- Citation/Abstract
- Dissertation or Thesis
Methodical evaluation of quality of service for heterogeneous networks
Abstract (summary)
Heterogeneous wireless networks expand the network capacity and coverage by leveraging the network architecture and resources in a dynamic fashion. However, due to the presence of different communication technologies, the Quality of Service (QoS) evaluation, management, and monitoring of these networks are very challenging tasks. Each communication technology has its own characteristics while the applications utilising them have their own QoS requirements. Most current methods are based on analysing the QoS of each application or access network separately. However, these methods do not combine the performance of all the applications and the radio access networks while reporting the QoS of the overall configuration. Therefore, it is hard to get any aggregate performance results using these methods.
To fill this gap, in this thesis, a methodical approach is adapted for the QoS analysis of these types of networks. At first, the approach uses a simple fixed weight-based method, and then moves to a more complex dynamic weight-based method and in the end integrates the concepts of fuzzy logic. The proposed methods consider the significance of QoS-related parameters, the available network-based applications, and the available Radio Access Networks (RANs) to characterise the network performance with a set of three integrated QoS metrics. The first metric denotes the performance of each available application on the network, the second one represents the performance of each active RAN on the network, and the third one characterises the QoS level of the entire network configuration.
Using the fixed weight-based method, the weights for the QoS-related parameters, applications, and RANs are defined based on their significance relevant to a given situation. Then, the dynamic weight-based method determines the weights of these entities dynamically by incorporating the changing circumstances of the network. Although, the dynamic weight-based method can account for the active changes of the network, it has some limitations. For instance, it cannot capture the underlying uncertainty of network dynamics. To overcome these limitations, the concepts of fuzzy logic are adapted for further enhancement of the QoS evaluations.
This results in a methodical approach that is particularly useful for QoS management, and monitoring of complex networks, consisting of different configuration settings, various network technologies supporting different applications. Therefore, it can quantify the performance of heterogeneous network-based service models by a unified QoS metric. This approach is also useful when some specific network-based service models are re-deployed from one region to another region. Each area has its own service requirements and technology availability. As a result, the service model, which demonstrates better performance in one context, is not necessarily going to have the same outcomes in another region. In such circumstances, it is possible to compare the resultant QoS level of any network-based service model with the expected QoS level by applying this approach.
To investigate the efficiency of the designed approach, a diverse range of simulation studies utilising different heterogeneous network-based service models are carried out. The simulation results indicate that the approach in this work facilitates better management and monitoring of heterogeneous network configurations and applications utilising them. The simulation studies also show that using the unified metrics, it is possible to choose a suitable network configuration for a particular application or service from among the range of available network configurations under investigation and classify them for their suitability to provide some specific services. Overall, the outcomes from the simulation results analysis clearly demonstrate that the proposed methods can significantly improve the QoS analysis of the heterogeneous networks.