Active sensing and inference methods and systems for millimeter wave applications

Abstract

Spatio-temporal sensing method and system for synthesizing structured virtual array manifolds for millimeter wave (mmWave) applications are described. An example method, based on the described embodiments, includes configuring multiple mmWave antennas to receive a signal associated with a direction of arrival that is within an angular range, and generating, based on processing the signal and a corresponding spatial filter associated with each of multiple analog combiners, digital measurements corresponding to the signal. In this example method, the multiple mmWave antennas are coupled to the multiple analog combiners, and each analog combiner is associated with a radio frequency chain and a spatial filter that spans the angular range. The method includes performing, using the digital measurements, at least one of a beam alignment operation, a channel estimation operation, or a multi-user detection operation. An example system uses one or more processors configured to implement the example method described above.

Description

International Patent Application Attorney Docket No.: 009062.8503.WO00, SD2024-084-2PCT ACTIVE SENSING AND INFERENCE METHODS AND SYSTEMS FOR MILLIMETER WAVE APPLICATIONS CROSS-REFERENCE TO RELATED APPLICATION

[0001] This patent document claims priority to and benefits of U.S. Provisional Patent Application No.63 / 593,497 entitled "ACTIVE SENSING AND INFERENCE FOR MILLIMETER-WAVE BEAM ALIGNMENT USING A SINGLE RADIO FREQUENCY CHAIN SYSTEM" and filed on October 26, 2023. The entire contents of the before- mentioned patent applications are incorporated by reference as part of the disclosure of this patent document. GOVERNMENT LICENSE RIGHTS

[0002] This invention was made with government support under Grant CCF-2124929 and Grant CCF-2225617 awarded by National Science Foundation (NSF). The government has certain rights in the invention. TECHNICAL FIELD

[0003] This patent document generally relates to millimeter wave communications, and more particularly, to active sensing and inference in millimeter wave applications. BACKGROUND

[0004] Millimeter wave (mmWave) technology is essential for expanding the existing capabilities of cellular networks. The availability of the large spectrum in the 30-300 GHz spectrum range, and the ability to place many more antennas in the same form factor on a device are very promising avenues for the next generation wireless systems. MmWave technology has numerous use cases, which include industrial-IoT, virtual / augmented reality, biomedical applications, and non-terrestrial networks. SUMMARY

[0005] Embodiments of the disclosed technology relate to methods, systems, and devices for active sensing and inference in mmWave applications. The described embodiments are applicable in a wide range of applications where a signal from multiple antennas is processed e.g., in millimeter-wave wireless systems such as UE and gNodeB for beam alignment, channel estimation, multi-user detection (MUD) and multi-user transmission, as well asInternational Patent Application Attorney Docket No.: 009062.8503.WO00, SD2024-084-2PCT millimeter wave radar systems.

[0006] In an example aspect, a system for wireless millimeter wave (mmWave) communication includes a first number of mmWave antennas, a second number of analog combiners coupled to the first number of mmWave antennas, and one or more processors. In this system, the first number of mmWave antennas is configured to receive a signal associated with a direction of arrival that is within an angular range, and each analog combiner is associated with a radio frequency (RF) chain and a spatial filter that spans the angular range. Furthermore, the one or more processors is configured to generate, based on processing the signal and a corresponding spatial filter associated with each of the second number of analog combiners, a plurality of digital measurements corresponding to the signal, and perform, using the plurality of digital measurements, at least one of a beam alignment operation, a channel estimation operation, or a multi-user detection operation. Herein, the second number is less than the first number, and a number of the plurality of digital measurements is greater than each of the first number and the second number.

[0007] In another example aspect, a method for wireless mmWave communication includes configuring a first number of mmWave antennas to receive a signal associated with a direction of arrival that is within an angular range, and generating, based on processing the signal and a corresponding spatial filter associated with each of a second number of analog combiners, a plurality of digital measurements corresponding to the signal. In this example method, the first number of mmWave antennas is coupled to the second number of analog combiners, and each analog combiner is associated with a radio frequency (RF) chain and a spatial filter that spans the angular range. The method further includes performing, using the plurality of digital measurements, at least one of a beam alignment operation, a channel estimation operation, or a multi-user detection operation. Herein, the second number is less than the first number, and a number of the plurality of digital measurements is greater than each of the first number and the second number.

[0008] In yet another example aspect, the above-described method may be implemented by an apparatus or device that includes a processor and / or memory.

[0009] In yet another example aspect, this method may be embodied in the form of processor-executable instructions and stored on a computer-readable program medium.

[0010] The subject matter described in this patent document can be implemented in specific ways that provide one or more of the following features.International Patent Application Attorney Docket No.: 009062.8503.WO00, SD2024-084-2PCT BRIEF DESCRIPTION OF THE DRAWINGS

[0011] FIG.1 is a block diagram illustrating the various operations in an example millimeter wave (mmWave) sensing system.

[0012] FIG.2 illustrates an example of a virtual uniform linear array (ULA) segment created using multiple snapshots (or coherence intervals).

[0013] FIG.3 is a block diagram illustrating the various operations in another example mmWave sensing system that includes feeding back the output of the system.

[0014] FIG.4 illustrates an algorithm listing for an example inference algorithm.

[0015] FIGS.5A and 5B are numerical results illustrating the efficacy of the described mmWave sensing system embodiments.

[0016] FIGS.6A–6C illustrate algorithm listings for an example adaptive SVAM beamforming algorithm.

[0017] FIG.7 is a flowchart of an example method for mmWave communication.

[0018] FIG.8 is a block diagram illustrating an example system configured to implement embodiments of the disclosed technology. DETAILED DESCRIPTION

[0019] Millimeter wave (mmWave) technology is essential for expanding the existing capabilities of cellular networks. The availability of the large spectrum in the 30-300 GHz spectrum range, and the ability to place many more antennas in the same form factor on a device are very promising avenues for the next generation wireless systems.

[0020] Section headings are used in the present document to improve readability of the description and do not in any way limit the discussion or the embodiments (and / or implementations) to the respective sections only.

[0021] 1. Introduction

[0022] In current wireless systems, more and more bandwidth intensive applications are emerging in daily routines of mobile users. Wireless data traffic is projected to skyrocket 10000 fold within the next 20 years. To tackle this incredible increase, one of the most efficient resolutions is to move the data transmissions into an unused nontraditional spectrum where enormous bandwidths are available, such as millimeter wave (mmWave). The impact of mmWave technology can be gauged from the numerous use cases enabled by mmWave technology, which includes industrial-IoT, virtual / augmented reality, biomedical applications, and non-terrestrial networks.International Patent Application Attorney Docket No.: 009062.8503.WO00, SD2024-084-2PCT

[0023] However, mmWave technology also faces many challenges. The mmWave channel incurs large propagation losses thereby restricting coverage per base station (BS), and requiring additional infrastructure compared to legacy cellular networks. A second challenge is the specular nature of mmWave channel rendering it to be sparse and requiring accurate beam alignment. This challenge is only further exacerbated by the narrow beamwidths and consequent large codebook size due to the large antenna array dimensions.

[0024] Several options are being considered for enhancing coverage at low-cost such as integrated access backhaul and intelligent reflective surfaces. On the other hand, reducing the beam alignment phase duration is a critical and active area of research. Hardware cost also impacts the ability of the transceivers to sense the mmWave channel. The large number of antenna elements are typically supported by only a few Radio Frequency (RF) chains, and thus necessitates for a low-dimensional projection of the received signal at the antennas. Beam alignment using such a low-dimensional signal is a challenging technical problem, and embodiments of the disclosed technology provide technical solutions which couple a better sensing approach with efficient inference mechanisms that exploit the array geometry and channel characteristics under the hardware constraints.

[0025] Some existing implementations formulate the beam alignment problem under a posterior matching framework, and actively learn the single-path direction of arrival (DoA), but assume that the small-scale fading coefficient (or complex path coefficient) is perfectly known. Subsequent efforts estimate both the DoA as well as the fading coefficient using, for example, Kalman filter-based posterior matching algorithms or variational hierarchical posterior matching algorithms, both of which require good a priori density information on the small-scale fading coefficient.

[0026] The described embodiments overcome the drawbacks of the existing implementations by considering the beam alignment problem from the perspective of active sensing for improved estimation performance, and do so without relying on additional information such as good prior knowledge about the small-scale fading coefficient. Some of the beneficial and advantageous features and aspects of the disclosed technology include:

[0027] – Sensing methodologies, inspired from Synthetic Aperture Radar (SAR), that uses limited RF chains and synthesizes a virtual Uniform Linear Array (ULA) manifold over spatial and temporal measurements. In this patent document, this method is referred to as a synthesis of virtual array manifold (SVAM) sensing (or SVAM sensing).

[0028] – When the channel small-scale fading coefficient is known, a reduction in theInternational Patent Application Attorney Docket No.: 009062.8503.WO00, SD2024-084-2PCT Cramer-Rao lower Bound (CRB) on the variance of estimation of the unknown dominant path angle (compared to a benchmark scheme), and the ability to incorporate the sensing methodologies within existing active beam alignment procedures, which results in lower training overhead for the integrated procedures.

[0029] – Beam alignment procedures that adapt the beamformer based on the current estimate of the posterior on the unknown angle. The proposed algorithms estimate a posterior on the small-scale fading along with the angular posterior. Furthermore, both flexible and hierarchical codebook-based beam alignment procedures are supported.

[0030] – Adaptive beam alignment procedures that have the ability to self- correct in case of premature misalignment during the early phase of the training period.

[0031] – For single RF chain systems, the described active sensing and inference methods can be implemented using conventional phased arrays and analog combiners. For such systems, the proposed sensing enables power-based high resolution beam alignment with just two measurements! This feature typically requires number of measurements that grow logarithmically with the required resolution.

[0032] FIG.1 is a block diagram illustrating an example active sensing mmWave system. As shown therein, an example system includes the operations of (1) designing a spatial filter, (2) determining design parameters for a virtual array manifold (VAM), (3) configuring analog combiners based on the spatial filter and VAM design parameters, (a) / (b) receiving measurements from the multiple (e.g., N) antennas using one or more (e.g., M) RF chains such that M is less than N, (4) synthesizing the VAM, and (5) implementing one or more receiver processing algorithms.

[0033] 2. Examples of active sensing and inference in mmWave systems

[0034] In some embodiments, a receiver (e.g., a base station or user equipment) with N antenna elements and R analog combiners (or alternatively, R radio frequency (RF) chains) are considered. In an example, the N antenna elements and R analog combiners (and RF chains) are configured as a fully-connected hybrid beamformer. Furthermore, it is assumed that there is a single dominant path between the transmitter and receiver (although this assumption is relaxed in some embodiments in Section 2.2 that consider multiple paths and / or multiple users), and that the channel remains coherent within the training duration.

[0035] It is noted that mmWave channels are often well characterized by a single dominant path – usually the line-of-sight (LOS) path or a strong reflection when LOS is blocked – and several other weaker paths. It has been shown that the contribution of the otherInternational Patent Application Attorney Docket No.: 009062.8503.WO00, SD2024-084-2PCT paths to noncoherent measurements appears as additional noise, so that optimal estimation based on a single-path model can be expected to perform well. As used in this patent document, the term “dominant path” (or “single dominant path”) corresponds the strongest path (e.g., based on the value of a metric like a signal-to-noise ratio (SNR) or the Received Signal Strength Indicator (RSSI)), which results in optimal estimation performing well.

[0036] The received signal at the ^^^^ antenna elements at instant ^^^^, ^^^^^^^^ ∈

[0037] Herein, ^^^^ denotes the total training duration, ^^^^^^^^ ∈ ℝ denotes the transmittedpower, ^^^^ ∈ ℂ is the unknown complex path gain (or coefficient), ^^^^^^^^(^^^^) is the (uniform lineararray (ULA) of size ^^^^) array manifold vector for a narrowband signal, ^^^^ = sin^^^^, ^^^^ ∈[−1,1), and ^^^^ ∈ [−^^^^⁄ 2 ,^^^^⁄ 2 ) denotes the incoming angle (e.g., the direction of arrival). Insome examples, the ULA is configured with half-wavelength (^^^^ / 2) spacing between antennaelements to avoid angular ambiguity. The noise ^^^^∈ ℂ is distributed asandindependent and identically distributed (i.i.d.) over time. The received signal is processedusing ^^^^ analog combiners, ^^^^^^^^,^^^^ ∈ ℂ^^^^ with ^^^^ ∈ {1, …and let ^^^^^^^^ =�^^^^^^^^,1, … , ^^^^^^^^,^^^^�. Theoutput, ^^^^^^^^ ∈ available for inference is given by:

[0038] The goal is to design ^^^^^^^^and infer ^^^^. Furthermore, ^^^^^^^^can be adapted over time to improve the inference (which is discussed in Section 3.1).

[0039] 2.1 Example sensing strategies based on the number of RF chains

[0040] The described technology provides the following three different sensing strategies depending on the available number of RF chains:

[0041] (1) Spatial-SVAM that exploits spatial degrees of freedom (DoF) using multiple yet fewer RF chains than antenna elements

[0042] (2) Temporal-SVAM that exploits temporal DoF using a single RF chain

[0043] (3) Spatio-Temporal-SVAM that exploits both spatial and temporal degrees of freedom to effect more than the provided number of RF chains

[0044] Each of these sensing strategies exploits the coherence across space or time to synthesize virtual array manifolds. The outcome is that it is possible to apply an unlimited number and less restrictive digital filters on the analog measurements gathered over space or time. The key aspect that allows one to do this is the synthesis of a structured array from theInternational Patent Application Attorney Docket No.: 009062.8503.WO00, SD2024-084-2PCT given uniform linear (or planar, or 3D) array. Said another way, the less restrictive digital filters can be applied to digital measurements in which the phase information of the analog measurements, which were digitized to generate the digital measurements, is preserved.

[0045] 2.1.1 Example spatial-SVAM embodiments

[0046] In some embodiments, a virtual uniform linear array of aperture size R (i.e., equal to the number of RF chains) is realized at the output of the beamformer in (2). Consequently, the measurements ^^^^^^^^become linear combinations of uniform linear array manifold vectors, and corrupted by noise. Similar ideas can be applied to realize either a uniform linear virtualarray of aperture size ^^^^^^^^,^^^^ ∈ ℕ (the set of natural numbers), or a non-uniform linear virtualarray (for example minimum redundancy array, nested array or co-prime array) post-combining. Herein, a spatial filter, f^^^^ ∈is designed to span the region of interest(RoI), which is range of angles that includes the direction of arrival (or equivalent, the range of interest can be defined as the narrowband signal being received at a range of angles thatincludes the direction of arrival). The analog combiner at time ^^^^ and RF chain ^^^^ ∈ {1, … ,^^^^} isdesigned as:

[0047] The proposed design in effect carries out a convolution operation on the signal at the antennas, and only preserves the steady state response to effect the reduction in the dimension at the output. Letwhich denotes the complex gain of the filter along angle ^^^^, be defined as: ^^^^ ^^^^^(^^^^) = f ^^^^^^^ ^^^^^^^^−^^^^+1(^^^^). (4)

[0048] The signal ^^^^ post-combining at the ^^^^-th RF chain can be expressed

[0049] Here, the measurements can be stacked to form ^^^^^^^^, which can be expressed as:

[0050] Herein, superscript S highlights that the measurement is under spatial-SVAM, and ^^^^^^^^(^^^^) is the synthesized virtual array manifold vector. The noise ^^^^^^^^can be ensured to remain white depending on the design of the beamformer ^^^^^^^^,^^^^and the filter f^^^^. It is noted that the assumption that channel remains coherent over time in (1) is not strictly needed. On the other hand, if that is the case, the virtual array manifold can be synthesized over both space and time, as discussed in Section 2.1.3.International Patent Application Attorney Docket No.: 009062.8503.WO00, SD2024-084-2PCT

[0051] 2.1.2 Example temporal-SVAM embodiments

[0052] In some embodiments, the fact that ^^^^ remains fixed (though unknown) over ^^^^ snapshots (coherence intervals) can be exploited to synthesize a virtual ULA over time. An important consequence is that, the measurements preserve phase information from the physical antenna, which captures rich information about the DoA of the incoming signal. Although this example assumes half-wavelength spacing, other virtual structured geometries can be similarly synthesized.

[0053] Let ^^^^^^^^denote the aperture size of the virtual ULA that is to be realized. FIG.2shows an example of a virtual ULA segment of aperture size ^^^^^^^^ = 4 that is created using 4snapshots. Therein, let ^^^^(^^^^) = ceil((^^^^ + 1) / ^^^^^^^^) denote the (ULA) segment index. Herein, aspatial filter, f^^^^(^^^^) ∈ ℂ^^^^−^^^^^^^^+1, is designed to span the region of interest (RoI). The analogcombiner at time ^^^^ is given by:

[0054] Within a segment, the spatial filter slides along the antenna aperture. In contrast to the previous case, where spatial degrees of freedom (multiple RF chain) were exploited to create virtual array manifold, here only a single RF chain is available and thus a ULA segment is synthesized over time (i.e., using temporal degrees of freedom). Let which denotes the complex gain of the filter along the angle ^^^^, be defined as:

[0055] The signal ^^^^^^^^,^^^^post-combining can be expressed as: ^^^^^^^^ = ^^^^^^^^^^^^ ^^^^^^^^ =�^^^^^^^^^^^^^^^^^^^^(^^^^)(^^^^) ∙ exp{^^^^^^^^^^^^ mod(1,^^^^^^^^)} + ^^^^^^^^ . (9)

[0056] It is noted that the complex gain ^^^^^^^^(^^^^)(^^^^) does not change within a segment, but the term “exp {^^^^^^^^^^^^ mod (^^^^,^^^^^^^^)}” varies within the segment. Here, the measurements within asegment can be stacked to form ^^^^^^^^ ={1, … , ^^^^ / ^^^^^^^^}, which can be expressed as:

[0057] Herein, superscript T highlights that the measurement is under temporal-SVAM, and ^^^^^^^^^^^^(^^^^) is the synthesized virtual array manifold vector. It is noted that since the noise in (1) is independent over time, the noise ^^^^^^^^is white, irrespective of the spatial filter design.

[0058] 2.1.3 Example spatio-temporal-SVAM embodiments

[0059] In some embodiments, multiple RF chains (but still numbering fewer than theInternational Patent Application Attorney Docket No.: 009062.8503.WO00, SD2024-084-2PCT number of antenna elements) are considered, and the fact that ^^^^ remains fixed (albeit unknown) is exploited to synthesize a virtual array manifold with aperture size more than the available number of RF chains. Although the described example assumes half-wavelength spacing, other virtual structured geometries can be similarly synthesized. Herein, an apertureof size ^^^^^^^^ = ^^^^^^^^,^^^^ ∈ ℕ is to be synthesized. Let ^^^^(^^^^) = ceil((^^^^ + 1) / ^^^^) denote the (ULA)segment index, and a spatial filter,region of interest(RoI). The analog combiner at time ^^^^ and RF chain ^^^^ ∈ {1, … ,^^^^} is given by:^^^^,^^^^ ^^^^^ ^^^^^^^^ ^^^^ =� ^^^mod(^^^^,^^^^)^^^^+^^^^−1^^^^�^^^^−mod(^^^^,^^^^)�^^^^−^^^^�. (11)

[0060] This embodiment utilizes the same spatial filter over Q snapshots. Using similar notation as for the single RF chain, the complex gain of the filter along the spatial angle ^^^^ isdenoted asand following the approach for the single RF chain case described in Section 2.1.2, the measurements can be stacked to form[^^^^(^^^^−1)^^^^ , … ,^^^^^^^^^^^^−1]^^^^ with ^^^^ ∈ {1, … , ^^^^ / ^^^^}, which can be expressed

[0061] Herein, superscript ST highlights that the measurement is under spatio-temporal- SVAM, and ^^^^^^^^^^^^(^^^^) is the synthesized virtual array manifold vector. Since the noise is independent over time, similar considerations as for spatial-SVAM are applicable here to ensure noise vector ^^^^^^^^is white.

[0062] 2.2 Example benefits and advantageous aspects

[0063] As discussed in the sensing strategies above, the measurements obtained are of a discrete-time complex exponential signal, which has been extensively studied and analyzed. These insights can be applied to each of the above-discussed sensing strategies. In this sense, the described embodiments helps to reduce the coupling between the sensing and inference problem. Some of the benefits and advantages of the sensing strategies include:

[0064] (1) Ensure informative measurements even when RoI is wide: The disclosed embodiments ensure that the measurements are informative about the unknown angle. It is noted that in the absence of prior knowledge on both α and u, at least two (spatial or temporal) measurements are needed to make meaningful inference on (α, u), even if the SNR is infinite. Moreover, the analog combiner over the two measurements ought to have different magnitude or phase response. This problem is avoided altogether in the described sensing strategies, as they ensure a variation in phase response across such measurements.

[0065] (2) SVAM for Adaptive Sensing (SVAM-AS): Some embodiments of theInternational Patent Application Attorney Docket No.: 009062.8503.WO00, SD2024-084-2PCT disclosed technology can be configured for adaptive sensing. The open loop system (e.g., shown in FIG.1) can be closed by using the outcome of the inference to reset the spatial filter. The synthetic aperture size and virtual array geometry can be adapted over time as well. Such adaptation can be used to improve the inference over time, as shown in FIG.3.

[0066] (3) Provide a structured framework for inference: As mentioned before, the described embodiments provide measurements of a discrete-time complex exponential signal. Such a measurement model allows for flexible beamspace processing. The flexibility is owing to the fact that these measurements are available for digital processing unlike the signal at the antennas (and are therefore referred to as “digital measurements”). A rich set of options for beamspace processing are possible because of the presence of the structured complex exponential signal.

[0067] In some embodiments, and in order to combat noise, multiple virtual (ULA) segments can be constructed while holding the spatial filter design f^^^^fixed. This duration is referred to as a slot (e.g., as shown in FIG.2). Let ^^^^^^^^denote the slot size, indicating the number of virtual ULA segments contained in one slot. The angle ^^^^ is inferred based on thefollowing average statistic (ℎ = 1, … , ^^^^ / (^^^^^^^^^^^^^^^^)):

[0068] Furthermore, in some embodiments, the slot duration can be adapted over time depending on the SNR.

[0069] (4) Allowing multiple paths: All three sensing strategies can be configured to operate even if more than one path exists between the transmitter and receiver as considered in (1). The superposition of multiple paths can be accommodated by considering the respective complex gain of the spatial filter along the multipath angles in (9). The spatial filter design problem then takes the passband filter design form.

[0070] (5) Multi-user detection OR multi-user transmission using reciprocity via virtual MIMO: The described embodiments can be configured to design precoders and combiners during the data transmission phase. The designed precoder / combiner in this manner (e.g., presented in (3), (7) and (11)) enables K ≥ 1 users to be served with the virtual array of a desired size. If the virtual array is of size M , then each user would transmit a symbol over M time intervals, for example using repetition coding. At the receiver using the proposed combiner, once the virtual array measurements are available, it is equivalent to anInternational Patent Application Attorney Docket No.: 009062.8503.WO00, SD2024-084-2PCT M antenna system with K users. The transmission strategy follows similarly using reciprocity. Thus, the proposed virtual MIMO scheme allows to multiplex independent data streams, using available spatial and temporal degrees of freedom.

[0071] 2.3 Example detection strategies

[0072] In some embodiments, and by leveraging the complex exponential signal component in (10), it is possible to apply much less restrictive digital filter to infer u. Thus, with little loss in terms of having zeros in the beamformer design, the limitations of having a single RF chain during the training phase can essentially be bypassed. This feature can be explicitly used within the detection strategy. By framing a detection question, the overall framework can be configured to operate in a low SNR regime. To this end, the RoI is dividedinto P partitions and P spatial filters ^^^^^^^^∈ ℂ with ^^^^ ∈ {1, … ,^^^^}, and each filter focusingon one such partition. The algorithm then compares power in the P narrow partitions within the RoI as follows: ^^^Dℎ = arg m^^a^^ xf^^^^ 2 ^ETℎd,e^^^t^� y�ℎ�. (14)

[0073] Although a large number of partitions allows to design narrow beamwidths with higher beamforming gain, it may also increase the probability of mis-detection.

[0074] 2.4 Examples estimation strategies

[0075] In some embodiments, the unknowns (^^^^,^^^^) are modeled as deterministic variables. Alternatively, both can be modeled as stochastic variables, especially if additional prior is available. The resulting MLE optimization is based on the well-known matching objective, which relies on the manifold vector ^^^^^^^^^^^^(^^^^′). This highlights the importance of the geometry that is synthesized, i.e., it decides the resolution. Sparse geometries such as nested arrays can synthesize large apertures in fewer snapshots.

[0076] 2.5 Example adaptation strategies

[0077] In some embodiments, the beam direction can be adapted based on the inference in the previous slot. For example, the beamwidth can be reduced by a deterministic amount (e.g., 0.5×previous beamwidth). In other embodiments, bounds on variance of estimation and concentration bounds can be used to adapt the beamwidth.

[0078] 2.6 Example inference algorithm for active sensing methods

[0079] In the described embodiments, the following parameters can be adjusted to achieve different sensing and inference objectives:

[0080] – the aperture size of the virtual ULA, ^^^^^^^^ ∈ {1, … ,^^^^}, andInternational Patent Application Attorney Docket No.: 009062.8503.WO00, SD2024-084-2PCT

[0081] – the spatial filter, which includes the beam direction and beam width.

[0082] Herein, ^^^^^^^^ = 1 reduces to the conventional beam design. Furthermore, and asdiscussed earlier, virtual ULAs with larger inter-element spacing and non-uniform linear arrays can be synthesized using the described embodiments.

[0083] FIG.4 illustrates an algorithm listing for an example inference algorithm, in accordance with the described technology. As shown therein, the input to the algorithm includes the total training duration (^^^^), the aperture size of the virtual ULA (^^^^^^^^), the slot size (^^^^^^^^), and either the number of spatial filters (P) for detection or a grid (^^^^grid) for estimation. Additionally, a new measurement is computed (in line 3) for each timestep in the training duration, an average statistic (e.g., as shown in (13)) is computed (in line 7) for each slot, and the spatial filter is computed (in line 10) using the Parks-McClellan filter design algorithm.

[0084] 2.7 Example numerical results

[0085] The efficacy of the above-described embodiments is shown in the numerical results illustrated in FIGS.5A and 5B. In FIG.5A, the root mean (over 200 realizations) squared error (RMSE) is plotted as a function of number of snapshots. Both curves assume αis known. The curve with the square markers uses the proposed SVAM sensing (^^^^^^^^ = 2) asopposed to the sensing methodology in a related existing implementation (Chui et al., “Active Learning and CSI Acquisition for mmWave Initial Alignment” in IEEE Journal on Selected Areas in Communications, August 2019). In this numerical example, posterior computations and the spatial filter design are similar to the existing implementation. As observed in FIG.5A, using the proposed SVAM sensing results in the algorithm requiring fewer snapshots for the posterior mode to align with the ground truth angle. In this experiment, 40 snapshots are sufficient for the proposed SVAM sensing based implementation, compared to the 90 snapshots required for sensing procedure in the existing implementation to ensure zero RMSE (i.e., no misalignment) for the beam alignment problem.

[0086] In FIG.5B, the RMSE (over 100 realizations) in u-space is plotted as a function of SNR in dB scale. Both α and u are unknown. The embodiment is configured to recover u on the grid and decide between two partitions (P = 2). In FIG.5B, the dashed curve for hierarchical posterior matching assumes a fixed prior of CN (α, 1) i.e., it assumes the mean is set to the correct value of α. As expected, when the SNR is low, the prior is more important and further helps to improve the estimation performance. At high SNR, the prior is less effective. Moreover, the initial measurements with fixed beam patterns are uninformative about the unknown angle, and adversely impact the performance of existing implementations.International Patent Application Attorney Docket No.: 009062.8503.WO00, SD2024-084-2PCT

[0087] As further observed in FIG.5B, simply replacing the sensing in the existing implementation with SVAM improves the performance (curve with square markers) indicating the effectiveness of the described technology. With α-unknown, the proposed sensing improves over the existing implementation with a fixed prior curve. Also, as observed, the detection strategy replaces the estimation strategy at low SNR. At high SNR, or after reliable detection, an estimation strategy may be used over the detection algorithm.

[0088] 3. Example embodiments for single RF chain mmWave systems

[0089] The temporal SVAM embodiments described in Section 2.1.2 are further detailed and discussed in this section (and using the same notation as therein). In particular, systems with a receiver (e.g., base station or user equipment) equipped with a ULA of size N and a single RF chain are considered. As in the previous section, it is assumed that there is a flat fading channel, with a single dominant path between the transmitter and receiver, and that the channel remains coherent within the training duration due to low receiver mobility.

[0090] In some embodiments, the beamformers can be designed as linear-phase Finite Impulse Response (FIR) filter using the Parks-McClellan algorithm.

[0091] In some embodiments, a virtual ULA can be constructed given just twomeasurements with ^^^^^^^^ = 2. This is equivalent to a contrived single snapshot measurementfrom a physical array of size 2. Owing to the rich (array) geometrical information preserved in the measurements, it is thus possible to estimate the dominant path DoA in a grid-lessmanner using existing techniques. In contrast, the beam scan operation using ^^^^^^^^ = 1 requiresas many measurements as the codebook size to detect the DoA.

[0092] 3.1 Examples of constructing a Sparse Linear Array (SLA)

[0093] In some embodiments, the framework described in Section 2.1.2 can be extended to form virtual ULA with more than ^^^^ / 2 spacing. More generally, a Sparse Linear Array (SLA) can also be realized as the virtual array geometry, for example, minimum redundancy arrays, nested arrays or co-prime arrays. These can help to increase the virtual aperture and improve resolution for the same segment duration.

[0094] Let ^^^^^^^^denote the number of antenna elements in the virtual SLA that is to beconstructed. Let ℙ = {^^^^^^^^: 0 ≤ ^^^^^^^^ < ^^^^,^^^^^^^^ ∈ ℤ, ^^^^ ∈ [^^^^^^^^]} denote the set of sensor positions inthe SLA ordered in an increasing manner, and ^^^^0 = 0 without loss of generality. Here, thelength of the beamformer f^^^^ to be designed is ^^^^ = ^^^^ − ^^^^^^^^^^^^−1. The analog combiner at time^^^^, in the case of SLA, is given by: ^^^^^^^^ =� ^^^^^^^^^^^^^^^^ ^^^^^^^^ mod(^^^^,^^^^^^^^)f^^^^0^^^^−^^^^−^^^^mod(^^^^,^^^^^^^^)�. (15)International Patent Application Attorney Docket No.: 009062.8503.WO00, SD2024-084-2PCT

[0095] The signal ^^^^^^^^post-combining can be expressed as:

[0096] Finally, the measurements within the ^^^^-th SLA segment can be stacked as:

[0097] Herein, ^^^^ℙ ∈ ℝ^^^^^^^^×^^^^ is a binary sampling matrix given by:[S 1 if ^^^^ = ^^^^^^ℙ] = � ^^^^^^,^^^^ 0 otherwise

[0098] 3.1 Examples of being agnostic to the adaptive scheme

[0099] Embodiments of the disclosed technology can be configured to remain agnostic to the adaptive scheme that is used in a system. In some embodiments, the beamformer design is deterministically modified after every ^^^^^^^^snapshots, involves a shift in space as described in (7) and (15), and requires the beamformer to be fixed for ^^^^^^^^snapshots. Furthermore, theSVAM beamformer has a size ^^^^ = ^^^^ − ^^^^^^^^ + 1. This configuration enables the beamformerdesign to be used with any adaptive scheme such that the adaptive scheme will choose beamformers close to the direction of the DoA, i.e., the described embodiments can remain agnostic to the adaptive scheme that is deployed in a mmWave system.

[0100] 3.2 Example beam alignment for an unknown complex path coefficient

[0101] In some embodiments, the beam alignment procedure begins with a wide regionof interest (RoI), e.g., ^^^^ ∈ [^^^^^^^^ ,^^^^^^^^];^^^^^^^^ < ^^^^^^^^;^^^^^^^^ , ^^^^^^^^ ∈ [−1,1), and assumes that the directionof arrival (DoA) lies within the RoI. Thus, the SVAM beamformer is initially set to span this region and suppress any interference coming from outside the RoI. This ensures any prior information about the DoA is incorporated. The approach is also practical, as base stations are typically deployed with dedicated antennas to serve a specific RoI. Measurements are collected over time and processed to compute an approximate posterior on the unknown angle. The SVAM beamformer is adapted once enough posterior mass around the mode of the posterior is accrued.

[0102] As discussed previously, since ^^^^ is unknown but assumed to be fixed during the training phase, it can be estimated along with the DoA given two or more measurements. An analysis of the CRB on the angle ^^^^ highlights a key requirement for being able to estimate both DoA and α simultaneously. The magnitude or phase response of the beamformer ^^^^^^^^should have variation over time, and it must vary differently for different angles. It is noted that this requirement is not satisfied for the existing implementations used in the numericalInternational Patent Application Attorney Docket No.: 009062.8503.WO00, SD2024-084-2PCT comparison discussed in Section 2.7, because that scheme is likely to capture all the initial measurements using a fixed codeword. However, this condition is naturally prevented in embodiments of the disclosed technology, as the phase response (over time) leads to thevirtual array manifold, which is sufficient to estimate both α and u when ^^^^^^^^ > 1 and given atleast two measurements.

[0103] The two basic prerequisites for adapting the SVAM beamformer are the next beam direction to steer towards, and the beamwidth. Since at least some of the described embodiments consider the single path scenario, the beam direction can be estimated by matching-based criteria which results in a maximum likelihood estimate (MLE), and it also yields a deterministic estimate for α. However, good beamwidth selection requires accounting for the uncertainty in the estimation procedure. The aforementioned CRB analysis can be used therefor. Herein, this is achieved by modeling both α and the unknown DoA, u, as stochastic variables. For the latter case, a uniform prior distribution is imposed in the wide RoI. Furthermore, α is modeled with a complex circular Gaussian distribution with a parameterized prior.

[0104] Next, a uniform grid, ugrid, of size ^^^^ within the RoI, [^^^^^^^^,^^^^^^^^], is introduced. Let ^^^^^^^^with ^^^^ ∈ [^^^^] denote the ^^^^-th grid point. The initial grid can be refined as the uncertainty of theestimate for u is successively reduced over snapshots. For each candidate DoA, ^^^^^^^^, a corresponding value for the complex path gain is estimated, and denoted ^^^^^^^^. Furthermore, theparameterized prior (^^^^^^^^~^^^^^^^^(0, ^^^^^^^^)) is imposed on ^^^^^^^^. In some embodiments, thehyperparameters of the imposed prior, namely ^^^^^^^^ with ^^^^ ∈ [^^^^] are determined in the MLEsense. The resulting estimate can be interpreted as normalized beamforming output power that is compensated for the noise power. The posterior probability on the DoA is then computed using marginalization and Bayes’ rule, and can be used to adapt the SVAM beamformer for the next snapshot. The aim is to ensure that the DoA lies within the passband of the SVAM beamformer so that effective received SNR for inference is high. A high effective SNR helps further to ensure a successful beam alignment.

[0105] FIGS.6A–6C illustrate algorithm listings for an example adaptive SVAM beamforming algorithm. As shown in FIG.6A, the beam direction of the SVAM beamformerf0 is initially set to the center of the RoI, and the beamwidth to ^^^^^^^^f = ^^^^^^^^initial < 2 in u-space, such that ^^^^^^^^initialcovers the RoI. The SVAM beamformer is only adapted if there is a sufficient posterior mass concentrated around the posterior mode. More specifically, the beamformer is adapted if the peak of the posterior mass in any contiguous span of a specificInternational Patent Application Attorney Docket No.: 009062.8503.WO00, SD2024-084-2PCT beamwidth, that includes the posterior mode, exceeds a fixed threshold ^^^^thresh. The algorithmbegins with setting the beamwidth for this search to ^^^^^^^^check = 0.5 × ^^^^^^^^f, i.e., half of thecurrent beamwidth for the SVAM sensing. If the threshold condition is not satisfied, ^^^^^^^^checkis doubled. This continues until the posterior threshold condition is met. Note that in the default scenario, the next beam resorts to the initial beam specifications. When the posterior threshold condition is met, the corresponding contiguous span of angular grid points is selected, and the next SVAM beamformer, f^^^^+1, is designed to cover the selected region. In some embodiments, the threshold parameter ^^^^threshmay be set to a fixed value; alternatively, is can be chosen dynamically. Herein, a lower value for ^^^^threshallows the SVAM beamformer to adapt often, whereas a higher value makes the adaptations more cautious. In Algorithm 1, the computation of the cumulative peak in lines 12 and 15 is performed using Algorithm 2, those listing is illustrated in FIG.6B.

[0106] In some embodiments, the described adaptive beam search mechanism has the ability to both refine, as well as correct erroneous adaptations. In the case when the beam is adapted in the wrong portion of the spatial region, the posterior mass along with the mode is expected to either shift or widen as more measurements are collected. The beam gets rectified within the proposed scheme as the search is carried out around the updated posterior mode, and it has the ability to widen the beamwidth beyond the ^^^^^^^^fpresently in use for the sensing.

[0107] In some embodiments, Algorithm 1 in FIG.6A adopts a flexible beam design which requires beam direction and beamwidth to design the SVAM beamformer. In some cases, such flexible beam designs may not be possible due to complexity, and a smaller codebook may be desired. Taking the example of the (binary) hierarchical codebook, the described embodiments can select a codeword based on the posterior (whose computation is discussed above). The procedure closely mimics the approach taken in Algorithm 1, while constraining the beam to belong to the hierarchical codebook.

[0108] Herein, the basic idea is to begin at one level below in hierarchy compared to the current level used for SVAM sensing, and traverse up the hierarchy to satisfy the posterior condition. At one-level below current, the algorithm begins with that node which contains the posterior mode. The principle in doing so, is to try and ensure that the DoA is included in the next beam. This beam search is summarized in Algorithm 3, illustrated in FIG.6C, and which can replace line 11−18 in Algorithm 1 (shown in FIG.6A).

[0109] In some embodiments, the beam search in Algorithm 3 can be initialized at a deeper hierarchical level, which can be favourable to converge early at high SNRs, but mayInternational Patent Application Attorney Docket No.: 009062.8503.WO00, SD2024-084-2PCT lead to premature beam focusing at low SNRs. This alternative initialization can also be incorporated into Algorithm 1.

[0110] 4. Example implementations of the disclosed technology

[0111] Embodiments of the disclosed technology provide a systematic mechanism for capturing highly informative measurements. The described sensing strategies exploit the mmWave channel statistics and the geometry of the antenna array to take measurements that carry information about the unknown angle of arrival and the small-scale fading coefficient. As discussed above, the described embodiments include sensing under extreme hardware constraints such as a single RF chain system, and can be easily utilized within a broad class of beam alignment algorithms.

[0112] FIG.7 shows a flowchart for an example method 700 for wireless millimeter wave (mmWave) communication. The method 700 includes, at operation 710, configuring mmWave antennas to receive a signal associated with a direction of arrival that is within an angular range.

[0113] The method 700 includes, at operation 720, generating, based on processing the signal and a corresponding spatial filter associated with each of multiple analog combiners, a plurality of digital measurements corresponding to the signal.

[0114] The method 700 includes, at operation 730, performing, using the plurality of digital measurements, at least one of a beam alignment operation, a channel estimation operation, or a multi-user detection operation.

[0115] In example method 700, the first number of mmWave antennas is coupled to a second number of analog combiners, each analog combiner is associated with a radio frequency (RF) chain and a spatial filter that spans the angular range, the second number is less than the first number, and a number of the plurality of digital measurements is greater than each of the first number and the second number.

[0116] The described features and aspects can be implemented to further provide one or more of the following technical solutions:

[0117] Solution A1. A wireless communication system, comprising: a plurality of antennas configured to receive at least one signal; a plurality of analog combiners, each analog combiner comprising a radio frequency (RF) chain and a spatial filter; and a processor, coupled to the plurality of analog combiners, configured to: convolute, using the plurality of analog combiners, the at least one signal, and generate, based on convoluting the at least one signal, a plurality of digital measurements corresponding to the at least oneInternational Patent Application Attorney Docket No.: 009062.8503.WO00, SD2024-084-2PCT signal, wherein the plurality of digital measurements are used to perform at least one of beam alignment, channel estimation, or multi-user detection.

[0118] Solution A2. The system of solution A1, wherein a number of the plurality of antennas is greater than a number of the plurality of analog combiners, and wherein the wireless communication system exploits spatial degrees of freedom.

[0119] Solution A3. The system of solution A2, wherein convolving the at least one signal produces an output that preserves the steady-state response in the at least one signal.

[0120] Solution A4. The system of solution A1, wherein a number of the plurality of analog combiners is one, and wherein the wireless communication system exploits temporal degrees of freedom.

[0121] Solution A5. The system of solution A5, wherein each of the plurality of antennas is associated with a common path gain.

[0122] Solution A6. The system of solution A1, wherein the wireless communication system exploits spatial and temporal degrees of freedom.

[0123] Solution A7. The system of any of solutions A1 to A6, wherein the plurality of digital measurements preserve a phase information of the at least one signal.

[0124] Solution A8. A wireless communication system, comprising: a sensing means configured to receive a plurality of analog signals; a synthesizing means configured to: process the plurality of analog signals using a set of analog combiners, and generate, subsequent to processing, a plurality of digital measurements corresponding to the signal; and a receiver processing means configured to perform at least one of beam alignment, channel estimation, or multi-user detection on the plurality of digital measurements.

[0125] Solution A9. A wireless communication method, comprising: receiving, using a plurality of antennas that correspond to a virtual uniform linear array (ULA), a signal characterized by a wavelength; synthesizing, based on the virtual ULA, a structured array; applying the structured array to the signal to generate a plurality of measurements comprising discrete-time complex exponential samples that correspond to the signal; and performing, using the plurality of measurements, at least one of beam alignment, channel estimation, or multi-user detection and transmission.

[0126] Solution A10. The method of solution A9, wherein a spacing between adjacent antennas of the plurality of antennas is equal to half the wavelength of signal.

[0127] The described features and aspects can be implemented to further provide one or more of the following additional technical solutions:International Patent Application Attorney Docket No.: 009062.8503.WO00, SD2024-084-2PCT

[0128] Solution B1. A system for wireless millimeter wave (mmWave) communication, comprising: a first number of mmWave antennas configured to receive a signal associated with a direction of arrival that is within an angular range; a second number of analog combiners coupled to the first number of mmWave antennas, wherein each analog combiner is associated with a radio frequency (RF) chain and a spatial filter that spans the angular range, and wherein the second number is less than the first number; and one or more processors configured to: generate, based on processing the signal and a corresponding spatial filter associated with each of the second number of analog combiners, a plurality of digital measurements corresponding to the signal, wherein a number of the plurality of digital measurements is greater than each of the first number and the second number, and perform, using the plurality of digital measurements, at least one of a beam alignment operation, a channel estimation operation, or a multi-user detection operation.

[0129] Solution B2. The system of solution B1, wherein the signal comprises a narrowband signal received over a channel comprising a single dominant path with a complex path coefficient.

[0130] Solution B3. The system of solution B2, wherein the one or more processors is configured, as part of performing the beam alignment operation, to: determine, based on the plurality of digital measurements, the direction of arrival; and determine, based on the direction of arrival, updated filter coefficients for the spatial filter for a subsequent transmission or reception using the first number of mmWave antennas.

[0131] In some embodiments (e.g., see Section 2.2(4)), the signal received at the first number of mmWave antennas comprises a narrowband signal received over a channel comprising multiple paths, and each of the multiple paths is associated with a complex path coefficient. In these cases, the one or more processors is configured, as part of performing the beam alignment operation, to determine, based on the plurality of digital measurements, the direction of arrival for each of the multiple paths, and determine, based on the directional of arrival for each of the multiple paths, updated filter coefficients for the spatial filter corresponding to each path of the multiple paths for a subsequent transmission or reception using the first number of mmWave antennas.

[0132] Solution B4. The system of solution B3, wherein determining the updated filter coefficients is based on using a Park-McClellan algorithm.

[0133] Solution B5. The system of solution B3 or 4, wherein determining the updated filter coefficients excludes relying on prior knowledge associated with the complex pathInternational Patent Application Attorney Docket No.: 009062.8503.WO00, SD2024-084-2PCT coefficient.

[0134] Solution B6. The system of solution B3 or 4, wherein determining the updated filter coefficients assumes a known value for the complex path coefficient.

[0135] Solution B7. The system of solution B2, wherein the second number is equal to one, and wherein generating the plurality of digital measurements exploits temporal degrees of freedom by assuming a known value for the complex path coefficient. Herein, the known value of the complex path coefficient corresponds to the complex path coefficient being substantially constant over the duration in which the plurality of measurements are generated. Additionally, or alternatively, the complex path coefficient (or alternatively, the multiple complex path coefficients when signals from multiple paths are received) can be accurately estimated during the generation of the measurements.

[0136] Solution B8. The system of solution B7, wherein the first number of mmWave antennas is configured as a uniform linear array, and wherein the plurality of digital measurements is representative of an output of a virtual uniform linear array (ULA) that is based on the first number of mmWave antennas and the second number of analog combiners.

[0137] Solution B9. The system of solution B8, wherein the first number is an integer N, wherein a size of an aperture of the virtual ULA is NS, wherein a length of the spatial filter is an integer M equal to N – NS + 1, and wherein M is a positive integer.

[0138] Solution B10. The system of solution B2, wherein the channel is coherent over a transmission and a reception of the signal by the first number of mmWave antennas.

[0139] Solution B11. The system of solution B2, wherein the second number is equal to one, wherein the angular range is divided into P non-overlapping partitions, wherein P is an integer, and wherein the one or more processors is configured, as part of performing the beam alignment operation, to: implement the spatial filter as P spatial filters such that each of the P spatial filters is associated with a corresponding partition of the P non-overlapping partitions.

[0140] Solution B12. The system of solution B2, wherein the second number is equal to one, and wherein the one or more processors is configured, as part of performing the channel estimation operation, to: determine, using a maximum likelihood estimation (MLE) procedure based on a matching objective, the direction of arrival and the complex path coefficient.

[0141] Solution B13. The system of solution B2, wherein the signal is associated with a steady-state response and a transient response, and wherein processing the signal and the corresponding spatial filter produces an output in which the steady-state response is preservedInternational Patent Application Attorney Docket No.: 009062.8503.WO00, SD2024-084-2PCT upon processing by a corresponding RF chain associated with each of the second number of analog combiners.

[0142] Solution B14. The system of solution B13, wherein generating the plurality of digital measurements exploits spatial degrees of freedom by preserving the steady-state response associated with the signal being received at each of the first number of mmWave antennas.

[0143] Solution B15. The system of solution B14, wherein generating the plurality of digital measurements exploits temporal degrees of freedom by assuming a known value for the complex path coefficient.

[0144] Solution B16. The system of solution B13, wherein processing the signal and the corresponding spatial filter comprises performing a convolution operation.

[0145] Solution B17. The system of solution B1, wherein, when performing the multi- user detection operation, the signal comprises multiple independent data signals from multiple users being multiplexed together.

[0146] Solution B18. The system of any of solutions B1 to B17, wherein each digital measurement of the plurality of digital measurements preserves a phase information of the signal received at a corresponding mmWave antenna.

[0147] Solution B19. The system of any of solutions B1 to B18, wherein the plurality of digital measurements comprises discrete-time complex exponential samples that correspond to the signal.

[0148] Solution B20. The system of any of solutions B1 to B19, wherein a spacing between adjacent mmWave antennas of the first number of mmWave antennas is equal to half a wavelength of the signal.

[0149] Solution B21. The system of solution B20, wherein the first number of mmWave antennas are arranged in a linear array, a planar array, or a three-dimensional array.

[0150] Solution B22. A method for wireless millimeter wave (mmWave) communication, comprising: configuring a first number of mmWave antennas to receive a signal associated with a direction of arrival that is within an angular range, wherein the first number of mmWave antennas is coupled to a second number of analog combiners, wherein each analog combiner is associated with a radio frequency (RF) chain and a spatial filter that spans the angular range, and wherein the second number is less than the first number; generating, based on processing the signal and a corresponding spatial filter associated with each of the second number of analog combiners, a plurality of digital measurementsInternational Patent Application Attorney Docket No.: 009062.8503.WO00, SD2024-084-2PCT corresponding to the signal, wherein a number of the plurality of digital measurements is greater than each of the first number and the second number; and performing, using the plurality of digital measurements, at least one of a beam alignment operation, a channel estimation operation, or a multi-user detection operation.

[0151] Solution B23. The method of solution B22, wherein the signal comprises a narrowband signal received over a channel comprising a single dominant path with a complex path coefficient.

[0152] Solution B24. The method of solution B23, wherein performing the beam alignment operation comprises: determining, based on the plurality of digital measurements, the direction of arrival; and determining, based on the direction of arrival, updated filter coefficients for the spatial filter for a subsequent transmission or reception using the first number of mmWave antennas.

[0153] Solution B25. The method of solution B23, wherein the second number is equal to one, and wherein generating the plurality of digital measurements exploits temporal degrees of freedom by assuming a known value for the complex path coefficient.

[0154] Solution B26. The method of solution B23, wherein the second number is equal to one, and wherein performing the channel estimation operation comprises: determining, using a maximum likelihood estimation (MLE) procedure based on a matching objective, the direction of arrival and the complex path coefficient.

[0155] Solution B27. The method of solution B23, wherein the signal is associated with a steady-state response and a transient response, and wherein processing the signal and the corresponding spatial filter produces an output in which the steady-state response is preserved upon processing by a corresponding RF chain associated with each of the second number of analog combiners.

[0156] Solution B28. The method of solution B22, wherein, when performing the multi- user detection operation, the signal comprises multiple independent data signals from multiple users being multiplexed together.

[0157] Solution B29. The method of any of solutions B22 to B28, wherein each digital measurement of the plurality of digital measurements preserves a phase information of the signal received at a corresponding mmWave antenna.

[0158] Solution B30. The method of any of solutions B22 to B29, wherein the plurality of digital measurements comprises discrete-time complex exponential samples that correspond to the signal.International Patent Application Attorney Docket No.: 009062.8503.WO00, SD2024-084-2PCT

[0159] Solution B31. An data processing apparatus for wireless communication comprising one or more processors configured to implement the method of any of solutions B22 to B30.

[0160] Solution B32. A computer-readable medium comprising instructions which, when executed by a computer, cause the computer to carry out the method of any of solutions B22 to B30.

[0161] FIG.8 shows an example of a hardware platform 800 that can be used to implement some of the techniques described in the present patent document. For example, the hardware platform 800 may implement the various modules and algorithms described herein. The hardware platform 800 may include a processor 802 that can execute code to implement a method. The hardware platform 800 may include a memory 804 that may be used to store processor-executable code and / or store data. The hardware platform 800 may further include an analog combiner 806 and a spatial filter design module 808, which can communicate with the processor 802. In some embodiments, the processor 802 may include one or more processors implementing at least a portion of the analog combiner 806 and the spatial filter design module 808. The processor 802 may be configured to implement convolution and other receiver processing algorithms. In some embodiments, the memory 804 may include multiple memories, some of which are exclusively used by the processor 802 when implementing the convolution and other receiver processing algorithms.

[0162] Implementations of the subject matter and the functional operations described in this patent document can be implemented in various systems, digital electronic circuitry, or in computer software, firmware, or hardware, including the structures disclosed in this specification and their structural equivalents, or in combinations of one or more of them. Implementations of the subject matter described in this specification can be implemented as one or more computer program products, i.e., one or more modules of computer program instructions encoded on a tangible and non-transitory computer readable medium for execution by, or to control the operation of, data processing apparatus. The computer readable medium can be a machine-readable storage device, a machine-readable storage substrate, a memory device, a composition of matter effecting a machine-readable propagated signal, or a combination of one or more of them. The term “data processing unit” or “data processing apparatus” encompasses all apparatus, devices, and machines for processing data, including by way of example a programmable processor, a computer, or multiple processors or computers. The apparatus can include, in addition to hardware, code that creates anInternational Patent Application Attorney Docket No.: 009062.8503.WO00, SD2024-084-2PCT execution environment for the computer program in question, e.g., code that constitutes processor firmware, a protocol stack, a database management system, an operating system, or a combination of one or more of them.

[0163] A computer program (also known as a program, software, software application, script, or code) can be written in any form of programming language, including compiled or interpreted languages, and it can be deployed in any form, including as a stand-alone program or as a module, component, subroutine, or other unit suitable for use in a computing environment. A computer program does not necessarily correspond to a file in a file system. A program can be stored in a portion of a file that holds other programs or data (e.g., one or more scripts stored in a markup language document), in a single file dedicated to the program in question, or in multiple coordinated files (e.g., files that store one or more modules, sub programs, or portions of code). A computer program can be deployed to be executed on one computer or on multiple computers that are located at one site or distributed across multiple sites and interconnected by a communication network.

[0164] The processes and logic flows described in this specification can be performed by one or more programmable processors executing one or more computer programs to perform functions by operating on input data and generating output. The processes and logic flows can also be performed by, and devices can be implemented as, special purpose logic circuitry, e.g., FPGA (field programmable gate array) or ASIC (application specific integrated circuit).

[0165] Processors suitable for the execution of a computer program include, by way of example, both general and special purpose microprocessors, and any one or more processors of any kind of digital computer. Generally, a processor will receive instructions and data from a read-only memory or a random-access memory or both. The essential elements of a computer are a processor for performing instructions and one or more memory devices for storing instructions and data. Generally, a computer will also include, or be operatively coupled to receive data from or transfer data to, or both, one or more mass storage devices for storing data, e.g., magnetic, magneto-optical disks, or optical disks. However, these are optional. Computer readable media suitable for storing computer program instructions and data include all forms of nonvolatile memory, media and memory devices, including by way of example semiconductor memory devices, e.g., EPROM, EEPROM, and flash memory devices. The processor and the memory can be supplemented by, or incorporated in, special purpose logic circuitry.

[0166] While this patent document contains many specifics, these should not beInternational Patent Application Attorney Docket No.: 009062.8503.WO00, SD2024-084-2PCT construed as limitations on the scope of any invention or of what may be claimed, but rather as descriptions of features that may be specific to particular embodiments of particular inventions. Certain features that are described in this patent document in the context of separate embodiments can also be implemented in combination in a single embodiment. Conversely, various features that are described in the context of a single embodiment can also be implemented in multiple embodiments separately or in any suitable subcombination. Moreover, although features may be described above as acting in certain combinations and even initially claimed as such, one or more features from a claimed combination can in some cases be excised from the combination, and the claimed combination may be directed to a subcombination or variation of a subcombination.

[0167] Similarly, while operations are depicted in the drawings in a particular order, this should not be understood as requiring that such operations be performed in the particular order shown or in sequential order, or that all illustrated operations be performed, to achieve desirable results. Moreover, the separation of various system components in the embodiments described in this patent document should not be understood as requiring such separation in all embodiments.

[0168] Only a few implementations and examples are described and other implementations, enhancements and variations can be made based on what is described and illustrated in this patent document.

Claims

International Patent Application Attorney Docket No.: 009062.8503.WO00, SD2024-084-2PCT WHAT IS CLAIMED IS:

1. A system for wireless millimeter wave (mmWave) communication, comprising: a first number of mmWave antennas configured to receive a signal associated with a direction of arrival that is within an angular range; a second number of analog combiners coupled to the first number of mmWave antennas, wherein each analog combiner is associated with a radio frequency (RF) chain and a spatial filter that spans the angular range, and wherein the second number is less than the first number; and one or more processors configured to: generate, based on processing the signal and a corresponding spatial filter associated with each of the second number of analog combiners, a plurality of digital measurements corresponding to the signal, wherein a number of the plurality of digital measurements is greater than each of the first number and the second number, and perform, using the plurality of digital measurements, at least one of a beam alignment operation, a channel estimation operation, or a multi-user detection operation.

2. The system of claim 1, wherein the signal comprises a narrowband signal received over a channel comprising a single dominant path with a complex path coefficient.

3. The system of claim 2, wherein the one or more processors is configured, as part of performing the beam alignment operation, to: determine, based on the plurality of digital measurements, the direction of arrival; and determine, based on the direction of arrival, updated filter coefficients for the spatial filter for a subsequent transmission or reception using the first number of mmWave antennas.

4. The system of claim 3, wherein determining the updated filter coefficients is based on using a Park-McClellan algorithm.

5. The system of claim 3 or 4, wherein determining the updated filter coefficients excludes relying on prior knowledge associated with the complex path coefficient.

6. The system of claim 3 or 4, wherein determining the updated filter coefficients assumes a known value for the complex path coefficient.International Patent Application Attorney Docket No.: 009062.8503.WO00, SD2024-084-2PCT 7. The system of claim 2, wherein the second number is equal to one, and wherein generating the plurality of digital measurements exploits temporal degrees of freedom by assuming a known value for the complex path coefficient.

8. The system of claim 7, wherein the first number of mmWave antennas is configured as a uniform linear array, and wherein the plurality of digital measurements is representative of an output of a virtual uniform linear array (ULA) that is based on the first number of mmWave antennas and the second number of analog combiners.

9. The system of claim 8, wherein the first number is an integer N, wherein a size of an aperture of the virtual ULA is NS, wherein a length of the spatial filter is an integer M equal to N – NS+ 1, and wherein M is a positive integer.

10. The system of claim 2, wherein the channel is coherent over a transmission and a reception of the signal by the first number of mmWave antennas.

11. The system of claim 2, wherein the second number is equal to one, wherein the angular range is divided into P non-overlapping partitions, wherein P is an integer, and wherein the one or more processors is configured, as part of performing the beam alignment operation, to: implement the spatial filter as P spatial filters such that each of the P spatial filters is associated with a corresponding partition of the P non-overlapping partitions.

12. The system of claim 2, wherein the second number is equal to one, and wherein the one or more processors is configured, as part of performing the channel estimation operation, to: determine, using a maximum likelihood estimation (MLE) procedure based on a matching objective, the direction of arrival and the complex path coefficient.

13. The system of claim 2, wherein the signal is associated with a steady-state response and a transient response, and wherein processing the signal and the corresponding spatial filter produces an output in which the steady-state response is preserved upon processing by a corresponding RF chain associated with each of the second number of analog combiners.International Patent Application Attorney Docket No.: 009062.8503.WO00, SD2024-084-2PCT 14. The system of claim 13, wherein generating the plurality of digital measurements exploits spatial degrees of freedom by preserving the steady-state response associated with the signal being received at each of the first number of mmWave antennas.

15. The system of claim 14, wherein generating the plurality of digital measurements exploits temporal degrees of freedom by assuming a known value for the complex path coefficient.

16. The system of claim 13, wherein processing the signal and the corresponding spatial filter comprises performing a convolution operation.

17. The system of claim 1, wherein, when performing the multi-user detection operation, the signal comprises multiple independent data signals from multiple users being multiplexed together.

18. The system of claim 1, wherein each digital measurement of the plurality of digital measurements preserves a phase information of the signal received at a corresponding mmWave antenna.

19. The system of claim 1, wherein the plurality of digital measurements comprises discrete-time complex exponential samples that correspond to the signal.

20. The system of claim 1, wherein a spacing between adjacent mmWave antennas of the first number of mmWave antennas is equal to half a wavelength of the signal.

21. The system of claim 20, wherein the first number of mmWave antennas are arranged in a linear array, a planar array, or a three-dimensional array.

22. A method for wireless millimeter wave (mmWave) communication, comprising: configuring a first number of mmWave antennas to receive a signal associated with a direction of arrival that is within an angular range, wherein the first number of mmWave antennas is coupled to a second number of analog combiners, wherein each analog combiner is associated with a radio frequency (RF) chain and a spatial filter that spans the angular range, and wherein the second number is less than the first number;International Patent Application Attorney Docket No.: 009062.8503.WO00, SD2024-084-2PCT generating, based on processing the signal and a corresponding spatial filter associated with each of the second number of analog combiners, a plurality of digital measurements corresponding to the signal, wherein a number of the plurality of digital measurements is greater than each of the first number and the second number; and performing, using the plurality of digital measurements, at least one of a beam alignment operation, a channel estimation operation, or a multi-user detection operation.

23. The method of claim 22, wherein the signal comprises a narrowband signal received over a channel comprising a single dominant path with a complex path coefficient.

24. The method of claim 23, wherein performing the beam alignment operation comprises: determining, based on the plurality of digital measurements, the direction of arrival; and determining, based on the direction of arrival, updated filter coefficients for the spatial filter for a subsequent transmission or reception using the first number of mmWave antennas.

25. The method of claim 23, wherein the second number is equal to one, and wherein generating the plurality of digital measurements exploits temporal degrees of freedom by assuming a known value for the complex path coefficient.

26. The method of claim 23, wherein the second number is equal to one, and wherein performing the channel estimation operation comprises: determining, using a maximum likelihood estimation (MLE) procedure based on a matching objective, the direction of arrival and the complex path coefficient.

27. The method of claim 23, wherein the signal is associated with a steady-state response and a transient response, and wherein processing the signal and the corresponding spatial filter produces an output in which the steady-state response is preserved upon processing by a corresponding RF chain associated with each of the second number of analog combiners.

28. The method of claim 22, wherein, when performing the multi-user detection operation, the signal comprises multiple independent data signals from multiple users being multiplexed together.International Patent Application Attorney Docket No.: 009062.8503.WO00, SD2024-084-2PCT 29. The method of claim 22, wherein each digital measurement of the plurality of digital measurements preserves a phase information of the signal received at a corresponding mmWave antenna.

30. The method of claim 22, wherein the plurality of digital measurements comprises discrete-time complex exponential samples that correspond to the signal.

31. An data processing apparatus for wireless communication comprising one or more processors configured to implement the method of any of claims 22 to 30.

32. A computer-readable medium comprising instructions which, when executed by a computer, cause the computer to carry out the method of any of claims 22 to 30.

Images