From vm.cnuce.cnr.it!abraham Thu Dec 17 20:48:18 1992 Date: Thu, 17 Dec 1992 19:05:43 MET Reply-To: Abraham From: Abraham Subject: Regional INformatics Network for AFrica (RINAF) Project To: "Robert D. Collet" , Tom Dunkenberger , SUSAN DINSMORE YOUNG , bshiflet@icm1.icp.net, skw@merit.edu, smr@merit.edu, gsp@merit.edu, mak@merit.edu, Rusty Schweickart , randy@psg.com (Randy Bush), Daniel Hagan Status: RO Hi! people here I am again. I am very happy that you people are interested in what we are doing here in PISA Italy. So here is a draft of a publication that can give you an idea on what we are doing. There are a lot of things done after this publication so little by little I will let you know. If you have other questions (specific) that you want to know you are wellcome. PUBLICATION RINAF: a network interconnection project of academic and research institutions in Africa L. Abba(*), S. Giordano(**), S. Trumpy(*) (*) CNUCE Institute of CNR Via S. Maria 36 56126 Pisa - Italy ph:+39 50 593211 fax: +39 50 589354 (**) Department of Information Engineering - University of Pisa Via Diotisalvi, 2 56126 Pisa - Italy ph: +39 50 568671 fax: +39 50 550560 E-mail: AFRICOM@ICNUCEVM.CNUCE.CNR.IT Abstract The Regional INformatics Network for AFrica (RINAF) Project was recently started by UNESCO, under Italian Government Funding. The aim of the project is to contribute to the interconnection of academic and research institutions in Africa among them and with the international research community. An investigation on the present status of computer utilisation and network facilities in each country is being conducted and information is also being collected on institutions capable of managing and maintaining the nodes of RINAF. Depending on the presence of organised groups of computer technicians, it will be possible to start the implementation of the network. The regional nodes will have human resources dedicated to the support and maintenance of the network and they will be connected with national nodes which will imply less responsibility and less expensive computer resources. 1.The RINAF Project 1.1 The aim The Regional INformatics Network for AFrica (RINAF) Project was conceived by the Intergovernmental Informatics Program (IIP) of UNESCO in 1985 (01). At that time, no initiatives to set up research network services in Africa existed. Due to delays incurred in gathering fundings and obtaining burocratical approval, the project was started late in 1991 with funding of about 1 million dollars from the Italian Government. The official opening of the project was held in DAKAR, in February 1992. At that time, a number of projects were started under the initiative of different governments, companies or institutions of the more developed countries; some initiatives were also started by the African countries themselves. For these reasons, the RINAF project decided to invest the funding available to promote the use of research network services by cooperation with the initiatives already existing. The aim of the project is to: - use new information and telecommunications technologies to favour exchanges between African countries; - remedy the isolation of development and research institutions in African countries and facilitate dialogue between researchers, academics and industrialists; - develop an operative process for the coordination, integration and upgrading of African networks, as well as exchange with other international networks. The IIP is seeking additional funding for the RINAF project from the governments and/or institutions in the more developed countries in order to reach a tangible improvement in the communications between scientific institutions in Africa and with the more developed world. The objective is to secure funding for an amount of 12 million dollars for the RINAF project. 1.2 The organisation Bearing this financial objective in mind, IIP has set up an organisational structure for the project which has a focal point from each of the African countries; this focal point, designated by a competent Ministry (for Scientific and Technological research, for State modernisation, for University or for other), is responsible for representing the computer communication needs of his country. A first general assembly of the focal points was held in Dakar on the 28-29 February 1992. Decisions regarding grants currently available from the Italian Government are taken by a Steering Committee, which will act during the two years period envisaged to carry out the objectives put into action by the Italian contribution. In Dakar an interim African Committee was set up to propose the actions to be taken in order to take the best advantage of funding available; the African Committee will represent a forum, whereby the regions of Africa define the status of the utilisation of information technology means in the countries and the requirements for RINAF. The interim African Committee is presently composed of representatives from North Africa (Algeria), East Africa (Kenya), West Africa (Senegal) and by the Technical Coordinator of the project. Stefano Trumpy, Director of the CNUCE Institute of the Italian National Council for Research (CNR), has been nominated Technical Coordinator for the RINAF Project. The task of the Technical Coordinator is to propose a set of initiatives and investments to the African Committee, based on the knowledge of the situation concerning network infrastructure and computer usage in the African countries. A structure has been set up in Pisa to support the Technical Coordinator in starting the initiatives, testing the technical solutions and activating the training activities. This structure is composed of : - the network infrastructure team from CNUCE; - the Networks Group set up at the Consorzio Pisa Ricerche (a research institutions and industries consortium located in Pisa), under the supervision of CNUCE; - associations representing the African national students from the university of Pisa; - representatives from the computer manufacturers and TLC providers giving support to the project (DATACONSYST, DIGITAL, HP, IBM ITALCABLE, OLIVETTI). 1.3 The project phases A first phase of the project will consist in an investigation of the situation in African countries regarding: - the status of telecommunications and computer utilisation; - the cooperation, where existing, with international networks; - the cooperation with projects, of a different nature, involving the realisation or the improvement of an existing nucleus of national or regional research networks; - the presence of sites where to invest in terms of hardware, software and training; - the existing cooperation with TLC carriers and computer manufacturers. The sources of information for this investigation are existing literature and specific questionnaires filled up by single countries, under the coordination of the national focal points. This analysis is being carried out with particular reference to cost. Different technical and topological solutions will be proposed depending on the cost and the quality of the services associated. Following this detailed information, the second step will be to decide who can manage and maintain the nodes of a RINAF network. Depending on the services provided, the RINAF nodes will be defined as regional nodes, with wider functions, and national nodes. The regional nodes, up to six, will be those supporting international or intercontinental links, and dedicating human resources to the support and maintenance of the network, as well as to training. These main nodes will be supported by developed countries in order to give a technical background in the field of networking and network management. The national nodes will have the task of improving the utilisation of their national network, or starting it, by using non expensive technology (i.e. switched lines) for access to the basic set of services. The return of the completed questionnaires is expected by the end of April and a proposal to start a first set of regional and national nodes will be prepared by the Technical Coordinator during the following month. The proposal will then be submitted to the African Committee. The second phase of the project will then be the realisation of the approved initiatives. 2 The available technical choices Today telecommunications technology offers a wide spectrum of feasible solutions characterised by different physical media, quality of services and costs. In the following we will summarise some technical choices suitable for African countries. 2.1 Data conveyance infrastructure and terminal equipment The primary component of each computer network architecture is the link which has to be used to establish a remote communication between users or devices located at different sites. The term "link" will be used to refer to channels, lines or virtual data path, depending on the technology used. The transmission systems considered in the framework of an African interconnection will be based on a public or private set of communication links using: - Switched lines typically used to connect a computer to a terminal, or to connect PCs and TTY terminals to an X.25 network. Switched lines can also be used to build up low cost international interconnections; - Dedicated or leased lines used to build up the framework of a much more reliable network, where the process of storing and forwarding data is usually supported by specialised communication hardware; - Microwave links an effective solution to implement long haul links or Radio Frequency backbones (in the range of GHz) in countries where a terrestrial link would be much too expensive; - Satellite links usually considered suitable as backbones for extremely wide area networks in the past, at present they can be used as an access point for low cost stand alone (isolated) computer stations (10); - Packet Radio links form a network developed by volunteers from the world of CB and Radio-amateurs which modulate on radio carriers (in the range of MHz) digital traffic with techniques similar to those used by packed switched networks; one of the special characteristics of these techniques is the possibility to easily move each station; - Fibre Optic links state of the art in developed countries for backbone links; although their cost is decreasing they are still very expensive and often not so easy to obtain from the PTTs. In order to obtain a much more efficient utilisation of the bandwidth of the communication links described above, we need to use particular modems. Modems can be divided into 3 main classes: - Radio modems (or modems for RTTY-Radio Teletype); - Asynchronous Modems; - Synchronous Modems. The main differences between them are related to speed and reliability: asynchronous modems can send packets of the size of a character (typically 7 or 8 bits long); there is no perfect bit synchronisation between sender and receiver; they are inexpensive and can be used on a common twisted pair. Synchronous modems can send longer packets (called frames which typically have a few thousand of bits length); the bit rate is synchronised between sender and receiver, they are faster and generally make use of a double twisted pair (one to transmit and one to receive). Synchronous modems are typically used on dedicated lines and asynchronous modems are used on switched lines. Radio modems are typically asynchronous and need to be connected to a radio transceiver: they are used for radio packet switching networks and have been used recently to store and forward files or electronic mail through satellite links from station to station. Although several network protocols include the error detection and recovery function, in recent time, inexpensive modems can perform similar functions through built-in hardware capabilities. These modems are particularly suitable for developing countries situation, where the signal to noise ratio is particularly low and unpredictable. Recently a new de-facto standard has been realised by multicarrier modems of Telebit which allow to maximise the throughput of a dial-up voice channel (07). 2.2 Network protocols The simplest way to build up a network supporting batch services is to let one system copy its files to another: this leads to a variety of protocols that are the basis for a great number of popular and inexpensive networks; they do not require a great deal of maintenance and can adopt dial-up (switched) lines. The typical applications supported by these networks are based on file transfer and the most used is electronic mail: each computer has a logical address and can be reached by another with store and forward techniques, based on protocols which can route each mail from one site to another. The process is as follows: each computer stores the mail composed by the users either directly from the keyboard of the computer connected to the network, or by dialling from home (using cheap asynchronous modems), and submitting a piece of mail prepared in local mode. During the night (or at predefined time intervals) the computer calls the nearest machine able to forward the mail to the destination computer which will store the file until the user will read it. These networks are very inexpensive and can be built up using Unix systems or less expensive MS/DOS PCs and Macintosh. Each computer needs only a modem and the programs necessary to manage the file transfer. The modems are typically asynchronous but very fast with error correction features, in order to reduce the time of the phone call that the file transfer requires, and thus the cost of the network paid by each node. The other most common networks are built using dedicated lines and special programs designed to let applications (for instance e-mail programs of message handling systems) talk to the operating system of another computer. These programs are sometimes very complex and are referred to as protocols. Depending on the protocols used, we have different networks; the most popular are DECnet SNA networks which are based on DEC and IBM products respectively, but also EARN, BITNET, and NORTHNET which were originally made using IBM protocols were later opened to several other computer and operating systems. Internet is based on TCP/IP which is a protocol designed to interconnect Local Area Networks to set up an heterogeneous wide area network (now the widest network in the world). While these networks were originated from Academic and Research community, PTT carriers have built up their own public data networks called Public Packed Switched Networks which are implemented on the X.25 recommendation of the CCITT and with different names for each country (ITAPAC in Italy, TYMENET in Usa TUNIPAC, SENPAC in Africa, etc.). 2.3 User services Networks may also be defined by the user services supported. It is absolutely essential that data communication services be integrated and made available to users. Computer network services are usually classified on two basic kinds of services (04): - Computer mediated communication services (CMC), which allow people to exchange information; - Resource sharing services, which allow the user to access computing resources (binary or text file transfer, databases access, and CPU power sharing, etc.). Both types of services may also be either batch or interactive. This means that, in an interactive service, a message may be delivered and read immediately or, in a batch service after a delay. CMC services may be primarily either one-to-one (mail), one-to-many (distribution lists or bulletin boards), or many-to-many (news or e-mail conference systems). Many researchers, in more developed countries, see the services offered by a network as an essential support to their research activities and believe that the services available at present are a minimum. They expect to see a great increase in the future. Due to the present situation and the funds available, the RINAF project decided to give priority to the implementation of electronic mail and database access services. 2.3.1 Electronic Mail Computer networks existing at present use dissimilar protocols at the network layer but, at the application layer, there is one service that is converted and interconnected almost universally: this service is the electronic mail service. The user of the e-mail service must have access to at least one of the research networks. This can be realised through a PC with a modem and a common telephone line or, obviously, with direct access to a computer centre connected to the networks. The text of the message is named mail body. The addresses of the recipients and other information form the header of the mail. Software called Mail User Agent is used to prepare the body of the mail, then a process called Local Agent or Mail Transfer Agent picks up the mail and sends it through the network to the recipients. Usually the E-mail user interface helps to: - keep track of correspondents' electronic addresses; - read, reply to, forward, print, and save incoming mail in notebooks, as well as creating, sending, and saving outgoing mail in notebooks. Incoming mail is stored by the system in the mailbox; - reread the mail saved in notebooks, and reply to, forward, print, and/or discard it. A great deal of more sophisticated services offered by computer networks are based on electronic mail and, due to the lack of standardisation at the application level, E-mail is often the only way to interoperate between different networks. It is a powerful way to start up electronic debates on an extremely wide range of subjects. These services are often called Bulletin Boards or E-Mail Conferences (04). They can be among a closed group or an open group of users. Each debate has a coordinator and these services are the best way to establish contacts with people with mutual areas of interest. It is possible to subscribe or sign-on to a list of users with the same interests, receiving in this way all the letters sent to the bulletin board or to the electronic debate. A user can of course sign off (or un-subscribe) from the list when he is no longer interested in the subject. 2.3.2 Data Base Access Network servers dedicated to distributing text files (11) (e.g. documentation), public domain software, and recorded mail of interest to the community are the typical data base systems present on a wide area network. By mail (02), and sometimes by real time messages, it is possible to submit queries to the data base and (by mail) they will be received at the user destination mailbox. Thus usual access to a data base is not in real time or interactive . The user has to submit the query and then wait for the system and for the network to receive the answer back. Apart from this slight inconvenience, remote access to automatic information and public domain software on the network is extremely appealing to the network user. With E-mail this is one of the most popular services among remote users. For developing countries this is a possibility of great interest as apart from interchanging information between network users, this is sometimes the best way to receive information of great interest about research, learning material and other social activities. Access to databases may of course be achieved in an interactive way with remote login to the site where the database is maintained (12). This application requires faster, more reliable links and also the possession of a user account which implies costs. 2.4 Computing equipment With the recent boom in low cost PCs and software packages to interconnect to international networks, the single user workstation has become the most common and low cost network nodes. A single workstation may also function as a mailbox to connect a number of end-users. Therefore the RINAF project may contribute to increase the utilisation of network services by implementing a number of mailboxes built up on standard MS/DOS PCs or workstations. In cases where some multiuser computer centres already exist, the RINAF project may contribute by connecting them to a network and improving the international connectivity. In the latter approach, the whole user community of the existing computing centres will immediately join the international user community. 3 Present status of network activities in African countries Some African countries are already connected, or plan to be connected, to the international research networks. Based on information from different sources, it appears that there are several countries in Africa with an X.25 network operating like Tunisia, Egypt, Senegal, Mauritius Is., South Africa, Ivory Cost, Gabon, Niger, Mozambique, Namibia, Zimbabwe, Togo and Chad. Several other network initiatives, or projects involving multinational data networks in different stages of realisation, are (13): Esanet (between Uganda, Tanzania, Zambia, Zimbabwe, Kenya), Rio (between Senegal, Mali, Niger, Burkina Faso, Togo, Cameroon, Congo, Ivory Cost), Ngonet (between Tunisia, Senegal, Kenya, Zimbabwe), Wednet (between Senegal, Burkina Faso, Ghana, Nigeria, Sudan, Kenya, Zambia, Zimbabwe), HealthNet-Satellife (between Uganda, Kenya, Tanzania, Zambia, Zimbabwe), Padisnet (based at the United Nations Economic Commission for Africa, UNECA, with 34 countries connected), Worknet (South Africa), Mango (Zimbabwe, South Africa), Arsonet (Ethiopia, Senegal, Kenya, Egypt), Uninet-ZA (South Africa, Botswana, Zimbabwe, Namibia, Zambia, Kenya and Ethiopia), Afrinet (Kenya, Zimbabwe, Tanzania, Uganda), Afrikanet (Cameroon), EARN (Tunisia, Marocco, Egitto, Algeria). Many of these networks are implemented using Fido Technology. Other solutions are achieved by means of mini (Unix) systems connected to an X.25 public network and running the TCP/IP (encapsulated) protocol on a packet switching network. Several personal computers are connected to a mini, sharing access to e-mail and all the other services available on Internet. Another interesting solution is offered by using UUCP on a Dos machine emulating the behaviour of a Unix system for all that concerns the store and forward of mail and files. The connection to UUCP gives the user access to the E-mail and file transfer services on Internet. 4 Some technological solutions for RINAF Following the description of some technical aspects of networking and the present status of network activities in Africa, we should like to highlight some considerations related to the current directions of the evolution of academic and research networks in developed countries. The task of building up an interconnected network (an internet) composed of different local area networks was pointed out by the academic and research community at Arpanet's time. The idea was to use a special protocol called an Internet Protocol, useful to mask the physical, topological and architectural heterogeneity of single networks in a logical, virtual, and unique network. Proprietary solutions from major computer manufacturers were oriented to separate directions with only willingness to migrate to OSI (Open System Interconnection) standards sometime in the future. Large, well known networks like EARN, BITNET, HEPNET, etc. were built up using a special set of protocols originally built for a particular system. Different systems could be connected to the network implementing the same stack of protocols, or in other words, having the same behaviour, from an external point of view, as the proprietary systems. Now a new process has emerged in the academic and research community as well as in the industrial community: it is founded on a multivendor, multiprotocol networking paradigm oriented to meet user requirements of any scale, configuration or application. The term "internet" will be used as the general term for a global, mainstream network architecture. Internet architectures will become the general-purpose network infrastructures providing the interconnection of LAN and WAN physically based on fibre, copper, coax, RF (radio frequency) interconnections characterised by a variety of data link techniques: the building blocks of such architectures are bridges, routers and gateways (06). We will try to realistically face the tasks aimed by the RINAF project in the framework of this multiprotocol networking paradigm. Depending on cost, services, expected quality of services, and the present computer network infrastructure of each different country we will try to reduce the wide spectrum of possible solutions for a network interconnection of African research and academic institutions. One of the most important parameters in trying to size a network interconnection, or any network access, are the link capacity and their expected utilisation. Link capacity is the most expensive part of a network and unfortunately the cost of a link and its capacity are not easily related. Moreover these costs differ for each country. Although in the future the allocation of bandwidth will be "on demand", at present the bandwidths are fixed. Leased connections are characterised by higher throughputs than dial up connections which usually correspond to voice channels (less then 4 KHz); another characteristic of dial up connections is the bad signal to noise ratio which can prevent the use of common modems at maximum speeds (07). By making use of complex signal modulation techniques it is possible to transmit data at speeds near to the theoretical limit imposed by the signal to noise ratio of the connection in an adaptive way. The use of error detection and recovery techniques, together with the use of efficient data compression methods performed by the modems, make it possible to reach data rates which, in the past, were an exclusive prerogative of dedicated connections (05). Those modems can use different modulation techniques by selecting a relevant subset of the carriers having the best signal to noise ratio. In terms of software and computer systems, it is possible to use special purpose programs oriented to a specific network services: the most common are electronic mail and file transfer. Using this software, a common stand alone PC with enough hard disk space can be considered as a node of a store and forward network, supporting e-mail and file transfer, implemented using fast modems and common switched lines or radio connections (04). The task of the RINAF project will be achieved making use of technical approaches which will differ for each different country, and be modular and flexible as the network grows. As mentioned before, our intention is not oriented to any special network architecture, but, depending on lines, systems and telecommunication devices available, we will try to present "de facto" standard solutions conforming to the academic and research activity of the rest of the world. At the very least, using only a telephone line, a good modem and a PC, it will be possible to build up a node of a wide area network. We call a node a system which can be accessed by more than one user for at least electronic mail applications. The minimal solution for the interconnection of an institution will be based on the so called "FIDO technology" (considering technology as the stack of programs utilised to build up a node using an MS/DOS system). The FIDO network (called FIDONET) is an extremely wide spread network based on store and forward capabilities for the transfer of files, texts and mail, using switched lines. Each node has a modem which lets users send or receive mail by logging in, without any charge, using a common VT100 terminal emulating program (like kermit, procomm, xterm, simpc, crosstalk, etc.).Using several start-stop file transfer protocols each user can also send or receive files from the node in the same way as a micro sends or receives files from a mainframe or a mini. To give to every user the opportunity to use the network the FIDONET node assigns each user a maximum time, after which the user is cut off by the system. Access to the system is in this way similar to that offered by BBS (Bulletin Boards Systems). Of course this kind of access can be cost effective only for those users very near to the FIDONET node. During the night, or in periods of time corresponding to lower billing, the FIDONET node automatically calls the nearest node or the regional node directly. The cost of these calls are lower in comparison to the remote access performed by the normal users, by data compression and efficient forwarding techniques. To reduce the cost of phone calls, and to automatize the file transfer procedures, remote users may have their PC configured as a FIDO POINT, which is a sort of sub node of a Fido node. Using a POINT the files are automatically transferred after a data compression process: in this way mail and files are sent and received in a faster way. FIDONET is organised in several regions corresponding to continents. In each region there is a node with the task of storing and forwarding the traffic produced by the entire continent to the other continents. The trasmission of mail and files can usually be achieved by the multiple storing and forwarding of files from a system to the nearest one reducing the cost of the calls performed by each system. Using a multicarrier modem it is also possible to specify directly to the system where to call giving an explicit control of the routing of files and mail. Fidonet is also supported by packet radio links or satellite connections which make it possible to use this technology where no telephone lines are available. Fidonet already has several interconnections to Internet via e-mail gateways. This "lower level" solution based on store and forward technique is similar to UUCP (Unix to Unix CoPy) solution which is at the base of widespread networks like: Eunet, Usenet, Junet, SDN, Auseanet, Pacnet. There are three principal classes of protocols in UUCP networks: the "g protocol" which breaks data into packets, uses checksums to detect errors and retransmits, when necessary, over common switched lines; the "f protocol" which follows the same procedure but over an X.25 network, and the so called "t protocol", used by UUCP networks based on TCP/IP connections. The limitation of this solution is that it is not possible to make up interactive sessions on remote systems (remote login). The access to remote data bases is only possible in a "batch" way. Mailing and file transfer capabilities are delayed considerably by the store and forward mechanism performed by UUCP or Fido systems. Interactive sessions can be activated on TCP/IP networks. We have looked at three possible types of approach to TCP/IP networking: using switched lines, using X.25 networks, and using leased lines. All of these solutions were developed in the framework of Csnet (Computer Science Network) (04). Now, taking costs into consideration we will propose a technical solution based on real TCP/IP connections on switched lines using specialised hardware Dial-up routers. Dial up routers can be linked to a modem pool connected to the public switched telephone network. Each routers are connected to one or more local area networks and when the address of the packet is for a remote network the router establishes a connection to another remote dial up router directly interconnected to a TCP/IP Wan. In this way one or more LANs are connected to the IP network and their users can continue to use standard application programs like Telnet (for remote login), Ftp (for binary or text file transfer) or e-mail with a common Internet address scheme. The router can also act as a terminal server for a single workstation or PC using SLIP (Serial line IP). When an IP connection reaches the maximum utilisation offered by a switched line and a fast modem, it will be possible to use the Dial up router as a common router (LAN to leased line router) using an optional card for 56Kbit/sec connections. If higher speeds are needed building up a backbone, the system can act as a backup router. These dial up routers also have the capability of line sharing over multiple dial- up lines. This means that leased lines can be simulated on a temporary basis over the public switched network. Due to the great diffusion of X.25 networks in some of the African countries, TCP/IP (or UUCP) networks can be based on X.25 connections. These solutions are similar to X25net a TCP/IP network built in Csnet for users able to use TELENET (a public X.25 network in Usa) to reach Internet (04). As an alternative solution, conforming with the choices of the research and academic community in terms of network interconnections, the use of multiprotocol routers (e.g. cisco, Proteon, Wellfleet, etc.) will be proposed in countries which have institutions or centres already based on different network architectures for interconnections within the countries. In summary, depending on the availability of lines, modems, PCs, Mini or Mainframes and know-how on different operating systems, RINAF will contribute to build up a multiprotocol architecture oriented to users services, flexible in terms of modularity, and upgradable where better telecommunication resources are available. >From very low cost solutions based on high quality multicarrier modems and stand alone PCs, the regional nodes will be based on solutions which give users the same opportunities as a direct leased line connection to the Internet world. TCP/IP protocols are extremely important for their diffusion among workstation users; they are widely adopted for CAD and X-Windows applications. High level interfaces will also produce a great diffusion of networks in countries without a strong background in network architectures and services. 5 Future actions and perspectives The RINAF project has the ambition of setting far more objectives in motion than those that can be reached with the Italian funds currently available. At present UNESCO IIP cannot as yet count on further funding and therefore the following considerations on perspectives concern the goals established for the two year period presently planned. The actual plan is to establish a number of regional nodes (six are planned) and a number of national nodes (at least six; the final number, not less then six, will also depend on the money available and the time to the carry out the project). The hierarchy of RINAF's achievements is only linked to the level of the services offered, to the personnel dedicated to this and to the presence or not of a coordination activity toward other African countries. Each RINAF achievement (national or regional node) will be treated as a stand alone subproject with its own technical solution, support plan from the group in Pisa, training plan, etc.. The Technical Coordinator in Pisa has already set up a support group which is also taking advantage of the contribution provided by a number of computer manufacturers and TLC providers. Following the indications of the Technical Coordinator, they will help in setting up some of the subprojects in which their technical solutions are particularly suited. Something must be said about the approach towards OSI standards. It is widely known that the network research environment is highly heterogeneous; therefore, the adoption of solutions that allow the heterogeneous systems to interoperate among them is mandatory. The international organisations, and UNESCO is among them, recommend utilising, as much as possible, the universally accepted standards, that is, OSI. In the real world, OSI applications are not yet available or not yet satisfying, from a price performance point of view. In some cases, OSI products are in use in the research networks and also in the Italian GARR network. However, the majority of the applications and services interoperate among them making use of standard solutions at levels lower than six and gateways or software emulators. The solutions proposed for RINAF will take the best of the existing situation, in terms of interoperability, and will also aim at installing robust and easy to use systems. The integration into the Internet environment will be considered as a priority issue; this will also ensure a gentle and progressive migration to standard solutions, as this is a primary goal of the research networking environment, and of Internet in particular. The present plan is to implement the RINAF subprojects in a two year period. It is evident to us that, in this period, it will be possible to contribute to improve only slightly the present situation of isolation of the research institutions in Africa. Therefore, in order to be more effective , it is necessary to secure a continuation of the project by making other funds available. It will be a task of UNESCO to secure this objective for the benefit of the research community in the African continent. References (01) G. Biorci , S. Trumpy, (1990): The UNESCO African Network Project in Computer Network and ISDN System, Vol. 19, n.3-5, North Holland November 1990, pp.189-190. (02) L. Abba, S. Giordano, G.A. Romano, S. Trumpy, D. Vannozzi (1990): ASTRA: a service to access databases over the EARN network, in Computer Network and ISDN System, Vol. 19, n.3-5, North Holland November 1990, pp.350-355. (03) R. Adams, D. Frey (1990): | % @ :: A Directory of Electronic Mail Addressing and Networks, O'Reilly & Associates, Inc., Sebastopol CA USA. (04) J.S. Quarteman (1990): The Matrix: Computer Networks and Conferencing Systems Worldwide, Digital Press (05) E. F. Casas, C. Leung (1991): OFDM for DATA Commistion over Mobile Radio FM channels - part 1, IEEE Transaction on Comm., Vol. 39 n.5, May 91. (06) C. A. Sunshine (1990): Network interconnection & gateways, IEEE JSAC Vol.8 n.1, February 90. (07) P. Baran (1991): The developing countries interest in data modems - standardization issues, Inter media (08) N. Lippis, J. Herman (1991): The internetwork decade, Data Communication Supplement, January 91. (09) S.O. Bradner (1991): Testing Multiprotocol routers, Data Communication, February 91. (10) G. Garriott (1991): VITASAT Global communications for development, Volunteers in Technical Assistence, Arlington. (11) E. Thomas (1988): Listserv database function - LSVDB MEMO file on every active LISTSERV/LISTEARN system. (12) NSF Network Service Center (1989): Internet Resource Guide, BBN Systems and Technologies Corporation. (13) CPR Networks Group (1992): Status of networking initiatives in Africa, Consorzio Pisa Ricerche, Pisa, April 1992. Cheers! Abraham Gebrehiwot e-mail: abraham@icnucevm.cnuce.cnr.it RINAF project  Info. Rinaf .