Toward 4G IP-based Wireless Systems: A Proposal for the Uplink Tony Ottosson 1 , Anders Ahl´ en 2 , Anna Brunstrom 3 , Mikael Sternad 2 and Arne Svensson 1 1 Dept. Signals and Systems, Chalmers Univ. of Techn., SE-412 96 G¨ oteborg, Sweden 2 Signals and Systems, Uppsala University, Box 528, SE-75 120 Uppsala, Sweden 3 Dept. Computer Science, Karlstad University, SE-651 88 Karlstad, Sweden Abstract A packet switched wireless cellular system with wide area cov- erage and high throughput is proposed. The system is designed to be cost effective and to provide high spectral efficiency. It makes use of a combination of tools and concepts: Smart an- tennas both at base stations and mobiles, provide antenna gain and improve the signal to interference ratio. The fast fading is predicted in both time and frequency and a slotted OFDM radio interface is used, in which time-frequency slots are al- located adaptively to different mobile users, based on their predicted channel quality. This enables efficient scheduling among sectors and users as well as fast adaptive modulation and power control. We here outline the uplink of the radio in- terface. Calculations based on simplifying assumptions illus- trate how the channel capacity grows with the number of simul- taneous users and the number of antenna elements. A high ca- pacity can be attained already for moderate numbers of users and base station/terminal antennas. 1 Introduction Current cellular networks, such as GPRS, provide high cover- age and mobility but with low throughput. On the other hand, wireless local area networks (WLANS) such as IEEE 802.11b and Hiperlan/2 provide high throughputs over limited distances and with limited mobility. Since the upcoming third genera- tion systems (WCDMA) will only allow a limited increase in throughput, the gap will remain between services and capacity offered by local area systems and by cellular wide area cover- age systems. The current trend in communications is toward integration of systems and services. Users of fixed networks will be ac- customed to high throughputs, good connections and innova- tive services. They might therefore become less willing to pay for the expensive and relatively low performance services that current and future cellular systems may offer. The current cri- sis for the telecommunications industry may be seen as one indicator of these worries. A remedy would be to develop new wide area coverage and high mobility systems with much increased throughput (at least ten-fold) at a cost that is not higher (but preferably lower) than todays cellular networks. However, even if all cutting edge techniques, such as iterative decoding, adaptive antennas, and space-time coding are used, the WCDMA system will pro- vide only a moderate increase in capacity (spectral efficiency) at the expense of an increase in complexity and thus cost. The Wireless IP project [1] within the Swedish research program for Personal Computing and Communications (PCC) studies problems that are important in the evolution of UMTS toward higher data rates, as well as in future 4G technologies for mobile systems. Our goal is to improve the spectral ef- ficiency for packet data, in particular IP traffic, with sufficient quality of service for various traffic classes. In the downlink we aim at a ten-fold improvement over existing technologies. In the uplink, which is more demanding due to the asynchronous behavior of the terminals, we aim for a somewhat lower capac- ity increase. To achieve this goal we propose to take advantage of the fast time and frequency varying channel properties, and to use multiple (smart) antennas in both base stations and mo- bile terminals. In [2, 3], these issues were discussed for downlink traffic. The present companion paper considers a radio interface for the uplink. In Section 2 we discuss various design trade-offs and present a proposal for the uplink where prediction, smart antennas and scheduling are discussed together with adaptive modulation and power control. Sections 3 provides a prelimi- nary analysis of this system in a special yet important scenario. 2 The Radio Interface Uplink We aim at a spectrally efficient and highly flexible packet switched system providing high throughput in both the down- link and the uplink. Our goal is to provide a capacity of Mbit/s per sector of each base station in the downlink and only slightly lower capacity in the uplink. In a system for packet data aimed at wide area coverage, we expect delay insensitive data and streaming traffic classes to constitute a large part of the traffic. We also expect a sig- nificant fraction of the users to be on the move, and thus to encounter time-varying channel properties. This can be trans- formed from a problem into an advantage, if we introduce a feedback system in which the channel properties are predicted and resources are allocated to users according to these channel predictions. Channels from different users will fade indepen- dently. With many users in the system, a channel could ideally almost always be allocated to users who at that time encounter favorable conditions. This effect is sometimes denoted mul- tiuser diversity. Fairness and delay sensitivity of course repre- sent constraints that reduce the attainable efficiency. In an ideal situation, without discrepancies in the local os- cillators of the mobile stations, the uplink could be designed in the same way as we proposed for the downlink [2, 3]. A syn- chronous system would thus allow us to use the downlink ap- proach. However, synchronization among terminals and base stations is a very difficult task that imposes tremendous re- quirements on the hardware and on the algorithms of the sys- tem. An asynchronous uplink would therefore be more desir- able since only a rough synchronization would be needed. To enable scheduling of resources to obtain multiuser di- versity in an asynchronous system, we propose the use of a