by Philippe Jacquet and Paul Mühletaler
Imagine mobile automatons in a factory, in space, or in any other locations hazardous for men. Imagine "intelligent" cars, or people wanting to stay "connected" when moving. All these scenarios have something in common: wireless communication.
But contrary to well known cellular radio telephone, these communications are "multimedia" like (voice, data, pictures) and must eventually operate without specific ground installation (beacons, central hubs). Furthermore, these communications require more efficiency from network facilities (higher throughput, reliability, promptness). The corner stone of such systems is the wireless digital local area network. In a first approach such stone does not seem too difficult to build. However this has never been realized. The reason for that is in the fact that it poses really difficult technical problem.
The first problem has concern with physical stuff. Radio waves echo on obstacles they meet, and this generally alternate the data they carry, when the bit rate is above the million per second. In order to cope with this problem one has to utilize massive coding/decoding devices which call for very powerful dedicated chipment.
The second problem has concern with algorithms and protocols. The problem is much more complex than with cabled local area networks. Indeed the network is now open and mobile, the communication medium is permanently subject to perturbations. These constraints call for specific medium access rules based on new algorithms; but these algorithms have to be connected to older interfaces and application devices commonly found on today market.
The third problem is normalisation. The frequency spectrum is a resource which is under high demand. Frequency allocation is a difficult task under the responsibility of European Community in cooperation with other international organisations. Since it is very unlikely that different kinds of wireless networks can coexist on the same medium, there is a need for the most concerted approach.
The European Community, in the frame of ESPRIT III project, is supporting and funding the LAURA project whose aim is to build a wireless local area network. INRIA collaborates to this project as full partner. The other academic partners are the universities of Bradford and Bristol. The industrial partners are Dassault (France), Elletronica (Italy) and Symbionics (Great Britain).
The targeted product is a wireless local area network similar to Ethernet like networks (IEEE 8802.3), with similar throughput (within 10 millions bits per second). The product could be connected with most existing network interfaces.
The research part of the project is shared between two main competences: Radio Frequency and Network technologies. The British universities have a strong experience in radio, INRIA have numerous results on network and protocol design. Three teams are concerned in the Institute: Algo, Codes and Reflecs. In order to give an outline of the current activities of these teams about LAURA, follows a brief description of their assigned tasks, understood that overlaps exist and enforce our trends.
Code team (Paul Camion) defines efficient error correction devices for data transmission on non reliable medium. This task is crucial because radio propagation is expected to provide uncompressible bit error rates. Increasing emission powers cannot improve things since it may increase the number of radio wave echoes. Contact person: Nicolas Sendrier.
Reflecs team (Gérard Le Lann) defines new medium access rules in order to match usual criterion of efficiency. The unstable nature of radio propagation calls for flexible and robust solutions, like, for example, contention protocols with collision detection, and learning routers in each connected node. Contact person: Paul Mühlethaler.
Algo team (Philippe Flajolet) works on design and optimization of the new algorithms. We know that the constraint of radio networks introduce efficiency loss more severe than with similar cabled networks. Thus there is a need of optimal tuning of the algorithms in terms of global performance at each layer of the protocol. Contact person: Philippe Jacquet
All these activities should give a product with targeted performance and costs to allow connection on personal or portable computers. We are working today on the first wireless local area network which will may be one of the elementary bricks of tomorrow technology.