Multistream Scheduling in Cellular Networks as a Coupled Optimization Problem

Cellular networks are more and more wide spread nowadays, moving from the current third generation UMTS to the forthcoming WiMAX and LTE 4G networks, promising ever increasing throughputs for new multimedia services. This thesis focuses on scheduling in these cellular networks, and proposes some multistream enhancements. Indeed, an efficient allocation of the available radio resources is one of the key parameters to provide larger throughputs to an ever increasing amount of users. To improve efficiency, multistream transmissions, where multiple users are allowed to transmit simultaneously, have been suggested. The resource allocation problem therefore becomes coupled through interferences, which we have to minimize. Moreover, this problem has multiple contradictory objectives, such as throughput, fairness, or respect of QoS guarantees. Both 3G's CDMA and 4G's OFDMA multiple access techniques are studied, for both SISO and MIMO transmissions. We have managed to decouple the resource allocation problem and limit interferences in CDMA through orthogonality factors, depending on users' channel dispersion in SISO, and on users' location in beamforming MIMO. As far as OFDMA is concerned, we have designed a priority-based scheduling algorithm which ensures some QoS guarantees, while remaining quite simple and requiring only a limited feedback, averaged over the full bandwidth. A MIMO extension of the algorithm, improving frequency reuse through a GoB strategy, has also been developed. The problem of WiMAX burst mapping has been deeply investigated, and a new mapping algorithm, seriously reducing resource wastes, has been designed.(DOCSC06) -- FUNDP, 201