The Adaptive Antenna Testbed

An adaptive antenna for DCS-1800 working in the uplink mode has been designed and constructed at the Signals and Systems Group in co-operation with Ericsson Radio Access AB. It has two multiplexing channels  allowing for trials with SDMA. It uses an DCS-1800 basestation as a host for generation of standarized DCS-1800 (GSM) signals for realistic bit error rate measurements. The architecture and results from measurements are presented in the conference papers listed below, and also in a forthcoming licentiate thesis.  The picture shows the antenna array to the left, consisting of ten patch antenna elements situated in a circular array configuration. To the right is the signal processing rack, consisting of receivers, A/D converters, weights and a DSP.
 
 

By performing measurements in the laboratory, the performance of the adaptive antenna was evaluated, and also the functionality was verified. The antenna front end was replaced by a 8*8 analog butler matrix to emulate a linear array with isotrop antenna elements. Two signal generators were connected to the butler matrix and by varying the carrier and the interferer power different measurements were made including BER measurements. The graph below shows thw suppresion of the interfering signal as a function of its power and the carrier power. This graph verifies the result that the adaptive antenna supresses the interferer to the noise floor, so the interferer suppression is increased linearly with incresing interferer power. However, as also can be seen in the graph, there is a limit of suppression, this is due to the quantisation of the hardware weights. The weight accuarcy is 1 dB in amplitude and 1 degree in phase. When the interferer power is above -30 dBm, the suppression drops abruptly, due to saturation of AD converters in the receiving chain. For more information, please download the papers at the end of this page.

To evaluate the adaptive antenna performance in an well defined signal environment, we went to an antenna measurement range at FFV Aerotech in Arboga, Sweden. Here, radiation pattern for "easy" scenarios could be measured. The field was open , so no multipath is present, also the signal sources were stationary, thus avoiding fading.  The antenna front end was mounted on top of a mechanical turntable, see the figure. The antenna front end was covered with a green plastic non-conducting "barrel", for weather protection.

 Below the setup there was a room for equipment setup as seen in the picture below.

 

Measurements were made on antenna radiation patterns for different power settings for the desired and interfering signal sources. Also the angle separating the two signal sources were varied. Radiation patterns are presented below:
 
 

The picture above show the  radiation patterns with two signal sources of equal power, desired direction marked with green and interfering direction red. As can be seen the adaptive antenna creates a sharp and deep null against the interfering signal, while the mailobe points at the desired signal. But what happens if the separation angle gets smaller? This was investigated as seen below.

These two radiation patterns shows how the antenna tries to null the interferer when the angle separation is very narrow, only 5 degrees. It is also interesting to measure the Bit Error Rate (BER) as a function of separation for the two mobiles. The results from these measurements can be seen in the figure below.
 
 

It is interesting to see how the smallest separation angle allowed is dependent of the C/I impinging on the antenna. For C/I = -20 dB the smallest separation angle to maintain approximately 0 % of BER is 10 degrees.