Jean-Paul Linnartz
Department of Electrical Engineering and Computer Sciences
University of California at Berkeley
A substantial technical problem confronted in attempts to converge to a uniform standard appears to be the lack of adequate methods for multiplexing various traffic streams efficiently in `unguided' communication means. For reasons of spectrum efficiency and user capacity, efficient design of packet-switched radio data networks essentially differs from circuit- switched cellular networks. This suggests that a "universal", i.e., integrated approach may be less desirable in wireless subscriber access links than in cable or fiber-optic backbone (ISDN) links, unless new efficient methods of merging traffic streams will be found.
Technical bottlenecks in the development of a ubiquitous (wireless) multi-media environment are the capacity of the wireless indoor (radio) link, its unreliability due to the adverse multipath propagation channel and the severe (statistically changing) interference from other transmissions. In the unrealistic scenario of supporting only a single multi-media terminal, the limitations of the channel can be overcome by appropriate signal processing techniques. However, the scarcity of the radio spectrum necessitates the shared use of the allocated bandwidth by multiple (spatially distributed) users, each generating and receiving bursty traffic.
Decades of research on sharing communication resources among multiple users and services on a wired networks has led to a wide variety of techniques for multiplexing, switching and multiple access to (guided) communication resources. The common goal of these scheme is the (static or dynamic) assignment of (a part of) the bandwidth during certain periods of time. Multiple-access techniques commonly used in wired local areas networks (LANs) were initially used in radio data networks. However, it soon appeared that the performance of many random-access schemes substantially differs for guided (wired) and unguided (radio) channels, being highly dependent on the physical characteristics of the channel. For realistic analysis of the performance of wireless radio networks, the common assumptions have been refined that
1) a data packet is always received successfully if no conflicting transmission (collision) occurs simultaneously on the same channel and that
2) data packets involved in a collision are always lost.
Moreover, the performance seen by each participating terminal is different from the network- average performance and highly depends on the location of the terminal.
The aspect of frequency reuse (in space) and that of allowing multiple users to share the same bandwidth - time resources have mostly been addressed separately. This is for instance illustrated by the fact that most existing mobile data networks use a fixed cellular frequency reuse pattern, and within each cell a random-access scheme is used independent from the traffic characteristics in other cells. Recent results show that cellular frequency reuse is far from optimum for wireless data networks. Significantly higher user capacity and smaller channel-access delay occur if each base station and each terminal can use the entire bandwidth. The corresponding high interference power levels from nearby other transmitters require a joint optimization (and dynamic management) of the spatial frequency reuse and the occupation of spectrum within cells. Methods to dynamically assign the space-time-bandwidth resources in radio channels are being developed, for instance STRMA.