Real-Time Caption Streaming over WiFi NetworkD. Maniezzo∗†, M. Cesana†, P. Bergamo†, M. Gerla∗, K. Yao†∗Computer Science Department†Electrical Engineering DepartmentUniversity of California Los Angeles - UCLALos Angeles, CA 900095-1594, [email protected], [email protected], [email protected], [email protected], [email protected]—The use of laptop and handheld computers ineducational environments has changed the nature of teaching,introducing new ways for students to interact with materials,teachers, and their classmates. Technological advances allowhandheld devices to be equipped with faster processors andwireless interfaces, making the performance comparable to lap-top computers. In this paper we propose a simple but effectivescheme according to which each student can visualize in real-time and store the captions of the ongoing lecture. The systemis based on IEEE802.11b Multicast protocols and implements aredundant transmission mechanism to mitigate the errors due tothe unreliable wireless channel.I. INTRODUCTIONTechnological advances in computing will make classroomsand laboratories much more accessible and effective. A newgeneration of high technology classrooms is becoming anecessity on college campuses, and most teaching institutionsrecognize the need to incorporate computers connected to theInternet into their students’ educational experience. Handhelddevices have been successfully used to encourage studentcollaboration and interaction in the classroom [1] [2] [3] [4][5].The rapidly decreasing price of handheld computers com-bined with wireless connectivity can boost the spreading oftechnology within the educational process. Students can beeasily given access to much richer and real-time informationuseful for their own learning process. For example, theycan access distributed resources such as electronic libraries,internet databases, personal file, database collections, real-timedata and laboratory equipment outside the classroom, all whilein a lecture, discussion, or group meeting. Mobile handheldor laptop computers with wireless connectivity to the Internetintroduce the opportunity of mobility for students and thechallenge of effectively using the tools for educators.Furthermore, technology can be extremely useful in thosesituations where a particular support to the education processis needed. As an example, video support during classes canhelp people with hearing impairments.In this paper, we propose a simple but effective schemebased on the IEEE802.11 technology according to whicheach student can visualize in real-time and store the captionsof the ongoing lesson (Figure 1). Such a solution can beuseful in many real situations: to help the comprehension offoreign students, students with hearing impairment, or just asa support to notes taking during classes (students can focuson understanding the lecture since notes are taken by theelectronic devices). Furthermore, our proposed scheme can beintegrated with the lecture slides, and a synchronized systemhaving slides and associated caption can be implemented.A similar application, named Multicast Power Point(MPPT), has been recently proposed by Microsoft [6]. Thebiggest difference between our solution and MPPT is thatin our case the information is transmitted in pure multicastmanner with one active sender and several passive receivers.In MPPT, on the contrary, some kind of signalling informationexchange is required from receivers to sender. For this reason,it is not well suited for implementation in a wireless broadcastscenario based on IEEE802.11 where multiple collisions at thesender side can happen if the receivers have to acknowledgethe sender transmissions.In this work, we focus on the caption streaming systemleaving the slide transfer to future studies. Key points of oursolution are:• Speech recognition algorithm running in the central host,• Multicast transmission via IEEE802.11b from the centralhost (sender) to the devices (receivers);We have implemented our solution in a testbed through whichwe are able to obtain some performance evaluation.The paper is organized in the following way: in section IIwe provide a detailed description of the basis of the proposedsolution, in section III we describe the testbed we implementedand give some performance evaluation. Finally, section IVcontains some concluding remarks.II. WIRELESS CAPTION DELIVERY SCHEMEA. Basic Idea and Protocol DescriptionWireless multimedia can be a powerful support to education.Today, almost every student owns a portable device whether itis a laptop or a PDA (Personal Digital Assistant). The generaltrend of many institutions is to introduce technology in theirclasses as an effective support to the learning process [7] [8][9].We propose a scheme which can easily be implemented inany classroom scenario. We will name our scheme WirelessCaptions Delivery (WCD). In our scenario, each student hasa IEEE802.11b wireless PCMCIA card to be plugged eitherin a laptop or in a PDA. The instructor is equipped with aheadset and an IEEE802.11b enabled station, running a speech0-7803-7724-9/03/$17.00c 2003 IEEE 500Fig. 1. Caption Streaming over WiFi Networksdetection software. The voice of the lecturer is captured by theheadset, converted into a text file by the speech recognitionsoftware and sent via a IEEE802.11b multicast transmissionto all the students. In the following, we present in details theproposed solution and its features.In this paper, we focus on the transmission part of oursolution and we make simplifying assumptions on the speechrecognition part.The lecturer’s host runs a client program which handlesthe multicast transmission and is wirelessly connected to thereceiving stations that run a server program able to receive thepackets and display the speech in a real-time fashion.The choice of multicast protocol is simply due to the point-to-multipoint nature of transmission. To understand our choice,let us consider the usual communication scheme. In order todeliver captions to students’ terminals, a number of unicastsockets (peer to peer reliable transport layer communications)equal to the number of students could be used. Since captionsare to be sent over each socket, the wasted bandwidth is verylarge. The actual problem is that such a scheme is not scalableand it could generate a dramatic congestion in situations ofvery noisy wireless channel.In the proposed scheme, there is a sender (the lecturer’shost) and
View Full Document
Unlocking...