Due to the success of DS-SS CDMA and OFDM in the early nineties, several approaches were developed to combine both schemes. Many papers which deal with multi-carrier techniques concentrate on signal processing in the frequency domain. This paper reviews research on time domain signal processing for multi-carrier spread-spectrum systems. Basic characteristics such as spectral shape, correlation behavior, and the need for a guard interval are discussed as well as questions about multiuser capability, receiver architectures, and implementation issues.
I. MOTIVATIONIn the early nineties, the breakthrough for DS-SS CDMA came with the commercialization of IS-95 as well as for OFDM with xDSL. Therefore most research effort is concentrated on combining the advantages of both techniques into a new one. One approach [l] called "multi-carrier spread-spectrum" (MC-SS) has the goal of designing a spread-spectrum system dual to DS-SS in the sense that DS-SS time-domain operations are MC-SS frequency-domain operations and vice-versa 121. That approach includes also the proposal [3] although the authors did not call their scheme MC-SS. In 141, 151, a system called "MC DS-CDMA" is presented. In that case a DS-SS signal is transmitted over multiple orthogonal carriers. A further approach [6] called "MC-CDMA" aims a t designing a spreadspectrum system based on an OFDM modem. In [7], (81 more general proposals are given to combine MC techniques with the CDMA scheme to enhance the system capacity. T h e third generation mobile communication standard cdma2000 also contains a multi-carrier component t o enhance the flexibility of the data rate. A detailed state-of-the-art overview is found in the proceedings of both MC-SS workshops held in 1997 [9] and 1999 [lo], respectively.In the research work, the questions to be answered are the following: 1. Is the envelope dynamic of the transmit MC-SS signal 2. Is a guard interval necessary for MC-SS systems? 3. Which receiver architecture should be chosen considering channel estimation, detection, and computational complexity? a problem like in OFDM? 4. How can MC-SS be implemented at low cost? Whereas many publications deal with signal processing in the frequency domain to solve these questions 1111, [U], 1131, [14], [15], the focus of this paper is to show that time domain processing can be applied as well. T h e paper 'This work was partly supported by Deutsche Forschungsgemeinschaft contract Fe 423/2is organized as follows. Section I1 reviews the MC-SS principle and explains basic characteristics in time and frequency. Section I11 discusses how t o reduce the envelope dynamic of the MC-SS signal. Section V deals with the correlation behavior. In section IV the need of a guard interval is investigated. Section VI summarizes some aspects concerning receiver architecture, in particular channel estimation, detection and implementation issues. Section VI1 presents two multiple access schemes to achieve multiuser capability. Finally, section VI11 concludes the work.
MC-SS: FREQUENCY vs. TIME DOM...