RECONFIGURABLE FULLY-CONNECTED BIDIRECTIONAL HYBRID BEAMFORMING TRANSCEIVER

Abstract

Disclosed herein is a new type of fully-connected, hybrid beamforming transceiver architecture. The transceiver described herein is bi-directional and can be configured as a transmit beamformer or a receive beamformer. A method and apparatus are described that allows the beamformer to operate in “carrier aggregated” mode, where communication channels in multiple disparate frequency bands can be simultaneously accessed.

Claims

1. A bi-directional beamforming transceiver comprising: a transmitter portion; and a receiver portion; wherein the transceiver is operated in carrier aggregation mode allowing independent data streams to be received or transmitted at multiple frequencies using a common antenna array.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0012] FIG. 1 shows, in View (a), a block diagram of a partially-connected beamforming receiver, and, in View (b), a block diagram of a fully-connected beamforming receiver.

[0013] FIG. 2 shows, in View (a), a block diagram of a partially-connected beamforming transmitter, and, in View (b), a block diagram of a first embodiment of the invention showing a fully-connected beamforming transmitter.

[0014] FIG. 3 shows a block diagram of a fully-connected beamforming transmitter using Cartesian weighting and splitting, representing a third embodiment of the invention.

[0015] FIG. 4 shows a block diagram of a fully-connected, hybrid beamforming transceiver configured to pass signals in both directions using the transmitter shown in FIG. 3.

[0016] FIG. 5 shows a block diagram of a fully-connected bidirectional hybrid beamforming transceiver using Cartesian weighting and splitting/combining heterodyne conversion.

[0017] FIG. 6 shows the spectra of waveforms in the signal chain in the receive, in View (a), and transmit, in View (b), paths of the bidirectional fully-connected hybrid beamformer.

[0018] FIG. 7 shows a block diagram of method and an apparatus for calibration and gain/phase mismatch detection and mitigation.

[0019] FIG. 8 shows block diagrams of different configurations of a bidirectional fully-connected hybrid beamformer.

DETAILED DESCRIPTION OF THE INVENTION

[0020] Novel aspects of the invention include disclosure of a FC-HBF transmitter, the combination of the FC-HBF transmitter with a FC-HBF receiver to form a FC-HBF transceiver, carrier aggregation in FC-HBF, and simultaneous transmit-receive communication with the FC-HBF transceiver. These aspects of the invention are discussed below.

[0021] FC-HBF Transmitter

[0022] A first embodiment of the FC-HBF transmitter described herein is illustrated in View (b) of FIG. 2. A PC-HBF transmitter is shown in View (a) of FIG. 2 for comparison. In the FC-HBF transmitter, independent baseband data streams are upconverted using a complex-quadrature mixer. Outputs y.sub.1 and y.sub.2 are then split into N identical paths. Complex-valued weights are applied to each upconverted streams in polar form using a combination of a programmable phase shifter and a programmable gain amplifier. In the FC-HBF transmitter, weighted signals from the individual streams are combined and then input to power amplifiers (PA). Each PA drives a single antenna element with the requisite amount of power. Implementation of the phase-shifters poses several difficulties including large size, high insertion loss, relatively low port impedances, limited bandwidth and non-linear control characteristics.

[0023] A second embodiment of the FC-HBF transmitter uses a similar architecture to the first embodiment but realizes the complex-valued weights using a vector modulator, as shown in the inset of FIG. 2. A vector modulator is a combination of a quadrature hybrid (QH) and a pair of programmable-gain active amplifiers or programmable-loss passive attenuators. While this approach overcomes some of the aforementioned shortcomings associated with the phase-shifter approach, it still suffers from large size, high loss, low-port impedances and limited bandwidth.

[0024] A third embodiment, which overcomes all the aforementioned challenges, is illustrated in FIG. 3. Here, each baseband stream is upconverted using a complex-quadrature mixer (shown schematically in the inset of FIG. 3) to the RF carrier frequency and split into N paths. Per-stream and per-antenna complex-valued weights (P.sub.s,k's) are applied using a pair (P.sub.k's) of programmable-gain amplifiers. Weighted signals from all upconverted streams are then combined and input to PA's which drive the antenna elements.

[0025] The operation of the third embodiment can be described mathematically as follows. We represent the s.sup.th baseband stream by its baseband envelope {tilde over (x)}.sub.s(t)=(x.sub.s,l+jx.sub.s,Q). The quadrature upconverted signal u(t) in the s.sup.th stream can be written as:

[00001]$\begin{array}{cc}{u}_s\left(t\right)=u_{s,I}+j{u}_{s,Q}={\tilde{x}}_s\left(t\right)e^{j{\omega }_{R{F}^t}}=\left(x_{s,I}+j{x}_{s,Q}\right)e^{j{\omega }_{R{F}^t}}=\left(x_{s,I}C-x_{s,Q}S\right)+j\left(x_{s,I}S-x_{s,Q}C\right)& \left(1\right)\end{array}$

[0026] For the s.sup.th stream and the k.sup.th antenna, the objective is to apply a programmable complex-valued weight P.sub.s,k=P.sub.Re-s,k+jP.sub.Im-s,k to the envelope of the signal (u.sub.s,l+ju.sub.s,Q), i.e.:

[00002]$\begin{array}{cc}{v}_{s,k}\left(t\right)=\mathrm{Re}\left[P_{s,k}.Math.{\tilde{x}}_s\left(t\right).Math.e^{j{\omega }_{R{F}^t}}\right]=\mathrm{Re}\left[\left(P_{\mathrm{Re}-s,k}+j{P}_{\mathrm{Im}-s,k}\right)\left(x_{s,I}+j{x}_{s,Q}\right)e^{j{\omega }_{R{F}^t}}\right]=P_{\mathrm{Re}-s,k}\left(x_{s,I}C-x_{s,Q}S\right)-P_{\mathrm{Im}-s,k}\left(x_{s,I}S-x_{s,Q}C\right)& \left(2\right)\end{array}$

[0027] Note that the last line of Eq (2) comprises only real-valued terms, which can be implemented using programmable gains P.sub.Re-k,s and jP.sub.Im-k,s. Finally, the weighted, upconverted signals from all streams are combined to produce the signal that drives the k.sup.th antenna:

S.sub.k(t)=Σ.sub.s=1.sup.Sv.sub.s,k(t)=Σ.sub.s=1.sup.SRe[P.sub.s,k.Math.{tilde over (x)}.sub.s(t).Math.e.sup.jω.sup.RF.sup.t]  (3)

[0028] Thus, it is seen that the third embodiment is a schematic representation of the above signal processing.

[0029] Bidirectional Transmit-Receive

[0030] The FC-HBF transmitter described above can be combined with the FC-HBF receiver architecture described in U.S. patent application Ser. No. 16/163,374 to realize a bi-directional FC-HBF transceiver. A schematic illustration of a transceiver based on a second embodiment of the invention as described above is shown in FIG. 4. Here, the programable-gain amplifiers used to realize the complex-valued weights are designed such that they can be configured to pass signals either in the forward direction only or in the reverse direction only.

[0031] Carrier Aggregation

[0032] Due to the availability of multiple downconversion chains, HBF's inherently able to support carrier aggregation (where independent data can be received at multiple frequencies), thereby increasing data rate. However, in a PC-HBF, a separate sub-array is required for each aggregated carrier. On the other hand, in a FC-HBF, the same antenna array can be used to transmit or receive several aggregated carriers. Specifically, an FC-HBF can support aggregation of as many carriers as the number of available frequency translation chains. This is a significant advantage of the FC-HBF. In addition, the FC-HBF achieves higher beamforming gain for each carrier-aggregated signal or stream.

[0033] In the FC-HBF described herein, direct conversion was assumed, which means that the baseband streams are translated to the RF carrier in a single step. In this architecture, a dedicated local oscillator (LO) generation circuit per stream is required for carrier aggregation. FC-HBF using a heterodyne architecture can be advantageous for carrier aggregation. By performing per-antenna-per-stream complex weighting using multiple frequency translation steps, functionality similar to a direct conversion FC-HBF can be obtained. An example of a heterodyne architecture using a single intermediate frequency (IF) is shown in FIG. 5. Here, complex-quadrature mixers are used for both frequency translation steps.

[0034] The operation of the heterodyne FC-HBF in carrier aggregation mode will now be described. Assume that the local oscillator is tuned to a frequency ω.sub.LO such that a high-band RF frequency ω.sub.RFA and a low-band RF frequency ω.sub.RFB can be accessed by frequency translation to/from baseband through the intermediate frequency ω.sub.IF. Note that the two RF bands are mutual images at this LO frequency. The spectra at various points in the signal chain are shown FIG. 6 using an example with two streams. The front-end beamforming weights are configured such that the array pattern in each stream is steered towards the direction of departure (or arrival) of one of the signals. The frequency translation chain in each stream is configured to translate one of the two signals to baseband while rejecting the other.

[0035] In the receive path, in each stream, the I/Q outputs of the first complex-quadrature stage comprise the two signals downconverted to the same IF, as shown by waveforms 2A and 2B in View (a) of FIG. 6, with one of the signals being attenuated by the beamforming array pattern programmed for that stream. Then, in each stream, the second complex-quadrature stage is configured to downconvert to baseband either the low-band or the high-band signal while rejecting (cancelling) the other signal, as shown by waveforms 3A and 3B in View (a) of FIG. 6.

[0036] In the transmit path, with reference to View (b) of FIG. 6, independent baseband signals are first upconverted to ω.sub.IF in each stream. The first complex-quadrature stage is configured for upconversion from ω.sub.IF to either the high band (ω.sub.RFA) or the low band (ω.sub.RFB). Following this upconversion, independent beamforming weights for each stream are applied to steer the beams towards the desired directions of departure.

[0037] Digital Calibration to Enhance Image-Rejection

[0038] In both the transmit and the receive modes, the gains and phases of the quadrature paths should be matched accurately. In practice, inevitable on-chip device and layout mismatches, which can be either random or systematic, cause mis-matches in the path gains and cause the quadrature phases to deviate from their nominal difference of 90°. Such mis-matches cause imperfect rejection of the image frequency signal which causes corruption of both the transmitted and the received signals.

[0039] Also disclosed herein is a method to calibrate such mismatches and mitigate their adverse effects. While the methods are applicable to any of the embodiments of the beamformer, it is particularly applicable to the second embodiment, shown in FIG. 3, the FC-HBF transmitter using Cartesian weighting and splitting.

[0040] In a single-antenna heterodyne receiver of the sliding-IF Weaver type, errors due to gain/phase mismatches in both mixing stages can be consolidated and corrected at baseband. However, in the case of the RF weighting HBF, the two mixing stages must be individually calibrated. The method and apparatus are shown in FIG. 7. The method comprises inserting a sinusoidal test signal at the input to the first complex quadrature stage. A quadrature error (QE) detector inserted at the output of this stage detects the deviation of the phase difference between these two signals from its nominal value of 90°. A robust embodiment of the QE detector that equalizes loading on the two inputs is shown in View (b) of FIG. 7. The output of the QE detector is lowpass filtered, digitized and then fed to a digital calibration which actuates a QE correction mechanism. Note that the low-resolution ADC, down to a single bit, can be used for digitization of the QE. Several mechanisms can be used for QE correction: (a) through the LO buffer, for example, via its resonant load, as shown in FIG. 7; (b) insertion of a Cartesian weighting circuit (i.e., a matrix rotation circuit), as shown in View (c) of FIG. 7; or (c) using statistical element selection within the complex quadrature mixer to tune mismatches using combinatorial redundancy.

[0041] Reconfiguration and Full-Duplex Beamforming

[0042] Multiple tiles of the transceiver described herein can be used to support simultaneous transmit and receive (STAR) operation, as shown in View (d) of FIG. 8. STAR operation in separate frequency bands for transmit and receive is called frequency division duplex (FDD), which STAR operation in the same frequency channel is called full-duplex (FD) operation. In both cases, the overarching challenge in full-duplex communication is the self-interference (SI) from the strong transmitted signal leaking into the receive path and causing severe corruption of the weak received signal. In the case of FDD, some attenuation of this leakage is achieved by using a diplexer which provides high attenuation of the transmit signal in the receive path. However, such a mechanism is not available in the FD case, and therefore, signal cancellation of the transmit signal leakage is the only viable option.

[0043] There are two variants of FD systems: shared-antenna FD, where each antenna element is shared between the transmit and receive paths, and separate-antenna FD, where transmit and receive paths use completely separate antenna arrays. While the shared-antenna approach has gained interest in sub-6 GHz FD communication, the separate-antenna is advantageous at mm-wave for the following reasons: 1) At mm-wave, many antennas can be packed in a small form-factor. Therefore, separate antenna arrays can be used in transmit and receive paths; 2) The separate-antenna approach avoids the use of a circulator. Recent innovations have made on-chip integration of circulators possible, but they are lossy, have limited linearity and bandwidth, and achieve inadequate transmit-receive isolation. Furthermore, they occupy a large die area and are difficult to integrate into beamformers with large numbers of elements in a cost-effective manner; 3) More importantly, due to small antenna spacing at mm-wave, adjacent antennas experience significant coupling, and hence, suffer from severe SI from one antenna to the nearby antennas in a shared-antenna approach. However, in a separate-antenna approach, SI due to antenna coupling can be greatly reduced by increasing the physical spacing between the transmit and receive antenna arrays.

[0044] In the reconfigurable architecture described herein and shown in View (a) of FIG. 8, seamless reconfiguration can be achieved between several modes including a fully connected receiver, shown in View (b) of FIG. 8, a fully-connected transmitter, shown in View (c) of FIG. 8, and a partially connected STAR mode with built-in SI-cancellation. To this end, the built-in SI-canceler can cancel two types of self-interferences: (1) SI due to a nearby reflection of the transmitted signal that leaks into the receiver through the receive antenna array; and (2) SI through antenna coupling from the transmit to the receive antennas.

[0045] The first kind of SI is canceled by directing a null towards the reflection paths in both the transmit and receive beamformer. Note that nulls in the transmit and receive array pattern can be steered towards different leakage multipath components or can be steered towards the same path to achieve higher rejection.

[0046] The second kind of SI is canceled by using independent per-element single-tap RF-domain SI canceler at the receiver (inside the LNA) by using a copy of the transmitted signal in each path of the receive array, as shown in View (d) of FIG. 8. It is important to note that, in the architecture described herein, no extra hardware is necessary to perform this per element SI-cancellation technique. As shown in View (d) of FIG. 8, in the bidirectional FC-HBF, one stream in the receive array can be repurposed to transmit a copy of the transmit signal that can be independently complex weighted in each element to cancel the incoming transmit signal leakage. Moreover, this technique can also be used to cancel leakage through nearby reflections that have low group delay. Note that this SI cancellation mechanism is not available in the PC-HBF.