Zur Ressourcenvergabe in einem selbstorganisierenden zellularen OFDM Mobilfunksystem (English shop)

Preview

Table of Contents, Datei (39 KB)
Extract, Datei (170 KB)

This thesis analysed the potential of self-organising resource allocation for cellular mobile radio systems based on OFDM transmission technology and proposed an SO-RRM concept for future 4G systems.

Owing to its efficiency and robustness against the multipath propagation of real mobile radio channels, OFDM transmission technology forms the transmission-related basis. The system proposal is based on a cell-spanning synchronisation of all BSs and MSs in time and carrier frequency. In a fully synchronised radio network, CCI does arise, but no adjacent channel interference. As a result, the CCI can be measured reliably and evaluated as a decision criterion for the self-organising resource selection.

In the proposed SO-RRM system, the BSs autonomously select the resources required to serve the MSs. The selection is based on measurements of the desired signal powers and the CCI carried out in the DL and UL. From these, the resource-specific SINRs are computed. Subsequently, the resources with the highest SINRs are selected. By applying a link adaptation technique, the number of selected resources is adapted to the channel conditions and to the intensity of the measured CCI. The MSs support the selection process by transmitting a proposal list containing the CCI measured in the DL to the BS. The BSs can carry out the resource selection jointly for DL and UL by merging their own ranking lists with the proposals of the MSs. Alternatively, the resource selection can be performed separately in DL and UL.

Self-organisation offers maximum flexibility in the use of the scarce radio resources. In an SO-RRM system, every BS has access to the entire system bandwidth. Resources can always be selected where they are currently needed. This makes it possible to serve asymmetric user loads such as hot spots highly efficiently. In addition, compared with conventional second- and third-generation mobile radio systems, there is no need for static advance planning of the resource demand and of the frequency bands occupied by the radio cells.

SO-RRM with constant data rates

First, Chapter 7 investigated the provision of constant data rates in the UL and DL. System-level simulations showed that the selection algorithms proposed in this thesis lead to stable network operation. The SO-RRM system was compared against a reference system with FCA. Even with a uniform distribution of users across the coverage area, a higher efficiency was achieved with SO-RRM. The SO-RRM system demonstrated its real strength in the hot spot scenario. Whereas in an FCA system no adaptation whatsoever to changes in the user load is possible, SO-RRM systems can adapt flexibly to the actual instantaneous traffic load. As a result, the efficiency of the SO-RRM system in asymmetric traffic scenarios such as hot spots increases considerably compared with conventional FCA-based concepts.

For the provision of constant data rates, both a joint and a separate resource selection for DL and UL were analysed. The better results were initially achieved with joint selection of the resources for DL and UL. Nevertheless, a separate resource selection is useful when different data rates are required in DL and UL. It could be shown that the system behaviour with separate selection can be improved by over-provisioning the resource demand. With the resource surplus, more robust PHY modes can be applied. This makes the transmissions more robust against possible increases in the CCI.

SO-RRM in the DL with variable data rates

In Chapter 8, the provision of variable data rates in the DL was considered as a second application case. For this purpose, a resource selection separated for DL and UL was carried out. As an extension of the SO-RRM concept, a protocol was proposed for the DL with which all resources selected for the DL can be redistributed among the MSs according to demand. The concept is based on the fact that this does not change the CCI perceived in the surrounding cells in the DL.

In the extended concept, the resource selection is controlled via the resource coverage ratio. This is determined in every access frame by a comparison between the available and the required total data rate. Resources are selected and released when specified coverage ratio thresholds are undershot or exceeded, respectively. Through a suitable choice of the thresholds, an over-provisioning of the resource demand can be achieved. Such a resource expansion already proved to increase efficiency in the provision of constant data rates. The occurrence frequencies of selection and release processes are reduced by a resource expansion, so that the measured CCI remains constant over a longer period of time.

By means of computer simulations it was shown that a resource expansion has a positive effect on all relevant QoS metrics, especially at low and medium system load. At high system load, however, the extent of the resource expansion should be reduced so that additional users can be served with the resources thus freed up. For this purpose, system-load-dependent coverage ratio thresholds were employed.

As an essential innovation, a coupling between the two processes of resource selection and resource assignment was introduced. The resources selected by the macro RRM are made available to the micro RRM in a resource pool. In every access frame, the micro RRM assigns the resources contained in the pool to the MSs according to demand. It then computes the coverage ratio resulting from the current assignment and reports this back to the macro RRM. The macro RRM in turn uses the fed-back coverage ratio as a control variable for its resource selection process.

For the assignment of the resources, a utility-based scheduling algorithm was proposed. The algorithm takes into account the instantaneous states both of the packet queues and of the frequency-selective and time-variant radio channels. This makes it possible to exploit the MUD present in the radio cells and to increase the system efficiency. In order to ensure fairness, the principle of diminishing marginal utility is applied. The utility of a further resource decreases with the number of resources already assigned to a user. It became apparent that, in particular, the evaluation of the waiting times accumulated in the packet queues is decisive for a demand-oriented assignment and for fairness within the cell. Through the simultaneous consideration of the instantaneous channel states, a high efficiency is achieved in the utility-based resource assignment.

The coupling of micro and macro RRM brings about a stable and efficient self-organisation at the overall system level and offers an effective mechanism for guaranteeing QoS with variable data rates.