Smart Antennas



The project on Smart Antennas aims at measuring and modelling the radio channel for handheld terminals with more than one antenna in operating conditions with interfering base stations or in macro diversity/soft-handover situations. The extra antenna (or antennas) potentially gives (give) a much-improved reception in operating conditions with either low received signal strength or much interference. Such improvement will be needed in order to supply users with services at higher data rates than the voice channel. As the influence of the user on the performance of the terminal is substantial we measure the radio channel with a number of test users. Each of them handles a small measurement handset during test runs. This handset mimics as realistically as possible nowadays-small handheld terminals, and is therefore connected with optic fibres to the receivers/data acquisition hardware [Kotterman2001a]. The advantage of the (non-conductive) optical connections is that these do not disturb the radiation pattern of a small terminal as coaxial cables certainly do [Kotterman2001b]. The earlier version of the measurement handset was fitted with two antenna branches that are simultaneouslymeasured. The new version has four antennas, switched two-by-two. The purpose of this larger number of antennas is mainly to find out which effect can be expected fromthe different separations between antennas. Monopoles or helical antennas are not preferred as they are too vulnerable on this handset. Besides, they are unlikely candidates for antennas on future terminals. During the test runs, three base stations are active and on each of the (active) antenna branches the signals of all three base stations are recorded simultaneously. We arranged two different base station configurations during the campaigns simulating two different interference situations. One configuration is with the base stations surrounding the measurement site; the second is with the base stations lined up. During 2001 we essentially made two different measurements campaigns: in Spring we measured in two different office buildings for sounding indoor-indoor channels, in Autumn we measured in the same buildings outdoor-to-indoor channels. Outdoor measurements in urban environment were performed in March 2002. The modelling done in 2001 is for the indoor-indoor environment. The power delay profiles of the channels were that short that a flat fading channel could be used. For the outdoor-to-indoor model this will certainly not be the case as the time dispersion is far too large. As we measured 6 channels (three base stations times two receive branches) simultaneously, the model we developed is a multi-channel model. Per channel a slow fading process and a fast fading process is generated. The resulting channel is the product of both. Standard deviation and distribution functions for both types of fading processes are derived from the measurements. Also derived from the measurements are the amounts of cross-correlation between the processes for the different channels. These values for the cross-correlations are incorporated in the fading generators in the model. The cross-correlation values for the slow fading processes were quite high, up to 0.94 (median 0.73), but those for the fast fading processes were relativelylow. For the measurements with monopoles, separated by 0.29 "lambda", the power cross correlation was at maximum0.3. For the measurements with patch antennas (separation 0.16"lambda") the power correlation coefficients stayed under 0.65 with a median value of 0.33 [Kotterman2002]. The final step in the modelling is adding a branch power difference to each antenna pair of channels, a random variable drawn from the measured distributions. This project was performed in cooperation with Nokia Research. For a more detailed description (Figures, Tables, references) please visit . (Wim A. Th. Kotterman, Kim Olesen, Devendra Prasad, Patrick C.F.Eggers)
Effektiv start/slut dato01/01/200031/12/2003