One of the biggest challenges for the next generation wireless system is the ubiquitous access to high data rate services. At the physical layer, Multi--Carrier (MC) transmission technology is envisaged, due to its high spectral efficiency and robustness to frequency selective fading. While MC transmission opens the field
for enabling such ambitious goals, the available radio resource needs to be efficiently distributed in order to perceive the gains at the system level. There are many design issues to be addressed. Two types of system architecture, the cellular and the relayed
system, envisioned for the next generation wireless system, are considered in this thesis. For each system, the main target is to produce radio resource allocation and scheduling algorithms that provide good performance with low complexity, making them suitable for practical implementation. While the common scheduling algorithms optimize one given metric, in reality, optimization of different types of metrics is desirable. The objective of the proposed algorithms is to enhance the fairness among users and
reduce service delays, without sacrificing system throughput. Another issue in MC systems is the explosion of the amount of Channel State Information (CSI) to report to the scheduling node. The higher the amount of CSI, the better the scheduling
performance but the larger the amount of signalling. Breaking with the usual approach where CSI minimization is viewed as a stand-alone problem, independent of the type of algorithm, in some of the proposed algorithms CSI reduction is included as one
of the design criteria. Adaptive CSI reduction schemes, which can be used for different types of algorithms, are also developed. The thesis is divided into two parts, each dealing with the considered system architectures.
Allocation algorithms are developed for cellular MC system, with a particular attention towards Proportional Fair Scheduling (PFS). While optimal PFS in the MC case is prohibitively complex, the
proposed method provides extremely tight bounds with reduced complexity. Several suboptimal algorithms are developed. As the common algorithms do, the first set of algorithms operate on a slot-by-slot basis, where a slot is equal to a symbol duration. However, the practical radio frames such as in the standards
contain a multitude of time slots for which the scheduling decision is made. The second set of algorithms enable multi-slot scheduling, and take advantage of the additional time dimension by
allocating different users in one subchannel-defined in this thesis as a group of adjacent subcarriers-, over different time slots, which is referred to as User Multiplexing. The advantage of these algorithms is that they are designed in such a way that reduces the amount of signalling required for describing
the user mapping within the frame, whereas the sequential application of slot-by-slot scheduling to all the slots of the frame results in an unacceptably high amount of signalling, dropping the overall throughput. Results show that the proposed algorithms achieve excellent throughput/fairness trade-off and
reduce service delays. Moreover, CSI feedback schemes are proposed, characterized by their flexibility to adapt to the required CSI which varies depending on the scheduler. The developed Differential encoding mechanism achieves large
feedback reductions, even for PFS for which higher amounts of CSI are required to ensure an acceptable level of fairness.
The second type of system is the relay-based MC system. The multitude of parameters in a multi-carrier system with multiple users and relays make the allocation problem very complex. The proposed algorithms require low complexity and minimal CSI
feedback as the allocation of relayed users is decoupled from the allocation of BS-originated transmissions. Higher throughput is achieved by the proposed request mechanism, where the relay
requests the BS to forward new packets for the users likely to be scheduled over the following frames, thereby optimizing the resource allocation between relayed and non-relayed users.
Results show that the algorithms achieve good throughput/fairness trade-off. In the case of multiple relays within a cell, the proposed adaptive schemes provide significant performance enhancement over static schemes. Finally, the issue of cooperative
diversity in MC systems is addressed. Approximations of the outage probabilities of designed schemes are theoretically derived. The results show the very high accuracy of the bounds. Simulations with practical OFDM channels confirmed the analysis, and one of
the schemes, the Average Best Relay Selection Scheme (AvgBRS), is the most promising due to its near-optimal performance while reducing the required CSI.
|Period||01-10-04 → 30-09-07|
|Research programme||<ingen navn>|