Zero division duplexing mimo radio with adaptable RF and/or baseband cancellation
11343060 · 2022-05-24
Assignee
Inventors
Cpc classification
H01Q21/28
ELECTRICITY
H04B1/525
ELECTRICITY
H04L5/1461
ELECTRICITY
H04B1/38
ELECTRICITY
H01Q21/24
ELECTRICITY
H04B1/54
ELECTRICITY
International classification
H04L5/14
ELECTRICITY
H01Q21/06
ELECTRICITY
H01Q21/28
ELECTRICITY
H04B1/525
ELECTRICITY
H01Q21/24
ELECTRICITY
H04B1/38
ELECTRICITY
H04B1/54
ELECTRICITY
Abstract
An intelligent backhaul radio is disclosed, which can operate by zero division duplexing for use in PTP or PMP topologies, providing for significant spectrum usage benefits among other benefits. Specific system architectures and structures to enable active cancellation of multiple transmit signals at multiple receivers within a MIMO radio are disclosed. Further disclosed aspects include the adaptive optimization of cancellation parameters or coefficients.
Claims
1. A radio comprising: a plurality of transmit radio frequency (RF) chains, wherein each transmit RF chain is configured to convert from a respective one of a plurality of transmit chain input signals to a respective one of a plurality of transmit RF signals; a plurality of adaptable RF transversal filter sets configured to convert one or more RF cancellation input signals derived from one or more of the plurality of transmit RF signals to a plurality of RF transmit cancellation signals, wherein each one of the plurality of adaptable RF transversal filter sets comprises: one or more adaptable RF transversal filters, wherein each adaptable RF transversal filter is configured to filter a respective one of the one or more RF cancellation input signals to provide a respective one of one or more adaptable RF transversal filtered signals; and an RF filtered signal combiner configured to combine the one or more adaptable RF transversal filtered signals within each one of the plurality of adaptable RF transversal filter sets to produce one of the plurality of RF transmit cancellation signals; a plurality of antenna elements, wherein each of a plurality of receive RF signals is derived from signals received from at least one of the plurality of antenna elements, wherein each of the plurality of receive RF signals comprises at least one or more transmitter related signals, and wherein each of the one or more transmitter related signals is derived from at least one of the plurality of transmit RF signals; a plurality of RF cancellation combiners, wherein each RF cancellation combiner is configured to combine a respective one of at least one of the plurality of RF transmit cancellation signals with at least one of the plurality of RF receive signals to provide a respective one of a plurality of receive chain input signals; and a plurality of receive radio frequency (RF) chains, wherein each receive RF chain is configured to convert from a respective one of the plurality of receive chain input signals to a respective one of a plurality of receive chain output signals; wherein the radio is configured to perform adaptable radio frequency (RF) cancellation for reduction of one or more transmit signals at a receiver within the radio.
2. The radio of claim 1, further comprising: one or more demodulator cores, wherein each demodulator core demodulates one or more receive symbol streams to produce a respective receive data interface stream; and a frequency selective receive path channel multiplexer to produce at least two receive symbol streams from at least two of the plurality of receive chain output signals.
3. The radio of claim 1, further comprising: one or more modulator cores, wherein each modulator core modulates a respective transmit data interface stream to produce one or more of a plurality of transmit symbol streams; and a transmit path channel multiplexer to produce the plurality of transmit chain input signals provided to the plurality of transmit RF chains from the plurality of transmit symbol streams.
4. The radio of claim 3, wherein the transmit path channel multiplexer is a frequency selective transmit path channel multiplexer.
5. The radio of claim 3, wherein the transmit path channel multiplexer utilizes transmit beam forming.
6. The radio of claim 1, wherein the radio is configured to insert transmit preambles that can be utilized at least to obtain an estimate of the level of the interfering signal component.
7. The radio of claim 1, wherein the radio is configured to estimate of the level of the interfering signal component by using at least a Fast Fourier Transform (FFT).
8. The radio of claim 1, wherein the radio is configured to determine the metric from at least one of a least squares or an MMSE calculation.
9. The radio of claim 1, wherein the metric comprises one or more of an RF transmit leakage metric, a transmit leakage metric, an RSSI metric, a metric involving a correlation with a reference signal, a metric involving a correlation with a transmit chain input signal, a metric derived from a receive RF chain, a metric associated with a desired receive signal, a metric associated with isolation between certain receive antenna array elements and certain transmit antenna array elements associated with the intelligent radio, a metric based upon measurements of signals, channel estimates, or cancellation performance, a receiver performance metric such as signal to noise ratio (SNR), bit error rate (BER), or frame error rate (FER), or a metric associated with the forward error correction (FEC) decoder.
10. The radio of claim 1, further comprising: one or more selectable RF connections, wherein each selectable RF connection is configured to selectively couple to at least one of certain of the plurality of antenna elements to derive at least one of certain of the plurality of receive RF signals.
11. The radio of claim 10, wherein at least one of the one or more selectable RF connections comprises circuitry to combine signals received from at least two of the certain of the plurality of antenna elements.
12. The radio of claim 11, wherein the circuitry to combine signals comprises at least an RF switch.
13. The radio of claim 1, wherein no channel bandwidth associated with any of the plurality of transmit RF signals at a time of transmission overlaps with any channel bandwidth associated with any of the plurality of receive RF signals at a time of reception, wherein the time of transmission and the time of reception can be coincident.
14. The radio of claim 1, wherein at least one channel bandwidth associated with any of the plurality of transmit RF signals at a time of transmission overlaps at least partially with at least one channel bandwidth associated with any of the plurality of receive RF signals at a time of reception, wherein the time of transmission and the time of reception can be coincident.
15. A radio comprising: a plurality of antenna elements; a plurality of transmit radio frequency (RF) chains coupled to at least one of the plurality of antenna elements; a plurality of receive RF chains coupled to at least one of the plurality of antenna elements; a plurality of RF cancellation combiners coupled to the plurality of receive RF chains; a plurality of adaptable RF transversal filter sets coupled to the plurality of RF cancellation combiners and coupled to the plurality of transmit RF chains; one or more modulator cores; and a transmit path channel multiplexer coupled to the plurality of transmit RF chains and the one or more modulator cores wherein the radio is configured to perform adaptable RF cancellation for reduction of one or more transmit signals at a receiver within the radio.
16. The radio of claim 15, further comprising: one or more selectable RF connections, wherein each selectable RF connection is configured to selectively couple to at least one of certain of the plurality of antenna elements.
17. The radio of claim 15, wherein each of the plurality of adaptable RF transversal filter sets comprise: one or more adaptable RF transversal filters; and an RF filtered signal combiner.
18. The radio of claim 15, wherein the radio is configured to perform adaptable RF cancellation for reduction of one or more transmit signals at a receiver within the radio by adapting at least one of the one or more adaptable RF transversal filters to reduce a level of an interfering signal component within a receive chain output signal, wherein the interfering signal component is derived from a transmitter related signal.
19. The radio of claim 18, wherein the radio utilizes a metric that accounts for at least the interfering signal component, wherein the metric comprises one or more of an RF transmit leakage metric, a transmit leakage metric, an RSSI metric, a metric involving a correlation with a reference signal, a metric involving a correlation with a transmit chain input signal, a metric derived from a receive RF chain, a metric associated with a desired receive signal, a metric associated with isolation between certain receive antenna array elements and certain transmit antenna array elements associated with the intelligent radio, a metric based upon measurements of signals, channel estimates, or cancellation performance, a receiver performance metric such as signal to noise ratio (SNR), bit error rate (BER), or frame error rate (FER), or a metric associated with the forward error correction (FEC) decoder.
Description
BRIEF DESCRIPTION OF THE DRAWINGS
(1) The accompanying drawings, which are incorporated into and constitute a part of this specification, illustrate one or more examples of embodiments and, together with the description of example embodiments, serve to explain the principles and implementations of the embodiments.
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
(11)
(12)
(13)
(14)
(15)
(16)
(17)
(18)
(19)
(20)
(21)
(22)
(23)
(24)
(25)
(26)
DETAILED DESCRIPTION
(27)
(28) For 3G and especially for 4.sup.th Generation (4G), cellular network infrastructure is more commonly deployed using “microcells” or “picocells.” In this cellular network infrastructure, compact base stations (eNodeBs) 516 are situated outdoors at street level. When such eNodeBs 516 are unable to connect locally to optical fiber or a copper wireline of sufficient data bandwidth, then a wireless connection to a fiber “point of presence” (POP) requires obstructed LOS capabilities, as described herein.
(29) For example, as shown in
(30)
(31) As described in greater detail in U.S. patent application Ser. No. 13/212,036, now U.S. Pat. No. 8,238,318, and Ser. No. 13/536,927 and incorporated herein, modem 624 of
(32) As described in greater detail in U.S. patent application Ser. No. 13/212,036, now U.S. Pat. No. 8,238,318, and Ser. No. 13/536,927 and incorporated herein, each of the M transmit chain input signals is converted to a transmit RF signal by respective ones of Tx1 . . . TxM 636 in
(33) In alternative embodiments, the M.sub.s RF Transmit Reference Signals (680) may be obtained directly from the M.sub.s RF Transmit Signals within Cancellation MUX 670, rather than from the Antenna Array 648. In various embodiments, the M.sub.s RF Transmit Reference Signals (680) may be utilized in analog or digital cancellation processing as will be described in further detail in relation to subsequent figures.
(34) The receive signal processing path in
(35)
(36) In an alternative embodiment, the IBR RF Switch fabric 812 may be bypassed for the transmission signals when the number of dedicated transmission antennas and associated Front-end Transmission Units (Q.sub.T) is equal to the number of transmit RF signals (e.g. Q.sub.T=M), resulting in directly coupling the transmit RF signals from respective Tx1 . . . TxM 636 to respective Front-end Transmission Units 809. In an associated embodiment, the IBR RF switch fabric 812 may be bypassed for the selection of the RF Transmit Reference Signals (680) coupled to the Cancellation MUX (670), by directly connecting the RF Transmit Reference Signals (1 . . . Q.sub.T) (805) directly to the RF Transmit Reference Signals (1 . . . M.sub.s) (680), when M=Q.sub.T. In an additional alternative embodiment, the IBR RF Switch fabric 812 may also be bypassed for the reception signals when the number of dedicated reception antennas and associated Front-end Reception Units (Q.sub.R) is equal to the number of receive RF signals (e.g. Q.sub.R=N), resulting in directly coupling the receive RF signals for respective Rx1 . . . RxN 640 to respective Front-end Reception Units 810. Alternatively, the IBR RF Switch fabric 812 may also comprise circuitry to combine signals from two or more Front-end Reception Units or to provide signals to two or more Front-end Transmission Units as described in greater detail in U.S. patent application Ser. No. 13/212,036, now U.S. Pat. No. 8,238,318, and Ser. No. 13/536,927 and incorporated herein.
(37) As shown in
(38)
(39) The transmit directive antenna elements depicted in
(40) The receive directive antenna elements depicted in
(41) Other directive antenna element types are also known to those of skill in the art of antenna design including certain types described in greater detail in U.S. patent application Ser. No. 13/536,927 and incorporated herein.
(42) Preliminary measurements of exemplary antenna arrays similar to those depicted in
(43)
(44) Note that within the current embodiment, associated with performing cancellation at various stages of the receiver chains, cancellation may be performed at analog baseband, intermediate frequency (IF), RF and/or digital baseband.
(45) In order to achieve the required performance of the IBR 1200, the signal to noise ratio of the Receive Streams (1 . . . L) must be sufficient so as to allow for acceptable demodulation error rate at IBR Modem (624). As discussed above, conventional radios utilize frequency duplexing or time duplexing to allow for sufficient isolation of the transmitter signals from the signal being received and demodulated. Associated with the exemplary embodiments of the ZDD enabled IBR 1200, isolation of the desired receive signals from transmitted signals is accomplished utilizing a combination of active cancellation and inherent isolation between the IBR Transmit Antenna Array (648A), and the IBR Receive Antenna Array (648B). The various isolating features and functions are referred to in the following discussion as Isolation Loops or Cancellation Loops. Embodiments of the Cancellation Loops generally include adaptation based on active measurements of signals, channel estimates, cancellation metrics, or other metrics. In the current embodiment depicted in
(46) The Isolation “Loop” L0 (1202) is mainly just indicative of the finite isolation between any two antennas (Tx to Rx, or Tx to Tx) which is a critical parameter for FDD and even more critical in ZDD. Some embodiments of L0 (1202) will not include adaptive adjustment or utilize active control, but may still be referred to as a Loop for the constancy of the terminology herein. Other embodiments of L0, are truly a “loop” to the extent that some feedback mechanism either moves a servo to an isolating structure or a tuning element that affects a different isolation transfer function between any two antennas or sets of antennas of interest for a particular operational mode. Such tuning of control may further include the optional antenna selection function of Rx Antenna/Channel Switch Matrix within 648B. The operation of the optional Rx Antenna/Channel Switch Matrix within 648B is equivalent to that of the receive portions of IBR RF Switch Fabric 812 of
(47) Design of the antenna array for minimal magnitude response by and/or capability of L0 tuning by feedback is a key enabler for simplifying the demands of L1, L2 and/or L3. Relative to L0 (1202), it is generally expected, and experimentally confirmed in specific embodiments allowing for testing, IBR antenna elements and orientations in a “vertical” array stack as shown in
(48) The Canceller Loop L1 (1204) samples the “actual” transmitted RF signal in each Tx chain (RF Transmit Reference Signals 1 to M.sub.s (680) labeled 680-1 to 680-M), and then generates for each Rx chain a modified “cancellation signal C1.sub.n” (also referred to in some embodiments as a RF transmit leakage cancellation signal) that when summed (via RF summers 1210-1 to 1210-N) with the Rx chain signals (678) before input to the downconverters (640) substantially cancels the Tx signals 1 through M (X.sub.1(t) to X.sub.M(t)) that have leaked into the Rx antennas contributing to each Rx chain “n”. Exemplary embodiments of L1 (1204) are realized by an analog equivalent to a complex FIR “filter” implemented at RF and described in relation to
(49) For some embodiments of ZDD-IBRs, L1 may be targeted at only the largest Tx to Rx coupling paths, and those that are fixed for a specific IBR RF switch fabric (812) selection. Such coupling paths are expected to involve timescales with variations typically of order 1 ns or less for example in some embodiments. In specific embodiments it is possible to determine the appropriate loop coefficients once (at a factory calibration in some embodiments) and then refine only very occasionally. Such an approach may be utilized in other embodiments where longer time delays with mode variation are addressed as well, as a combination of a fixed or slowly adapting L1, and a parallel of sequential L1 cancellers which address the longer delay, lower magnitude, and more highly varying coupling paths. Such embodiments may include a “primary” loop to address the most significant magnitude response components and once cancelled by the primary loop (typically including L1 but possibly using part of L2 or L3 instead or in addition to L1), this should allow a secondary loop (probably within L2 or L3) to track shorter variations of much smaller magnitude. Details of exemplary L1 embodiments are described in further detail in connection with
(50) The Canceller Loop L2 (1206) also samples the transmitted RF signals (RF Transmit Reference signals 680). In some embodiments, it may be preferable not to sample directly as indicated but instead sample by taking a set of interim output signals from within L1 (there are M×N such interim cancellation signals) and then applying the additional fine resolution processing of L2 (1206). L2 is distinctly different from L1 (1204) in that L2 processes Tx signals in the digital baseband domain using FIR digital “filter” techniques which enables L2 to practically cancel far lower power signals with substantially longer delays than with L1 (1204). Optionally the L2 cancellation signal can be upconverted back to RF (“C2R.sub.n” as shown) or used as a cancellation signal at digital baseband (“C2D.sub.n” as shown and referred to as a baseband transmit leakage cancellation signal in some embodiments). In embodiments utilizing digital baseband cancellation via L2 or L2 and L3 together, summer nodes (alternatively referred to as cancellation combiners) 1212-n will be respectively utilized to sum the C2D.sub.n signal with receive chain output signals (674) from respective receive chains to provide respective baseband cancelled receive signals. A major issue with L2 is the noise and distortion added by the act first of downconverting and digitizing the sampled Tx RF signals, and then further upconverting and leveling if the optional “C2R.sub.n” signals are to be generated. Aspects of L2 embodiments addressing such impairments will be discussed. Note that exemplary embodiments of such processing are depicted and described in further detail in connection with
(51) The Canceller Loop L3 (1208) is very different from L1 (1204) or L2 (1206) in that it takes as an input a digital baseband representation of each Tx chain signal Tx.sub.1(t) to Tx.sub.M(t). In L3 (1208) all processing of the Tx chain signal can be done using digital FIR “filter” techniques similar to L2. Ideally the input to L3 would be after the digital low pass channel filter 1214-m. In embodiments where the filter outputs are not accessible the digital filters (1214) can also be replicated in L3 to generate a better estimate of the actual Tx signal. Such replication provides for matched channel impacts thus reducing the overall need for the L3 channel estimation to include compensation for filters 1214-1 to 1214-M. Eliminating the requirement to estimate this filter will allow for a faster or less complex estimation of L3 cancellation coefficients in some embodiments. Similarly, the effective response of the low pass analog filter (1216-m) following the DAC, and the BPF (1218-m) following the upconverter can also be included in L3 to improve the cancellation accuracy, efficiency, or convergence speed. Additionally, it is contemplated that other filters may be included in L3 as well, for instance those in the receiver such as a bandpass filter (not show) following summers 1210-n, and lowpass filters 1221-n, and 1222-n. Further, certain intermod products in the Tx chain or as created in the Rx front-ends (or in the analog portions of L1 or L2) can also be modeled within L3 (or possibly within L2) for cancellation by the L3 output. Such processing will be further described in connection with embodiments depicted in
(52) However, L3 (1208) cannot cancel Tx chain noise as L1 can generally or L2 can subject to certain limitations. Typically the L3 cancellation signals “C3D.sub.n” would be used at digital baseband (referred to as baseband transmit leakage cancellation signals in some embodiments) as shown but optionally these signals can be upconverted as “C3R.sub.n” and applied at RF (at summer 1210-n for instance and wherein C3R.sub.n comprises an up-converted baseband transmit leakage cancellation signal which acts as a digitally generated RF transmit leakage cancellation signal in some embodiments). It is also possible to generate both “C3D.sub.n” and “C3R.sub.n” simultaneously and in some embodiments with different emphases on the various undesired signal components to be cancelled.
(53) The ZDD Canceller Loop Coefficients Generator (ZCLCG) (675) is notional as shown in
(54) Referring to the IBR RF (632) and specifically the receive portions (640), the receive gain control, in some specific embodiments, must take into account the remaining transmit signal following the L1 cancellation process at the summers 1210-n. Some embodiments may control the AGC from the ZDD Canceller Loop Coefficients Generator (675) or from the IBR Modem or IBR Channel Mux instead of from within the Rx chain as shown in
(55) IBR embodiments utilizing ZDD may be utilized in at least two variants. The first variant is “co-channel” ZDD (“CC-ZDD”) wherein the Tx chain channels at least partially overlap (if not completely coincide with) the Rx chain channels. Embodiments utilizing CC-ZDD are theoretically possible of achieving twice the spectral efficiency for a single link at the physical layer, relative to systems not utilizing ZDD approaches. When coupled with available MAC efficiencies due to the reduction of delay and increased automatic retransmission (ARQ) efficiency utilizing an ACK/NAK protocol or the equivalent, additional efficiencies are possible. In particular, the delay reduction relative the TDD based systems is significant. This CC-ZDD mode is ideal for licensed band operation where interference is well controlled and spectral efficiency most highly valued.
(56) The second ZDD variant can be called “single band” or “co-band” ZDD (CB-ZDD) wherein both the Tx chain channels and the Rx chain channels are within a single band (using a single band pass filter) utilized during both transmission and reception, but the channels do not overlap. Such an arrangement results in minimal spectral efficiency improvement (though FDD is typically better than TDD) but is still highly desirable at least as a fallback mode for unlicensed band operation. One advantage of CB-ZDD in unlicensed bands results from the interference seen at the receiver of one IBR, being dramatically different (and frequency dependent) compared to its peer device such that operating in a similar frequency range allows for a more optimized channel frequency choice under significant interference conditions. Other advantages include the flexibility for a CB-ZDD device to operate in bands which operation would otherwise not be possible without expensive and fixed band pass filters. In some ways, “co-band” ZDD (“CB-ZDD”) is simpler because the analog and/or digital baseband low pass filters (as well as the FFTs in the IBR channel multiplexer) can perform much of the Tx signal cancellation. But in other ways since the requirement to cancel the Tx chain noise within the Rx chain channel remains the same as CC-ZDD, then CB-ZDD has the additional complication that the “easily” detectable Tx signal that drives the ZCLCG process is not at the right channel frequency to ensure noise cancellation in the Rx chain under normal operating conditions. However other metrics for Tx noise cancellation may be utilized mitigating much of this disadvantage, such as adapting the loops based upon receiver performance metrics such as signal to noise ratio (SNR), bit error rate (BER), frame error rate (FER) or metrics associated with the forward error correction (FEC). Such approaches would be most applicable to L1 or L2 with C2D.sub.n adaptation (or “L2D”) for noise cancellation.
(57) The aforementioned CC-ZDD (and potentially CB-ZDD) has specific applicability in use in an ZDD Aggregation End IBR (AE-IBR) when operating in a point to multi-point (PMP) mode in communication with multiple non-ZDD Remote End IBRs (RE-IBRs) operating each in a TDD mode. Such a configuration allows for the ZDD enabled AE-IBR to be transmitting to one or more RE-IBRs, whilst receiving from one or more other RE-IBRs. In these embodiments, no ZDD cancellation is required at the RE-IBRs, but a doubling of the overall network efficiency is realized relative to AE-IBRs not utilizing ZDD. To enable such an embodiment, a time multiplexing of one or more of the AE-RE links and RE-AE links must be arranged and scheduled such that the RE-IBRs are time multiplexed with their transmission and reception periods, and offset relative to other RE-IBRs, at least to the extent that TDD operation is achievable at each RE-IBR individually, not considering other multipoint multiplexing approaches such as frequency multiplexing (such as Orthogonal Frequency Division Multiple Access (OFDMA), Single Carrier Frequency Division Multiplexing (SC-FDM), and the like). Such techniques may be using in combination with the aforementioned ZDD multiplexing approaches.
(58) Other embodiments of the aforementioned ZDD techniques may be utilized for repeaters to be used interposed between IBRs, or in conjunction as a feature of a particular ZDD enabled IBR. Such repeater embodiments may utilize ZDD approaches to allow for reception and transmission of signal simultaneously allowing for higher efficiency relative to TDD based repeating approaches, or more spectral efficiency relative to FDD based repeating approaches. ZDD repeater embodiments may perform the repeating function in a number of approaches including: at an RF and un-demodulated level, a modulation symbol by symbol level, a stream level, an FEC Block Level, a MAC Frame Level, or potentially higher levels. When performed at a stream level, beam forming techniques may be utilized to allow for a spatially rich propagation environment allowing for an increased performance network.
(59) Embodiments of a Loop 1 (L1) (1204) canceller of cancellation multiplexer (670) are depicted in
(60) The transfer function G.sub.Sm (1306-m) is intended to indicate a “zero-th” order match between the magnitude and delay from X.sub.m(t) to an average Rx chain at the first summer and a known frequency response. The delay might be realized in a cable of preset length that approximates the delays in the Tx-Rx path. The transfer function might be strongly influenced by one or more band pass filters in each Rx chain, or other filters in either the transmit or receive paths which are “in the loop”, and so it may be advantageous to place an identical band pass filter (or other such “in the loop” filters) within each G.sub.Sm (1306-m). This has the additional benefit of keeping out of band Tx spurs from being injected into the Rx chains. The transfer function also includes the effect of the coupling to X.sub.m(t), by a line coupler near the Tx antenna feed point in some embodiments. It is desirable to sample a signal highly correlated to the actual X.sub.m(t) transmission and to have sufficient signal such that the sampled noise floor is far above the equivalent input noise of L1.
(61) In some embodiments of the “FIR” structure depicted in
(62) The exemplary embodiment depicted in
(63) In practice, the RF delay line (1310) depicted in
(64) Referring now to
(65) In one greatly simplified and exemplary embodiment addressing the impacts of phase noise (and other sources of noise), a single shared VCO is utilized for all up converters and down converters for the transmitters, receivers, sampling receivers, and any C3R up converters. Such an arrangement does not require the RF “spatial multiplexing” discussed above in order to address phase noise concerns as the transmitter phase noise when received using a C2D sampling receiver (1406, alternatively referred to as transmit RF reference receive chains in some embodiments) will have a common noise as the transmitter signal when received on the signal receivers (1225-1 through 1225-N). Other arrangements may share VCOs between all up converters (that generate C3R.sub.n) and transmit chains (636), and between all down converters within C2D sampling receivers (1406-m of 1404-1 through 1404-M, and alternatively referred to as transmit RF reference receive chains in some embodiments) and signal receivers (1225-1 through 1225-N). Further embodiments may utilize common VCOs between pairs of down converters in a respective sampling receiver and signal receiver, or similarly utilize common VCOs for pairs of up converters for a respective signal transmitters and C3R up converter chain.
(66) Returning now to the exemplary embodiments of
(67) In some embodiments, the inputs from X.sub.m(t) can be a line coupled input as described for L1, or can be from the same G.sub.Sm (1306-m) (or a parallel G.sub.Sm for L2 only) as described for L1. In alternative embodiments, the inputs for each X.sub.m(t) might be derived from the set of C1.sub.m,n (where n=1 to N). This could be the average of the N cancellation signals, the maximum of them or some weighted blend where the weights are determined dynamically by the ZCLCG.
(68) Note that with respect to embodiments of
(69) In one embodiment in which Cm (1318-n) signals are utilized respectively as inputs (1402-1 to 1402-M, where M=N) to each sampling receiver 1406-1 to 1406-M, sets of weights (1311) may be chosen so as to allow for a “spatial” multiplexing between the transmitter signals and the separation of each of the transmitter signal components from one another including associated transmitter noise and specific sampling receiver (1406-m) noise as discussed above. As noted previously the phase noise of both the Tx chains and the Rx chains are particularly problematic for the operation of L2D processing. However in specific implementations, utilizing transmitters and receivers each with two or more shared voltage controlled oscillators and frequency references, such noise impact may be compensated for in such embodiments. In such a configuration, all M transmitters will be received by all N sampling receivers (1406-1 to 1406-M, when M=N). As noted, by appropriately choosing the L1 weights (1311) associated with each RF transmitter reference signal 1302-m, each combination (M by N) of transmitter and receiver noises will be sampled and recoverable via the mentioned spatial multiplexing approach. To the extent that pairs (or more) of receivers have common frequency local oscillators (VCOs and frequency references) individual transmitter and receiver noise components may be recovered and or be compensated for so as to allow for base band digital cancellation of the noise included in the Tx to Rx leakage signals from each receive chain. This is enabled by pairing each signal receiver (1225-n) with an L2D sampling receiver (1406-m) with a common LO frequency as discussed. The demultiplexing may be performed utilizing specific weights and processing within the FIR filters of
(70) In the embodiments associated with
(71) The preceding embodiment indicates a conventional complex FIR filter structure for CFIR2 (1408). It may be desirable when L2 is being used to cancel significant variations in overall delay of the Tx to Rx chain coupling paths to, in some embodiments, have a complex CFIR2 with a smaller time delay for certain of the taps and longer for the remainder, or to have parallel complex FIR structures with different time delay steps and then sum them together.
(72) In some embodiments each sampling receiver chain (1406-m) may be realized in practice using conventional components or incorporated in an RFIC. Each CFIR2.sub.m,n (1408) and digital domain summation (1410-n) can be constructed using conventional digital circuits in either an ASIC or FPGA, or realized in software on a DSP.
(73) Referring to
(74) However, L3 (1208) cannot cancel Tx chain noise as L1 can generally or L2 can subject to certain limitations. Techniques to mitigate such noise impacts may be addressed associated with L1 or L2 as discussed. One such approach combining C2D.sub.n and C3R.sub.n will be discussed associated with
(75) Returning to the exemplary embodiment of
(76) In the exemplary embodiment of
(77) The above discussion indicates the use of a conventional FIR filter structure for CFIR3 (1516-n for example). It may be desirable when L3 is being used to cancel signal components with significant time delay variations to either have a complex CFIR3 with a smaller time delay for certain of the taps and longer for the remainder, or to have parallel complex FIR structures with different time delay steps and then sum them together.
(78) The exemplary embodiment of
(79) In addition to IMGTx.sub.m, other intermods of potential interest include those created in the RF LNA of each receive chain n and those created within the other cancellers (such as L1 per the indication above). Each RF LNA may generate intermods of all M Tx signals each of which may be uniquely transferred to each Rx.sub.n. Thus, cancellation of such intermods requires at least a bank of M CFIR3Rx.sub.m (1502-m) for each Tx.sub.m, each comprising N CFIRs of length D3RF taps and weights (not shown). The intermods are then modeled based on M inputs at each respective IMGRx.sub.n (1504-n) wherein certain weights (not shown) may be passed to optimize the model in view of observed results. Similarly, for intermods created within each canceller branch of L1 (or of L2, not shown), these can be estimated for each X.sub.m(t) by applying Tx.sub.m(t) to IMGC1.sub.m (1512) as shown above and then applying the result to a bank of N CFIR2C1-m each of D3C1 taps in length with weights (not shown) supplied by the ZCLCG (675).
(80) In specific embodiments, the various operational and design parameters are chosen such that none of the intermod cancellers depicted above are necessary. However, this will not always be the case particularly for high transmit powers associated with longer range operation.
(81) Note the preceding discussion shows both C3R.sub.n and C3D.sub.n but in practice only one of these would normally be present from a single CFIR3 bank as depicted. If it were desired to have both, then there would need to be additional CFIR circuitry present (not shown) to account for the difference in magnitude, phase and delay between the two types of outputs in order to effect cancellations, or alternatively the CFIR3 bank would be replicated separately (i.e. CFIR3R & CFIR3D) with separate weights (not shown) for each to effect a “L3R” and a “L3D” in parallel.
(82) All of the CFIR, IMG and summation circuits (1514, 1520, etc.) depicted in
(83)
(84) In one embodiment, a basic process of subsequent cancellations based on an RRC setting (first order) and RLC setting (second order) is employed. In one exemplary embodiment of a ZDD enabled IBR, a start-up technique is for both ends of an IBR link in a given channel to start-up initially in a TDD service mode that allows the IBRs to exchange key RRC data and establish common frame sync.
(85) Then, in one embodiment based on RRC, L0 “coefficients” (possibly embodied as just selectable settings) are chosen first, typically from a non-volatile memory set by design or factory calibration (and possibly, or alternatively, within the Coefficients Memory (1606)). To the extent that such L0 coefficients or settings have multiple valid values for given RRC parameters, such values can be tested by inserting Tx-m chain preambles (serially or in parallel as described below for L1), and then applying FFT (or even just an RSSI, not shown, but available from the sampling receiver chain (1225-n) in one embodiment, or elsewhere in further embodiments) to each (or all) Rx.sub.n to measure the effects on undesired Tx.sub.m signal leakage into each Rx.sub.n. This can be iterated until a minimum leakage value is determined. Alternatively, the L0 settings may be balanced with an identified desirable IBR receive signal. In one embodiment, such a balance comprises determining settings achieving a target threshold receive signal strength, or SNR, or other metric for a desired receive signal, and a target L0 isolation level. In such an embodiment, compromises between receive desired SNR (or other metric), and L0 isolation requires balance and may be achieved by optimizing the value (maximum or minimum) of a formula defining a metric such as the following:
Vo=max(f.sub.DS(W.sub.0(i))*f.sub.ISO(W.sub.0(i))), for all i,
(86) In the equation above, f.sub.DS(W.sub.0(i)) is a function of the desired receive signal, where W.sub.0(i) are L.sub.0 coefficients sets as a function of i, and in one embodiment include the RRC antenna selection settings of the IBR antenna array (648). Note that i is an index which ranges in value from 1 to the total number of possible W.sub.0(i) setting combinations.
(87) In the equation above, f.sub.ISO(W.sub.0(i)) is a function of the isolation between each transmit chain and each receive chain, where W.sub.0(i) are L.sub.0 coefficients sets as a function of i, and in one embodiment include the RRC antenna selection settings of the IBR antenna array (648).
(88) The functions f.sub.DS, and f.sub.ISO in some embodiment are linear functions, while in other embodiment are nonlinear functions, or a combination. In one embodiment, the functions are a comparison to one or more thresholds, or fuzzy logic processing. In other embodiments the functions take the form of a global optimization function determining the maximum link throughput between two or more IBRs. Embodiments of such a function balance the target IBR received throughput and reliability and the current IBR's receive throughput and reliability as function of W.sub.0(i) settings. Examples of some of such embodied optimization algorithms associated with the L0 setting process, or associated with the other loop settings may be found in Systems Engineering in Wireless Communications, by Koivo and Elmusrati (ISBN 0470021780).
(89) Second, in one embodiment holding the L0 coefficients constant, based also on RRC primarily, an initial guess for the L1 coefficients would be read from a non-volatile memory set by design or factory calibration (and possibly within the Coefficients Memory (1606)). An iterative process would commence wherein Tx.sub.m preambles (either serially or in parallel initially) are inserted into each Tx.sub.m and the effect on each IQ magnitude bin for each Rx.sub.n is considered to modify the coefficients to minimize observed Tx.sub.m in each Rx.sub.n. There are numerous techniques that the Coefficients Generator Processor (1610) and/or Coefficients Calculator (1608) can use, such as iteration by steepest descents (or alternatives such as those disclosed within Koivo and Elmusrati), to determine the coefficients in view of the previous values of W1.sub.m,n,d and previous IQ magnitude bins given a new observed set of IQ magnitude bins for a new tested set of W1.sub.m,n,d. Such processing may be performed as time domain based algorithm, or utilizing frequency domain based processing, or as a combination of both domains. In some embodiments, the Tx.sub.m preambles may be inserted in parallel to optimize the W1.sub.m,n,d in view of Tx.sub.m1 to Tx.sub.m2 leakage that passes through L1 and creates an additional leakage path independent of L0 into each Rx.sub.n from each Tx.sub.m. In an exemplary embodiment, the Critical Timing Unit (1616) schedules the preamble insertions and FFT sampling in view of the Tx Symbol Clock (for precise preamble symbol insertion) and the Tx Frame Sync (so that, for example, Tx.sub.m preambles are sent when other IBRs in receive antenna view of the instant IBR in the same Rx channel have their Tx signals substantially inhibited in power). Such timing may also be coordinated via the IBMS Agent (700) in cooperation with other IBRs or an IBMS Server as described in co-pending U.S. application Ser. No. 13/271,057 by a common inventor and assignee.
(90) In an embodiment where L1 is not present but L2 is, the above process would be performed in a similar manner for L2 in some embodiments. To the extent that the delay range of such an L2 includes long delay cancellations, the optimization of such W2.sub.m,n,d would in some embodiments, be performed similar to the process described below for L3 in some embodiments. Alternatively or in addition, the L2 coefficients may be calculated by closed form approaches rather than using iterative algorithms. Such algorithms may include so called MMSE and LMS based approaches in some embodiments, which will be described in more detail.
(91) After the L1 coefficients have converged, reflection cancellations are made using either L2 or L3 (for example, using L3, as in L3D only in certain embodiments as depicted associated with
(92) The above process assumes CC-ZDD wherein some or all of the Tx-m channel overlaps with the Rx-n channel. For CB-ZDD wherein the Tx and Rx channels are adjacent, it may be preferable to use the above approach with the Rx channel BW in the downconverter chains temporarily set to cover some or all of the Tx and Rx together to more efficiently determine various cancellation coefficients. For CB-ZDD with disparate Tx and Rx channels, it may be preferable to temporarily force the RRC to tune the signal receiver chains (1225-n) to the Tx channel so that various cancellation coefficients can be efficiently determined from the Tx leakage signals. However in CB-ZDD, even if neither of the above options is exercised, it should still be possible to determine cancellation coefficients simply from Tx noise (and out of channel IM signals if present) using substantially the same procedure described above for CC-ZDD.
(93) After both ends of a particular link have converged to an initially acceptable set of canceller coefficients, it is also necessary to update and maintain these coefficients in view of changing environmental parameters such as temperature or voltage internal to each IBR or channel obstructions external to each IBR. It is expected that L0 parameters would not be part of this update process. One exemplary procedure for this update process would be to periodically, such as once every L3.sub.U frames (where L3.sub.U may be 1 to 10 depending on the environment), have the Critical Timing Unit send one or more Tx.sub.m chain preambles, while the other IBRs in view of the Rx antennas at the Rx channel frequency substantially inhibit their Tx power, and effectively repeat the W3.sub.m,n,d coefficient calculation process if the IQ magnitude bins have increased beyond some threshold from the previous test stored in memory. Similarly, every L1.sub.U frames (where L1.sub.U may be 100-1000 depending on the environment), the W1.sub.m,n,d coefficients may be re-tested with C2 and/or C3 inhibited to recalculate W1 if a change beyond a threshold has occurred. If so, recalculation of W2 and/or W3 (and/or IM parameters) would need to follow. Alternatively, rather than every L1.sub.U frames, this W1 test may be scheduled only if the L3 (and/or L2) update process fails to keep the residual Tx signals in one or more Rx.sub.n below a particular threshold value. In some ZDD IBR installations, the channel dynamics may be far slower or faster changing than in others. Thus, it may also make sense to have L1.sub.U, L2.sub.U, and/or L3.sub.U into parameters that are adaptive to the estimated channel dynamics. For example only, if current values of L1.sub.U, L2.sub.U, and/or L3.sub.U consistently produce insubstantial changes to their respective W1, W2, and/or W3 coefficients, then increase the given values of L1.sub.U, L2.sub.U, and/or L3.sub.U until some predetermined maximum is reached. Conversely, if substantial changes to coefficients do result, then reduce the given values of L1.sub.U, L2.sub.U, and/or L3.sub.U until some predetermined minimum is reached.
(94) As will be discussed, in other embodiments a closed form LMS or MMSE calculation may be made to determine L2 or L3 weights (or other weights for that matter).
(95) In one embodiment associated with calculating L2 weights, a least squares closed form based approach may be utilized in a C2D (or analogously for calculating L3 weights in a C3D) cancellation wherein the sampled RF Transmit Reference Signals (X.sub.S1 to X.sub.SM) corresponding to signals associated with 680-1 through 680-M, using sampling receivers (1406-m), are compared with the receive chain output signals (674) prior to summer (or alternatively referred to as combiners) 1212-n (or at 1212-n with C2D and/or C3D temporarily inhibited) to calculate the coefficients W2.sub.m,n,d, utilizing a closed form calculation. Some embodiments will utilize a least squares calculation, or MMSE calculation which when performed in the time domain requires a large matrix inversion. Alternatively such calculations may be performed in the frequency domain to achieve a closed form solution. For example in one embodiment, the following process is followed for each receiver chain n, where n varies from 1 through N:
(96) 1) A vector of sampled receive data (Z.sub.Sn) is assembled from S.sub.R complex valued time domain samples of Z.sub.Sn(t) at the output of receiver chain 1225-n, taken from the input (674-n) of cancellation combiner (alternatively referred to as cancellation summers) 1212-n. Note that Z.sub.Sn is arranged to have the dimensions S.sub.R by 1, and that S.sub.R≥D2, where D2 is the number of taps in each CFIR2 (1408). Next, the Fourier transform of Z.sub.Sn is taken to provide FZ.sub.Sn (also having dimensions S.sub.R by 1). Such a Fourier transform may be realized in any one of known conventional approaches, including utilizing FFT processing block 1614, with additional coupling to signals at reference points 674 to obtain time domain samples of Z.sub.Sn (not shown in
(97) 2) M vectors are respectively assembled for each block 1404-m coupled to RF Transmit Reference Signals 680-m, where m=1 through M. Each vector (X.sub.Sm) of sampled RF Transmit Reference Signals data is assembled from S.sub.R complex valued time domain samples of X.sub.m(t) at the output (1409-m) of the sampling receiver chain (1406-m) of block 1404-m, for each reference signal. Note that each vector X.sub.Sm is arranged to have the dimensions S.sub.R by 1, and that S.sub.R≥D2, where D2 is the number of taps in each CFIR2 (1408). Next, the Fourier transform of X.sub.Sm is taken to provide FX.sub.Sm (also having dimensions SR by 1). Such a Fourier transform may be realized in any one of known conventional approaches, including utilizing FFT processing block 1614, with additional coupling the signals at reference points 1409-m of each sampling receivers 1406-1 through 1406-M to obtain time domain samples of X.sub.Sm (not shown in
(98) 3) Next, in the frequency domain, and on a bin-by-bin basis, a Least Squares estimation is performed. For a reference on the least squares estimator see eq. 3.33, Smart Antennas For Wireless Communications, Rappaport; ISBN 0-13-71987-8 and derivation 3L, Linear Algebra and its Applications, Strang; ISBN 0-15-551005-3. Also see “Least Squares Problems with several Variables” as one example utilizing complex mathematics for the current embodiment, though alternative solutions are contemplated. Adapting the MMSE and Least Squares approach to the current application yields Eq. 14-1.
FW2.sub.n.sup.EST(b)=[FXs(b).sup.HFXs(b)].sup.−1FXs(b).sup.HFZs.sub.n(b) (Eq. 14-1)
(99) where b indexes frequency bins 1 through S.sub.R, and where the “H” superscript refers to the Hermitian (conjugate) transpose of a matrix, and the superscript “−1” refers to a matrix inversion and, where, FZs.sub.n(b) is a complex scalar and FXs(b) is comprised of a 1 by M vector, and FW2.sub.n.sup.EST(b) is an M by 1 vector, where each element comprises the b.sup.th complex frequency domain bin value of the frequency domain estimation of the filter taps associated with CFIR2.sub.m,n.
(100) 4) Following computation for all frequency bins from b=1 through S.sub.R, the newly determined (M by S.sub.R) matrix FW2.sub.n.sup.EST may be combined utilizing a weighted averaging, or other filtering approaches, on an bin by bin basis with previously utilized or estimated versions of matrix FW2.sub.n.sup.EST resulting in matrix FW2.sub.n.sup.Filt. Alternatively, the frequency domain based filtering may be performed in a later time domain step, and vector FW2.sub.n.sup.EST may be substituted for the subsequent step referencing the matrix FW2.sub.n.sup.Filt.
(101) 5) Next an inverse FFT of FW2.sub.n.sup.Filt is performed on a row wise basis (over S.sub.R samples, for each row m) resulting in W2.sub.n.sup.EST which comprises estimates of the time domain FIR filter weights as a vector of dimension M by S.sub.R, where S.sub.R is ≥D2 (the number of filter taps per FIR2 (1408)).
(102) 6) Finally W2.sub.n is calculated by performing a truncation of W2.sub.n.sup.EST from M by S.sub.R to a M by D2, where for each receive chain the most significant magnitude coefficients will be contained in the first D2 values of the M rows of W2.sub.n.sup.EST. Additionally, a time domain sample-by-sample averaging, weighted averaging (FIR), recursive (IIR) filtering may be performed with previous calculated versions of W2.sub.n.sup.EST or W2.sub.n. Note that each row W2.sub.n corresponds to the weights, W2.sub.m,n,1 to W2.sub.m,n,D2 for the corresponding CFIR.sub.m,n (1408). Note that this processing equally applies to the calculation of the L3 coefficients as well, with the exception that such calculation be processed on the output of the cancellation of C2 (if present), and only while the W1, W2, and W3R weights (if present) are held static. Additionally the transmit reference signals would be based upon the pure digital reference signals Tx1 . . . M (672) input to the transmit chains, rather the RF Transmit Reference Signals (680).
(103) Referring now to
(104) Note that L0 (1202) includes the effects of each transmit antenna signal X.sub.m(t) on every receive antenna signal Y.sub.q(t), as well as on every other X.sub.m(t). Note also that every antenna, X.sub.m(t) and Y.sub.q(t), may be subject to receiving unwanted interference I(t) as well as the desired signal from another IBR and various undesired leakage signals from the instant IBR.
(105) The goal of L0, L1 and L3 is that each Rx.sub.n(t) does not have a level of any X.sub.m(t) (or other undesired signal introduced by the instant IBR and not the outside world radiating into each receive antenna) that causes appreciable (typically more than 1 dB) desensitization at the demodulator relative to an equivalent receiver where each X.sub.m(t) at the instant IBR has zero power.
(106) Referring now to
(107) However, in some embodiments utilizing commercially available chip set implementations, VCOs are shared only among pairs of transmitters and pairs of receivers. As discussed previously, the phase noise combinations may be recovered and compensated for in further cancellation processing at base band during C2D or C3R processing.
(108) Referring now to
(109)
(110) The sampled Tx.sub.m(t) signals utilized by L1 are predominantly G.sub.Sm×X.sub.m(t). However, they may also include unwanted leakage of other transmit signals on a theoretically infinite number of propagation paths as shown in Eqs. 19-3 and 19-4 of
(111)
(112) At the inputs to each downconverter chain, Z.sub.n(t) (Eq. 20-2), the signal is composed of all 8 Y.sub.q(t) signals multiplied by respective complex transfer functions and subject to a small delay plus the equivalent input noise for the downconverter chain and the L1 cancellation signal, as well as any intermod products generated in the circuits from antenna q to downconverter n (especially in view of Tx leakage from X.sub.m that may be much larger than Y.sub.Rxq). Note that within 648B the effect of the switch matrix and the LNA power down when a particular receive antenna is not selected typically results in the selected q.sub.n mapping having a magnitude of G.sub.Rxq,n that greatly exceeds that of all unselected mappings.
(113) The L1 cancellation signal can be described as a summation across the two sampled transmit signals X.sub.Sm (which each comprise X.sub.1 and X.sub.2 due to finite Tx to Tx antenna isolation) weighted by complex transfer functions W1.sub.m,n,d at delays of approximately “d*Δt.sub.1” (but not necessarily uniform delay steps in some embodiments) and G1.sub.m,d (in some embodiments these might all be unity but in practice with RF delay lines these will vary somewhat in other embodiments), plus undesired intermodulation products created within L1 as indicated.
(114)
(115) This representation of Z.sub.n(t) (Eq. 21-00) illustrates five constituent components for each Rx downconverter input chain signal:
(116) (21-01) This is the “desired Rx chain signal” component that will be eventually channel multiplexed, equalized, and demodulated in the baseband Rx PHY. This is dominated by the switch matrix selection from q to n.
(117) (21-02) This is the “Rx chain input noise floor” (which includes interference at the Rx antennas) component that leads to finite SINR for the receive chain. This should also be dominated by the switch matrix selection from q to n in consideration of the thermal noise floor and RF front-end noise figure. Ideally the dominant G.sub.Rxq,n×N.sub.Rxq is much greater than N.sub.DCn as well in good RF design practice.
(118) (21-03) This is the “undesired Tx leakage” component that reflects the cumulative effects of finite isolation in L0 from any Tx antenna m to any receive antenna q in view of the switch matrix selections.
(119) (21-04) This is the L1 cancellation component (or “L1 output signal”) which in an ideal world would be exactly equal to, but opposite in sign to, the undesired Tx leakage component. In reality, this is not possible, but if L1 sufficiently cancels certain key sub-components of the Tx leakage, then L2 or L3 may be able to sufficiently cancel the rest for CC-ZDD. In the case of CB-ZDD, then L1 needs to cancel only the Tx leakage noise (and IM signals if present) in the Rx channel to be acceptably low relative to the noise plus interference component.
(120) (21-05) This is the “undesired Rx intermods” (of the Tx leakage signals) produced in the RF front-end (typically the LNA) of each path from Rx antenna q to downconverter input n. Ideally, the L0 isolation and the linearity of the LNA and/or switch matrix would cause such intermods to be acceptably low relative to the noise plus interference component by design such that this component is negligible. It is generally not possible to cancel this particular intermod component by L1. However, in some cases it may be feasible to model the generation of this intermod component relative to the Tx.sub.m digital baseband signals and thus cancel it in L3.
(121)
(122) This representation of Z.sub.n(t) now illustrates nine constituent components for each Rx downconverter input chain signal:
(123) Equation components (22-01) and (22-02) are equivalent equation segments (21-01) and (21-02) of
(124) Equation Segment (22-03) is the first of three subcomponents of the “undesired Tx leakage” component from the previous slide. This particular subcomponent is representable in terms linearly related to the original Tx chain baseband digital signal Tx.sub.m(t).
(125) Equation Segment (22-04) is the second of three subcomponents of the “undesired Tx leakage” component from the previous slide. This particular subcomponent comprises the Tx chain output noise.
(126) Equation Segment (22-05) is the third of three subcomponents of the “undesired Tx leakage” component from the previous slide. This particular subcomponent comprises the Tx chain output intermods.
(127) Equation Segment (22-06) is the first of three subcomponents of the “L1 output signal” component from the previous slide. This particular subcomponent is the actual L1 cancellation signal that is complicated by the finite Tx-Tx antenna isolation as represented in L0tx.sub.m1,m2,i. This “L1 cancellation signal” would ideally cancel all three of the undesired Tx subcomponents.
(128) Equation Segment (22-07) is the second of three subcomponents of the “L1 output signal” component from the previous slide. This particular subcomponent comprises the interference at the Tx antennas and the noise generated by L1. It is not possible to cancel the noise term in this sub-component. Therefore, L1 and the Rx front-end must be designed such that the noise term in this sub-component is acceptable relative to the “Rx chain input noise floor” component. In practice, this involves minimizing the noise output from L1 and raising the gain in the Rx front-end without causing other problems (such as downconverter saturation or Rx front-end undesired intermods). It is also not possible to cancel the interference term in this subcomponent at least from within the ZDD subsystem of the IBR. However, this interference I.sub.Txm is likely correlated to I.sub.Rxq in practice and thus an IBR with redundant Rx chains may be able to cancel such interference in the Rx channel multiplexer. Note that in practice, W1×G1×G.sub.Sm is likely substantially lower in magnitude that G.sub.Rxq,n for the selected q of a given n.
(129) Equation Segment (22-08) is the third of three subcomponents of the “L1 output signal” component from the previous slide. This particular subcomponent arises from intermods of X.sub.m(t) (or effectively Tx.sub.m(t) if the EVM is small) generated within L1. Preferably, these intermods are acceptably low relative to the “Rx chain input noise floor” component by design and hence effectively negligible. In some cases, it may be feasible to model the generation of this intermod component relative to the Tx.sub.m digital baseband signals and thus cancel it in L3.
(130) Equation Segment (22-09) is equivalent to Equation Segment (21-05) of
(131)
(132) This representation of Z.sub.n(t) now illustrates eight constituent components for each Rx downconverter input chain signal:
(133) Equation components (23-01) and (23-02) are equivalent equation segments (21-01) and (21-02) of
(134) Equation component (23-03) is the first of three subcomponents of the “residual Tx leakage” component which is expressed in view of the three sub-components of the “undesired Tx leakage” in combination with analogous sub-components of the L1 cancellation signal of the previous slide. This particular subcomponent is representable in terms linearly related to the original Tx chain baseband digital signal Tx.sub.m(t) (across all Tx chains). To the extent that the magnitude of [G.sub.Txm×Tx.sub.m] times L0.sub.m,q,i exceeds the equivalent input noise (plus interference) for any [t.sub.m,q,i+t.sub.Rxq,n]>[t.sub.sm+t.sub.D1], then at least some L3 cancellation will be required or the receive chain will be desensed.
(135) Equation component (23-04) is the second of three subcomponents of the “residual Tx leakage” component. This particular subcomponent comprises the residual Tx chain output noise (with contributions from all Tx chains). It is imperative (in some embodiments) that t.sub.D1 be chosen such that [t.sub.sm+t.sub.D1]>[t.sub.m,q,i+t.sub.Rxq,n] for any “i” wherein the magnitude of [N.sub.Txm×L0.sub.m,q,i] exceeds the equivalent input noise (plus interference) of each receive chain and further that D1 be large enough in number of taps such that any frequency dependence of the effective transfer function of L0 can be sufficiently approximated. This is necessary because L3 in this exemplary embodiment cannot cancel any residual Tx noise in a receive chain.
(136) Equation component (23-05) is the third of three subcomponents of the “residual Tx leakage” component. This particular subcomponent comprises the Tx chain output intermods (with contributions from all Tx chains). Ideally in the current embodiment, t.sub.D1 would be chosen such that [t.sub.sm+t.sub.D1]>[t.sub.m,q,i+t.sub.Rxq,n] for any “i” wherein the magnitude of [X.sub.IMTxm×L0.sub.m,q,i] exceeds the equivalent input noise (plus interference) of each receive chain and D1 would be large enough in number of taps such that any frequency dependence of the effective transfer function of L0 can be sufficiently approximated so that further cancellation of Tx IM products in L3 of this exemplary embodiment is not required.
(137) Equation components (23-06, 07, and 08) are equivalent to (22-07, 8, and 9) of
(138) Note that the relationship of W1.sub.m,n,d to G.sub.Rxq,n, G1.sub.m,d, G.sub.Sm, L0tx.sub.m1,m2,i, L0.sub.m,q,i, and the delays is theoretically identical for Tx.sub.m(t), N.sub.Txm(t) and X.sub.IMTxm(t). Thus in CC-ZDD solving for W1.sub.m,n,d such that the residual Tx signal is minimized should also minimize the residual Tx noise and intermods at a given Rx chain input n. For CB-ZDD, this is still true for residuals in the Tx channel, but not necessarily true for residuals in the Rx channel if W1.sub.m,n,d was determined by minimizing residual Tx signal in the Tx channel. In particular, for non-overlapping CB-ZDD, the frequency dependence of L0 may make the W1.sub.m,n,d that minimizes the residual Tx noise in the Rx channel somewhat different than that which minimizes residual Tx noise (or residual Tx signal) in the Tx channel. However, determining W1.sub.m,n,d may be far simpler in the Tx channel by minimizing residual Tx signal so this may be a good interim step despite the likely requirement of temporarily operating the Rx chain at the Tx channel (and note that to the extent iteration is used, the first guess may still preferably come from a table set by factory calibration and/or design as discussed previously). Final optimization of W1.sub.m,n,d in CB-ZDD may require processing gain by averaging or multiple iterations across many residual signal tests due to the statistical variations of the noise.
(139)
(140) The first component is the desired Rx chain signal (24-11). The second component (24-12) is the Rx chain noise floor plus interference from sources other than the instant ZDD IBR. Because L1 connects the Tx antennas in a path to each receive chain independent of the Rx antennas, the interference at the Tx antennas is also in each receive chain per this component. However, if I.sub.txm is correlated with I.sub.Rxq, then in certain receive architectures such interference may be cancellable by the Rx FDE or other mechanism within the Rx channel multiplexer assuming redundant Rx chains.
(141) The third component (24-13) is the residual Tx signal (with contributions from all Tx chains due to Tx-Tx antenna coupling). The L3 cancellation signal must be optimized by finding W3.sub.m,n,d3 coefficients such that this residual Tx signal is sufficiently small relative to the Rx chain noise floor in CC-ZDD. Because the residual Tx signal is linear with respect to Tx.sub.m(t), then for sufficiently large D3 and t.sub.D3>t.sub.m,q,i for any L0.sub.m,q,i of interest, it should always be possible to find W3.sub.m,n,d3 such that this residual Tx signal is sufficiently minimized. For non-overlapping CB-ZDD, this component may not be important as the frequency dependence of G.sub.DCn may greatly attenuate this component relative to the desired Rx signal and the Rx noise floor, and the residual may be effectively discarded by the FFT in the Rx channel multiplexer.
(142) One or more of the methodologies or functions described herein may be embodied in a computer-readable medium on which is stored one or more sets of instructions (e.g., software). The software may reside, completely or at least partially, within memory and/or within a processor during execution thereof. The software may further be transmitted or received over a network.
(143) The term “computer-readable medium” should be taken to include a single medium or multiple media that store the one or more sets of instructions. The term “computer-readable medium” shall also be taken to include any medium that is capable of storing, encoding or carrying a set of instructions for execution by a machine and that cause a machine to perform any one or more of the methodologies of the present invention. The term “computer-readable medium” shall accordingly be taken to include, but not be limited to, solid-state memories, and optical and magnetic media.
(144) Embodiments of the invention have been described through functional modules at times, which are defined by executable instructions recorded on computer readable media which cause a computer, microprocessors or chipsets to perform method steps when executed. The modules have been segregated by function for the sake of clarity. However, it should be understood that the modules need not correspond to discreet blocks of code and the described functions can be carried out by the execution of various code portions stored on various media and executed at various times.
(145) It should be understood that processes and techniques described herein are not inherently related to any particular apparatus and may be implemented by any suitable combination of components. Further, various types of general purpose devices may be used in accordance with the teachings described herein. It may also prove advantageous to construct specialized apparatus to perform the method steps described herein. The invention has been described in relation to particular examples, which are intended in all respects to be illustrative rather than restrictive. Those skilled in the art will appreciate that many different combinations of hardware, software, and firmware will be suitable for practicing the present invention. Various aspects and/or components of the described embodiments may be used singly or in any combination. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the invention being indicated by the claims.