Synchronous Learning

Transcript

“Synchronous e-Learning: A QoE Perspective” George Liodakis, Lecturer TEI of Crete- Department of Electronics Laboratory of Broadband Communications & Electromagnetic Applications May 24, 2013 What is synchronous e-Learning? • Synchronous learning is live, real-time (and usually scheduled), facilitated instruction and learning-oriented interaction (as the interaction is essential to learning, “learning-oriented interaction” is incorporated to the definition in order to differentiate synchronous learning from lecture, product demonstrations, and other “knowledge dispersal” activities). • Synchronous e-Learning is synchronous learning that takes place through electronic means (you may know synchronous e-Learning by another name, or by one of the myriad modes that it can take: virtual classroom, Webcasting, Web conferencing, videoconferencing, Webinars, live e-Learning, eConferencing, ... ). Synchronous vs. asynchronous e-Learning Roots of synchronous e-Learning Webcasting   Webcasting as a term was derived from the concept of broadcasting over the Web. As this etymology implies, the expression originally referenced audio and video sent from a single source to multiple passive receivers, either live or on demand Webcasting utilizes streaming media to transmit audio/video efficiently over the Internet. These media streams are encoded and decoded using a common system format (e.g. Windows Media, RealMedia, Flash Video, QuickTime,or DivX). Webcasting (cont’d)   Webcasting can refer simply to one-way audio/video streams. However, numerous services and tools have emerged, providing more sophisticated communication options that are synchronized with the audio-video stream — most commonly presentation slides, real-time text captioning, text Chat, polling, and file downloads. In other words, fuller features and opportunities for interactivity have been introduced into some Webcasting services, challenging its differentiation from Web conferencing. Webcasting is principally utilized for presentation-style, knowledge-dispersal types of learning. Webcasts are typically most practical for reaching large volumes of learners simultaneously, so the opportunities for complex interaction with learners are intentionally restricted. Web conferencing     What is it? Highly interactive, Internet-based applications with a rich collaboration feature set (e.g. audio/video from presenters and learners, application sharing, whiteboarding and markup tools, breakout rooms, polling, quizzing, hand raising and emoticon responses, slides and media, Web site tours, public/private text Chat). It is capable of scaling from small groups to hundreds or thousands of simultaneous users. When people discuss “synchronous e-Learning,” they are typically referring to Web conferencing. Web conferencing excels in the development of “higher order” learning skills (such as synthesis, analysis, socialization, acculturation). Synchronous e-Learning and media quality impairments   Media quality impairment continues to be a combination of both network induced degradation (loss, delay, etc.) as well as network independent parameters (encoding, compression, audio-video synchronization, etc.). The synchronous e-learning educational process is highly affected by the various media quality impairments (QoL- Quality of Learning). From QoS to QoE • The notion QoS (Quality of Service) is a central research topic in communication networks with a technical view on service quality. The adoption of the QoE (Quality of Experience) … • Has redirected the focus towards the end user and trying to quantify her subjective experience gained from using a service. • Extends QoS and may be defined as overall acceptability of an application or service as perceived subjectively by the end user. • QoE has also been referred as end-to-end QoS or end-user perceived QoS QoE extends the current QoS perspective towards the actual end user including …         Technical QoS as well as the expectations of the end user The content of the service The importance of the service for the end user The characteristics of the device The usability of the human – computer interface The joyfulness of interaction The perception of security and Maybe even the price of the service QoE and network-based applications    The QoE term is applied to any kind of network-based applications such as Web navigation, multimedia streaming, Voice over IP, etc. According to the application it has different meanings. For a voice over IP application a positive QoE relates to the sound fidelity and ability to smoothly take turns in a conversation. A remote multimedia streaming application has a high QoE if the video image is large and clear when presented to the user. For a Web surfer good QoE means that Web content is retrieved fast enough before getting bored and clicking a link to another Web site. In general QoE is expressed in human terminology rather than technical metrics. Rough classification of applications and their relations to QoE   Transactions-oriented applications: characterized by request/response flows corresponding to bidirectional data transfers (e.g., Web servers, databases access, etc.). For these applications, the QoE is mainly related to the delay in the reception of an answer after a request has been submitted. Throughput-oriented applications: characterized by bidirectional data transfers requiring non-fixed network resources (e.g., FTP transfers, applications updates, etc.). QoE for these applications is usually related to the duration of the transaction for a given volume of data. Rough classification of applications and their relations to QoE (cont’d)  Streaming-oriented applications: like in the case of transactional applications, the streaming applications are characterized by request/response flows. The difference resides in the large difference between the request and response transfer sizes, and mainly in the time-delivery constraints associated to the consumer side of the application. The QoE associated to streaming applications largely depends on the usability of the received data while the transfer lasts, as the consumption is done in real time, independently of the actual network’s load (e.g., voice, radio or video over IP, etc.). Streaming applications generally require an assured minimum amount of server and network resources. Categories of methods for assessing the QoE   Objective methods are based on algorithms in order to characterize the quality of media and predict the user’s behavior. Example objective QoE metrics (full-reference metrics, noreference metrics, reduced-referenced metrics): POLQA: a next-generation mobile voice quality testing standard according to recommendation ITU-T P.863. It has been especially developed for super wideband (SWB) requirements of HD Voice, 3G, VoLTE (4G), VoHSPA and VoIP. PSNR (Peak Signal to Noise Ratio): a pixel-by- pixel comparison of a processed frame with that of the original. MDI (Media Delivery Index) Subjective methods require the user to describe his experience in a survey. Media Delivery Index (MDI)     MDI measurements gives an indication of expected quality, ultimately, users’ QoE based on network level measurements. It is useful to see the relationship between jitter and buffering. Jitter is the is a change in end-to-end latency with respect to time. Packets with irregular arrival rate exhibit non-zero jitter. If the instantaneous data arrival rate does not match the rate at which the destinations is consuming data, the packets must be buffered upon arrival. MDI: Delay Factor (DF)     The more severe the jitter, the larger the buffers need to be in order to eliminate it. The price to pay for having larger buffers is that introduce delay. Furthermore, buffers are of finite size, and excessive jitter will cause them to either overflow or underflow. Both of these are undesirable, and they degrade the QoE. The DF is the time value indicating how many msec of data the buffers must be able to contain in order to eliminate jitter. MDI: Media Loss Rate (MLR)    The MLR is simply defined as the number of lost or out-of-order media packets per second. Out-of-order packets are important because many devices make no attempt to reorder packets before presenting them to the decoder. Any packet loss, represented as a non-zero MLR, will affect QoE. Subjective methods for assessing the QoE    Not easy to obtain reliable results since the users who are surveyed may be influenced by various parameters (personal opinion, feeling, previous knowledge of the subject). A common way of subjectively characterizing QoE continues to be the mean opinion score (MOS). MOS is representative of the average human response (say, on a scale of 1 to 5) to a given video flow. In other words, it is a mapping of human inference of distortion on a pre-defined quality scale. MOS calculations tend to be computationally intensive, cumbersome, not repeatable, and often hard to adapt to real time quality assessment as most existing schemes assume the availability of the original frame for reference. Subjective methods for assessing the QoE (cont’d)   MOS is the most popular measure of QoE. The basic definition of MOS can be found in ITU-T Rec. P.10: “the mean of opinion scores, i.e., of the values on a predefined scale that subjects assign to their opinion of the performance of the telephone transmission system used either for conversation or for listening to spoken material.” The definition from ITU-T Rec. P.10 adheres to voice telephone services, but the MOS scale is currently used for evaluation of other services, especially video. ITU-T Recs P.800 and P.800.1 define a five point MOS scale. Example scales for measuring subjective response (qualitative vs quantitative scale) Objective and Subjective QoE assessment of 18 video samples POLQA (Perceptual Objective Listening Quality Assessment) Case study: Network-based Music Collaboration and Performance (NMCP) in DIAMOUSSES project The advent of broadband networking  Advancements in computer and music technologies has stimulated the interest of:  Professionals in music technology  Network and content providers  General audience  Research community  Characteristics of NMCP and Technical Challenges     A highly demanding application in terms of bandwidth requirements, synchronization issues, delay constraints Psychoacoustic, perceptual and artistic aspects are introduced Emphasis on QoS and QoE issues A research and development area with a high degree of interdisciplinarity Scenario 1: Music Rehearsal (Analogy: Collaboration in a synchronous elearning environment) Scenario 2: Concert (Analogy: Webcasting in a synchronous e-learning environment) Scenario 3: Master class (Analogy: Instructor-student in a synchronous e-learning environment) Technical challenges for DIAMOUSES or Why NCMP is a Premium Service    Bandwidth demand for CD-quality multi-channel audio transmission (at least 1,41 Mbps - 44,1 KHz sampling, 16 bits/sample for stereo music). Clock synchronization of multiple participants in a network music performance (order of microseconds) A 20-30 millisecond round-trip delay is tolerable for traditional ensemble performance Testing of DIAMOUSES system (Scenario 1) Testing of DIAMOUSES system for scenario 3 Parameter Average Minimum Maximum Lost packets (%) 0,01 0,00 0,04 MDI:DF (ms) (Media Delivery Index: Delay Factor) 6,99 0,00 26,46 Throughput (Mbps) 1,47 1,48 1,48 Out-of Sequence Packets 0 0 3 An overview of factors influencing QoE Grade of Service (GoS)    As meant by ITU-T, GoS applies to circuit switched networks and describes all phenomena occurring during connection in the context of telephone networks (ITU-T Rec. E.720, E.721, E.771, E.493). Currently, it applies especially to circuit switched optical services. GoS parameters are, for example: connection set up delay, probability of end-to-end blocking, delay in authentication, and probability of breaking an active connection (forced or unpredictable tear down). If an admission control mechanism is considered, an example of the GoS parameter would be the blocking (request rejection) probability. Quality of Resilience (QoR)   Resilience, i.e., network survivability against failures, has traditionally been perceived as one of the dimensions of QoS. So far, reliability-related metrics are agreed in SLAs under a general QoS umbrella. It concerns mainly the availability, i.e., the probability that a service is operational. Nevertheless, recently this aspect of quality provisioning has been recognized as an independent field. One of the approaches related to QoR is related to definitions of reliability-based metrics. They describe the influence of failures on a network and client service, taking into account different survivability mechanisms. For instance, ITU-T Rec. E.800 determines basic measures such as the reliability function, availability, downtime, failure rate, etc. Moore’s theory of transactional distance   It offers a framework for the current research into student perceptions of synchronous e-learning. According to transactional distance theory, distance is considered a pedagogical phenomenon. The “sense of distance” a learner feels during the learning process transcends geography and is concerned with student interaction and engagement in the learning experience. Transactional distance theory consists of three elements: dialogue, structure, and learner autonomy, all of which interrelate across learner-instructor, learner-learner, learner-content, and learner-interface interactions. Moore’s theory of transactional distance (cont’d)    Dialogue is defined as two-way communication and interaction in its many forms. Structure refers to course organization and the impact this has upon student engagement. Learner autonomy represents the learner’s perception of both independent and interdependent participation in the course and is directly related to the student’s level of self-directed learning. Moore (1993) suggests that instructors need to pay attention to all three elements of transactional distance theory in order to reduce the “distance” experienced by the student. Conclusion Taking into account the availability of technology (broadband communications infrastructure, software technology, etc.), the following question arises: How can we decrease the “distance” envisaged by Moore’s theory in a synchronous e-learning context?