Monday, 28 August 2023: 1:45 PM
Great Lakes BC (Hyatt Regency Minneapolis)
The Airborne Phased Array Radar (APAR) concept consists of four phased array antennas to be installed on a National Science Foundation C-130 aircraft that is operated by the National Center for Atmospheric Research (NCAR). The antennas are active electronically steered arrays (AESAs) consisting of 56 rows of elements, with each row feeding a digital receiver. Each row is therefore a subarray. This architecture enables the acquisition of a slice of radar data by transmitting a fan beam that is broadened in elevation followed by digitally forming multiple simultaneous elevation beams during reception.
Signals from the elements in each row of the APAR AESA are summed using analog RF power combiners whose outputs are digitized to produce I/Q data streams, whereupon the row data streams are summed digitally in a processor. This type of hybrid analog/digital beamforming has been used for many years in arrays that have analog subarrays of equal size, but the APAR array is different because the analog subarrays (rows) have varying numbers of elements, as shown in the figure below. The traditional naive digital summation of row signals produces distorted antenna patterns in the present case because the analog subarray combiners sum signals in power while the digital beamformer sums voltages linearly. In effect, this approach mixes units and produces an erroneous result. An analysis of this hybrid beamforming case, where signals from unequal-sized analog subarrays are combined digitally, does not appear to have been treated in the existing literature.
While the coherent signals from elements in a row add in power in the RF power combiner, it is known that a portion of the incoherent thermal noise power is dissipated in the combiner’s internal resistance, reducing the noise power exiting the combiner. We show that this dissipation, together with the need to present excess thermal noise to the ADCs to maintain a low overall noise figure, determines the electronic RF gain used behind each element. We then show that, if the r-th row contains M_r antenna elements, a correction factor of sqrt(M_r) must be applied digitally to its I/Q data stream prior to summation to obtain correct array performance. This factor may be combined with the usual complex weight that steers the beam in elevation and applies a taper for sidelobe control. (When subarrays contain the same number of elements, on the other hand, M_r is a constant that simply scales the overall array gain. Since this digitally applied gain does not affect SNR, it may be set to one for convenience, explaining why naive digital beamforming of equal-sized analog subarrays is successful.) The final part of this presentation uses these weights to derive expressions for the overall SNR and dynamic range of the AESA.
Signals from the elements in each row of the APAR AESA are summed using analog RF power combiners whose outputs are digitized to produce I/Q data streams, whereupon the row data streams are summed digitally in a processor. This type of hybrid analog/digital beamforming has been used for many years in arrays that have analog subarrays of equal size, but the APAR array is different because the analog subarrays (rows) have varying numbers of elements, as shown in the figure below. The traditional naive digital summation of row signals produces distorted antenna patterns in the present case because the analog subarray combiners sum signals in power while the digital beamformer sums voltages linearly. In effect, this approach mixes units and produces an erroneous result. An analysis of this hybrid beamforming case, where signals from unequal-sized analog subarrays are combined digitally, does not appear to have been treated in the existing literature.
While the coherent signals from elements in a row add in power in the RF power combiner, it is known that a portion of the incoherent thermal noise power is dissipated in the combiner’s internal resistance, reducing the noise power exiting the combiner. We show that this dissipation, together with the need to present excess thermal noise to the ADCs to maintain a low overall noise figure, determines the electronic RF gain used behind each element. We then show that, if the r-th row contains M_r antenna elements, a correction factor of sqrt(M_r) must be applied digitally to its I/Q data stream prior to summation to obtain correct array performance. This factor may be combined with the usual complex weight that steers the beam in elevation and applies a taper for sidelobe control. (When subarrays contain the same number of elements, on the other hand, M_r is a constant that simply scales the overall array gain. Since this digitally applied gain does not affect SNR, it may be set to one for convenience, explaining why naive digital beamforming of equal-sized analog subarrays is successful.) The final part of this presentation uses these weights to derive expressions for the overall SNR and dynamic range of the AESA.

