search:

3   Optical Networks

3.1   Research and Development of CEF Networks

In 2005, international research made efforts at practical verification of new networking possibilities based on the use of optical fibres and their new lighting technologies and on cheap and fast free space optical or microwave transmissions. CEF networks, as fibre networks controlled by end customers, have a significant position in this effort. CESNET has the advantage of having focused on this area since 1999. Several European NRENs jointly managed to establish dark fibres as the basis for the new GÉANT2 pan-European network; other telecommunication services are used only temporarily in locations where no fibres are available for lease or the offers are not acceptable. GÉANT2 is thus essentially a CEF network although it does not label itself as such. The use of dark fibres and other services in the GÉANT2 network is shown in Figure 3.1.

Moreover, the grid idea aims at enabling the end users to control and scale their mutual connection and network resources allocation in a heterogeneous environment consisting of multidomain and multivendor networks. Significant is also the requirement to eliminate the inequality in access to high-speed networking services among various regions, countries and continents (the digital divide) and thus to contribute to the strengthening of research capacities and advances of prosperity and culture.

It is widely accepted that existing production networks (including the research and education networks) do not provide sufficient space for networking research, thus slowing down the development of networking technologies and postponing solutions to well-known problems. This is mainly due to the requirement of high reliability of production network services that strongly limits the possibilities for experimenting with new technologies and methods. The same requirement also increases the investment and operational costs of networks and thus limits the networking possibilities in developing regions.

This situation can be changed, for example, by differentiating reliability and price of lambda services within a single network. This idea was used by the NLR (National Lambda Rail) network in the United States: "core wavelengths" are more expensive and have higher guaranteed availability than "flexible wavelengths". For certain research applications, the lambda availability of e.g., 97 % is perfectly acceptable and "four nines" 99.99 % availability means just waste of money for them. Dark fibre providers usually do not guarantee 99.99% availability anyway and the availability of a lambda service can hardly be higher than the availability of the underlying fibre. While availability may be increased through redundancy (usually by the deployment of another, topologically different route), only few end premises have such an option and so the "many nines" service availability may be out of reach anyway. The concept of reliability/price differentiation opens new possibilities for networking research, even if the requirement of a non-destructive character of tests on physical level, on the level of pure optical lambda connection (i.e., without conversion to electrical signal) and on the level of optoelectrical lambda connection must usually be observed. The opportunity for tests in production networks is mostly realised by having redundant network elements (e.g., a separate fibre route may be used for pilot tests of transmission technologies).

Research in technologies for global, continental or national networks requires, apart from analytic, simulation and laboratory studies, sufficiently large-scale experimental testbeds and network facilities that allow setting up networks and optical routes with various parameters for periods ranging from several days to several months. In the USA, we can observe an increasing trend towards making the research and experimental networks independent of the existing national production networks for research and education. The Directorate for CISE (Computer and Information Science and Engineering) of the National Science Foundation is planning an initiative called GENI (Global Environment for Networking Investigations) "to explore new networking capabilities that will advance science and stimulate innovation and economic growth. The GENI Initiative responds to an urgent and important challenge of the 21st Century to advance significantly the capabilities provided by networking and distributed system architectures." See the GENI web pages for more details.

Wavelengths in production networks can also be used for creating virtual testbeds that enable experiments at higher layers of the network. This way, geographically large-scale experiments may be carried out with acceptable costs. However, this approach also limits the potential for investigating all-optical networks (a virtual testbed only uses lambdas without implementing them).

Since the beginning of 2003, CESNET has been participating in GLIF (Global Lambda Integrated Facility). On the national scale, we are building the CzechLight CEF testbed that, among other purposes, also enables its users to access GLIF. This way we combine tests of new networks with network services. In contrast to other countries with highly developed networks, we also concentrate on the issue of cost-effective networks and network services, partly due to our own needs and partly due to the increasing demand for such technologies that we see on the international research scene.

Together with TERENA and other NRENs, we are participating in the SEEFIRE project (supported by EU since March 2005). Within the scope of this project, we help the deployment of CEF networks in south-east Europe - see the project web page. In the second half of the year 2005, we joined the preparatory phase of another EU project - Porta Optica Study. This project promotes the construction of networks based on dark fibre for the needs of NRENs in Eastern Europe. The project was approved and is supposed to start in February 2006.

The second international CEF Networks workshop was held in Prague in May 2005. For three days, the representatives of national research and education networks from 26 countries and also of the most important equipment suppliers were exchanging experience in customer networks design and operation. We mainly appreciated the open discussion on ideas and intentions carried out by researchers from almost the whole Europe and the US, which helps us to use our research potential effectively. We also gave a summary presentation of metropolitan academic CEF networks in the Czech Republic (Brno, Praha, Plzeò, Olomouc, Hradec Králové). These networks were among the first in the world to use dark fibre. The workshop was very successful, its participants highly appreciated the presentations and the preparation of the workshop and recommended to continue this form of the support to CEF networks and exchange of information.

The existing strong interest in dark fibre cross-border connections between NRENs in the neighbouring countries (CBF, Cross-Border Fibre) is of great importance to network research. The cross-border dark fibre connection between CESNET and SANET from Brno to Bratislava has been in use since 2003. At that time it was one of the very first international connections utilising dark fibre. Nowadays, there are already several such international NREN connections in Europe (see GN2 JRA4) and also between states and regions in the USA (cross-border fibre connection of RONs - Regional Optical Networks), see The Quilt. Using such CBFs, it is possible to implement wavelengths connecting NREN/RONs or traversing them. Consequently, universities and research institutions throughout the continent may be potentially interconnected on demand by lambdas. Of course, this will take some time, probably months or even years. Meanwhile, it is necessary to investigate options for integrating these cost-effective and flexible networking possibilities into the existing infrastructure. Another open question is the future role of the successors of the NLR, Abilene and GÉANT2 networks, as they now use their own overlay fibre structure covering the corresponding continent. The costs of fibre lease or ownership contribute usually over 60 % to the costs of wide area networks and so optimisation will be of a great importance. We presented this issue at the Fall 2005 Internet2 International Task Force meeting, in the session "Interconnecting RONs and NRENs and national infrastructure: emerging models for dark-fibre based networks and other optical networking trends in the global R&E community" - see the presentation. We cooperate on this issue with other partners in the GN2 pan-European project.

We also continue our work towards extending and improving the quality of our fibre infrastructure and supplementary telecommunication services for the CESNET2 network. Figure 3.2 shows the current topology of the CESNET2 network.

For the needs of the CESNET2 network and the CzechLight testbed, we investigated possibilities for leasing fibre that could connect premises of our customers with insufficient connection capacity or special requirements. We expect the customer connections to become gradually multi-colour meaning that, beside the standard lambda IP service, some of the participants will also use other lambdas provided by the CESNET2 and GÉANT2 networks, CzechLight testbed and GLIF. From the 22 evaluated cases it is obvious that it is usually rather expensive to lay the fibre all the way to end user premises, although offers in the range of 35,000 EUR per kilometre of optical cable are available (outside urban areas, the price is even lower). This means it may be economical to set up gigabit or ten-gigabit lines for few large universities and research institutions, but not for a higher number of small institutions or regional premises. Other countries in the EU and North America appear to be in a similar situation. Out of the 22 locations that were included in the investigation, only two did not require high setup costs. Two other offers came from one supplier who is able to build the first fibre mile for a setup fee but does not lease fibres and instead offers service from 10 Mbps to 1 Gbps.

Long microwave links with transmission speeds 10 Mbps and 34 Mbps, used so far in the CESNET2 network for connecting small premises, are not fully satisfactory from the viewpoint of reliability (due to atmospheric events, power supply failures at intermediate locations etc.). We are thus trying to acquire L1 or L2 circuits with transmission speed at least 10 Mbps mostly on fibre lines and with the first mile implemented by a relatively short microwave link or free space optics backed up by microwave. Out of the demanded localities, 7 lines at 10 Mbps are now being realised. Their transmission speed may later be increased if needed. At the same time, we started negotiations with the suppliers concerning the cooperation on the development of affordable first mile solutions at 100 Mbps and more, utilising our previous experience in this area.

3.2   GLIF and CzechLight

GLIF turned out to be a unique and very useful environment for network research, services and applications. Users of this environment are various international teams preparing experiments that usually have both network and application part. Such experiments thus usually involve researchers from both transit and "terminal" countries. Existence of such a team is a prerequisite for a successful use of GLIF for experiments. The task is much more complicated than the use of standard network services, but makes possible to verify new ideas and carry out activities and tests that wouldn't be feasible on "stable" networks any time soon. This possibility of testing technologies in advance, when used properly, may obviously be of great significance from the viewpoint of the development of science, research, technology and business.

Another prerequisite is the connection of the participating sites to GLIF with a transmission speed and parameters that enable participation in the experiment. The necessary transmission speed is usually 1 Gbps or 10 Gbps and other properties such as low delay or jitter may also be required. Therefore, it is usually necessary to connect the sites by fibre or lambda to the the GLIF core.

In 2005, the Optical Networks activity coordinated or supported the creation of an environment for preparing and carrying out experiments, participated in these experiments and handled the connection of Czech participants.

A new PoP in Brno was deployed for the CzechLight testbed and 10 Gbps connection between Prague and Brno was set up. A prototype of CLA DI 01 link amplifier has been deployed on that link since the end of March 2005. The link is used for single-colour transmission with an option to increase the number of colours to 8. The original attempts in early 2005 to span 295.8 km using the "Nothing in Line" (NIL) approach failed due to the length and probably also the line inhomogeneity (various fibre types were used together). In August 2005, successful negotiations with fibre providers led to shortening the line to the current total length of 283.9 km.

The reliability of this link was the main precondition for Masaryk University in Brno to join the preparation and realisation of the demonstration of a multipoint videoconference in HDTV quality at the iGrid conference in September 2005 among MU Brno, San Diego and Baton Rouge (Louisiana State University). This experiment used a 10 Gbps connection via Prague (CzechLight), Amsterdam (NetherLight) and Chicago (StarLight) by means of GLIF. The demonstration was successful and was received very positively. See Section 10.2.1 for more details about the experiment. This link was also later reused by MU Brno and CESNET for subsequent events.

By the end of 2005, a new CzechLight fibre link between Masaryk University in Brno and Cieszyn in Poland was set up. On the Polish side it was further connected to the PIONIER network. This route will enable testing new transmission technologies between Prague and Poznañ and also offer Polish researchers access to GLIF. The fibres for CzechLight were mainly obtained as extra bonuses related to the connectivity contracted for the CESNET2 network. The development of CzechLight and its connection to GLIF is shown in Figure 3.3. The CzechLight and GLIF services will also be available via lambda services of the CESNET2 network.

Regional Computing Centre for Particle Physics in Prague was connected to CzechLight. The Centre provides computing and storage capacity for demanding computations of the DO experiment that uses the TEVATRON accelerator at FNAL. Two other experiments involving the LHC accelerator at CERN - ATLAS and ALICE - are being prepared for 2007. For the ongoing DO experiment, the Centre performs simulations of the detector response and experimental data processing. Preliminary simulations are also already being performed for the ATLAS and ALICE experiments. The Goliath computing farm featuring 200 processors and 40 TB of disk space is also a part of the Centre. The Centre is integrated into the environment of international LHC Computing Grid (LCG).

In 2005, CESNET and PASNET finished a dedicated 1 Gbps network connecting several laboratories in Prague that collaborate on particle physics experiments. The throughput was tested by the iperf program (version 1.7.0) that was run each time on one of Goliath cluster servers and, at the same time, on a server inside each of connected institutions (or directly on their access router). Details about the interconnections and equipment used at individual sites and the performance tests can be found in the technical report [Ben05]. The routes we realised using dedicated dark fibres that may be later upgraded to 10GE with multi-colour connections as needed, also for GÉANT2 and CESNET2 lambda services that are being prepared. For example, a lambda connection between the Institute of Physics (FZU) in Prague and Tier1 Centre in Karlsruhe might be set up via the GÉANT2 network. In general, the experiments and services that are already available on CESNET2 and GÉANT2 will not be offered by GLIF and CzechLight. Reservations of GLIF and CzechLight facilities may be terminated due to, e.g., tests of new technologies on the physical layer. Also, the facilities are not allocated to experiments permanently, but according to an agreed plan. For example, it is expected that a 10G lambda service of the GÉANT network will be used for connecting CzechLight and NetherLight in 2006. For particle physics research and experiments, two international 1 Gbps lambdas over GLIF and CzechLight to Tier1 centres have been put into operation. The first of them leads to Fermi National Laboratory in the USA (FNAL), the second to ASGCC (Academia Sinica Grid Computing Centre) in Taipei, Taiwan. Both routes pass through NetherLight in Amsterdam. The route to FNAL has been used since mid 2005, whereas the connection to Taipei was established later, in December 2005. This delay was mainly due to the reservation of full 10 Gbps capacity between Prague and Amsterdam for iGrid95 and the following experiments.

The connection to Fermi National Laboratory has been used intensively for input data and, in the opposite direction, for the results of simulations and reconstructions within the DO experiment. The connection to ASGCC has been used for data transmission between FZU and Tier1 centre in ASGCC within the LCG project. For the next period, we plan to set up a central data repository for a joint project with the University of Alberta (Canada) and make it accessible via GLIF and CzechLight.

Experimental use of GLIF and CzechLight is open to all research areas that may benefit from such an environment. In particular, this includes applications in the fields of visualisation, geographic information systems, cartography and graphics processing. Beside particle physics, another promising area seems to be health care. Together with the "Medical Applications" activity (see Chapter), we are preparing experiments in this field.

One of the first top-level health care institutions that is interested in joining the tests of new applications is the Central Army Hospital (UVN) in Prague. Anther key institution is the Thomayer Teaching Hospital (FTN) in Prague that is interested in and capable of becoming the communication centre for visual data processing in the Prague region along the lines of the MeDiMed project in Brno. Another cooperating subject is the Masaryk Hospital in Ústí nad Labem (MNUL). This hospital is equipped with a system for digital graphics processing and tries to stay at the leading edge in this area. It is also participating in the MediGrid Information Society project of the Academy of Sciences. The three hospitals mentioned are already interconnected by dark fibre, which opens the opportunities for application testing.

For example, FTN is interested in providing consultation services from its sophisticated neuroscience lab at UVN. As both hospitals are equipped for filmless PACS (Picture Archiving and Communication Systems) processing, their interconnection might be very useful. However, such a connection has to meet at least the following two requirements: sufficient parameters of the connection between FTN and UVN and safety of data transmission against wiretapping. The lambda service satisfies both aspects: it provides transmission with negligible latency and, moreover, the use of a dedicated wavelength ensures the separation of this traffic. During the pilot operation, we will analyse the options for establishing an appropriate security policy, and also the alternatives for a direct interconnection of disc arrays of the medical data archive using the FC (Fibre Channel) protocol.

A different situation is in the case of the planned interconnection between MNUL and UVN, where the network presence and back-up of PACS data repositories have been considered since the very beginning. At the two sites, we will test the quality and reliability of data transmissions and scenarios for emergency situations such as outages of the data repository due to equipment failure or service procedures. An important advantage is the geographical distance between the two sites, approximately 100 km, which nearly eliminates the danger of both systems being simultaneously damaged, even if the worst is taken into account. This is crucial especially for UVN, which is a part of NATO structures.

The connection of UVN to CzechLight and GLIF allows to prepare this hospital for joining the world-wide project Biomedical Informatics Research Network (BIRN), whose coordination centre is at the University of California, San Diego. This project currently includes neuroscience research teams from 30 universities and 21 other organisations in the USA and Europe.

3.3   Methods of Data Transmission in CEF Networks

By means of proprietary simulation software, we analysed the possibilities for transmitting multiple (DWDM) signals at 10 Gbps (the main focus being 10GE) over distances up to 1000 km. Laboratory verification was limited by the fact that we had only 4 EDFA amplifiers at our disposal. For further tests we plan to use the CzechLight amplifiers (CLA). Theoretical and experimental results will be utilised for further development of the CzechLight experimental network, particularly for the fibre link between Brno and Poznañ.

Our experimental evidence shows it is possible to transmit 4 DWDM channels at 10 Gbps over the distance of 250 km with G.652 fibre and no in-line equipment. The block diagram of this experiment is given in Figure 3.4. Using the same setup, we also tested fixed and tunable fibre Bragg gratings (FBG) from TeraXion.

By compensating for chromatic dispersion using the FBG, we were able to replace expensive compensation fibre modules and, at the same time, decrease the number of necessary EDFAs. First results of multi-channel 10GE transmissions utilising the NIL method were presented at the ONDM conference in Milan. At the same conference, researchers from TU Copenhagen presented similar results with 10GE transmissions for the same distance (252 km), but with the use of an in-line amplifier.

A prototype of the CLA PB 01 amplifier that was installed on the link between Prague and Hradec Králové operates in a unique configuration known as OSA (One Side Amplification). The status of this amplifier was changed from experimental to production in March 2005 and has been working without any problems since then. Further deployments are planned on the CESNET2 link Prague-Ústí nad Labem. Another prototype CLA DI 01 was successfully deployed on the CzechLight line Prague-Brno as a single in-line amplifier. Currently, licensed production of the CLA equipment is under preparation.

We started the construction of a tunable dispersion compensator based on multichannel tunable FBGs. We tested the possibilities for FBG deployment on the CzechLight line Prague-Brno. Furthermore, we devised a new solution for the links Brno-Bratislava and Brno-Vienna utilising chromatic dispersion compensation by means of FBG. For these links and other applications, another EDFA amplifier prototype of the CLA family is being worked on - a booster.

In addition, we experimented with increasing the reach of transmission equipment operating in the 1310 nm band. The primary motivation is to eliminate the expensive transponders that are necessary for connecting powerful servers and clusters via 10GE LR transceivers. With the use of Raman distributed amplification in the transmission fibre, the nominal reach of the transceiver - 10 km - was extended up to 135 km. Moreover, with the use of a single in-line praseodym-doped fibre amplifier (PDFA) the reach was extended further to 200 km. We also tested optical semiconductor amplifiers (SOA) that are becoming easily available and less expensive than the original PDFAs. These results were, for example, presented at the ICTON international conference in Barcelona and the OK 2005 conference in Prague.

We also experimented with a fully optical conversion of modulated signals from 1310 nm band to 1550 nm band using a semiconductor-based optical amplifier. The block diagram is shown in Figure 3.5.

We tested InPhenix SOA converters IPSAD1301 and IPSAD1303. The signal conversion succeeded at both 1 Gbps and 2.5 Gbps speed. At 10 Gbps, the eye-diagram of the 1550 nm signal was closed with the rise and fall edges not steep enough. These SOA types are not designed for 10 Gbps bit rates and we will thus test other SOAs, mainly for use with 10GE server network cards. We evaluated SOAs from various manufacturers (InPhenix, CIP, Covega) as in-line amplifiers in CWDM systems.

We also developed another important element of CEF networks - the pumping source for Raman fibre amplifiers. The architecture of this pumping source is again based on CLA and uses a commercial module with four semiconductor lasers by Amonics with the total output power of 500 mW. This amplification module is intended for the band from 1530-1560 nm. The spectral dependence of the distributed Raman amplifier (measured with a 37 km long NZ DSF fibre) is shown in Figure 3.6.

Yet another component that is necessary for fully optical networks is an optical switch. During the year 2005, we invested a lot of effort in searching for suitable components. We decided to buy an optical switch from DuPont - an 8×8-port fully optical switching module that is controlled through an RS-232 interface. Currently, we are working on the first prototype and plan laboratory tests.

We also experimented with duplex optical transmission over single fibre, both in an entirely passive configuration (without any optical signal amplification) and in a cost-effective configuration where the amplification is applied only on one side of the link (OSA - One Side Amplification).

In the passive configuration, we evaluated the use of optical splitters, circulators and filters. For less demanding applications, optical splitters turned up to be a good choice. Although their insertion loss is higher compared to circulators and filters, splitters are simple and cheap components. The deployment of these splitters together with 1GE CWDM transceivers (1530 nm/1550 nm, reach approximately 120 km) enabled us to span the distance of 95 km using a single fibre. Optical circulators turned out to be less suitable for this application and the maximum distance we reached is the same as for splitters. Due to their high sensitivity, the receivers in these transceivers are influenced by the backward Rayleigh scattering. The distances up to 130 km (on single fibre) may be spanned using identical CWDM transceivers together with CWDM filters in three-port configuration; the block diagram is shown in Figure 3.7.

In order to further increase the reach, we combined OSA method with the use of EDFA. In this configuration, we used both the CWDM transceivers mentioned above - due to EDFA performance, these are only suitable for a single duplex transmission - and 1GE DWDM transceivers that allow for multiple duplex channels. In the first case, the nominal reach of 120 km on a pair of fibres was increased to 180 km on a single fibre, while for DWDM transceivers the reach was increased from nominal 105 km to 175 km. For distances exceeding 150 km, another filter had to be deployed in front of the receiver in order to suppress the strongly amplified spontaneous emission. The block diagram is shown in Figure 3.8; the DWDM filters we used are so-called reflexive 3-port filters. We will continue these experiments and we plan to use newer transceivers with the transmission distance up to 160 km.

3.4   High-Speed Free Space Transmissions

Most high-speed networks have to cope with the first mile problem. While optical fibre is certainly the ideal solution, in many cases it is too expensive or practically impossible. That's why we investigate alternative transmission technologies based on free space optics (FSO) and radio connections. Although the transmission speeds are lower compared to the optical fibre, the prices are lower as well. Each technology has its advantages and disadvantages:

• FSO systems generally offer higher transmission speeds, often 100 Mbps or more, but are sensitive to meteorological influences such as fog or vast temperature changes, especially when used over longer distances.
• In contrast, radio devices are able to work also in point-to-multipoint modes and do not require direct visibility. On the other hand, their throughput is usually lower. When considering systems in the price range up to 100,000 CZK (3,500 EUR) operating in the unregulated 2.4 and 5 GHz bands, we must also take into account possible signal disturbances.

That is why we usually recommend to combine a primary FSO system with a back-up radio connection.

The year 2005 marked a significant change in the free frequency band legislation. The general authorisation VO-R/12/08.2005-34, issued by the Czech Telecommunication Office, now allows transmission in the 5 GHz band. The frequency range 5.15-5.35 GHz may be used only inside buildings and 5.470-5.725 GHz and 5.725-5.875 GHz bands also outdoor. The 5.4 GHz band is particularly important for building first mile links, since the radiation output in the outdoor deployment can be 10 times higher than in 2.4 GHz band and 100 times higher than in the previously permitted 5.725-5.875 GHz band. The equipment for the 5.4 GHz band is significantly more expensive than indoor access points - prices start at 25,000 CZK (900 EUR) - and will thus mainly be offered for outdoor environments. This fact, together with different regulations for indoor and outdoor networks and 11 bands of non-overlapping channels, should lead (at least according to optimistic forecasts) to much less troubles with mutual network disturbances, as we know them from the 2.4 GHz band. The devices in the new category also usually offer a higher level of data security.

In 2005, we followed the developments in both categories of wireless technologies. As for FSO, we focused on the development of our own low-cost prototype and made a technical comparison with existing products and prototypes. In the area of radio devices, we monitored the emerging 5 GHz band market, and deployed one of the available sets - the WinLink product. Detailed results of this evaluation are presented in the technical reports [WC05] and [CW05].

The work on the prototypes of low-cost devices for free space optical transmission mainly focused on the reconstruction of the mechanical part of the original 10 Mbps link in order to increase its robustness and stability, and also on looking for and evaluating functional prototypes of the electronic subsystem for 100 Mbps and potentially higher transmission speeds. Their deployment on the first mile of L2 connections depends on replacing the current optical-Ethernet converter with an optical-optical converter and creating an autonomous system with wireless back-up and a possibility for monitoring the status of the link.

We actively searched for, compared and tested 100 Mbps FSO devices that are available on the market or under development (e.g., by the original authors of 10 Mbps FSO devices). As a result, we were also able to identify both suitable and problematic elements of potential solution.

From the evaluated FSO devices, Elspeedy 100 seems to offer the best price/performance ratio and technical parameters. In comparison to commercially available devices, we achieved quite interesting results with this device for 100 Mbps transmission over the distance of 3000 m. The price of this device, after the development is finished, might be about 80,000 CZK (2,800 EUR). Elspeedy 100 requires significant modifications though, because its current mechanical design is relatively expensive for production and also the optical bond cannot be easily focused for longer distances. For production, it would be more appropriate to use special aluminium profiles, which could also lower the end-user price of the device.

On the basis of these tests and analyses, we designed new electronics called "LightShuttle" for the redesigned mechanical part of the prototype. We designed printed circuit boards and selected parts for the final modular electronic subsystem for 100 Mbps and started their gradual activation. The tests of the new prototype are planned for the first half of 2006. The device we work on has a four times larger surface of the receiving lens than Elspeedy 100, more precise focus and increased robustness for "long-haul" connections. So far, the results of the development and tests seem promising, also from the perspective of speeds over 100 Mbps.

Obviously, production of such devices on our own is beyond the scope of the research intent. Instead, the most effective way seems to be to finish the development of a certain prototype and then grant production licences to third parties. The advantage of CESNET is that it is able to test both prototypes and resulting products on the CzechLight testbed or on a pilot infrastructure of the CESNET2 production network.

previous

contents

next