System level simulator for hybrid broadcast/cellular networks

As the focus of modern telecommunication systems shifts increasingly toward wireless mobile data transfer, the systems’ capacity of operation in unstable channel conditions is constantly improved. From a system simulation point of view this presents challenges both in terms of simulator complexity and the need to develop accurate system level models for mobile data transfer to analyze systems in different protocol layers.

Usually simulations are performed for one-user point-to-point link. Overall system performance analysis requires thousands of such trajectories. It is evident that bit-true simulations for all users result in unacceptable simulation duration.

Simulator consists of three parts:

• Mobility model

This part of the simulator moves users across the service area and provides the user velocity. Mobility around the cell accounts for need to include shadowing into the model.

• Coverage estimation

The goal of this task is to provide a sufficiently accurate coverage map of the scenario under study. The accuracy of the propagation models depend on available cartography and its resolution (pixel size), 3D cartography being the most detailed one, including terrain height, terrain morphology and shapes of buildings. In simplest case, it is possible to user attenuation models such as Okamura-Hata. Better results are obtained with the use of professional coverage planning tools (e.g., Atoll) and 3D cartography to obtain a coverage map. If possible, coverage predictions should be compared with the measurements.

• Packet channel model

To construct efficient simulation models of complicated telecommunication systems, it is valuable to have error models that can be applied directly in simulating a specific protocol layer. This can be realized using packet, or block, error models, where given data protocol packets are labeled as erroneous or error-free according to suitable criteria. By nature, such models are realizable using discrete-time stochastic processes. However, determining the structure and parameters of such processes is a nontrivial task; even assuming that accurate physical layer channel models were available for a given transmission scenario, it is often infeasible to derive analytical expressions for the resulting high-protocol layer error process. Our research is concentrated on analyzing the performance and methods of evaluating the model parameters of finite-state packet error models. Specific topics in packet channel modeling are:

• Model parameterization, i.e. determining the parameters of a high-protocol layer packet error model as functions of measurable physical reception conditions such as the signal strength and vehicle speed.
• Including byte error approximation in packet error models. That is, for cases where more detailed analysis in needed, packet error model should be able to generate approximation of byte or bit level error distribution.

The general structure for system simulator is radio access network (RAN) independent. Adding more RANs to the simulator requires only creating suitable packet error models. Currently the system simulator is being constructed for DVB-H. Project participants are encouraged to add other systems (3G, WiMAX, etc.)

Simulator can be used to analyze for example

• How different RANs could be combined to provide improved performance for users having multimode terminals.
• Quality of Service (QoS) or Quality of experience (QoE) of a video streaming service for users
• Performance of a hybrid cellular and DVB-H IP IPDC system for different RRM strategies
• Initial performance analysis of candidate 4G systems.

Contacts

Team leader: Jarkko Paavola

Team supervisor: Valery Ipatov

Students involved in the project: Jussi Poikonen