Multi-service traffic modelling for wireless communication systems

This thesis presents generic traffic models which can be used for integrated voice and data traffic for networks with various characteristics. The main focus is on the impact of buffering voice and data packets having various traffic loads and analyzing the effects of mobility issues on system performance. Analytical models for single and multi-channel systems are developed and their solution techniques for finite and infinite queuing capacities are presented. For single server systems a new product form solution approach is only provided for the static stations. For more complex systems, when multi-server systems and mobile stations are considered, it may not be possible to find a product form solution because the increased number of transitions makes the solution intractable. Therefore, Kolmogorov equations are used effectively to find the solutions for the finite queuing capacities. On the other hand, Spectral Expansion is used for systems with in- finite queuing capacities. The results obtained are validated using a discrete-event simulation (DES) tool. In integrated voice and data wireless networks, the most important consideration is to utilize the resources in an efficient way and maintain the required quality of service (QoS). Because of the increasing demand in using wireless networks for multiple services, efficient Buffer Management (BM) analysis, Call Admission Control (CAC) techniques and also threshold analysis in order to find the minimum and maximum values for the system resources (i.e. buffer and channel capacities) are required for integrating different services such as voice and data in order to provide the expected QoS to users. Unlike traditional networks, it is possible to take advantage of a single network for multiple services in the current wireless networks (WLAN, EDGE, 3G, HSDPA, HSPA, LTE). The integration of services prevents the need for many overlaying networks, which makes network management difficult. Service requirements are different for voice and data traffic. While voice packets are delay-sensitive, data packets cannot tolerate blocking (i.e. being dropped). In order to provide high QoS at all times, it is important to consider the trade-off between voice and data traffic. Current wireless network systems are indeed getting faster and faster. Therefore, it may be possible to queue voice calls. In addition, channel reservation is widely accepted as the only way to provide acceptable QoS for Wide Area Networks (WANs) as different characteristics of services are integrated. In this thesis, firstly, we developed a performance model of a cell for WLAN and cellular network systems where the impact of buffering voice packet is considered for integrated voice and data services. Following this, service and buffer management analysis is carried out to minimize the cost of data packets over voice packets and therefore prevent data packets to be lost. Furthermore, the effect of handoff arrival is included. Algorithms to determine the buffer thresholds for voice packets and the overall capacity are developed. Traffic constraints and mobility issues are integrated and a generic traffic model is proposed. The proposed generic model can be applied to all networks using different policies and algorithms. The results obtained from this study clearly show that it is possible to queue voice packets with a small portion of buffering. This reduces the blocking probability and hence allows for better operation of both traffic services. Moreover, apart from the prioritization of traffic classes, it is also important to optimize the buffering scheme for the overall system performance. Furthermore, this work points out that channel reservation, which is commonly used, is not the only way to provide acceptable QoS for integrated service networks. Different analytical solution techniques are used in order to obtain results which are validated using discrete-event simulation techniques. This study sheds a light to better understand the trade-off between different types of services to provide high QoS for wireless networks. Hence, the proposed models in this study can be considered as part of a generic model in the design and development of future wireless systems and therefore can be modified to support not only voice and data services but also video applications over several networks.