Low-complexity performance evaluation methodologies for OFDMA-based packet-switched wireless networks
Cellular wireless networks have to be accurately planned to provide Quality of Service (QoS) to the user and to achieve revenue for the operator. Therefore, estimates of key performance indicators (KPIs) depending on parameters like the user scheduling, the data traffic and the data traffic load are necessary for planning of cellular wireless networks. Today’s cellular wireless networks are based on orthogonal frequency division multiple access (OFDMA) and a packet-switched network architecture like in, e.g., Worldwide Interoperability for Microwave Access (WiMAX) or Long Term Evolution (LTE) of Universal Mobile Telecommunications System (UMTS). OFDMA and a packet-switched network architecture are suitable for many challenges that today’s cellular wireless networks have to cope with, e.g., a large system bandwidth to provide high data rates or different traffic services. Resources in terms of bandwidth and time can be allocated for transmission of data in a very flexible manner and users are scheduled for transmission, e.g., depending on channel state information or QoS requirements. The consideration of all these parameters for evaluation of KPIs leads to a high computational complexity. Usually, system level simulations are required with simulation durations of days to obtain accurate results for KPIs.
In this project, a framework is derived for evaluation of KPIs in OFDMA based packet-switched wireless networks. Based on this framework, this project gives two novel evaluation methodologies. At first, a snapshot based simulative system level analysis methodology is proposed. Short and independent snapshots are considered for the evaluation of KPIs. During each snapshot, it is assumed that the location of a user does not change significantly so that large-scale propagation loss only has to be calculated once per snapshot. To get independent snapshots that are randomly selected from the busy hour, constraints are derived in this project, e.g., for the number of active users and the buffer occupancy at snapshot start. Probability density functions (pdfs) are given in this project for the constraints considering different data traffic models.
During each snapshot, the arrival of data packets in the transmitter, user scheduling and the frequency selective and time variant small scale fading channel are modeled in detail. The simulative system level analysis methodology is evaluated assuming a cellular wireless network, different channel models for the transmission of data and users that utilise different data traffic services. It only takes several hours to achieve KPI results with the simulative system level analysis methodology using a standard personal computer instead of more than a day with state-of-the-art dynamic system level simulations. The accuracy of the KPI results using the simulative system level analysis methodology is comparable to the accuracy achieved with state-of-the-art dynamic system level simulations. The simulative system level analysis methodology is suitable for accurate performance evaluation of a specific network configuration.
Secondly, an analytical system level evaluation methodology is defined in this project that contains an analytical calculation of KPIs for packet-switched wireless networks so that results for KPIs are achieved even faster than with the proposed simulative system level analysis methodology. Based on the developed framework for evaluation of KPIs in packet-switched wireless networks, properties in OFDMA based packet-switched wireless networks are described analytically. Derivations for pdfs of, e.g., the number of active users, the change in signal to interference ratio (SIR) and the number of allocated resources due to user scheduling are derived in this project so that KPIs are calculated analytically. Evaluations of the analytical system level evaluation methodology assuming a cellular wireless network indicate that KPI results are obtained within seconds at the expense of an error in KPI results below 15 % compared to system level simulations. Therefore, the analytical system level evaluation methodology is able to enhance planning tools for OFDMA based packet-switched wireless networks by providing fast estimates for KPIs. Nevertheless, system level simulations shall be used to obtain precise KPIs for the final network layout.
The impact of data traffic and user scheduling on KPIs is evaluated using the developed methodologies. Among others, it is shown that the average user throughput is almost proportional to the inverse of the average number of active users or that for a constant ratio between the average number of active users and the server data rate similar average cell throughput results are achieved. Furthermore, thresholds are derived indicating the performance that can be achieved for three different user scheduling strategies. Finally, KPI results are presented for a public reference scenario representing real world scenarios for European cities indicating that the developed simulative system level analysis methodology and the developed analytical system level evaluation methodology are both able to give KPIs for real world scenarios and can be applied in the planning process of OFDMA based packet-switched wireless networks.