Computational Complexity of Drop Based Radio Channel Simulation

Radio channel models intended for system performance assessment need to balance between accuracy and complexity. Computational complexity of channel realisation generation tend to grow due to increasing bandwidth, number of antennas, number of users, as well as with advanced network topologies. This paper investigates complexity related to radio channel parameterisation of a drop based MIMO simulation, where a number of user terminals are dropped into a network topology and short segments of the radio channel are simulated. The focus is on pre-processing associated with antenna embedding, e.g. interpolation of antenna field patterns, and stochastic propagations parameter determination, e.g. random delays, powers and azimuth angles. Complexities are calculated based on WINNER II channel model. Two modelling principles are compared, namely correlation matrix based and sum of rays based methods. Also variation of the bit error ratio (BER) performance on different drops was measured with a 2times2 MIMO hardware demonstrator. This study show that drop based simulation does not necessarily increase the complexity compared to the traditional fixed models. It also shows that geometry based simulation has lower pre-processing complexity than correlation based simulation.