A Low-Complexity Time-Domain Beamforming Method and System Based on a Dual-Sampling Ultra-Wideband Folded Receiver
By using single-branch array element grouping and least common multiple period space-time alignment, the problem of array element-level replication sampling branches is solved, achieving low-complexity broadband coherent synthesis, reducing hardware requirements and maintaining the desired directional response.
Patent Information
- Application Number
- CN202611636630.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-10-22
- Publication Date
- 2026-11-17
AI Technical Summary
Existing dual-sampling folded arrays require replicating sampling branches at the element level, which increases the hardware scale and channel consistency calibration workload, and makes it difficult to directly achieve broadband coherent synthesis for elements with different sampling rates.
A method of single-branch array element grouping, least common multiple period spatiotemporal alignment, and two-level joint beamforming is adopted. By undersampling different array element groups at different sampling rates and aligning snapshot data on a common timestamp, broadband beamforming is achieved by using a constraint matrix for intra-group and inter-group phase compensation.
While reducing hardware complexity, coherent synthesis of different array elements was achieved, reducing the size of samplers, interfaces and buffers, and maintaining the desired directional response at multiple frequency points.
Smart Images

Figure CN123122579A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of digital beamforming technology, and in particular relates to a low-complexity time-domain beamforming method and system based on a dual-sampling ultra-wideband folded receiver. Background Technology
[0002] Ultra-wideband direction finding and interference-resistant reception require processing a wide radio frequency band simultaneously. If the analog-to-digital converter is configured according to the highest signal frequency, the sampling rate, interface bandwidth, storage capacity, and back-end computational load will all be large. Folded receivers utilize bandpass undersampling or direct RF sampling to fold signals from different Nyquist zones to a lower frequency band, and then combine multi-sampling rate information to eliminate frequency ambiguity, which can reduce the rate requirement of a single sampling branch.
[0003] A common dual-sampling structure connects two sampling branches in parallel after the same array element, with the two branches operating simultaneously at different sampling rates. For an M-element array, this structure typically requires 2M sampling branches, along with corresponding demultiplexing, clocking, buffering, and digital interface resources. As the number of array elements increases, the hardware scale and channel consistency calibration workload also increase. Another type of structure uses different undersampling rates for different array elements, avoiding the duplication of two sampling branches after each element. However, the two sampling rates correspond to different sampling times, and directly splicing the samples from each array element will disrupt the phase relationship of the broadband signal; furthermore, matching the folded spectrum only in the frequency domain cannot directly obtain the coherent array gain oriented towards the desired direction.
[0004] Therefore, a processing scheme is needed that is suitable for different array element groups using different sampling rates. Under the condition that only one sampling channel is retained in each array element, the two sampling sequences are aligned to a unified timestamp, and the spatial phase and temporal phase are compensated within and between groups, so that the two array elements can participate in the same broadband beamforming process. Summary of the Invention
[0005] The purpose of this invention is to solve the problem that existing dual-sampling folded arrays require replicating sampling branches at the element level, and that it is difficult to directly achieve broadband coherent combining after allocating the two sampling rates to different array elements. The invention proposes a low-complexity time-domain beamforming method and system based on a dual-sampling ultra-wideband folded receiver. It adopts a method that combines single-branch array element grouping, least common multiple period space-time alignment, and two-stage joint beamforming. In this method, each array element is connected to only one sampling channel, which does not mean that the entire receiving array contains only one array element channel.
[0006] This application provides a low-complexity time-domain beamforming method based on a dual-sampling ultra-wideband folded receiver, the method comprising:
[0007] The receiving array consisting of M elements is divided into a first array element group and a second array element group. Each array element in the two array element groups is connected to only one sampling channel. The first array element group undersamples the received signal at a first sampling rate, and the second array element group undersamples the received signal at a second sampling rate different from the first sampling rate, thus obtaining two sets of spectral folding sampling sequences.
[0008] Determine the least common multiple period of the sampling periods corresponding to the first sampling rate and the second sampling rate. Within each least common multiple period, collect a first number of snapshots from the first array element and a second number of snapshots from the second array element. Align the two sets of snapshots with timestamps to form a first spacetime snapshot vector and a second spacetime snapshot vector, respectively. The first number is determined by the coprime sampling factor of the second sampling rate, and the second number is determined by the coprime sampling factor of the first sampling rate.
[0009] Based on the preset frequency point and the desired beam direction, the constraint matrix of the first array element group and the constraint matrix of the second array element group, which include the element angle phase and tap delay, are constructed respectively. The covariance matrix is estimated by the first space-time snapshot vector and the second space-time snapshot vector respectively, and the beamforming weights in the first group and the second group are calculated to obtain the first group output and the second group output.
[0010] After aligning the common timestamps, the first and second sets of outputs are buffered according to their greatest common divisor. Based on the equivalent phase centers of the two arrays and the buffer tap delay, a two-level constraint matrix between the groups is constructed. The two sets of outputs are then subjected to two-level joint beamforming using the two-level constraint matrix to obtain the joint beam output.
[0011] Furthermore, the least common multiple period is determined by the following method:
[0012] Let the common reference sampling period be T. s If the first coprime sampling factor is r1 and the second coprime sampling factor is r2, then the first sampling rate is f. s1 =1 / (r1T s The second sampling rate is f. s2 =1 / (r2T s The greatest common divisor is N. gcd =gcd(r1, r2), η1=r1 / N gcd η2 = r2 / N gcd η1 and η2 are two coprime numbers with no common divisor; the least common multiple period is T. c =η1η2T s The first quantity is η2, and the second quantity is η1.
[0013] Furthermore, the first and second spacetime snapshot vectors are determined using the following method:
[0014] In the q-th least common multiple period, q = 0, 1, ..., (N) gcd -1), the time qT corresponding to the γ1 snapshot data of the m-th element in the first array element group. c +γ1η1T s γ1=1,2,…,η2; the γ2 snapshots of the nth element in the second array element group correspond to time qT. c +γ2η2T s γ2=1,2,…,η1; γ1 and γ2 are the time indices of two sets of array elements in the least common multiple period. The snapshot data is arranged in a fixed order according to the time indices, resulting in dimensions of γ1 and η2 respectively. and The first spacetime snapshot vector and the second spacetime snapshot vector, where and The number of array elements in the first array element group and the second array element group are respectively. M is the total number of array elements in the receiving array; m and n are the array element indices of the first array element group and the second array element group, respectively.
[0015] Furthermore, for the frequency f in the first array element and the second array element l And the desired direction θ0, construct the spacetime steering vector A1(f l ,θ0) and A2( f l ,θ0), l=1,2,…,L, where l is the frequency index, L is the total number of constraint frequencies, and the elements of the spacetime steering vector are e j2πfl(γi-1)ηiTs The spatial manifold components and coprime sampling delays of the corresponding array elements are given, where i∈{1,2}; γi corresponds to γ1 and γ2, which are the time indices of the two sets of array elements in the least common multiple period; the space-time guide vector A of L frequency points is used. i ( f l Arrange the elements (θ0, θ0) column-wise to obtain the constraint matrix C of the i-th element group. i =[A i (f1,θ0), A i (f2,θ0),…, A i (f L , θ0)],i∈{1,2},i represents the index of the constraint matrix of the first array tuple and the constraint matrix of the second array tuple.
[0016] Furthermore, the beamforming weights within the i-th array element are determined as follows: w i = , where R i Let be the covariance matrix of the i-th element. This represents the inverse operation. For conjugate transpose operation; C iLet f be the constraint matrix of the i-th tuple, and f be the frequency response constraint vector.
[0017] Furthermore, the frequency response constraint vector is used to ensure that the two array elements have the same linear phase reference, and the l-th element is e. −j2πflTs(η1η2-1) / 2 η1 and η2 are two coprime numbers without common factors.
[0018] Furthermore, the second-level joint beamforming includes: continuously preserving the greatest common divisor N. gcd The first and second sets of outputs are alternately arranged to form a joint snapshot vector; based on the equivalent phase center, common time interval, and target direction of the two array elements, an inter-group constraint matrix C3=[A3(f1,θ0), A3(f2,θ0),…, A3(f2,θ0),…,A3(f3,θ0),…,A3(f2 ... L , θ0)], where Let A3(f) be the signal manifold vector of two array elements. l ,θ0) includes a 1,q (f l , θ0) and q = 0, 1, ..., N gcd -1 is the time index of the least common multiple period, a 1,q (f l , θ0) and The receiving frequency f of the first element in the first array element group and the second array element group in the q-th least common multiple period is respectively. l The manifold of the signal with incident angle θ0; the phase compensation weight w3 is determined by C3, and the joint beam output is calculated. Where y1(N) gcd ) is the first element in the Nth array gcd Beamforming output with least common multiple of the period, y2(N) gcd ) is the second array element in the Nth position. gcd The beamforming output is a least common multiple of the period. Furthermore, before constructing the first and second space-time snapshot vectors, amplitude correction, phase correction, and integer sampling point delay correction are performed on the two sets of spectral folding sampling sequences using the amplitude and phase calibration coefficients of each sampling branch and the fixed time delay calibration value.
[0019] This application also provides a low-complexity time-domain beamforming system based on a dual-sampling ultrawideband folded receiver, the system comprising:
[0020] A receiving array, comprising a first array element and a second array element;
[0021] The sampling module has each array element connected to only one sampling channel in the sampling module. The sampling channel corresponding to the first array element group works at the first sampling rate, and the sampling branch corresponding to the second array element group works at a second sampling rate different from the first sampling rate.
[0022] The common timestamp alignment module is used to determine the least common multiple period of the two sampling periods. Within each least common multiple period, it collects snapshot data of the number of coprime sampling factors of the other sampling rate and forms the first space-time snapshot vector and the second space-time snapshot vector.
[0023] The first-stage beamforming module is used to construct broadband constraint matrices containing angle phase and tap delay respectively, calculate the beamforming weights of two groups within the group and output the first and second groups of outputs.
[0024] The second-stage beamforming module is used to cache two sets of outputs on a common timestamp and perform inter-group phase compensation and secondary joint beamforming based on the equivalent phase centers of the two array elements.
[0025] Furthermore, the sampling module includes M sampling channels, each corresponding to one of the M array elements; compared to the dual-channel structure where each array element has a first sampling channel and a second sampling channel, the system does not have an array element-level sampling channel splitter.
[0026] Furthermore, the common timestamp alignment module, the first-stage beamforming module, and the second-stage beamforming module include field-programmable gate arrays, radio frequency chips, and digital signal processors; the common timestamp alignment module includes a dual-port memory or shift register group that is cyclically addressed according to the least common multiple period, and generates two sampling clocks and a common timestamp synchronization mark according to a common reference clock.
[0027] The significant advancement of this invention compared to existing technologies lies in:
[0028] First, with the same number of array elements, the number of array element-level dual sampling branches is reduced from 2M to M, the array element-level sampling splitter is eliminated, and the size of the sampler, interface and buffer is reduced;
[0029] Second, two spatiotemporal snapshots are constructed using complementary phase samples within the least common multiple period, so that the outputs of different sampling rates have a definite common time reference.
[0030] Third, the constraint matrix simultaneously represents the spatial phase and the sampling time phase, which can maintain the desired directional response at multiple frequency points;
[0031] Fourth, the two-level structure first completes intra-group dimensionality reduction and then completes inter-group coherent synthesis, which facilitates parallel pipeline implementation in field-programmable gate arrays or RF chips.
[0032] To more clearly illustrate the functional characteristics and structural parameters of the present invention, further explanation is provided below in conjunction with the accompanying drawings and specific embodiments. Attached Figure Description
[0033] Figure 1 This is a schematic diagram of the overall structure of the low-complexity time-domain beamforming architecture based on a dual-sampling ultrawideband folded receiver of the present invention.
[0034] Figure 2 This is a schematic diagram of the array element grouping and the single sampling branch of each array element in this invention.
[0035] Figure 3 This is a schematic diagram of the sampling timing of the two sampling rates of the present invention within the least common multiple period.
[0036] Figure 4 This is a schematic diagram of the two-stage combined beamforming processing structure of the present invention.
[0037] Figure 5 This is a flowchart of the combined beamforming method of the present invention.
[0038] Figure 6 A schematic diagram of the optional computational structure when using alternating decimation and constraint matrix dimensionality reduction for secondary beamforming. Detailed Implementation
[0039] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0040] This invention divides M array elements into a first array element group G1 and a second array element group G2, with the number of array elements being respectively... and , The first array element uses the first sampling rate f. s1 Undersampling is performed, and the second array element is sampled at the second sampling rate f. s2 Undersampling is performed. Each array element corresponds one-to-one with one RF sampling channel, and two sampling channels are not set after the array element.
[0041] Let the common reference sampling period be T. s If the first coprime sampling factor and the second coprime sampling factor are r1 and r2 respectively, then the two sampling rates satisfy:
[0042] ;
[0043] ;
[0044] Greatest common divisor N gcd =gcd(r1, r2), η1= r1 / N gcd η2 = r2 / N gcd η1 and η2 are two coprime numbers with no common divisor, and their least common multiple is the period:
[0045] ;
[0046] Each least common multiple period T c Within the array, the first array element collects η² temporal samples, and the second array element collects η¹ temporal samples. The first and second array elements form arrays with dimensions of... and The space-time snapshot vector. By using common timestamp markers, cache addresses, and time delay phases relative to the common timestamp, samples from different sampling times are unified to the q-th common timestamp qT. c .
[0047] For multiple constrained frequency points f1, f2, ..., f L Given the desired direction θ0, construct constraint matrices C1 for the first array element and C2 for the second array element. Each column of the constraint matrix contains both the angular phase caused by the element position and the delay phase of the snapshot data relative to the common timestamp. Estimate the covariance matrices R1 and R2 based on the two sets of snapshots, and calculate the beamforming weights within each group:
[0048] ;
[0049] Where R i Let be the covariance matrix of the i-th element. This represents the inverse operation. This is the conjugate transpose operation. (C) i Let f be the constraint matrix of the i-th tuple, and f be the frequency response constraint vector; i∈{1,2}; Corresponding to as well as ;
[0050] The outputs of the first stage are as follows: and q is the least common periodic time index. Then, two sets of N are stored on the common timestamp. gcd Output value y i (1:N gcd For i∈{1,2}, construct the inter-group constraint matrix C3 based on the equivalent phase centers of the two array elements, determine the inter-group phase compensation weight w3, and obtain the second-level joint beam output:
[0051] ;
[0052] Where y1(N) gcd ) is the first element in the Nth array gcd Beamforming output with least common multiple of the period, y2(N) gcd ) is the second array element in the Nth position. gcd Beamforming output with least common multiple of the period.
[0053] The present application will be further described below with reference to the accompanying drawings.
[0054] like Figure 1 and Figure 2 As shown, the receiving array comprises M elements. The first element group G1 includes... Each array element, the second array element group G2 includes Each array element is grouped into two groups. The processor stores the coordinates of each element, so changes in the grouping method are reflected through spatial guidance components, and the two groups are not required to have the same number of elements or be arranged adjacently.
[0055] Each array element is connected sequentially to a low-noise amplifier, an optional bandpass filter, a sample-and-hold circuit, and an analog-to-digital converter. For sample-and-hold circuits that can directly handle the input RF bandwidth, the analog down-converter can be omitted, and direct RF undersampling can be performed.
[0056] The common reference clock frequency is f s =1 / T s An integer frequency divider generates two sets of sampling clocks using r1 and r2 as division factors. r1 and r2 are coprime; or they have a greatest common divisor N. gcd At this point, coprime factors η1 and η2 are used for time counting. The common timescale generator counts every T. c A synchronization flag is generated to reset or verify the read and write addresses of the two sets of circular caches.
[0057] Suppose there are P far-field broadband signals in the monitoring scenario, and the direction of arrival of the p-th signal is θ. p The desired direction is θ0. The continuous-time received signal of the m-th array element is written as:
[0058] ;
[0059] In the formula, a m (f, θ p ) represents the array element m at frequency f and direction θ p The spatial manifold under s p (t) represents the p-th incident signal, n m (t) represents noise. When element amplitude and phase errors exist, the calibration coefficients are incorporated into a. m (f, θ p ).
[0060] like Figure 3 As shown, in the q-th least common multiple period, q = 0, 1, ..., (N) gcd -1), the sampling time of the first array element is qT. c +γ1η1T s γ1=1,2,…,η2; the sampling time of the second array element is qT c +γ2η2T s γ2 = 1, 2, ..., η1, where γ1 and γ2 are the time indices of the two sets of array elements within the least common multiple period. The processor writes samples within the same least common multiple period into the corresponding circular buffer.
[0061] The first spatial snapshot vector x1(q) is concatenated with the data elements x1 and q in sequence according to their time indices. m,γ1 (q), m=1,2,.., γ1 is the element index of the first array element G1, and γ1 is the time index of the first array element in the least common multiple period:
[0062] ;
[0063] For the conjugate transpose operation, the second spacetime snapshot vector x2(q) is constructed according to the same rule:
[0064] ;
[0065] x2(q) is derived from the second array element G2. It consists of several array elements, and and The total number of array elements M that together constitute the receiving array, so the array elements in x2(q) are from The index starts with +1 and ends with M.
[0066] In x2(q), the second subscript of each parameter is the time index within the group. In engineering implementations, an element index-first arrangement can also be used, as long as the constraint matrices and weights are in the same order. For real-time stream processing, the buffer is implemented using shift registers or dual-port memory; in offline processing, equivalent grouping is performed based on timestamps.
[0067] For the first array element, l=1,2,…,L is the frequency index, and the frequency f l and direction θ p The guiding vector corresponding to the m-th element and the γ1-th element is:
[0068] ;
[0069] a m (f l , θ p ) represents the phase difference between array elements, ej2πfl(γi-1)η1Ts This indicates the time delay phase between taps.
[0070] For the second array element, the guiding vector is:
[0071] ;
[0072] Arrange the corresponding equivalent guidance components in order to obtain two spacetime guidance vectors A1(f) with desired directions θ0. l ,θ0) and A2( f l ,θ0), and then arrange the L frequency points column-wise to obtain the constraint matrix of the i-th matrix element:
[0073] ;
[0074] The constraint frequency points can be uniformly set within the operating frequency band, or adaptively set within the signal-occupied frequency band. The dimension of the frequency response constraint vector f is L, which is the same as the number of constraint frequency points, and remains consistent between the two sets. Preferably, the l-th element of f is e. −j2πflTs(η1η2-1) / 2 This allows both groups to use the same linear phase reference.
[0075] Calculate the two sets of weights and obtain... x i In (q), i∈{1,2} because x i (q) already contains multiple time phase information within the least common multiple period, and the weights within the group simultaneously complete array element synthesis and time phase delay compensation.
[0076] like Figure 4 As shown, the first stage is to and The spacetime snapshot signals are compressed into vector outputs, and the processor continuously stores N. gcd Two sets of outputs y based on a common timestamp i (q), followed by a second-level joint, the second-level input vector x3 is represented as:
[0077] ;
[0078] The inter-group spatial phase is determined based on the equivalent phase center of the two array elements, and qT is determined based on the tap index. c The time phase on the scale is determined, and an inter-group constraint matrix C3 is formed. The normalized steering vector is used as a fixed phase compensation weight, or w3 is calculated based on the joint covariance matrix and the minimum variance criterion of linear constraints, and the second-level output is obtained.
[0079] When two groups have different numbers of array elements, noise power, or effective channels, a group gain normalization coefficient is added to C3 or w3 to prevent the group with more array elements from having excessive weight simply due to dimensionality differences. If a sampling branch fails, simply delete the corresponding element from the corresponding snapshot vector and constraint matrix; there is no need to change the sampling clock of the other group.
[0080] like Figure 6 As shown, after beamforming and common timescale alignment within the aforementioned groups are completed, the historical taps of the two scalar outputs are alternately decimated. Let the secondary buffer length N be... gcd =2J-1, where N gcd For an odd number not less than 3, J = (N gcd +1) / 2. The first group retains the relative tap index S1={0,1,3,…,N}. gcd -2}, the second group retains S2={0,2,4,…, N} gcd -1}, each group retains J values, therefore the second-level joint snapshot is changed from the original N gcd The dimension changes to J; the initial tap 0 is retained in both groups. Tap indices are based on the common timescale qT. c Measurement, with an interval of T c And by combining the S1 and S2 indices, C3 is divided into C... 3,1 and C 3,2 Two sets of extracted guide vectors. And for C... 3,1 and C 3,2 Perform eigenvalue decomposition:
[0081] ;
[0082] Where U and V are the left and right characteristic matrices corresponding to the singular values, and Σ is the diagonal matrix formed by the non-zero singular values.
[0083] Take the first r non-zero singular values to form Σ r Take the corresponding left and right singular vectors to form U respectively r and V r, The constraint response vector f corresponding to the frequency domain constraint matrix C3 is reconstructed as follows:
[0084] ;
[0085] f r This is the constraint response vector after dimensionality reduction.
[0086] Both extraction and dimensionality reduction involve trade-offs between degrees of freedom and frequency response, and their use is determined by cache, computing resources, and response error requirements.
[0087] The present invention will be further described below with reference to specific embodiments.
[0088] Example
[0089] In one non-limiting embodiment, the receiving array has 16 elements. = =8. Common reference clock frequency f s Given a GHz frequency of 2.4 GHz, and taking r1=3 and r2=4, the first sampling rate f is... s1 800MHz, second sampling rate f s2 For 600MHz, N gcd =1, η1=3, η2=4, least common multiple period T c The value is 5ns, corresponding to a public timestamp rate of 200MHz.
[0090] The first array group is in each T c The first array element group collects four time snapshots for each element, with a first spatiotemporal vector dimension of 32. The second array element group collects three time snapshots for each element, with a second spatiotemporal vector dimension of 24. L constraint frequency points covering the bandwidth occupied by the monitoring signal are selected, and the weights within the 32-dimensional and 24-dimensional groups are calculated respectively. Each group outputs a 200MHz common timestamp sequence, which is then multiplied by N. gcd Each tap performs inter-group phase compensation and secondary synthesis. Compared to a structure with two sampling branches of 800MHz and 600MHz per array element, this embodiment uses 16 sampling channels instead of 32. This comparison only considers the number of sampling branches; actual power consumption and logic resources are also related to the analog-to-digital converter model, bit width, cache depth, number of constrained frequency points, and weight update cycle.
[0091] In the above 16-element embodiment, N can also be selected. gcd =5 secondary common timescale taps. The first group retains taps {0,1,3}, and the second group retains taps {0,2,4}, reducing the dimension of the secondary snapshot from 10 dimensions to 6 dimensions. If L constraint frequency points are set, singular value decomposition is performed on the reconstructed 6×L constraint matrix, and a compressed rank not exceeding 6 is selected according to a preset threshold. The frequency band response is checked before the corresponding weights are applied. This option does not change the timescale of each period T. c The steps involve collecting 4 and 3 original temporal samples, respectively, from the first and second array elements.
[0092] In the RF system-on-chip implementation, the two sets of sampled data enter independent phase buffers and matrix-vector multiplication pipelines. The covariance matrix and weights can be updated frame by frame, and the real-time data path only performs complex multiplication and accumulation; a common time-stamp synchronization marker is used to initiate the reading of the two sets of outputs and secondary synthesis. For fixed-beam scenarios, the three weights can be pre-calculated and stored in on-chip memory; for adaptive anti-interference scenarios, the processor periodically updates the weights according to the new snapshot.
[0093] The two sampling rates are not limited to the values mentioned above. For the generalized coprime sampling factor, it is only necessary to calculate the common time scale period according to the formula and set the η1 and η2 time phase positions respectively. The number of array elements can also be extended to more than three groups. In this case, the least common multiple of all sampling periods is used as the common time scale, and the number of phases in each group is equal to the ratio of the common time scale period to the sampling period of the group. Then, layered or tree-shaped joint beamforming is used.
[0094] The method of this invention can be used for uniform linear arrays, area arrays, conformal arrays, and distributed arrays. The spatial steering vector should be calculated based on the actual array element coordinates and the far-field or near-field propagation model. For near-field signals, the spatial phase of the plane wave is replaced with the spherical wave propagation phase from the array element to the target position, while the least common multiple period space-time alignment method remains unchanged.
[0095] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.
Claims
1. A low-complexity time-domain beamforming method based on a dual-sampling ultrawideband folded receiver, characterized in that, The methods include: The receiving array consisting of M elements is divided into a first element group and a second element group, with each element in the two element groups connected to only one sampling channel. The first array element undersamples the received signal at a first sampling rate, and the second array element undersamples the received signal at a second sampling rate different from the first sampling rate, thus obtaining two sets of spectral folding sampling sequences. Determine the least common multiple period of the sampling periods corresponding to the first sampling rate and the second sampling rate. Within each least common multiple period, collect a first number of snapshots from the first array element and a second number of snapshots from the second array element. Align the two sets of snapshots with timestamps to form a first spacetime snapshot vector and a second spacetime snapshot vector, respectively. The first number is determined by the coprime sampling factor of the second sampling rate, and the second number is determined by the coprime sampling factor of the first sampling rate. Based on the preset frequency point and the desired beam direction, the constraint matrix of the first array element group and the constraint matrix of the second array element group, which include the element angle phase and tap delay, are constructed respectively. The covariance matrix is estimated by the first space-time snapshot vector and the second space-time snapshot vector respectively, and the beamforming weights in the first group and the second group are calculated to obtain the first group output and the second group output. After aligning the common timestamps, the first and second sets of outputs are buffered according to their greatest common divisor. Based on the equivalent phase centers of the two arrays and the buffer tap delay, a two-level constraint matrix between the groups is constructed. The two sets of outputs are then subjected to two-level joint beamforming using the two-level constraint matrix to obtain the joint beam output.
2. The method according to claim 1, characterized in that, The least common multiple period is determined by the following method: Let the common reference sampling period be T. s If the first coprime sampling factor is r1 and the second coprime sampling factor is r2, then the first sampling rate is f. s1 =1 / (r1T s The second sampling rate is f. s2 =1 / (r2T s The greatest common divisor is N. gcd =gcd(r1, r2), η1= r1 / N gcd η2 = r2 / N gcd η1 and η2 are two coprime numbers with no common divisor; the least common multiple period is T. c =η1η2T s The first quantity is η2, and the second quantity is η1.
3. The method according to claim 2, characterized in that, The first and second spacetime snapshot vectors are determined using the following method: In the q-th least common multiple period, q = 0, 1, ..., (N) gcd -1), the time qT corresponding to the γ1 snapshot data of the m-th element in the first array element group. c +γ1η1T s γ1=1,2,…,η2; the γ2 snapshots of the nth element in the second array element group correspond to time qT. c +γ2η2T s γ2=1,2,…,η1; γ1 and γ2 are the time indices of two sets of array elements in the least common multiple period. The snapshot data is arranged in a fixed order according to the time indices, resulting in dimensions of γ1 and η2 respectively. and The first spacetime snapshot vector and the second spacetime snapshot vector, where and The number of array elements in the first array element group and the second array element group are respectively. M is the total number of array elements in the receiving array, and m and n are the array element indices of the first array element group and the second array element group, respectively.
4. The method according to claim 3, characterized in that, The constraint matrices of the first array tuple and the second array tuple are determined by the following method: For the frequency f in the first array element group and the second array element group l And the desired direction θ0, construct the spacetime steering vector A1(f l ,θ0) and A2( f l ,θ0), l=1,2,…,L, where l is the frequency index and L is the total number of constraint frequencies; the elements of the spacetime steering vector are e j2 πfl(γi-1)ηiTs The spatial manifold components and coprime sampling delays of the corresponding array elements are given, where i∈{1,2}, and γi corresponds to γ1 and γ2, where γ1 and γ2 are the time indices of the two sets of array elements in the least common multiple period; the space-time steering vector A of L frequency points is then used. i (f l Arrange the elements (θ0, θ0) column-wise to obtain the constraint matrix C of the i-th element group. i =[A i (f1,θ0), A i (f2,θ0),…, A i (f L ,θ0)],i∈{1,2},i represents the index of the constraint matrix of the first array tuple and the constraint matrix of the second array tuple.
5. The method according to claim 4, characterized in that, The beamforming weights within the i-th array element are determined as follows: w i = , where R i Let be the covariance matrix of the i-th element. This represents the inverse operation. For conjugate transpose operation; C i Let f be the constraint matrix of the i-th tuple, and f be the frequency response constraint vector.
6. The method according to claim 5, characterized in that, The frequency response constraint vector is used to ensure that the two array elements have the same linear phase reference, and the l-th element is e. −j2πflTs(η1η2-1) / 2 ; η1 and η2 are two coprime numbers with no common factors.
7. The method according to claim 6, characterized in that, Second-level joint beamforming includes: continuously preserving the greatest common divisor N. gcd The first and second sets of outputs are alternately arranged to form a joint snapshot vector; based on the equivalent phase center, common time interval, and target direction of the two array elements, an inter-group constraint matrix C3=[A3(f1,θ0), A3(f2,θ0),…, A3(f2,θ0),…,A3(f3,θ0),…,A3(f2 ... L , θ0)], where Let A3(f) be the signal manifold vector of two array elements. l ,θ0) includes a 1,q (f l , θ0) and q = 0, 1, ..., N gcd -1 is the time index of the least common multiple period, a 1,q (f l , θ0) and The receiving frequency f of the first element in the first array element group and the second array element group in the q-th least common multiple period is respectively. l The manifold of the signal with incident angle θ0; the phase compensation weight w3 is determined by C3, and the joint beam output is calculated. ;where y1(N gcd ) is the first array element in the Nth position. gcd Beamforming output with least common multiple of the period, y2(N) gcd ) is the second array element in the Nth position. gcd Beamforming output with least common multiple of the period.
8. The method according to claim 1, characterized in that, Before constructing the first and second space-time snapshot vectors, amplitude correction, phase correction, and integer sampling point delay correction are performed on the two sets of spectral folding sampling sequences using the amplitude and phase calibration coefficients of each sampling branch and the fixed delay calibration value.
9. A low-complexity time-domain beamforming system based on a dual-sampling ultrawideband folded receiver, characterized in that, The system is used to implement any one of the methods of claims 1 to 8, and the system comprises: A receiving array, comprising a first array element and a second array element; The sampling module has each array element connected to only one sampling channel in the sampling module. The sampling channel corresponding to the first array element group works at the first sampling rate, and the sampling branch corresponding to the second array element group works at a second sampling rate different from the first sampling rate. The common timestamp alignment module is used to determine the least common multiple period of the two sampling periods. Within each least common multiple period, it collects snapshot data of the number of coprime sampling factors of the other sampling rate and forms the first space-time snapshot vector and the second space-time snapshot vector. The first-stage beamforming module is used to construct broadband constraint matrices containing angle phase and tap delay respectively, calculate the beamforming weights of two groups within the group and output the first and second groups of outputs. The second-stage beamforming module is used to cache two sets of outputs on a common timestamp and perform inter-group phase compensation and secondary joint beamforming based on the equivalent phase centers of the two array elements.
10. The system according to claim 9, characterized in that, The sampling module includes M sampling channels, each corresponding to one of the M array elements; the system does not have an array element-level sampling channel splitter.
