---

Networking in Latin America: A View from the Fringe
By Ermanno Pietrosemoli
ermanno@ula.ve

Merida is a small university town perched on a mesa between two rivers at the foot of the Sierra Nevada, the northern end of the Andes range.

Although the view of the snowcapped mountains has been a centuries-old source of inspirations for academics, at 700 kilometers (434 miles) by mountain roads from Caracas it hardly qualifies as a technology Mecca. Yet in the past decade it has seen the birth of the first academic computer network in Venezuela and a Wide Area Broadband Wire-less Data Communication Network that was recognized by SuperComm'98 as the best in the Remote Access category.

It has also spawned EsLaRed, the Latin American Networking School, which since 1992 has trained a thousand professionals in various aspects of networking and applications. These accomplishments are the very proof that the Internet can empower the residents of many backwater regions to take a more active role in the global arena.

Ten years ago, the University of the Andes had 40,000 students and 3,000 professors, yet interaction with the rest of the world was very limited. The postal system was slow and unreliable, international phone and fax very expensive, and the university's libraries made journals available only several months after they were published; books took even longer. Although important research was carried out, such as research that led to the development of the Orimulsion (a technique to make viable the tapping of shale oil), it was limited to a few well-established research groups that were able to attract financing. Nowadays, thanks to the Internet, a plethora of extramural relationships has brought new life to the university and the city that boasts 60 cybercafes.

Five members of the faculty--Luis Nunez, Cheo Silva, Edgar Chacon, Edmundo Vitale, and Ermanno Pietrosemoli--began the task of building a computer network, and pretty soon UUCP (Unix to Unix Copy, an early form of computer communication) was used to make a daily dial-up connection to the Internet. The magic of e-mail, news, and file had arrived. Early on, we realized that although the technical and economic hurdles to the growth of the Net were daunting, the most formidable was the lack of enough trained people to manage the network. So we started by offering to a group of our brightest students the opportunity to follow certain special extracurricular training in Unix and network management. The students formed the core of our network administrators. We did not have enough expertise in all of the areas required, but Nunez had just attended the First International School on Computer Network Analysis and Management, held by ICTP (the International Center for Theoretical Physics) in Trieste, Italy, so he proposed that we should try to do something similar in our town. I must confess that I argued that such an enterprise was well beyond our financial and technical capabilities, but I was finally convinced by the enthusiasm of my colleagues. Taking advantage of my forthcoming sabbatical leave of absence, a plan was drawn for me to spend three months at ICTP during the preparation and realization of the second edition of the Networking School. The rest of my leave I spent at SURANET (the Southeastern Universities Research Network) in College Park, Maryland, working with Prof. Glenn Ricart and at Bellcore in Morristown, New Jersey, with Andres Albanese.

Ricart introduced me to Saul Hahn of the Organization of American States, who was involved in an ambitious program to provide Internet connectivity in the hemisphere and who immediately endorsed the idea of a training activity in Venezuela, pledging a substantial financial contribution.

Back home, the rest of our group was also very busy, gathering support from several organizations that provided networking gear, and thus during three weeks in Nov-ember 1992 we were able to assemble 15 international experts who donated their efforts to train 45 participants from 10 countries in the region, aided by a contingent of students from our university who helped in the lab activities.

There was consensus among the participants that those training activities were worth repeating, and so the following year we helped in the organization of Escuela Latinoamericana de Redes-Taller de Lima in Peru, and in 1995 we had the Second EsLaRed, the third in 1997, and the fourth in 1999--all of them in Merida, which had proved fertile ground for training activities due to its ideal climate and relaxed atmosphere. The number of participants kept growing steadily, so that in 1999 we had 200 from 17 countries, distributed in six different tracks, including one specially dedicated to librarians, under the leadership of Steve Cisler. Gradually, the teaching responsibility was shifted to Latin Americans, too many to mention, but I cannot overstress the contributions of Raul Echeberria from Uruguay and Sylvia Cadenas from Colombia. A complete list can be found at (www.eslared.org).



The author (foreground) and engineer Javier Trivino with a multisectored antenna at La Aguada, a 3,500-meter mountain overlooking Merida.


One of the instructors of EsLaRed'92, Ben Segal of CERN in Geneva, suggested that we contact the Internet Society (ISOC) which was conducting similar training activities. In 1997 Vitale attended in Montreal the Internet Society workshop for countries in the early stages of Internet connectivity and met George Sadowsky, then ISOC president of education. The two agreed that EsLaRed should organize a Spanish version of the Internet workshop, then being taught in English and French. So in 1998 we went to Rio de Janeiro, where Jose Luis Ribeiro Filho from the Rede Nacional de Pesquisa sponsored WALC'98 (Workshop para America Latina y el Caribe) at the Universidade Federal de Rio. WALC'99 was held in Merida merged with EsLaRed'99, while WALC2000 took place at the Universidad Autonoma de Mexico (UNAM) in Mexico City, spearheaded by the president of the Mexican chapter of the Internet Society, Alejandro Pisanty.

While all of these training activities were taking place, the University of Merida computer network, REDULA (Red de Datos de la ULA), was vigorously expanding, making use of 26 kilometers of single-mode fiber-optics cable that supports both a 100 Mbit/s TDM (time division multiplexing) data and voice network and a 155 Mbit/s ATM (asynchronous transfer mode) that span the city of Merida. It is worth noting that we installed the university fiber-optic network when the Venezuelan PTT had no fiber in Merida.

But what about the neighboring villages and even those places in town out of reach of the fiber cables? Access by telephone modem is provided, but in a country with such a limited telephonic infrastructure that the number of cellular phone lines amounts to triple the number of land lines, wireless is clearly the way to go.

During the lab sessions of EsLaRed'92, participants were able to get acquainted with wireless data networks based on radio amateur equipment making use of KA9Q, a software suite that implements TCP/IP over radio and that was developed by Phil Karn. Ham radio bandwidth is limited to 25 kHz in the VHF band and 100 kHz in the UHF band, which limits attainable speeds to 9,600 bit/s in VHF and 56 kbit/s in UHF. For faster speeds-up to 2 Mbit/s--another approach was to use Wavelan radio-modem cards built by NCR, which worked at 915 MHz using the spread--spectrum technique, albeit with a more limited range. The participants were able to compare both of the wireless data solutions and choose the one better suited to their needs.

We also made use of these techniques to provide access to REDULA and the Internet, taking advantage of a nearby 3,500-meter (11,483 foot) mountain, La Aguada, on which we installed a repeater that commands the city of Merida and its environs. The repeater initially worked at VHF for greater range, but by fitting the Wavelan cards with Yagi antennas built in the university's communications lab by Prof. Nestor Angulo, we were also able to reach up to 10 kilometers at 2 Mbit/s from the same site.

By 1996 Fundacite Merida, a government organization chartered to promote scientific activities in our state, started chipping in to provide Internet access for remote towns and villages. Our early efforts were focused on spread-spectrum techniques, both at 915 MHz and at 2.4 GHz. Some of the limitations we encountered were interference in these unlicensed bands and diminished throughput as the number of users increased. Clearly, a better solution was called for. In the quest to surmount these shortcomings, our attention was called to a start-up in Nashua, New Hampshire, that claimed a bountiful 10 Mbit/s throughput per sector on 6 MHz channels in a multisectored narrow-band frequency reuse solution. Upon visiting Spike Technologies, we were convinced that Spike's approach was the best suited to our data communications needs, because its patented antenna solved the interference problems, or its IP switching-based scheme allowed for easy integration into our data network and the low latency was well suited to streaming video and IP telephony services. With funding from Fundacite, a trial system was set up in 1997 with four subscriber stations in Merida, the base station at La Aguada, and one repeater 40 kilometers away that gave service to the city of Tovar, 41 kilometers from the repeater, and 80 kilometers from Merida. The performance of the system impressed the governor of the state of Merida, who gave funds for a full-blown system with 400 remote units that would be deployed across the state to provide Internet access for schools, health care facilities, libraries, nongovernmental organizations, and government agencies. The subscribers enjoy LAN speeds with a geographic coverage of hundreds of kilometers. Several concurrent video streams are supported for applications like telemedicine, distance learning, and remote surveillance. The base station has 11 sectors, each with a 15-degree beamwidth. Adjacent sectors have different frequencies, but the frequency is reused in the farther-away sector. A total of four frequency pairs that are 6 MHz wide are used (including repeaters), for an aggregated throughput of 220 Mbit/s. Each sector is an independent LAN, interconnected by a high-speed switch. This network received the SuperQwest Award at SuperComm'98 as the best in the Remote Access Category.

So now people like me who have been waiting 20 years for a telephone line without getting it, can nevertheless access a high-speed data network and communicate with the world.

This article is scheduled to appear in an upcoming issue of ConnectWorld.

For more information
(www.eslared.org)
(www.ula.ve)
(www.walc2000.unam.mx)
(www.wavelan.com)
(www.spikebroadband.com)