Some experiments with array data collected in actual urban and suburban environments

We compare the performance of several algorithms for signal separation based on actual mobile cellular radio data. The data were collected by base stations in two different environments: using an eight element linear array on a hillside overlooking a suburban area, and using a four element square array at the top of a ten story building in a dense urban area. Calibration data were available for one of the arrays used for the suburban data, allowing for performance comparisons between DOA-based and blind signal separators. Experiments were conducted with three different types of signal waveforms: sinewaves, simple FM waveforms, and /spl pi//4-DQPSK signals, each of which necessitated the use of a different performance metric. We focus on the results obtained for cases where the data were approximately low rank, with little delay spread (although not all data sets from the urban environment were of this type).