fashion (XIF). This method is both flexible and expandable. As the technologies evolve, the existing techniques (in the tool-kit) can be updated, and many other new and improved techniques can be easily incorporated in the tool-box without incurring any significant overheads. 8. Acknowledgements The authors would like to thank Mary Barnes and Roni Even for encouraging discussions and guidance. The authors also wish to thank Romans Krzanowski for reviewing and providing comments on version 0 of the Internet Draft. 9. References 9.1. Normative References [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate Requirement Levels, BCP 14", March 1997. 9.2. Informative References [1] ITU-T, Rec., "P.NAMS, Non-intrusive parametric model for the Assessment of performance of Multimedia Streaming", November 2009. [2] ITU-T, Rec., "P.NBAMS, Non-intrusive bitstream model for the Assessment of performance of Multimedia Streaming", January 2011. Khasnabish, et al. Expires January 4, 2014 [Page 15] Internet-Draft End-point based Multimedia QoE Management July 2013 [3] Serral-Graci, A. and R. Barlet , "An Overview of Quality of Experience Measurement Challenges for Video Applications in IP Networks, 8th International Conference on Wired/Wireless Internet Communications (WWIC)", June 2010. [4] ITU-T, Rec., "Recommendation, BT 500", January 2011. [5] ITU-T, Rec., "P.800, Recommendation", August 1996. Appendix A. Appendix A.1. Factors Influencing QoE The QoE depends on several factors. This includes the type of service, type of content and the terminal device intended for use of the content. A.1.1. Service For any given entertainment service there is a specific minimum level of service expected by the viewer. This would determine the MOS. One could generalize and state that if a payment is made for an entertainment service then there is a much higher expectation of QoE than if it were to be a "free" service. In the free service category one would also include ad-driven services where the viewer's QoE expectations are lower. A.1.2. Content The type of content will influence the user's QoE expectations. The requirements for drama are different to those for sports content. The QoS parameters contributing towards the MOS would be different depending on the type of content. A.1.3. Terminal Device Used for Consuming Content The viewer's QoE expectations vary depending on the terminal device used for viewing the content. For example, the expectations of a viewer would be much higher if the content were being viewed on a 50- inch HDTV display as opposed to on a 4-inch smart phone screen. A.1.4. View/Listener Characteristics Each human view or listener can have a different impact on the QoE measurement. In order to make the QoE measurement practical from an operations perspective it is important to take this (impact of the Khasnabish, et al. Expires January 4, 2014 [Page 16] Internet-Draft End-point based Multimedia QoE Management July 2013 listener and viewer) out of the QoE measurement. This can be done by taking as wide a sample of viewers and listeners in the MOS calculation stage. [See ITU-T Ref.] A.2. Range of Network Architectures TBD A.2.1. Managed vs. Unmanaged Networks Today with network based media services there is a clear distinction made between how such services are provided depending on whether it is over a managed or unmanaged network. A managed network has the ability to configure, manage and monitor the network, thereby enabling greater control over how data is transported, and who has access to the data. It is also possible to generalize to some extent by stating that managed networks support reservation of bandwidth for each user. Unmanaged networks, by implication, do not have such reservation capabilities. A.2.2. Unicast vs. Multicast Delivery Unicast delivery services pose less of a problem with respect to maintaining QoE when compared to multicast delivery. The problems experiences with multicast delivery can be as complex as those for unmanaged networks. This document will focus on both forms of delivery -- unicast and multicast, as well as on managed and unmanaged networks. A.2.3. Range of Service Offerings A.2.3.1. Traditional Closed Network operators Traditionally, network operators, such as cable MSOs, satellite operators, "Telcos" and terrestrial operators have, controlled the networks over which services are provided. They either own the network infrastructure, or they have close business relationships with the owners of the network infrastructure. Starting from traditional (i.e., non-IP) delivery methods they have migrated to providing IPTV services. Other than the fact that IP and invariably UDP protocols have been introduced to the network stack there has been no fundamental shift in the service offerings. A.2.3.2. Over-the-Top Service Offerings More recently, companies like Netflix and Apple have started to provide media service to subscribers using network infrastructures which are neither owned or in which they have any business Khasnabish, et al. Expires January 4, 2014 [Page 17] Internet-Draft End-point based Multimedia QoE Management July 2013 relationship. These services are termed "over-the-top." The traditional network operators (cable, satellite, Telco and terrestrial) have also started to provide over-the-top media services. These over-the-top media services use broadband delivery, and for the traditional network operators this is a component of their overall offering. Over-the-top services cannot assume any reservation of bandwidth, and hence such services are provided over unmanaged networks. Authors' Addresses Bhumip Khasnabish ZTE USA, Inc. 55 Madison Avenue, Suite 160 Morristown, New Jersey 07960 USA Phone: +001-781-752-8003 Email: vumip1@gmail.com, bhumip.khasnabish@zteusa.com Gerard Fernando Compression Labs, Inc. 303, Twin Dolphin Drive, Suite 600 Redwood City, CA 94065 USA Phone: +1-650-704-9862 Email: gerardmxf@yahoo.co.uk, gerard.fernando@compressionlabs.com Lin Ya ZTE Corporation 22/F, ZTE Research and Development Building, Nanshang District Shenzhen 518057 P.R.China Phone: +86-755-26776850 Email: lin.ya@zte.com.cn Khasnabish, et al. Expires January 4, 2014 [Page 18]