Satellite Technologies for Broadband Internet Access Onboard High Speed Trains David SANZ SNCF - Direction de l'Innovation et de la Recherche, Paris, France [email protected] Abstract This paper presents last years SNCF work concerning broadband Internet service delivery to passengers onboard high speed trains. First of all, the reasons to the deployment of this kind of services are presented. Then, the different candidate technologies are listed, before to focus on satellite solutions advantages and drawbacks. Finally, the different projects and experimentations carried out by SNCF onboard high speed trains (TGV) are introduced. The paper concludes with some brief conclusions and by giving some perspectives of future researches SNCF is willing to carry. Introduction Last years have witnessed the constant and unstoppable growth of Internet accesses. People can use their Company's Internet access when they're in the office, their broadband DSL/cable connections at home (9,5 M high speed Internet subscribers in France in December 2005!), 2,5G and 3G cellular communication networks when they're on the move, and of course, WiFi Internet accesses in public hotspots like airports, hotels and train stations,… People are becoming used to have Internet access wherever they are and all the time… that's the well- known concept of "Always-on, "Always-connected", "Anyplace-Anytime-Anydevice", … Trains should not be an exception! First of all, because traveler's needs and expectations concerning this kind of value added services really exist, but also because train competitors (plane, car, bus,…) don't stop innovating in this area (the ConnexionByBoeing service is an example). That's why Railway Operators' interest in broadband communication systems for trains has been growing during last years, and that's also why the firsts products are, little by little, arriving to this new market. Pointshot Wireless, Icomera, 21Net, Telespazio, Raysat, Viasat, Qinetiq (and BroadReach Networks), Nomad Digital, Pulz8,... are some of the suppliers proposing products/solutions on that domain. In addition, intercity trains are a very favorable environment: people have more than 1 hour of “profitable” time (in average), and some times many hours… and they're installed in a comfortable way (a seat, a small table, in a quiet place,…). That's why Intercity train passengers are a “captive audience” and that this kind of service represent a real interest for them. TGV high speed trains alone transported 100 M passengers in 2005, a huge “captive” potential market for those kind of services. What kind of services these passengers are waiting for ? They're waiting for all kinds of Internet-related value-added services they are already used to have: § the basic internet services: web-surf, send/receive emails, access to generic information (news, weather, …), … § other more business-oriented services: in particular, to connect to their company's Intranet, through the implementation of VPN/IP-SEC secure connections, § other more leisure-oriented services like chat, radio and TV streaming, Peer-to-Peer, … and, of course, railway operator services like real-time information about the travel, tickets information and online purchasing, train timetables, … Internet connection service for passengers will probably be "the" onboard service of the future years, the one that should make the difference! And from SNCF's point of view, it will also be "the" service that will make possible for Railway Operators to deploy the needed effort in order to bring those kind of broadband connections onboard trains… But Internet connection is just one of the many possible applications we can consider! A broadband train-ground communications link should of course also be used for operational purposes… § To "connect" the train crew. Today, crew are "disconnected" the most of the time. Even when they're equipped with GPRS or UMTS mobile phones (with data-communications capabilities), the very poor French territory cellular coverage makes that often they cannot use those phones in order to get real-time traffic information, real-time seat-reservations information,… § To "connect" the trains themselves for the transmission of real-time information to/from control centers: trip and traffic information, proactive maintenance, machine-to-machine,… Those broadband train-ground connectivity will allow for Railway Operators to improve: § The delivered service to their passengers § The efficacy in the operation of their trains § Their "brand image", because of the technical innovation, but most of all because of the service innovation! thus, giving Railway Operators a competitive advantage with respect to other transport means, which, as said before, are also very active in that domain. Why satellite ? Many kinds of telecommunication technologies can be considered in order to deliver data train-ground connectivity. Each of them has its own advantages and drawbacks, but any of them can answer to all the constraints imposed by the very rough railway environment and by the foreseen services themselves (quality of service required to connect 30-100 people simultaneously). In Europe, and without searching to be exhaustive, the following terrestrial technologies could be considered: § Cellular mobile-phone-oriented technologies: GSM and its railway version GSM-R, GPRS, HSCSD, EDGE, UMTS, and soon HSDPA and HSUPA. § Terrestrial broadcasting technologies: DAB-T, DMB, DVB-T,… § Wireless LAN/MAN technologies: WiFi, HiperLAN2, WiMAX, and other single-constructor more or less proprietary solutions like Flarion's FlashOFDM , Arraycom/Kyocera's iBURST, NOMAD DIGITAL modified WiMAX solution,… Current cellular mobile-phone-oriented technologies are suitable to deliver medium data-rate connectivity for a reduced number of simultaneous users and are fully dependent on mobile operators choices (of implementation, of coverage,…). For example, in France, some mobile operators have launched the GPRS service allowing just two 4+1 channels per base-station. That means that the total available throughput in that cell will be of just around one hundred kbps for the downlink and some tens of kbps for the uplink… so even the aggregation of channels will not allow a broadband connectivity. The territory coverage proposed by mobile operators is optimised to cover the population areas and they're, hence, not intended to correctly cover the rural areas crossed by the 31000 km of SNCF railway tracks. The result is that when you take a train in France, you'll have (in average) a good signal only during half of your travel time… In addition, the modulation schemes of those technologies has been chosen in order to support slow mobility, and are not well adapted to the high speed trains environment (the most important market of SNCF, which is, with its more than 400 TGV trainsets which runs up to 320 km/h, one of the most important high speed operators in the world). Actually, the Doppler effects (frequency shift and spread on the received signals) are present with any terrestrial technology, are caused by the relative movements between transmitter and receiver and are proportional to their relative speed, so the bigger the speed, the bigger the Doppler effect, and only the characteristics (at conception level) of the communications technology and the capacities of the receiver will allow to cope with this effect. That means that some technologies will support more easily the speed than others. SNCF online tests (at high speed) have shown that the available throughput decreases with the speed of the train on a significant way. HSDPA and HSUPA UMTS-based technologies (respectively 3.5G and 3.75G) will make possible in the next years to establish up to 20 Mbps downlink and 5 Mbps uplink communications, but the problems of Doppler and coverage (since at the beginning of 2006, UMTS is only available in large population areas) will remain. Terrestrial broadcasting technologies are by definition transmitting-only technologies. Internet service needing a return link, those technologies alone cannot be a candidate. However, the use of broadcasting techniques for the downlink and a suitable technology for the uplink can be considered since commercial services of this kind already exist in various domains (for example for cars: DVB-T-based Internet access with mobile-phone return channel for interactivity). As any other terrestrial technology, terrestrial broadcasting suffers of the Doppler effects presented above and, more important, need the installation of infrastructures along the railway tracks… which means long and hard deployments/installations (conditioned to the permission of the Railway Infrastructure Owner) and often very expensive. The Wireless LAN/MAN technologies are very interesting technologies because they have been conceived to deliver high data rate connectivity, the first one (LAN) over medium coverage areas (about 100-500 meter), the second ones (MAN) over large coverage areas (2 to 50 km depending on the kind of technology, on the mobility of the terminals,…). Since they're terrestrial technologies, we'll have again the same deployment constraints explained before. NEC showed in October 2003 [1] in Japan that WiFi (IEEE 802.11b standard) can be used, through the implementation of a software that allows for a fast sign-on and hand-over between standard access points, placed at 500 meter intervals, in order to implement connectivity to a mobile (a Porsche car) running at 330 km/h! WiFi 20 MHz direct sequence spread spectrum