Real-Time Howling Detection for Hands-Free Video Conferencing System

Transcription

1 Real-Time Howling Detection for Hands-Free Video Conferencing System Mi Suk Lee and Do Young Kim Future Internet Research Department ETRI, Daejeon, Korea {lms, Abstract: This paper presents howling detection method in two-way hands-free communication system. Though hands-free communication is convenient for multi-user system such as audio/video conference, it is very common for the open microphones and loudspeakers produce acoustic feedback in a closed loop, which results in howling. This obviously prevents any useful conversation between participants. In this paper we propose a real time howling detection method based on the long-term average spectral power and howling information of the previous frame. Performance tests show that the proposed algorithm provides quick and stable howling detection results. Keywords-component; howling detecion; howling suppression; hands-free video conferencing I. INTRODUCTION As high speed internet connectivity has become more easily available, video conferencing is being used more and more everyday life for business and educational purpose, not just for the personal use. Recently, many educational institutions including universities are interested in distance education by using the video conferencing technology for students who are separated by time and distance. In the fields of business, various types of video conferencing system has been adopted, because it enable individuals in distance locations to participate in meetings with time and money saving. In video conferencing system, hands-free communication is more convenient for users. But it is very common to produce acoustic feedback in a closed loop, resulting in howling. This obviously prevents any useful conversation between participants. A typical closed loop path (dashed line) can exist when both parties are using hands-free video conferencing system is illustrated in figure 1. This type of closed loop which includes microphone, network, and loudspeaker is most common in real time hands-free communication system. If the loop gain is greater than unity at one or more frequencies the system becomes unstable and produces oscillations. If the oscillation signal has the frequency in the audible range of human hearing, it is called howling or squealing. It not only disturbs normal communications, but also damages power amplifier for overload. In two-way hands-free communication, acoustic feedbacks from loudspeaker into the microphone are traditionally cancelled using acoustic echo canceller (AEC). When it has reached enough convergence, it provides protection against howling because it reduces acoustic feedback. Until then, or when the acoustic path is changed or high loudness is required, the system exposed to howling because of improper function of AEC[1,2]. One of This research was funded by the MSIP (Ministry of Science, ICT & Future Planning), Korea in the ICT R&D Program The most popular techniques for supplementing the AEC to prevent howling in hands-free communication systems is notch filter based howling suppression (NHS) methods. NHS methods consist of howling detection and howling suppression by using a notch filter. The design of notch filters for howling suppression is based on well-established filter design techniques. Hence, it is critical to detect howling signal rapidly and exactly in NHS methods. In this paper we propose a howling detection method ) (ISSN : ), Volume No : 12 Issue No : 1..Page 1

2 based on the howling information of the previous frame and long-term average spectral power for real time twoway hands-free video conferencing system. This paper organized as follows: Section 2 briefly reviews a notch filter based howling suppression method. In section 3, we discuss characteristic of howling signal recorded in our two-way hands-free video conferencing system. Section 4 introduces a proposed howling detection method for NHS and Section 5 presents performance test results. A B Figure1. Acoustic feedback in packet based video conferencing system NOTCH FILTER BASED HOWLING SUPPRESSION II. Notch filter based howling suppression (NHS) is one of the most popular methods for acoustic feedback control in public address and hands-free communication systems[3]. The NHS methods consist of howling detection and howling suppression by using a notch filter. And NHS can be divided into two categories, i.e., onestage and two-stage, depending on whether the howling detection and notch filtering are performed jointly or separately. The adaptive notch filter (ANFs) based methods are typical one-stage method. Two-stage NHS methods are the most popular method for acoustic feedback control. The two-stage NHS in hands-free communication system can be outlined as shown in figure 2. The microphone signal is first processed by a howling detection algorithm, which forwards a set of design parameters of notch filter. When howling has been detected, a notch filter has been activated to suppress the howling signal before transmit the input signal to the remote site. In the two-stage NHS method, howling detection is crucial to get reliable howling suppression results because the notch filter design method is well defined. Figure 2. Two-stage notch filter based howling suppression III. HOWLING SIGNAL IN TWO-WAY HANDS-FREE VIDEO CONFERENCING ENVIRONMENT In order to analyze the howling signal in two-way hands-free video conferencing environment, we set up an experimental environment as shown in figure 1. The hands-free video conferencing system consists of personal computer, display, microphone and loudspeaker. The characteristic of the microphone and loudspeaker is as follow: o Loudspeaker: BOSE Companion 2 multimedia speaker o Microphone: ETM-003 of Edutige o Condenser / Omnidirectional / Boundary microphone o Sensitivity: -23dB o Frequency response: 50~18 khz ) (ISSN : ), Volume No : 12 Issue No : 1..Page 2

3 Figure 3 shows a block diagram for audio signal processing in our video conferencing system. The microphone signal of site A is processed by AEC and encoded with G wideband encoder [4] and transmitted to remote site B after RTP packing. In site B, the received packet is unpacked and decoded by G decoder. The decoded signal is played out through a loudspeaker. Since the loudspeaker and microphone are located in same room, the speaker output is captured by a microphone again and transmitted to site A. A B Figure 3. Audio signal flow in two-way hands-free video conferencing Figure 4 and 5 present time domain signal and frequency spectrum of howling signal recorded at 16 khz sampling rate without AEC block both in site A and B, respectively. The frequency range of each spectrum in those figures is limited to 1,000 Hz. From the figures, we can see the amplitude of howling signal is globally increase with time but has some fluctuation. Figure 4. Howling signal recorded in site A ) (ISSN : ), Volume No : 12 Issue No : 1..Page 3

4 Figure 5. Howling signal recorded in site B IV. PROPOSED HOWLING DETECTION METHOD Figure 6 presents the high-level block diagram of the proposed howling detection method. Basically, it is operated in a 10 msec frame based manner. The microphone input signal is stored in a buffer and this time domain signal is transformed into a frequency domain signal by a FFT analysis. FFT is run once every 10 msec and the FFT window size is 16 msec (corresponding to 256 samples at 16 khz sampling). After howling candidate selection and detection, the related parameters are update for the next frame processing. Figure 6. High level block diagram of the proposed howling detection A. Howling Candidate Selection The magnitude of the howling signal is globally increased over time, but it can be locally fluctuated by the delay caused in communication environment. In this case, it is difficult to detect howling signal consistently in some segment which have relatively decreased magnitude. In order to solve this problem, we use the howling information of the previous frame when selecting howling candidate. To select a howling candidate, first, N frequencies which have highest spectral power are chosen by using peak picking. We select a frequency which has maximum spectral power from N as a howling candidate of the current frame. However, if the selected howling candidate is different from the howling frequency of the previous frame, the howling candidate can be replaced with the howling frequency of the previous frame according to the following procedure. First, it is checked that there is a same frequency with the howling frequency of the previous frame in N-1 frequencies except howling candidate. Second, if one of them is same with the howling frequency of the previous frame, then compare the spectral power of the current howling candidate and the previous howling frequency. The howling candidate of the current frame are replaced with the howling frequency of the previous frame if the ) (ISSN : ), Volume No : 12 Issue No : 1..Page 4

5 spectral power at previous howling frequency is greater than the weighted spectral power of current howling candidate. Here, the weighing factor is set to a value less than 1. B. Howling Detection The howling detector judges that the howling candidate is a true howling signal. We use not only the peak to short-term (frame) average spectral power ratio but also long-term average spectral power ration of the howling candidate of current frame and howling frequency of the previous frame. We detect the howling candidate as a true howling signal in two cases. First, it is decided the howling frequency candidate is true howling signal when both ratio of the spectral power of the howling candidate to a short-term average spectral power and the long-term to short-term average spectral power ratio of the howling candidate is greater than the predetermined threshold values. Second, it is detected the howling candidate is a true howling signal when the previous frame has howling signal and the long-term average spectral power of the howling candidate is greater than the weighted long-term average spectral power of the howling signal of previous frame. The weighing factor is set to a value less than 1. V. EXPERIMENTAL RESULTS In order to test the performance of our proposed method, we recorded two types of signal without AEC operation at 16 khz sampling rate. One is a pure howling signal without speech activity and second is a howling signal with a speech activity from one of the sites. Figure 7 shows a test result for a pure howling signal in time domain and frequency domain with a limited frequency range. A microphone input signal (blue wave) and howling suppressed signal (green wave) by using a 2nd order IIR notch filter is shown in figure 7. And the howling detection results also depicted with black line overlapped with the waveform and spectrum. Figure 7 shows that the proposed method is able to detect the howling signal very quickly and consistently even though the magnitude of howling signal is fluctuated. Figure 7. Experimental results for a pure howling signal. Figure 8 shows a performance test results for a howling signal mixed with speech signal. The time domain waveform of input signal(blue wave) and howling suppressed signal(green wave) are presented with howling detection results(black line) in first block of figure 8. Also, the spectrum of the microphone input signal and howling suppressed signal is illustrated with limited frequency range. It indicates that the proposed howling detection is providing stable performance when the speech signal is presented. ) (ISSN : ), Volume No : 12 Issue No : 1..Page 5

6 Figure 8. Experimental results for a howling signal mixed with speech signal. VI. CONCLUSION In this paper, we propose a howling detection method based on the howling information of previous frame and long-term average spectral power. The performance of the proposed method is tested with the sample data recorded in real-time two-way hands-free video conferencing environment. The test results shows that the proposed method quickly detect the howling frequency and also gives stable results for the howling signal mixed with speech. REFERENCES [1] Senthil Kumar Mani and Sowmya Mannava, Robust and High quality Howling Suppression for Real Time Hands-Free Communication systems, IEEE 2011 [2] Franck Beaucoup, A Novel Loop Stabilisation Technique for Full-Duplex Speakerphones, IEEE CCECE/CCGEI, May, 2006, pp [3] T.van Waterschoot and M.Moonen, comparative spectral analysis of howling detection criteriia in notch filter based howling suppression. AES 126th conv., Munich, Germany, May, 2009 [4] ITU-T Recommendation G.711.1, Wideband embedded extension for ITU-T G.711 pulse code modulation, sept ) (ISSN : ), Volume No : 12 Issue No : 1..Page 6

7 Applying Delphi Methodology for Lifelong Learning Activities Miguel Doctor, Rafael Mompó European University of Madrid Madrid, Spain David de la Mata, Judith Redoli University of Alcalá Alcalá de Henares, Spain Abstract : Nowadays we are immersed in a highly competitive job market where demonstrating good technical skills is not in most cases a guarantee of a success. In addition, the economical context is provoking a diaspora of highqualified workers from their countries to others with lower unemployment rates. In this scenario, applicants need to demonstrate not only a strong background in their specialization fields, but also, a set of soft skills and abilities, which allow them, highlight themselves in the hard task of finding a job. In this paper we present an evolution of a methodology supported by a web tool, aims to helping working adults trainers and university professors to develop soft-skills in engineering, as well as technical competencies provided by classical engineering training programs. Delphi Learning Package (DLP) tool is based on Delphi strategic consulting. Firstly, a short introduction about what the Delphi method consist of, will be performed. Then, a description about the adjustments and modifications made on the methodology in order to fix it to educational environments will be discussed as well as an evaluation of the effects triggered by them. Afterwards, the technological environment created for hosting the tool and supporting the methodology is described. A justification about why a tool to automatize the methodology is required will be provided. Also, how students develop different soft skills such as critical thinking, synthesis ability, inference, and argumentation following our method is discussed. Finally, we assess the methodology and the tool by discussing about the results obtained during two pilot projects that have been developed involving both, students with a major in Telecommunications Engineering and software engineers from a software department of, a well known food company. This tool can be downloaded and used freely by contacting the authors of this article or by accessing the website Keywords-component; Delphi, lifelong learning, e-learning, collaborative/cooperative learning I. Introduction The current socio-economic context triggers high levels of instability and uncertainty over the high- qualified positions job market. The ever faster technological evolution gives engineers and specialized workers, through their skills and abilities, the key to support their companies in the current high-competitive scenario. Because of this, companies are starting to consider engineering training as one of the most important issues in their strategic company dossiers. Nevertheless, the quick evolution of development processes and the frequency of appearance of new technologies, affect even the best and most experimented workers, who observe how their skills are getting obsolete in few months. As a consequence, the value of the technical background is loosing ground with other more generic skills or soft skills, which means that it is becoming more and more important that engineering students and professionals workers develop soft skills [1]. ) (ISSN : ), Volume No : 12 Issue No : 1..Page 7

8 The methodology proposed aims to force students and professional workers to develop personal skills in order to solve intricate problems, to manage large amounts of information and to get the ability to extract useful knowledge from huge amounts of available documentation [2]. Collaborative/cooperative learning is the approach we have chosen as axe of our methodology. Since the proposal was to develop technical and soft skills at the same time, it was considered necessary to investigate how to adapt classic methodologies of strategic consulting based on making forecasts, to be used for engineering learning purposes. Among the different methodologies analyzed (such as Genius Forecasting [3], Trend Extrapolation [4], Consensus Methods (e.g., Delphi) [5], Simulation Methods [6-7], crossed impact matrix method [8], scenario [9], or decision trees [10]) one of them, which belongs to the group of consensus methodologies, was chosen. The most famous method from this group is the Delphi [11-12] one. The reasons to choice this methodology over other previously mentioned is because Delphi allows us, in a flexible way, to structure the communication flow efficiently. II. Special Issues Regarding Working Adults Learning Every year huge amounts of money are invested in programs and activities oriented to keep updated workers knowledge and skills in many engineering companies around the world [13]. Nevertheless, it is estimated that no more than 10% of this investment is returned to the companies like a real transference of knowledge or improvements in their production procedures [14]. The explanation of this low success ratio could be find in the techniques and methods used in the training courses and also, in the gap existing between programs or services offered by the training provider organizations (universities or training consultancy companies) and the real requirements demanded by the engineering companies [15]. The dynamic world in which engineers develop their activity presents them with new demands and provides new challenges almost everyday. This vertiginous technological evolution provokes that technical competencies and skills could become obsolete even before that recent graduated engineers could find their first employment. Because of that, technical background is losing importance compared to other kind of skills or abilities. This change started to be observed at the end of the 20th Century. In those years an evolution started from a hard-engineering model to a soft-engineering one. In the past, engineers were focused only in technical issues about projects, but today they are more interested in a softengineering working model, in which engineers need assume management resources tasks as well as marketing and commercial responsibilities [16]. This tendency (the importance of soft-skills as a part of engineers background) has risen on the first decade of the 21st Century, and currently an engineer has to demonstrate basic competencies in the next transversal areas [17]: Social Science: Communication skills, Social skills, Presentation skills, Interpersonal skills. Business/Management: Leadership skills, Business management skills, Team-working skills, Financial skills Computer/Technology: Computer skills, Programming skills, Design skills Mathematics/Science: Problem solving skills, Research, self-learning and development skills, Analysis/synthesis skills Traditionally, these competencies were acquired after some years of working on real projects, but modern companies require the application of some processes oriented to develop these soft-skills as well as modern technical skills [18-19]. In this paper, a methodology, which develops the soft-skills listed below in two different target populations (working adults and university students) is presented: ) (ISSN : ), Volume No : 12 Issue No : 1..Page 8

9 Synthesis skills, Communication and presentation skills, Argumentation Capacity, Critical Thinking, Inference Capacity, Building Customized Knowledge from a Certain Amount of Information, Self-learning by discovery. III. Delphi Methodology Background Initially, the Delphi methodology was designed to support forecast Escalation Dynamics on different conflict scenarios [20], but it was quickly applied to other fields, such as medicine, information systems, or company organization [21 23]. Delphi methodology aims to make future forecasts to improve decision making in the present. In this article, the use of Delphi methodology as a collaborative learning methodology to solve complex problems approached daily by engineering professionals and students is proposed. A Delphi process consists of selecting a group of experts that are asked about their opinion on certain matters pertaining to future events. The group could be made up by experts on the matter, affected or/and with an interest, in such a way that because of their level of information and extent of knowledge they can contribute different ideas and points of view to the problem at hand. The experts are asked about the topics by filling in the basic element within a Delphi process: the questionnaire. In this way we avoid experts meetings, or face-to-face debating which means that we can increase significantly the number of consulted experts. Between rounds, the experts are given controlled feedback, which results from all experts forecasts. The product we really get with a Delphi method is a collective vision built on each group communication structure. In other words, at the end of the process we will obtain a set of opinions with different levels of consensus. IV. Delphi Methodology Based Learning In this section our proposal of adaptation of the Delphi methodology to collaborative learning is described. First of all, a coordinating group (professors or trainers) is set. This group is in charge of summarizing the concepts and contents that will be studied during the course, creating questionnaires that efficiently contribute to obtain results and defining the group of people that will compose the panel of experts. This panel of experts will be the students enrolled in the subject. It is considered that the prior knowledge acquired in the first years of the university program qualifies them sufficiently to analyze the information that the professor provides and to propose ideas, key aspects, or solutions to the case study presented, according to the focus determined by the professor. The questionnaires are the core elements of the process: initial questions and feed- backs are generated by the coordinating group (professors), that are sent to the panel of experts (students, or professionals attending to a training program) so as they can express their ideas and opinions. This process makes it possible to reach a consensus and to obtain conclusions that help solve the problem under study. Nevertheless, it is necessary to make certain adjustments to the method, as certain limitations are en- countered when applying it to a teaching environment, for example, the time of application (a semester or a number of hours purchased), or the selection of the panel of experts (always composed of students). Thus, Delphi methodology applied to learning establishes the following line of work: First step: previous study of the state of the subject matter by the participants. The learners receive information on the theme to be studied. Some of the information will be provided by means of attending class, but it is fundamental that the learners have access subsequently to the material presented in class, as well as to additional information (documents, presentations, etc.). Second Step: open questionnaire. The professor draws up a questionnaire in which he/she requests students to contribute ideas for the resolution of the problem. ) (ISSN : ), Volume No : 12 Issue No : 1..Page 9

10 Third step: closed questionnaire (feedback). Subsequently, the professor carries out a selection of the better ideas contributed. With these ideas the professor generates a closed questionnaire (feedback) in which the students, by means of allocation of a score to each idea of the closed questionnaire, assess the degree of importance. It is a matter of sorting the ideas according to its importance taking into account a criterion previously established by the professor. Fourth step: public discussion. Subsequently, a public discussion is opened through virtual forums, in which the students argue their points of view on the importance of the ideas. The professor should promote that discussion. Fifth step: final feedback. After the debate of the fourth step, students can modify the scoring assigned in the third step, taking into account the arguments that have prevailed as more serious and coherent in the debate. Sixth step: evaluation. Depending on the scenario (university students or working adults attending to a course) the evaluation can take different ways. For university courses an evaluation grade for each student is generated automatically. This grade is calculated taking into account the deviation of each student s answers from the average in the third and fifth steps, as well as the number of proposed ideas by the student in the second step (which were chosen by the professor due to their quality). This evaluation could be omitted in working adults learning programs, because in some cases the goal of the training is not to score the workers, but to demonstrate that the team is capable to build a work planning using the concepts and skills learned during the course. All these steps an their elements are related among themselves as it is indicated in Figure 1. In this manner, a process of communication is established, according to the phases of a Delphi method. Figure 1. Delphi learning communication process We have to pay attention to the number of questions included into the questionnaire. In practice, for a group of fifty students, the application of this methodology means that between 100 and 400 ideas are generated in the second step. From them all, the professor should select between 10 and 40 to generate the feedback questionnaire of the third step. It has been proven that a questionnaire with more than 30 questions is not effective due to the loss of concentration in the answers as the length of the questionnaire increases [24]. V. A software Tool is necessary: DLP In order to make the application of the methodology Delphi viable, it is necessary that both the professor and the learner have a software tool that facilitates the management and selection of the ideas generated by the learners, ) (ISSN : ), Volume No : 12 Issue No : 1..Page 10

11 and maintains each idea associated with the participant who has proposed it. Likewise, the software tool simplifies the preparation of the questionnaires to be answered by the students. Finally, the tool should show the professor the progress of the learner along the diverse steps and should facilitate the generation of final grading for the evaluation, according to the contributions of each participant. Therefore, a collaborative learning software tool has been developed based on the Delphi methodology applied in learning called DLP (Delphi Learning Package) and developed with PHP (PHP: Hypertext Preprocessor). The interface is user-friendly and fast for both the professor and the student. The DLP software has been developed as a two separated modules WEB application. The first module is a common web application that implements the MVC (Model View Controller) [25] pattern through the framework CakePHP (cakephp.org). The principles followed for its design and subsequent development have been the following: sturdiness, efficiency, scalability, and usability. The second module, named interface module is a piece of software, which allows us running the application within a LMS (Learning Management System), concretely Moodle ( This layered architecture aims to allow install the application in other Web platforms (Learning Manager Systems like Sakai, other Web Portal Solutions like Liferay or Drupal, or even Social Networking sites like Facebook) just creating a new interface module to connect with the new platform. There are two objectives in using the selected technology. On one hand, to integrate the tool developed within a LMS (Learning Management System), which results in all the Delphi activity being carried out in the Virtual Class, and on the other hand, to obtain independence from the e-learning platform that is to be utilized. To carry out the Delphi activities both the DLP tool and LMS (or other community support application) are necessary. The solution adopted is an application based on CakePHP technology working on Apache Server ( that communicates with a LMS Moodle also based on PHP technology. The platform makes use of a version of MySQL with which they form a self-sufficient architecture, entirely interoperable and modular. VI. Experimentation Learning activities cannot be considered as isolated products with well-defined con- tents and duration, but as a permanent process, which should be linked to the professional activity of the learner and adapted to their professional needs and challenges [26]. Since this process should start before young engineers get access to the job market, the proposed methodology has been applied to two groups of engineers involved in two different stages of their professional careers [16][27]. The first pilot experience has been performed with students of fourth year of Telecommunications Engineering program. The objective of this experiment was to verify in practice the functionalities of the tool, its suitability to the needs of professors and students, as well as the adequacy of the methodology for the attainment of the educational objectives. When it was finalized, a poll among the students was conducted and gathered feedback from the professors. From the functional point of view, the purpose was to: Verify whether the tool is user friendly to carry out each one of the activities both by professors as well as by students. Test the functionality of the tool that allows the selection of ideas and generation of the closed questionnaire. This is the most complex task undertaken by the tool and it is necessary to verify its behavior with a high number of participants. ) (ISSN : ), Volume No : 12 Issue No : 1..Page 11

12 Detect improvements in the evaluation functionality both for the professor as well as for the students. The goal is to generate grading along the process, which will allow a continuous monitoring of the students performance. From the educational methodology point of view, the purpose was to: Verify the suitability of the Delphi activities to attain the learning objectives proposed by the professor. Design different types of activities that are adequate to be carried out by means of an educational Delphi activity. The second pilot was developed with a software development team working for a software development department of a food company. They had to approach the maintenance of a web portal developed using PHP technology but they didn t have background using it. In this case, the purpose from the functional point of view was to: Test whether the tool is flexible enough to be used with working adults, considering the special features of this target population (training combined with professional tasks, family responsibilities, etc.) Detect how to improve the learning experience by making the access to the learning activities and tasks easier. From the educational methodology point of view, the purpose was to: Provide them with enough knowledge about PHP in order to allow them make the maintenance of a real application. The different roles of the participants were considered in order to offer a useful training. Test the methodology for lifelong training in a real software development team. Evaluate if the Delphi methodology improves some of the soft-skills in a real software development team. Detect improvements in the methodology and in the tool in order to avoid dropout cases during the course. VII. Results The results of both experiments proved that the Delphi method adapted to the learning environment for the development of skills in engineering useful. After analyzing the pool of students and also after a discussion meeting with the engineers enrolled in the second pilot, it can be concluded that, in addition to the technical skills gained from the specific study content (which will vary with the subject), the Delphi method allows developing a series of soft skills that are discussed below. Building customized knowledge from a certain amount of information, Self-learning by discovery and Synthesis Ability: In the current complex modern professional world, professionals need to build their own knowledge from huge amounts of information by filtering contents that can be useful to their daily work. The Delphi activities are designed to increase the student s capacity for self-learning and building of knowledge. To do this, the activity proposes an initial approach on a multiple-solution problem, study, or technical documentation. The student has different sources of information that allow him/ her to examine the problem from different points of view. Working on this information, participants generate answers to the initial approach, based on prior knowledge and on the given material. By analyzing the proposals of their fellows it makes it possible to advance a step further in the comprehension of the problem. This feature resulted very interesting and productive in the second pilot, when participants with different roles retrieved from common information sources, the contents more related to their work activities. Regarding Synthesis ability development, each participant has a limited space to present his/her solution or proposal. Therefore, he/she has to exercise synthesis in such a way as to express in own words the solution to the problem. ) (ISSN : ), Volume No : 12 Issue No : 1..Page 12

13 Inference Capacity, Argumentation Capacity and Critical Thinking: The inference capacity allows to establish relations among the available information (that initially is not related), and to use it to resolve certain approaches or problems. In the Delphi activities, the participant has multiple documents and sources of information to approach the problem, and he/she is challenged to propose creative solutions based on them. Likewise, in the debate phase, he/she is obliged to confront his/her opinions with those of his/her colleagues. This exercise develops his/her inferential capacity at the same time than his/her argumentation capacity, because the participant has to defend his/her proposal. Regarding the second pilot, the debate phase became a typical kick-off project meeting, in which participants with different roles tried to convince the others about the best way to approach the project. It was detected how participants with the same role, presented closer opinions than members of different role groups. Finally, critical thinking abilities were worked out by reasoning the pros and cons of different proposals. VIII. Conclusions and future work lines The results of both experiments proved that the Delphi method adapted to the learning environment for the development of skills in engineering useful. The pilot experience performed in the context of university programs made it possible to demonstrate that the methodology encourage the participation and collaboration among students. It was observed that some soft skills were exercised during the process (a poll offered to the students at the end of the pilot reveals that critical thinking, learning by discovery and capacity of argumentation are skills that the students consider promoted because of using this methodology). It also proved that for the practical implementation of the methodology, a software tool that allows easy processing and managing of information contributed by students, as well as monitoring of students, is needed. The support of a software tool is essential; it would be very difficult to apply the methodology without the help of a software tool. Moreover the experience reveals that the evaluation procedure followed in the methodology is fair and represents properly the amount of work performed for each participant. The second pilot worked in a different scenario, and in consequence, led to different conclusions. The software engineers who participated in the working adults training paid more attention to the final average questionnaires, as well as the ideas contributed, than to the evaluations. Since the participants in this experience have different roles (managers, programmers, administrators, designers, etc.), their feedbacks for the different stages of the process were strongly influenced by their role. This is very interesting and indicates that this methodology is an excellent resource for working adults learning activities because each student applies their background and improves their daily work. The participants really appreciated the use of a virtual platform because they could fill in the questionnaires at anytime even after training hours. Nevertheless, some of them mentioned that we should work in order to add more time-space flexibility to the platform. Concretely, they considered that offering a smartphone-tablet compatible version would increase the participation and would allow them to use some useless gap times for learning purposes. REFERENCES [1] J. Mills, M. Ayre, D. Hands and P. Carden, Learning About learning styles: Can this im- prove engineering education? MountainRise, vol.2, pp 33-48, ) (ISSN : ), Volume No : 12 Issue No : 1..Page 13

14 [2] R. M. Felder, D. R. Woods, J.E. Stice and A. Rugarcia: The future of engineering education II. Teaching methods that work, Chem Eng Educ, vol 34, pp 26 39, [3] H. Kahn, Thinking about the unthinkable, Horizaon Press, New York, [4] H. Kahn, World economic development: 1979 and beyond, Morrow Quill Paperbacks, New York, [5] A. Fink, J. Kosecoff, M. Chassing and R.H. Brook, Consensus methods: Characteristics and guidelines for use Am J Public Health, vol.74, pp , [6] M. B. Blake, A Student-enacted simulation approach to software engineering education, IEEE Trans Educ, vol.46, pp , [7] F. J. Jiménez-Hornero, J. V. Giráldez, A. M. Laguna and J. E. Jiménez-Hornero, An educational computer tool for simulating longterm soil erosion on agricultural landscapes, Comput Appl Eng Educ, vol.17, pp , [8] W. Schultz, Scenario building: The Manoa approach, Available at: [9] P. Bishop, A. Hines and T. Collins, The current state of scenario development: An overview of techniques, Foresight, vol.9, pp 5 25, [10] J. Buckley and T. Dudley, How Gerber used a decision tree in strategic decision-making, Graziadio Business Report, [11] N. Dalkey and O. Helmer, An experimental application of the Delphi method to the use of experts, Manage Sci, vol.9, ,1963. [12] H. A. Linstone and M. Turoff, The Delphi method: Techniques and applications, Addison-Wesley, London, [13] P.H. Wu, G. J. Hwang, H.C. Chu, C.C. Tsai and Y.M. Huang, A Computer-Assisted Collaborative Approach for Developing Enterprise e-training Courses on the Internet, Journal Of Research And Practice In Information Technology, vol.41 (4), pp , [14] E.A. Awoniyi, O.V. Griego, and G.A. Morganm, Person-environment fit and transfer of training, International Journal of Training and Development, vol.6, pp 25-35, [15] M. Salas-Velasco, The transition from higher education to employment in Europe: the analysis of the time to obtain the first job, Higher Education, vol.54 (3), pp , [16] G. Guest, Lifelong learning for engineers: a global perspective, European Journal of Engineering Education, vol.31 (3), pp , [17] J. Farr, D. Brazil, Leadership Skills Development for Engineers, Engineering Management Journal, vol.21 (1), pp 1-8, [18] V. Garousi, Applying Peer Reviews in Software Engineering Education: An Experiment and Lessons Learned, IEEE Transactions on Education, vol.53 (2), pp , [19] J. Arco-Tirado, F. Fernández-Martín and J.M. Fernández-Balboa, The impact of a peer- tutoring program on quality standards in higher education, Higher Education, 62 (6), pp , [20] F. E. Morgan, K.P. Mueller, E. S. Medeiros, K. L. Pollpeter and R. Cliff, Dangerous threshold. Managing escalation in the 21st century, RAND Corporation, Project Air Force, Santa Mónica, [21] P. Holsappele and K. Joshi, Knowledge manipulation activities: Results of a Delphi study, Inf Manage 39, pp , ) (ISSN : ), Volume No : 12 Issue No : 1..Page 14

15 [22] V. Lai and W. Ching, Managing international data communications, Inf Manage vol.45, pp 89 93, [23] J. Landeta, J. Barrutia, People consultation to construct the future: A Delphi application, Int J Forecasting, vol.27, pp , [24] R. M. Groves, F. J. Fowler, M. P. Couper, J. M. Lepkowski and E. Singer, Tourangeau, R.: Survey methodology, John Wiley, New York, [25] A. Leff and J. T. Rayfield, Web-application development using the model/view/controller design pattern, In: Proceedings of the 5th IEEE International Conference on Enterprise Distributed Object Computing (EDOC 01), Washington, [26] J. J. Rodríguez-Andina, L. Gomes, S. Bogosyan, Current Trends in Industrial Electronics Education, IEEE Transactions on Industrial Electronics, vol.57 (10), pp , [27] V. Mayorova, Integration of educational and scientific technological areas during the process of education of aerospace engineers, Act Astronautica, vol.69 (7 8), pp , ) (ISSN : ), Volume No : 12 Issue No : 1..Page 15

16 A pilot e-tutoring program for students of secondary education Spyros Doukakis, Cleo Koutroumpa The American College of Greece - PIERCE Athens, Greece sdoukakis@acg.edu, cleo@acg.edu Abstract : This paper presents a pilot e-tutoring program on mathematics and ancient Greek realized through Blackboard Collaborate and focusing on the support of 58 students of lower secondary education. Two teachers for each subject provided daily two hour support to students. In addition the results of a research conducted using a questionnaire answered by the students involved in the program after the latter s completion are also presented. The results show that 6 out of 10 students participated in etutoring sessions. Their satisfaction exceeded 90%, while the program s acceptance exceeded 94%. Moreover the results indicate no correlations between gender, personal involvement with web 2.0 tools and knowledge of computer use. Finally, the students related their satisfaction to the e-tutor s role as a facilitator, something that puts forward the need for further investigation of the educator s role. Keywords-component; e-tutoring, secondary education, satisfaction I. INTRODUCTION The integration and incorporation of digital tools to the educational procedure has led educators to the redefining of their way of teaching. Within this framework, e-tutoring (electronic tutoring) is a digital media of supporting students, which utilizes possibilities offered by the internet and web 2.0 tools in order to enhance the cooperation between students and educators. E-tutoring has the characteristics of traditional teaching in a classroom, in the sense that there is a teacher who facilitates students to acquire further knowledge, develop capacities and modify attitudes towards the subject taught [1]. The difference lies on the environment via which the cooperation between teacher and student is realized. E-tutoring is realized via an online environment, where an internet site or platform is used [2]. These environments dispose a series of interactional and co-operational possibilities that contribute to teaching, learning and students assessment. Furthermore, they provide a synchronous discussion system which permits face to face contact in rooms, dispose a multifunctional whiteboard with graphics, chat, application sharing, students assessment tools, and offer the possibility of recording the etutoring course for further and later use, etc. [3]. Finally, taking into consideration the easy access to and use of the e-tutoring environment, as well as the fact that teachers and students are at home and not at their work place (or wherever they wish, for that matter), e-tutoring environments modify students, teachers and parents perspective of teaching and learning. In this paper, primarily, the frame of e-tutoring utilization in the international educational community is presented. Afterwards, the pilot operation of the e-tutoring program in the school is described. Sequentially, the research conducted on the students is reported and the data and relative descriptive measures, followed by the research s results, are shown. Finally, the paper concludes with the outcome and discussion on issues of etutoring implementation, whilst ideas for future research are cited. II. OVERVIEW The increase of internet use and the integration of web 2.0 tools to the educational procedure has led to the growing utilization of e-tutoring environments. According to the relative literature, e-tutoring is an ) (ISSN : ), Volume No : 12 Issue No : 1..Page 16

17 individualized service providing support to a student or a group of students from an educator who uses the internet as their mean of communication [1], [2], [4]. Researcher Prensky supports that e-tutoring can function more effectively than traditional teaching, due to the frequency of interaction, the immediate feedback and the personal style of teaching and learning [5]. Lately, e-tutoring is offered internationally by public, private, and non-profit institutions [6]. A case of integrated e-tutoring is the Homework Help program which was created in 2008 by the Ontario Ministry of Education and in 2011 it covered about students ( According to Jopling s research, several studies of e-tutoring programs have been published [7]. From these, 11 studies concern exclusively elementary and secondary education. Furthermore, 9 out of the 11 study the possible improvement of the students school performance. Particularly, from the studies is shown a) the greater students involvement in the learning procedure when they participate in an e-tutoring program [8], b) the possibility provided to educators to take into consideration their students style of learning and thinking and to students to bring forth their interests [9] and c) the opportunity provided to educators and students to use pedagogical tools which couldn t be utilized in the traditional classroom [10]. On top of that, from the research of Dekhinet et al. is shown that through e-tutoring the students further develop their initial motives for learning, as their involvement in the program modifies their perspective of learning [11]. According to the results of Beal et al., most benefited was the weakest student of the group that participated in the e-tutoring program [12]. Finally, Gabriel and Kaufield put forward that teaching through an e-tutoring environment provides bidirectional learning opportunities for both the educator and the student and contributes to the student s participation in a community of learning, reinforcing students that were isolated in the traditional teaching and learning environment [13]. However, in the Greek educational community no studies related to e-tutoring environments in elementary and secondary education have been recorded. Thus, in this paper we will try to show the way the available etutoring environment within the educational unit was utilized and, thereafter, the research conducted to students who used e-tutoring as an educational procedure, as well as its results. III. OPERATION FRAME OF E-TUTORING PROGRAM The e-tutoring platform used in the educational unit is Blackboard Collaborate (BC). Primarily, during a 6 hour training on the environment and the available tools, the e-tutors discussed with the students the way to utilize and the netiquette for appropriate use of the e-tutoring program. The students were also notified that they needed headphones with a speaker. Daily each student received two internet addresses (one for ancient Greek and one for mathematics) and was able to connect to e-tutoring of the subject s/he wished. The pilot operation of the program lasted seven weeks, during the first of which students training was realized. E-tutoring was available for 2 hours per day, 4 evenings per week. The digital tools used for teaching were various. Mostly, suitable PowerPoints presentations uploaded on Whiteboard through an appropriate tool available from Blackboard Collaborate. On PowerPoint presentations students and the e-tutor were able to intervene, adding comments, erasing, marking or underlying. Furthermore, direct writing on Whiteboard was utilized, using tools from the control panel (Figure 1). Mathematics teachers used the dynamic mathematical software Geogebra, which, through Application Sharing or Print Screen, produced appropriate images. Application Sharing also proved valuable in solving problems on technical matters. Along those, the possibility to assess students through wrong/right answers and multiple choice questions was applied. Finally, particularly useful proved the possibility to divide groups of students into private working rooms. All e-tutoring courses were recorded, in order for them to be available both to students and educators. ) (ISSN : ), Volume No : 12 Issue No : 1..Page 17

18 Figure 1. Blackboard Collaborate platform, PowerPoint uploading and Whiteboard utilization During the pilot operation of the program, students participation ranged from 0 to 15 students per twohour session. Each of the participants was connected to the program for a time period ranging from 10 minutes to 2 hours. Nonetheless, sometimes the program lasted more than two hours, since there were still questions unanswered or issues unresolved. Overall, 22 e-tutoring meetings for every subject and 137 students connections took place. With the completion of the program s pilot operation, students were asked to fill anonymously a questionnaire in order to a) research the degree of their satisfaction from e-tutoring, b) relate their answers to gender, grade, knowledge of computer, internet and web 2.0 tools use. IV. RESEARCH APPROACH A survey questionnaire was used in this study. It consisted of two sections; the first section required that participants provide demographic and educational information (gender, grade, knowledge of computer use and internet, existence of a Facebook account and/or use of Skype program), and the second section included items which measure the degree of students satisfaction from the whole program and the learning benefits they might have gained from it. Satisfaction has been defined as the perception of pleasurable fulfillment of a service [14]. For the development of the questionnaire, questions adapted from previous studies were used [1], [15], [16]. The questionnaire was distributed to the students after the completion of the pilot operation. Initially it was given to two students (a boy and a girl) one of whom, must be noted, even though given the chance, did not use e-tutoring who were asked to complete it, in order to track down any problems. The responses showed no misunderstandings. Then, the questionnaire was filled out by the total of students that had the possibility to use etutoring. 58 questionnaires were completed. Students connected at least once to e-tutoring were asked to answer other questions as well, concerning the way of communication during the use of e-tutoring, the degree of satisfaction from the environment, the e-tutors and the material. The questionnaire is included in the appendix of the present paper. The data gathered was analyzed with SPSS software. With its use, both the descriptive statistics of data and the correlations were measured, as well as the reliability analysis (Cronbach s alpha), in order to evaluate the level of internal consistency of its elements [17]. V. RESULTS A. Descriptive statistics 58 students participated in the research. 33 (57%) of the participants were boys and 25 (43%) were girls. The students were originally asked if they possess a Facebook and/or Skype account and, furthermore, were asked to define their knowledge on computer and Internet use. From the whole number of students in question 50 (86%) have a Facebook account, while 34 (59%) have a Skype account. In table 1 the profile of the research students is presented. ) (ISSN : ), Volume No : 12 Issue No : 1..Page 18

19 TABLE I. THE PROFILE OF THE RESEARCH STUDENTS Students profile No % Gender Male 33 57% Female 25 43% grade 28 52% 2nd 30 48% Grade 1st grade According to the research results, 45 students declared excellent or very good knowledge of computer use, whilst 11 declared good or average knowledge (2 students didn t answer). Also, 54 students declared excellent or very good, while 4 declared good knowledge of internet use. From the study of the recorded meetings is shown that 23 students (39.6%) didn t participate or were never connected to the e-tutoring program. From the remaining 35 students, 15 (25.8%) were connected at least once to both subjects; 9 (15.5%) only participated in ancient Greek and 10 (17.2%) only in mathematics sessions. As a sum, 137 students connections occurred, from which 49 were in ancient Greek and 88 in mathematics. Only students connected at least once to e-tutoring were called upon to answer questions concerning the environment and the educational procedure in which they took part. The results showed high student satisfaction, of the program in which they participated. More particularly, 32 out of the 35 students who used the program were satisfied (Figure 2), while 33 out of 35 deemed it successful. At the same time, 28 out of 35 students declared that e-tutoring responded to their needs. On top of that, 80% of the students declared that through e-tutoring they received individualized support aimed at learning. As far as the program s environment is concerned, 30 students declared that it was easy to use and 33 that it was friendly. Furthermore, 33 out of 35 students said that the environment was safe. Also, roughly 89% of the students declared that through the environment they were provided with the help necessary to master the subject s material. However, only 20 of the 35 students admitted that e-tutoring helped them control the progress of their learning. In this context, a rough 89% of the students said that through e-tutoring easy communication between the e-tutor and students was developed, but only 52% of them said that it was easy to communicate with their classmates. Approximately 89% of the students declared that the material provided through e-tutoring by the etutors responded to their needs and 85% of them considered this material sufficient. Figure 2. Overall satisfaction of e-tutoring ) (ISSN : ), Volume No : 12 Issue No : 1..Page 19

20 B. Reliability analysis Regarding the reliability, Cronbach s alpha indicators was applied [17]. According to Fornell and Larcker, Cronbach s alpha value greater than 0.7 indicates a high reliability [18]. The result of the test revealed acceptable indices of internal consistency is C. Correlation analysis A correlation analysis followed using X2 method, in order to detect substantial statistic differences between the students assertions. In the analysis the correlation concerning gender, grade, knowledge of computer and Internet use, the possession of Facebook and Skype account was investigated. From the data analysis is demonstrated that there is no apparent correlation between the use of e-tutoring and the level of knowledge of computer and Internet use, nor gender and age. VI. DISCUSSION E-tutoring programs constitute a contemporary approach to teaching and learning, aiming to the coverage of students needs. The present research didn t detect statistically important differences between the participants gender, age and knowledge of computer and Internet use. It appears that the research s students don t have differences as far as the use of both digital environments (Facebook and Skype) and educational environment of e-tutoring are concerned. The results agree with resent researches that refer generally to e-learning environments [19], [20]. According to the above mentioned researches, the differences between boys and girls are rare, since digital environments and new technologies have been integrated in young people s daily routine. Also, from the data analysis was shown that that the students participation in the e-tutoring program did not depend on the possession of a Facebook and/or Skype account. The results show high student satisfaction and great acceptance of the e-tutoring pilot program. Students satisfaction and acceptance of similar programs on international level is also high [22]. For the students it was an innovative program they hadn t used in the past and, according to them, the learning benefits they acquired contributed to the improvement of their school performance. E-tutors provided students with material that met their personal needs, as these were defined by the students themselves. The students connected to the environment informed the e-tutor of the material of the morning course on which they wished to be supported, and the e-tutor with appropriate presentations and targeted questions, tasks and exercises tried to help students overcome the learning impediments they had spotted. In this context, it was very important that students defined their personal needs, which enhanced their self-awareness concerning the knowledge they had obtained during the morning course or their study at home, as well as the deficiencies they had located and wished to be helped to overcome. Nonetheless, from the research s results it was shown that only 6 out of 10 students thought that e-tutoring helped them control their progress of learning, a fact that demonstrates that the rest of the students considered e-tutoring as a tool that contributes to the overcome of daily learning problems and not to the overall improvement of their learning. The above results are in agreement with the research [12] who state that with e-tutoring programs is succeeded the improvement of some students and especially those with low performances. Through the environment was reinforced a form of communication between e-tutor and students which is difficult to achieve within the classroom context, since personal contact of each student separately and the meeting of his/her needs was considered important by the students. As a result, they consider crucial the e-tutor s role as a facilitator to overcome the learning obstacles they had spotted themselves. This role appears also at Brychan et al. research, where the e-tutor needs to possess special communication, feedback to students and organizational ) (ISSN : ), Volume No : 12 Issue No : 1..Page 20