Analysis of Radio Channel Measurements Using Multiple Base Stations

Abstract: Future wireless communication systems will utilize the spatial properties of the wireless channel to improve the spectral efficiency and thus increase capacity. This is realized by deploying multiple antennas at both the transmitter and receiver. Utilizing the spatial properties of the channel demands channel models that properly reflect these characteristics. Due to the unpredictable nature of the wireless channel, a common approach is to model its effects in a statistical fashion. A few large world wide cooperations, like the 3GPP or WINNER projects, have developed channel models intended for reference and standardization use. These models are partly based on bulk parameters which describe the characteristics of the channel over larger areas of several wavelengths. These parameters include shadow fading, angle spread, and delay spread among others, and are within the WINNER project called large scale (LS) parameters. Considering the variations of these parameters, it is likely that they are correlated between closely located areas. Therefore it is of key interest to model their correlation properties instead of randomizing them independently for each new simulation. In multi-cell scenarios, when considering macro diversity or evaluation other properties such as interference and/or handover schemes, the intra site correlation of the shadow fading parameters should be taken into account for proper simulations. Neglecting this correlation the interference could be over/under estimated, hence the system capacity can be over/under estimated. Further, reliable estimations of the interference levels will increase the frequency reuse, and may even permit reuse within the same cell, on spatially separated links. The thesis focuses on modelling the statistics and variations of these parameters to increase accuracy of channel models. Previous studies have looked at the autocorrelation properties of the shadow fading in outdoor environments and found it to follow an exponential decay. This means that an first order autoregressive function would be sufficient to model this effect. Research of the angular spread parameter is more scarce, and therefor this issue is studied herein. In addition to the autocorrelation properties the intra-site correlation between the different large scale parameters is of interest. It is herein found that the shadow fading is negatively correlated with angle spreads, while the angle spread at the base station and the angle spread at the mobile show positive correlation. Finally, the inter site correlation between one mobile and multiple base stations is studied as a function of the angular separation between the links. For closely located base station, ie with small angular separation as seen from the mobile, substantial inter-site correlation is found. It is further seen that the correlation of the shadow fading is more prominent than the correlation of angle spread.