A Tele-Education Oriented Experiment based on an Integrated Terrestrial/Satellite Network *Davide Adami, **Piero Castoldi, °Mario Marchese, °°Andrea Morelli, °°°Luca Simone Ronga CNIT - Italian National Consortium for Telecommunications * CNIT Pisa Research Unit, University of Pisa, Via Diotisalvi, 2, 56100, Pisa (Italy). ** CNIT National Photonic Networks Laboratory, Via Matteucci 34/L, 56100 Pisa (Italy). ° CNIT Genoa Research Unit, University of Genoa, Via Opera Pia 13, 16145, Genova (Italy). °° CNIT Parma Research Unit, University of Parma, Parco Area delle Scienze 181/A, 43100 Parma (Italy). °°° CNIT Firenze Research Unit, University of Firenze, Via di Santa Marta 3, 50139 Firenze, (Italy). E-Mail: davide.adami, piero.castoldi, mario.marchese, andrea.morelli, luca.ronga@cnit.it Abstract The paper presents the architectural solutions and the experimental results of the integration of two networks, a Campus terrestrial Network and a Geographic satellite Network, which have been designed and deployed by the CNIT (Italian National Consortium for Telecommunications). The two networks were developed in different contexts and use a different approach, but they have the common aim to demonstrate the feasibility of a tele-teaching system using standard and ad-hoc software application. The result of the integration is a nation-wide distributed tele-education network with a large number of clients. It is composed of an ATM/ADSL network within a Campus and of a TCP/IP satellite network, which acts as a backbone for the geographic distribution. On one hand, the ATM portion of the network allows reserving specific network resources for each flow (even if it is not necessary in the experiment performed due to the high capacity available), on the other hand, through a satellite network, whose bandwidth is limited to 2 Mbits/s, a TCP/IP suite is used. TCP/IP protocols offer a best-effort service but, in this case, a guaranteed service should be offered. There are two methods in the literature to reserve resources in a TCP/IP-based network: the Integrated Services and the Differentiated Services. The former has been chosen in this environment, on the base of previous experimental studies and of the test-bed dimension. The experiment of integration has been investigated by using both subjective and objective performance parameters. Network users (students) have been asked to fill in reports to measure the Perceived Quality of Service, in order to get a Mean Opinion Score (MOS) about the level of user perception of the overall tele-teaching service. Measurements of packet loss and jitter have been also performed in order to assess the Quality of Service of the integrated network. I. INTRODUCTION There is an increasing interest for the delivery of multimedia services to the home and business user. Services fall within different frameworks such as education (tele-teaching, tele-tutoring, courseware on-line), entertainment (hi-fi radio broadcast, jukebox music, movies, pay per view, tv channels, network games), public services and administration (e.g. tele-diagnosis, trading on line, tele-banking, tele-meeting). Many of them require a high level of interaction and involve different media as video, audio, document sharing, data broadcast from sensors. The offer of different and heterogeneous services implies the support of a suited telecommunication network. Some types of networks, e.g. ATM (Asynchronous Transfer Mode) have been designed to support QoS (Quality of Service) for specific traffic flows. A statistic investigation to verify the availability of the resources to guarantee a fixed level of service is performed before accepting a new call entering the network. Call Admission Control (CAC) mechanisms are used in this case. On the other hand, the Internet is a “best-effort” network, (TCP/IP protocols [1] have not been designed to provide guaranteed quality of service). Both the approaches offer advantages: ATM networks are "structured" to support QoS through simple and efficient methods to reserve bandwidth. TCP/IP networks have the advantage of being a standard de facto. Most applications to transmit data, voice and video are based on the TCP/IP suite and this protocol stack simplifies the use and development of commercial applications. Moreover, in more restricted contexts, as well as in private networks designed to support specific services (e.g. a tele-education network oriented to academic audience, as in this paper), it allows a simple interconnection with other existing networks, as, for instance, the Internet. In the same private environment, also different approaches to the distribution can be used. A wireless local loop can be used to serve users that are concentrated in a limited area; a satellite coverage can be deployed to serve scattered users on a geographic area; solutions based on coax cable or fiber optics can be used for users within a campus metropolitan area. Actually, a satellite backbone offers some advantages, with respect to a cable network, as scalability, wide land coverage and multicast service. Due to this, satellite links are also seen as interesting solutions to provide service to the terminal user [2]. On the other hand, many terrestrial networks are already operative and they offer a good service to the connected users. In this