Scalable method for digital beamforming without channel state information
A scalable algorithm may be used for determining beam/nullforming weights with minimal computation and minimal coordination between elements or subarrays. The algorithm may beamform large numbers of elements without channel state information and without knowing the signal of interest. The algorithm may optimize the signal-to-interference-plus-noise ratio of signals received at an array. The algorithm may remove the requirement for explicit channel state information (CSI) or for iterative co-estimation of CSI, remove the requirement for coordination with or knowledge of signals transmitted from null targets, and/or remove the requirement for knowledge of the number of targets by estimating and optimizing overall SINR, rather than jointly optimizing each received signal. The algorithm interprets the N-element arrays view of the signal-to-interference-plus-noise ratio as a N-dimensional gradient ascent. The search explores this N-D space to find a direction of ascent and applies a simultaneous search and track loop.
1 . An adaptive digital beamformer comprising:
N-number of analog-to-digital converters (ADC 0 to ADC N-1 ), wherein the N-number of analog-to-digital converters (ADC 0 to ADC N-1 ) are configured to generate N-number of in-phase and quadrature signals (s 0 (t) to s N-1 (t));
a search loop, wherein the search loop is configured to multiply the in-phase and quadrature signals (s 0 (t) to s N-1 (t)) with search gains (g S,0 [k] to g S,N-1 [k]), wherein the search gains (g S,0 [k] to g S,N-1 [k]) are complex gains based on search phases (Θ S,0 [k] to Θ S,N-1 [k]), wherein subsequent search phases (Θ S,0 [k+1] to Θ S,N-1 [k+1]) are defined by the search phases (Θ S,0 [k] to Θ S,N-1 [k]) plus a product of search step directions (D S,0 [k] to D S,N-1 [k]) and a search phase precision (φ S [k]), wherein the search loop is configured to determine a search signal-to-interference-plus-noise ratio (SINR S [k]), and
a track loop, wherein the track loop is configured to multiply the in-phase and quadrature signals (s 0 (t) to s N-1 (t)) with track gains (g T,0 [k] to g T,N-1 [k]), wherein the track gains (g T,0 [k] to g T,N-1 [k]) are based on track phases (Θ T,0 [k] to Θ T,N-1 [k]), wherein subsequent track phases (Θ T,0 [k+1] to Θ T,N-1 [k+1]) are defined by the track phases (Θ T,0 [k] to Θ T,N-1 [k]) plus a product of track step directions (D T,0 [k] to D T,N-1 [k]) and a track phase precision (φ T ), wherein the track loop is configured to determine a track signal-to-interference-plus-noise ratio (SINR T [k]);
wherein the search step directions (D S,0 [k] to D S,N-1 [k]) and the track step directions (D T,0 [k] to D T,N-1 [k]) are random discrete variables, wherein the search loop is configured to update the subsequent search phases (θ S,0 [k+1] to Θ S,N-1 [k+1]) and the search step directions (D S,0 [k] to D S,N-1 [k]) and the track loop is configured to update the subsequent track phases (Θ T,0 [k+1] to Θ T,N-1 [k+1]) and the track step directions (D T,0 [k] to D T,N-1 [k]) using the search signal-to-interference-plus-noise ratio (SINR S [k]) and the track signal-to-interference-plus-noise ratio (SINR T [k]) as feedback.
2 . The adaptive digital beamformer of claim 1 , wherein the adaptive digital beamformer is configured to form B-number of beams and M-number of nulls, where a sum of the B-number of beams and the M-number of nulls is at most one less than the N-number of in-phase and quadrature signals (s 0 (t) to s N-1 (t)).
3 . The adaptive digital beamformer of claim 1 , wherein the adaptive digital beamformer is configured to parallelize a plurality of search loops and a plurality of track loops.
4 . The adaptive digital beamformer of claim 1 , wherein phase adjustments and not magnitude adjustments are performed on the in-phase and quadrature signals (s 0 (t) to s N-1 (t)) in the search loop and the track loop.
5 . The adaptive digital beamformer of claim 1 , wherein the search phase precision (φ S [k]) is larger than the track phase precision (φ T ).
6 . The adaptive digital beamformer of claim 1 , wherein the search step directions (D S,0 [k] to D S,N-1 [k]) and the track step directions (D T,0 [k] to D T,N-1 [k]) are chosen from a set of {−1, 0, or 1}.
7 . The adaptive digital beamformer of claim 1 , wherein the search signal-to-interference-plus-noise ratio (SINR S [k]) and the track signal-to-interference-plus-noise ratio (SINR T [k]) are determined without measuring a channel state information and without measuring an interfering signal to be nulled.
8 . The adaptive digital beamformer of claim 1 , wherein the search loop is configured to determine a search phase-adjusted-signal power (S S [k]) and a search phase-adjusted-interference power (I S [k]), wherein the search signal-to-interference-plus-noise ratio (SINR S [k]) is the search phase-adjusted-signal power (S S [k]) divided by the search phase-adjusted-interference power (I S [K]);
wherein the track loop is configured to determine a track phase-adjusted-signal power (S T [k]) and a track phase-adjusted-interference power (I T [k]), wherein the track signal-to-interference-plus-noise ratio (SINR T [k]) is the track phase-adjusted-signal power (S T [k]) divided by the track phase-adjusted-interference power (I T [k]).
9 . The adaptive digital beamformer of claim 8 , wherein the search phase-adjusted-signal power (S S [k]) and the track phase-adjusted-signal power (S T [k]) are powers of a first in-phase and quadrature signal (s 0 (t)) after phase adjustment by a first search gain (g S,0 [k]) and after phase adjustment by a first track gain (g T,0 [k]), respectively;
wherein the search phase-adjusted-interference power (I S [K]) and the track phase-adjusted-interference power (I T [k]) are powers of a remainder of the in-phase and quadrature signals (s 1 (t) to s N-1 (t)) after phase adjustment by the remainder of the search gains (g S,1 [k] to g S,N-1 [k]) and the remainder of the track gains (g respectively.
10 . The adaptive digital beamformer of claim 1 , wherein the track loop is configured to set the track phases (Θ T,0 [k] to Θ T,N-1 [k]) to the search phases (Θ S,0 [k] to Θ S,N-1 [k]) and subsequent track step directions (D T,0 [k+1] to D T,N-1 [k+1]) to new random discrete variables if the search signal-to-interference-plus-noise ratio (SINR S [k]) is greater than the track signal-to-interference-plus-noise ratio (SINR T [k]).
11 . The adaptive digital beamformer of claim 1 , wherein the track loop is configured to set the track phases (Θ T,0 [k] to Θ T,N-1 [k]) to respective of previous track phases (Θ T,0 [k−1] to Θ T,N-1 [k−1]) and update subsequent track step directions (D T,0 [k+1] to D T,N-1 [k+1]) to new random discrete variables if the track signal-to-interference-plus-noise ratio (SINR T [k]) is less than or equal to a previous track signal-to-interference-plus-noise ratio (SINR T [k−1]);
wherein the track loop is configured to set the subsequent track step directions (D T,0 [k+1] to D T,N-1 [k+1]) to the track step directions (D T,0 [k] to D T,N-1 [k]) if the track signal-to-interference-plus-noise ratio (SINR T [k]) is not less than or equal to the previous track signal-to-interference-plus-noise ratio (SINR T [k−1]).
12 . The adaptive digital beamformer of claim 1 , wherein the search loop is configured to set the search phases (Θ S,0 [k] to Θ S,N-1 [k]) to respective of previous search phases (Θ S,0 [k−1] to Θ S,N-1 [k−1]) and set subsequent search step directions (D S,0 [k+1] to D S,N-1 [k+1]) to new random discrete variables if the search signal-to-interference-plus-noise ratio (SINR S [k]) is less than or equal to a previous search signal-to-interference-plus-noise ratio (SINR S [k−1]);
wherein the search loop is configured to set the subsequent search step directions (D S,0 [k+1] to D S,N-1 [k+1]) to the search step directions (D S,0 [k] to D S,N-1 [k]) if the search signal-to-interference-plus-noise ratio (SINR S [k]) is not less than or equal to the previous search signal-to-interference-plus-noise ratio (SINR S [k−1]).
13 . The adaptive digital beamformer of claim 12 , wherein the search loop is configured to set a subsequent search phase precision (φ S [k+1]) to a previous phase precision (φ S [k−1]) if the search signal-to-interference-plus-noise ratio (SINR S [k]) is less than or equal to the previous search signal-to-interference-plus-noise ratio (SINR S [k−1]).
14 . The adaptive digital beamformer of claim 12 , wherein the search loop is configured to update a subsequent search phase precision (φ S [k+1]) if the search signal-to-interference-plus-noise ratio (SINR S [k]) is not less than or equal to the previous search signal-to-interference-plus-noise ratio (SINR S [k−1]).
15 . The adaptive digital beamformer of claim 14 , wherein the search loop is configured to update the subsequent search phase precision (φ S [k+1]) based on a maximum of a search beamforming error (φ S,BE [k]) and a search nullforming error (φ S,NE [k]).
16 . The adaptive digital beamformer of claim 15 , wherein the search beamforming error (φ BE [k]) is defined by real components of a quotient of a search phase-adjusted-signal power (S S [k]) over a input signal power (S IN [k]); wherein the search nullforming error (φ NE [k]) is defined by imaginary components of a quotient of a search phase-adjusted-interference power (I S [K]) over a input interference power (I IN [k]); wherein the input signal power (S IN [k]) is a power of a first in-phase and quadrature signal (s 0 (t)); wherein the input interference power (I IN [k])) is a power of a remainder of the in-phase and quadrature signals (s 1 (t) to s N-1 (t)).
17 . The adaptive digital beamformer of claim 15 , wherein the search loop is configured to normalize the search beamforming error (φ BE [k]) and the search nullforming error (φ NE [k]) to the N-number of the in-phase and quadrature signals (s 0 (t) to s N-1 (t)).
18 . A radio system comprising:
an adaptive digital beamformer comprising:
N-number of analog-to-digital converters (ADC 0 to ADC N-1 ), wherein the N-number of analog-to-digital converters (ADC 0 to ADC N-1 ) are configured to generate N-number of in-phase and quadrature signals (s 0 (t) to s N-1 (t));
a search loop, wherein the search loop is configured to multiply the in-phase and quadrature signals (s 0 (t) to s N-1 (t)) with search gains (g S,0 [k] to g S,N-1 [k]), wherein the search gains (g S,0 [k] to g S,N-1 [k]) are complex gains based on search phases (Θ S,0 [k] to Θ S,N-1 [k]), wherein subsequent search phases (Θ S,0 [k+1] to Θ S,N-1 [k+1]) are defined by the search phases (Θ S,0 [k] to Θ S,N-1 [k]) plus a product of search step directions (D S,0 [k] to D S,N-1 [k]) and a search phase precision (φ S [k]), wherein the search loop is configured to determine a search signal-to-interference-plus-noise ratio (SINR S [k]), and
a track loop, wherein the track loop is configured to multiply the in-phase and quadrature signals (s 0 (t) to s N-1 (t)) with track gains (g T,0 [k] to g T,N-1 [k]), wherein the track gains (g T,0 [k] to g T,N-1 [k]) are based on track phases (Θ T,0 [k] to Θ T,N-1 [k]), wherein subsequent track phases (Θ T,0 [k+1] to Θ T,N-1 [k+1]) are defined by the track phases (Θ T,0 [k] to Θ T,N-1 [k]) plus a product of track step directions (D T,0 [k] to D T,N-1 [k]) and a track phase precision (φ T ), wherein the track loop is configured to determine a track signal-to-interference-plus-noise ratio (SINR T [k]);
wherein the search step directions (D S,0 [k] to D S,N-1 [k]) and the track step directions (D T,0 [k] to D T,N-1 [k]) are random discrete variables, wherein the search loop is configured to update the subsequent search phases (Θ S,0 [k+1] to Θ S,N-1 [k+1]) and the search step directions (D S,0 [k] to D S,N-1 [k]) and the track loop is configured to update the subsequent track phases (Θ T,0 [k+1] to Θ T,N-1 [k+1]) and the track step directions (D T,0 [k] to D T,N-1 [k]) using the search signal-to-interference-plus-noise ratio (SINR S [k]) and the track signal-to-interference-plus-noise ratio (SINR T [k]) as feedback.
19 . The radio system of claim 18 , comprising an antenna array and a radio-frequency front end, wherein the radio-frequency front end is coupled between the antenna array and the adaptive digital beamformer, wherein the N-number of analog-to-digital converters (ADC 0 to ADC N-1 ) are coupled to the radio-frequency front end.
20 . The radio system of claim 18 , comprising a software-defined radio, wherein the software-defined radio is configured to receive a combined digital signal (s(t)) from the track loop, wherein the combined digital signal (s(t)) is a sum of the product of respective of the in-phase and quadrature signals (s 0 (t) to s N-1 (t)) and respective of the search gains (g S,0 [k] to g S,N-1 [k]).