Polarization transmitter using zero crossing avoidance

Abstract

A polarization transmitter using zero crossing avoidance is provided. A polarization transmitter and method thereof generate a filtered IQ waveform representing an input bit stream in an IQ space. The filtered IQ waveform is modified to avoid zero crossing regions by intermittently adding a zero crossing avoidance signal to the filtered IQ waveform, wherein the zero crossing avoidance signal has a spectrum including at least a first tone and a second tone, wherein the first tone and the second tone define a first peak and a second peak on opposite sides of a center frequency trough. A polarization signal including a polarization amplitude and a phase is generated based on the modified IQ waveform. An RF carrier is modulated using the polarization signal.

Description

Technical Field

[0001] The present disclosure relates generally to digital modulation, and more particularly to phase shift keying (PSK) modulation using polar transmitters. Background Art

[0002] In the PSK modulation scheme, the bit stream to be modulated is divided into n-bit sequences, and each n-bit sequence is represented by a specific phase shift of the binary code mapped to the n-bit sequence. n The PSK scheme maps n-bit codes to 2 n phase shifts. For example, in an 8PSK system, each of the 8 3-bit sequences is mapped to a different phase shift (sequences 000, 001...110, 111 can be represented by phase shifts 0, π / 4,..., π, -3π / 4). Each phase shift can be represented as a coordinate in the IQ (In-Phase-Quadrature) plane, where I is the abscissa of the IQ plane and Q is the ordinate of the IQ plane. Meanwhile, the quality metric for PSK modulation is the Error Vector Magnitude (EVM), where EVM is a measure of how far the actual IQ constellation points deviate from their ideal positions.

[0003] In a Cartesian modulator, separate I and Q signals of an IQ pair are generated and applied to respective mixers. At the receiver, the I and Q signals are recovered and used to obtain an n-bit sequence.

[0004] Polar transmitters are an alternative to Cartesian modulators and operate by converting I and Q values ​​into amplitude (A) and phase (P) signals. Polar transmitters can achieve significant power savings by allowing the use of switching power amplifiers and removing the need for mixers and baseband filters of Cartesian modulators.

[0005] However, when an n-bit sequence is represented by a phase shift of π, this results in a "zero crossing" event where the IQ trajectory in the IQ space passes through (or very close to) the origin of the IQ coordinate system. At the zero crossing, the corresponding amplitude A in the polar transmitter decreases to near zero. When the modulation trajectory passes through the origin, this results in an undesirable large instantaneous FM modulation, which can lead to higher bit errors on the receiving side. Some wireless protocols have narrow channels that make it difficult to tolerate wide FM modulation. Bluetooth is one such example, where Bluetooth EDR (Enhanced Data Rate) 3 packets use 8PSK modulation and the channel is 1MHz wide. Implementing large instantaneous FM modulation via a phase-locked loop (PLL) is challenging for all systems and is particularly difficult for non-narrowband systems. Therefore, polar architectures are typically reserved for narrowband systems. For narrow channels, high instantaneous FM modulation must be precisely time-aligned with AM modulation to avoid spectrum leakage into other channels.

[0006] Therefore, attempts have been made to design polar modulators that avoid zero-crossing events by steer- ing the IQ trajectory around the origin, thereby reducing the FM range. Zero-crossing avoidance also has additional benefits for the AM modulation path in polar transmitters: reducing the AM modulation bandwidth / slew rate, reducing the necessary dynamic range for the AM path, and reducing the required time alignment between the AM modulation and the large instantaneous FM modulation. However, current zero-crossing avoidance schemes exhibit the drawback of EVM degradation, making them unsuitable for applications with specified EVM limits. Summary of the invention

[0007] According to the inventive concept, a polar transmitter and method thereof for generating a zero-crossing avoidance signal having frequency characteristics capable of improving performance in terms of EVM, frequency offset, and reducing power consumption are provided.

[0008] In an embodiment, a method for modulating a bit stream includes: generating a filtered in-phase (I) and quadrature-phase (Q) ("IQ") waveform representable in an IQ space based on a stream of in-phase (I) and quadrature-phase (Q) symbol pairs, wherein each symbol pair represents a portion of the bit stream. Modifying the filtered waveform to avoid zero-crossing regions by intermittently adding a zero-crossing avoidance signal to the filtered waveform, wherein the zero-crossing avoidance signal has a spectrum including at least a first tone and a second tone, wherein the first tone and the second tone define a first peak and a second peak on opposite sides of a center frequency trough. Based on the modified IQ waveform, generating a polarization signal including a polarization amplitude and a phase. Modulating an RF carrier using the polarization signal.

[0009] In an embodiment, a polar transmitter includes: a symbol mapper configured to receive an input bit stream and generate a stream of IQ symbol pairs to obtain an IQ waveform, wherein each symbol pair represents a portion of the bit stream. A pulse shaper performs low-pass filtering on the IQ waveform to obtain a filtered IQ waveform that can be represented in an IQ space. A zero-crossing avoidance signal generator generates a zero-crossing avoidance signal, wherein the zero-crossing avoidance signal is applied during a window containing a π phase shift transition and has a characteristic defined by: a main peak at a zero-crossing time that coincides with a zero-crossing of the filtered IQ waveform, a first trough at a time between a start time of the window and a zero-crossing time, and a second trough at a time between a zero-crossing time and an end time of the window. A delay element delays the filtered IQ waveform to provide a delayed IQ waveform. An adder adds the zero-crossing avoidance signal to the delayed IQ waveform to obtain a modified IQ signal. An IQ to polarization conversion circuit converts the modified IQ signal into a polarization signal, and a polarization modulator modulates an RF carrier using the polarization signal. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] The above and other aspects and features of the inventive concept will become more apparent from the following detailed description taken in conjunction with the accompanying drawings, in which like reference numerals represent like elements or features, and in which:

[0011] Figure 1 is a functional block diagram of an embodiment of a polar transmitter according to the inventive concept.

[0012] Figure 2 is available in Figure 1 Functional block diagram of an example polar modulator used within a polar transmitter.

[0013] Figure 3 is used to explain Figure 1 Signal diagram of example signals of multiple aspects of zero crossing detection, IQ waveform filtering, and polarimetric signal generation in a polarimetric transmitter.

[0014] Figure 4 Show Figure 3 Example of the progression of the unfiltered signal in the IQ coordinate system.

[0015] Figure 5A is a graph depicting an example of a filtered IQ waveform in IQ space.

[0016] Figure 5B Is a description Figure 5A Graph of an example of a filtered IQ waveform after being corrected by a zero crossing avoidance signal generated according to an embodiment.

[0017] Figure 6 An example comparison of a filtered Cartesian waveform and a Cartesian waveform with signal correction avoided according to an embodiment is shown.

[0018] Figure 7 Shown based on Figure 6 Samples of the I and Q waveforms produce exemplary polarization signals.

[0019] Figure 8 is a graph showing a frequency spectrum comparison between a conventional "filtered pulse" type avoidance signal and an example "multi-tone" avoidance signal in accordance with the present inventive concepts.

[0020] Fig. 9 Show Figure 8 Time domain comparison of the two avoidance signals.

[0021] Fig.10 is a flow chart of an example method for digital modulation according to the inventive concept.

[0022] Fig.11is a flow chart of an example method for modifying a filtered IQ waveform by adding a zero crossing avoidance signal according to an embodiment, and is a method for implementing Fig.10 An example method of operation 1008 is provided.

[0023] Fig.12 Example traces of a filtered IQ waveform and an avoided signal corrected (and filtered) IQ waveform are shown in accordance with an embodiment.

[0024] Fig.13 Shown is a graph depicting the maximum frequency offset of an uncorrected polar transmitter relative to a polar transmitter according to embodiments disclosed herein.

[0025] Fig.14 is a graph showing simulation results showing power spectral density of an amplitude path versus frequency for an exemplary embodiment disclosed herein.

[0026] Fig.15 Yes Figure 1 A functional block diagram of an example electronic device for a polar transmitter. DETAILED DESCRIPTION

[0027] The following description with reference to the accompanying drawings is provided to assist in a comprehensive understanding of certain exemplary embodiments of the inventive concepts disclosed herein for illustrative purposes. The description includes various specific details to assist those of ordinary skill in the art in understanding the inventive concepts, but these details are to be considered merely illustrative. For the purposes of brevity and clarity, descriptions of known functions and structures may be omitted when their inclusion may obscure an understanding of the inventive concepts by those of ordinary skill in the art.

[0028] Embodiments of the inventive concept provide a zero-crossing avoidance scheme for a polar transmitter that provides lower power consumption than a Cartesian modulator and can exhibit specific advantages over polar transmitters that employ conventional zero-crossing avoidance schemes. These benefits may include: avoiding excessive EVM degradation; reducing the requirements for timing alignment between the AM (amplitude modulation) path and the PM (phase modulation) path of polar modulation; and reducing the bandwidth of the amplitude path signal. At the same time, the embodiments can still meet the strict ACP (adjacent channel power) requirements for specific protocols (e.g., Bluetooth EDR 3). Therefore, the embodiments can simplify the implementation of polar transmitters, provide reduced power consumption and clean on-channel signals.

[0029] Figure 1is a functional block diagram of an embodiment of a polar transmitter 100 according to the present inventive concept. In brief, the polar transmitter 100 PSK modulates an input bit stream by first generating a stream of I and Q symbol pairs ("IQ pairs") in an IQ space and pulse shaping the stream of IQ pairs (e.g., by low-pass digital filtering) to provide a filtered IQ waveform. The filtered IQ waveform is then modified to avoid zero-crossing regions by intermittently adding a zero-crossing avoidance signal (interchangeably, just "avoidance signal") to the IQ waveform. The avoidance signal may have a spectrum having at least a first tone and a second tone, wherein the first tone and the second tone define a first peak and a second peak on opposite sides of a zero frequency (DC) trough (see later discussed). Figure 8 ). A polarization signal (amplitude and phase) is then generated based on the modified IQ waveform, and the polarization signal is used to modulate the RF carrier. The unique characteristics of the zero-crossing avoidance signal facilitate the generation of the modified IQ waveform, and the corresponding polarization signal, having characteristics that result in improved error vector magnitude (EVM) compared to polar transmitters using conventional avoidance signal methods.

[0030] To this end, the input bit stream of the polar transmitter 100 is applied to a symbol mapper 110 that maps the bit stream to PSK symbols (i.e., IQ pairs). Below, an example of 8PSK modulation will be given to explain the concepts herein, but these concepts are applicable to any modulation scheme (e.g., 4PSK, 16PSK, etc.) that is expected to achieve zero crossing avoidance in IQ space. The IQ pairs can be generated in real time (with some processing delays) substantially synchronously with the bit stream, and thus may be referred to herein as a stream of symbol streams or IQ pairs.

[0031] In the symbol mapper 110, IQ pairs with π transitions can be detected because these symbols will likely cause transitions of the IQ waveform close to the origin. The symbol mapper 110 can generate a "π transition signal" indicating an IQ pair with π transitions, wherein the "π transition signal" is passed to the avoidance signal generator 140. The symbols are also passed to the pulse shaper 120, wherein a suitable pulse shape is applied to the symbol stream using digital filtering. The output of the pulse shaper 120 may be referred to herein as a pulse-shaped IQ waveform or a filtered IQ waveform, examples of which will be presented later. The filtered IQ waveform is provided to the delay element 130, wherein the filtered IQ waveform is delayed at the delay element 130 while the avoidance signal (if necessary) is generated by the avoidance signal generator 140. As shown in the path 122, the filtered IQ waveform may also be provided directly to the avoidance signal generator 140 so that the avoidance signal (as described later) can be generated with high accuracy.

[0032] The avoidance signal output of the avoidance signal generator 140 and the output of the delay element 130 are provided to the adder 150, wherein the adder 150 adds the delayed and filtered IQ waveform to the avoidance signal, thereby modifying the filtered IQ waveform. The latter is output to the IQ to polarization conversion block 160, wherein the IQ to polarization conversion block 160 converts the latter into a polarization signal having a time-varying amplitude A(t) component and a phase P(t) component. Since the modified IQ waveform can be provided in the form of digital samples, the IQ to polarization conversion block 160 may include a digital-to-analog (D / A) converter (not shown) to generate the polarization signal in analog form. The polarization modulator 170 modulates the RF carrier using the polarization signal and outputs the modulated RF signal carrying the bit stream information to the antenna for transmission to the receiving device.

[0033] Figure 2 is a functional block diagram of an example polar modulator 170. In this case, polar modulator 170 has a two-point PLL architecture and may include a power supply modulator 180, a phase locked loop (PLL) 190, a phase differentiator 172, an adder 174, a voltage controlled oscillator (VCO) 176, and a power amplifier 182. PLL 190 may include a divider control circuit 188, a divider 184, a phase frequency detector (PFD) and time to digital converter (TDC) block 186, and a divider loop filter 178.

[0034] The phase signal P(t) is applied to the phase differentiator 172 in the main path, wherein the phase differentiator 172 differentiates the phase signal P(t) to provide a time-varying frequency signal. The addition of the avoidance signal in the following IQ waveform modification scheme enables the frequency signal to exhibit a reduced offset because the phase changes are not as fast as they would otherwise be. The phase signal P(t) is also applied to the divider control circuit 188 of the PLL 190. The adder 174 adds the frequency signal from the phase differentiator 172 and the PLL feedback signal output by the divider loop filter 178, and provides the total output to the VCO 176 to control its oscillation frequency. In the main path, the output signal of the VCO 176 is applied to the power amplifier 182 as an input signal. In this example, the VCO output is fed directly to the power amplifier 182, but a driver amplifier (not shown) may be optionally included between the VCO 176 and the power amplifier 182. In the PLL feedback path, the VCO 176 output signal is fed back to a divider 184, where the divider 184 divides the frequency of the VCO 176 output signal (e.g., in half). The divider 184, controlled by a divider control circuit 188, sends the divided frequency output signal to a PFD and TDC block 186, where the PFD and TDC block 186 compares the frequency of the divided frequency output signal with the frequency of the PLL reference frequency applied to the PFD and TDC block 186. The PFD and TDC block 186 outputs a difference signal indicative of the difference between the two input signals to a divider loop filter 178, where the divider loop filter 178 filters the signal and provides it as a second input to the adder 174.

[0035] At the same time, the amplitude signal A(t) is sent to the power modulator 180, which modulates the power supply of the power amplifier 182 using the amplitude signal to produce amplitude modulation. The output of the power amplifier 182 is then sent to the transmit antenna. Timing alignment circuits (not shown) may be included in the AM and / or PM paths.

[0036] Other suitable configurations for the polar modulator 170 are available. For example, the polar modulator 170 can be optionally configured with an envelope elimination and restoration (EE&R) modulator architecture that does not modulate the VCO. In this design, which is often referred to as a Kahn EE&R transmitter, an envelope detector is used in the power supply modulator and the power supply to the power amplifier is modulated with a variable voltage based on the envelope detection. The phase signal modulates the RF carrier to provide a modulated RF signal that is applied to a limiter (without a PLL). The limiter converts the input to a constant envelope phase signal that is applied to the input of the power amplifier. This type of design exhibits similar Figure 2The two-point PLL architecture has similar power efficiency without the complexity of modulating the VCO.

[0037] Figure 1 and Figure 2 The block diagram of only shows the main components of the polar transmitter. Other components may be included as needed to fully realize a working transmitter. It is also noted here that the symbol mapper 110, the pulse shaper 120, the delay element 130, the adder 150, the avoidance signal generator 140, and the IQ to polarization conversion block 160 may each be a separate circuit. Each of these elements may therefore be referred to as a "circuit". Each of these circuits may be part of a processor that executes instructions read from a memory (not shown) to perform their specific functions. The processor may be a general-purpose processor or a special-purpose processor.

[0038] The polar transmitter 100 can be part of any electronic device that requires PSK modulation. It should also be noted that any PSK receiver compatible with the same protocol can be used to receive the RF modulated output of the polar transmitter 100. For example, if the polar transmitter 100 is a Bluetooth compatible transmitter, any conventional or non-conventional Bluetooth compatible receiving device can be used to receive the transmitted RF modulated signal.

[0039] Figure 3 is a signal diagram with example signals for explaining aspects of zero crossing detection, IQ waveform filtering, and polarimetric signal generation in polarimetric transmitter 100 . Figure 4 A progression of an example signal (unfiltered) in an IQ coordinate system ("IQ space") is shown. In this example, the input bit stream is divided into 3-bit sequences for 8PSK modulation. The input bit stream is mapped to symbols to produce a symbol stream of IQ pairs that can be understood as a time-varying IQ waveform (also interchangeably referred to as an IQ trajectory). Separate I and Q waveforms can also be considered.

[0040] Assume that at the initial time t synchronized with the transition of the data clock CLK-D CLK-0 Start receiving an input bit stream, where the first sequence of the bit stream is 101. Assume that the IQ phase constellation point is initially set to 0 degrees corresponding to the IQ pair coordinate (1,0). In the 8PSK modulation example, the 101 sequence is associated with a phase shift (phase increment) of -π / 2. Therefore, at time t after the end of three clock cycles of the bit stream, 1 , the constellation phase has shifted by -π / 2, resulting in a current value of -π / 2. (Due to the pulse shaper 120, the IQ / polarization "sample time" t 0 ,t 1 ,t 2 , ... are assumed to be the corresponding data clock transition times t from the end of the associated n-bit sequence CLK-0 ,tCLK-n ,t CLK-2n , ... delayed by the time of Delay DL, where n=3 in the current example. For illustrational simplicity, Delay DL is shown as being approximately half a symbol period (SP) in duration, but can be longer or shorter. In some cases, Delay DL can be in the range of 1 symbol to 5 symbols). Figure 4 As the path starts at time t 0 The IQ coordinates of the time t 1 The corresponding unfiltered and uncorrected I and Q waveforms are shown in Figure 3 The next bit sequence 001 corresponds to a phase change of π / 4, causing the 2 The transformation of the coordinates (0.717, 0.717).

[0041] The next bit sequence 110 is associated with a phase change π, which changes the phase constellation from time t 2 The -π / 4 is moved to the time t with IQ coordinates (-0.717, 0.717) 3 As seen in IQ space, this results in the trajectory passing directly through the origin of IQ space at (0,0), i.e., at time t ZC (Approximately at time t 2 and t 3 As previously described, if the zero-crossing event is brought into polar coordinates and applied to the polar modulator, performance degradation may result due to the inability to achieve the required large instantaneous frequency or the inability to synchronize the frequency offset with the amplitude modulation very accurately. The performance degradation may be error vector magnitude and adjacent channel interference. When the polar transmitter 100 bypasses the zero-crossing event, these potential problems are prevented. The symbol mapper 110 of the polar transmitter 100 detects the π transition signal and outputs the π transition signal to the avoidance signal generator 140 when detected, so that the avoidance signal generator 140 generates the π transition signal at time t. 2 and t 3 The avoidance signal is generated between the two to modify the IQ trajectory so as to avoid crossing the origin (as discussed below). Figure 5B However, before applying the avoidance signal, the pulse shaper 120 filters the IQ waveform to smooth out sharp transitions. Figure 3 The filtered I waveform F is shown i A portion of the filtered Q waveform F q Without the avoidance signal, the filtered IQ waveform (F i , F q ) may still pass through time t ZC The origin is (0,0).

[0042] like Figure 3 As shown, if the IQ pair is converted into a polarization signal without avoiding signal correction, the polarization amplitude A(t) at time t ZC will be zero. That is, the polarization amplitude 1 (equal to (I 2 +Q 2 )^1 / 2) quickly changes to 0 and then back to 1. The polarization signal amplitude waveform "A" corresponding to the filtered IQ waveform f ” also transitions from 1 to 0 (or close to zero) and then back to 1. Similarly, the unfiltered polarization phase P(t) undergoes an instantaneous 180 degree transition in the unfiltered case, but the phase trajectory can be smoothed by filtering by the pulse shaper 120 to provide a smoother phase waveform P f Although pulse shaping in the IQ domain limits the bandwidth, an instantaneous or near-instantaneous 180 degree transition can be made in the filtered waveform. This is due to the nonlinear nature of the conversion from Cartesian to polarization, which extends the bandwidth of both the amplitude and polarization paths.

[0043] Figure 5A is a description of the pulse shaper 120 output having the same Figure 3 The bitstream shown in Figure 4 The filtered IQ waveform (F i , F q ). Figure 5A As can be seen in FIG, filtering serves to smooth the sharp transitions between phase shifts and also serves to slightly reduce the amplitude of the IQ pair at the extremes. ZC Indicates the time at which the filtered waveform passes through or is closest to the origin of IQ space.

[0044] Figure 5B Is a description Figure 5A 1 is a graph of an example of a filtered IQ waveform after being corrected by the zero crossing avoidance signal generated in the avoidance signal generator 140. The corrected waveform (CF i , C.F. q ) is shaped into ZC Passing through the outside of a predetermined hole H centered at the origin (0,0) or passing along the periphery of the predetermined hole H. In addition, the overall trajectory is usually time t 2 and t 3 The zigzag trajectory between. Figures 8 to 11 An example is provided of how the avoidance signal and the corrected IQ waveform may be calculated.

[0045] Figure 6 An example comparison of a filtered Cartesian waveform and a Cartesian waveform after avoiding signal correction is shown in a graph of amplitude versus time. Figure 3 multiple symbol periods SP (wherein the time interval between the midpoints of adjacent symbols (t 1 -t 0 )、(t 2 -t 1 ), each time interval in ... is a symbol period SP) is shown on Figure 3 The filtered I waveform F i and the filtered Q waveform F q (referred to as a Cartesian signal). The waveform F of the output signal of the pulse shaper 120 may be i 、F q An exemplary corrected waveform CF which may be the output signal of adder 150 is i CF q Avoiding the I and Q components of the signal AS (ie, the signal AS i and AS q ) is shown at a scale of 10 times relative to the filtered and corrected waveform. The signal component AS will be avoided i ,AS q With the filtered signal waveform F i 、F q Add together to get the corrected waveform CF i CF q It can be observed that at the zero-crossing time t ZC Nearby waveform (CF i Comparison between Fi and CF q Contrast F q ), which results in IQ trajectories like Figure 5B Note that in this example, the avoidance signal AS is applied for more than one symbol period SP. Specifically, in this example, the avoidance signal AS is applied for approximately 1 1 / 2 symbol period, for example, from time t 2 About 1 / 4 symbol period before (time (t 2 -SP / 4)) to time t 3 After about 1 / 4 symbol period (time (t 3 +SP / 4)). In other cases, the avoidance signal may be applied for a shorter or longer duration, which may depend on optimization criteria. Typically, when applied to avoid zero crossings for isolated π shifts, the avoidance signal may be applied for a duration in the range of 1-2 symbol periods centered around the symbol period associated with the zero crossing. Figure 6A horizontal axis numbered from 0 to 70 is also shown to illustrate that the I and Q waveforms can be digitally calculated and stored as time samples. The samples are converted to polar signal samples by the IQ to polar conversion block 160, where the polar signal samples are D / A converted to output analog polar signals.

[0046] Figure 7 Shown based on Figure 6 The I waveform and Q waveform samples generate an exemplary polarization signal. Specifically, Figure 3 The filtered polarization signal (by amplitude A f and phase P f ) and the polarization signal after filtering and signal correction (represented by the amplitude CA f and phase CP f It can be seen that at the zero crossing time t ZC Nearby Amplitude CA f Higher than A f , and the waveform CP f These slight changes are used to produce significant performance improvements in EVM and frequency offset, which will be detailed below.

[0047] Figure 8 is a graph showing a frequency spectrum comparison between a "filtered pulse" type avoidance signal and an example "multi-tone" avoidance signal in accordance with the present inventive concepts. Fig. 9 A time domain comparison of these two types of avoidance signals is shown. First, consider Figure 8 The frequency characteristics of the different avoidance signal types are illustrated in each case as a graph of DFT (Discrete Fourier Transform) amplitude versus frequency. The DFT amplitude at a discrete frequency is understood to be a calculated amount of spectral energy at that frequency. The conventional filtered pulse avoidance signal 815 is at the zero crossing time t ZC (like Fig. 9 The filtered IQ waveform is pushed out at zero crossing with a single pulse at a frequency (as shown) and gradually decreases to a signal of zero energy over time on both sides of the pulse. Avoidance signal 815 has a "mountain-shaped" frequency characteristic with peak spectral energy 813 at 0 Hz ("DC", and is an example of a center-band frequency) and gradually decreases symmetrically with frequency. When the spectrum is later applied to modulate the carrier during polar modulation, the carrier is (partially) shifted with the positive and negative frequencies of avoidance signal 801, so that Figure 8 The 0Hz in can be understood as being consistent with the carrier frequency.

[0048] The example multi-tone avoidance signal 801 may have a spectrum having a local null value N at 0 Hz, a local zero value N at positive frequencies f 1There is a first tone T1 at a position (wherein the first tone T1 has a DFT level higher than the zero value N level), and at a negative frequency -f 1 There is a second tone T2 at (wherein the second tone T2 also has a level higher than the zero value N level). Figure 8 In the example of , both tone T1 and tone T2 have energy levels approximately three times the null value N at DC, but other ratios are possible. The multi-tone avoidance signal 801 may be higher than f 1 The frequency f 2 The third tone T3 has a level higher than the zero value N level (and may be higher than the T1 level). The avoidance signal 801 may be at a frequency -f 2 There is a fourth tone T4 at , wherein the fourth tone T4 has a level equal to the level of T3. Figure 8 In the example of , tone T3 and tone T4 have levels between three and four times the level of the null value N, but other ratios may be set.

[0049] In an alternative embodiment, the avoidance signal 801 may include only the higher tone T3 and tone T4 and omit tone T1 and tone T2 such that the spectrum includes a null value N and rises linearly to the level of T3 on the positive frequency side and to the level of T4 on the negative frequency side. In either embodiment, the null value N at 0 Hz rising to the higher level tone on both sides of the spectrum defines a general characteristic of having a center frequency trough. Thus, in either embodiment, the avoidance signal 801 may have a spectrum including at least a first tone and a second tone, wherein the first tone and the second tone define a first peak and a second peak on opposite sides of the center frequency trough.

[0050] Tones T1 through T4 may all be "out-of-band" tones. A signal bandwidth B may be specified for the transmitted RF modulated signal. In an example, the signal bandwidth B may be determined as (1 / symbol period). At a receiving device that receives the RF signal, a bandpass filter may be used to recover information carried only by spectral energy within bandwidth B. Thus, frequency f 1 、-f 1 、f 2 and -f 2 Each of may be outside the range of -B / 2 to B / 2. For example, in a Bluetooth protocol application, a narrowband signal channel bandwidth B of + / -0.5 MHz is established for each point-to-point communication (note also that the symbol period can be 1 microsecond, so that 1 / symbol period = 1 MHz). In this case, B / 2 = 0.5 MHz, and the frequency f 1 and f 2All are above 0.5 MHz. In a specific example, the frequency set (f1, -f1) is + / -1 MHz and (f2, -f2) is + / -2 MHz. In another example implementation of Bluetooth, tone T1 and tone T2 are omitted, and the frequency set (f2, -f2) of tones (T3, T4) is + / -1 MHz. The latter design is particularly advantageous for higher power Bluetooth applications.

[0051] It has been observed that while the filtered pulse type avoidance signal limits the bandwidth of the zero crossing avoidance, it results in EVM degradation. On the other hand, the multi-tone avoidance signal contemplated by the present invention has been found to cause minimal EVM degradation, especially when the tones are all placed outside the signal bandwidth.

[0052] Fig. 9 FIG. 1 shows a time domain comparison of a conventional filtered pulse avoidance signal and a multi-tone avoidance signal according to the present invention. In this example, the filtered pulse avoidance signal 915 is compared to Figure 8 The signal 815 corresponds to the signal 815 , and the multi-tone avoidance signal 901 corresponds to the signal 801 .

[0053] As shown, the filtered pulse signal 915 has a zero-crossing time t ZC (indicated as 0 μs) has a peak amplitude near t ZC This produces a bubble-shaped IQ waveform trace near the zero crossing with minimal change to the rest of the trace.

[0054] For each zero crossing event, the multi-tone avoidance signal 901 may be applied for a time window having a duration optimized for a particular application (the avoidance signal 901 may also be considered to have zero amplitude throughout the time periods between the windows to which it is applied). For example, the avoidance signal 901 may be applied for a duration in the range of (1-2) symbol periods or (1 / 2-2) symbol periods centered around the symbol period containing the zero crossing. Fig. 9 For example, for a symbol period of 1 μs, avoid applying a signal of approximately 1 1 / 2 The optimized time for one symbol period (from -0.75 μs to +0.75 μs) can be used. By applying the avoidance signal longer than one symbol period, a long window function can be used, allowing more filtering of the tones. This limits the spread of their energy to the signal bandwidth or higher frequencies. For example, referring to Figures 3 to 7 The bit stream and the mapped symbol timing example can be avoided from the signal start time (AS start time) (t 2 -SP / 4) to avoid signal end time (AS end time) (t 3+SP / 4). In general, avoidance signal 901 may be applied from approximately the π transition start time to the π transition end time, which may be said to correspond to a portion of the filtered IQ waveform that is substantially consistent with the π phase shift.

[0055] The multi-tone avoidance signal 901 is at time t ZC There is a peak energy "PL" near the zero crossing point, which is referred to as the "main peak" (note that Figure 1 The delay element 130 allows the avoidance signal 901 to be added to the filtered IQ waveform at the correct timing so that at time t ZC The rest of the trace as a function of time differs from the filtered pulse signal 915 by including a first trough region 912 between the AS start time and the main peak and a second trough region 914 between the main peak and the AS end time. The first trough region 912 is at time t μ1 The second valley region 914 has a first valley zero value at time t, which has a level N1 lower than the level S1 at the start time of AS. μ2 has a valley zero value, which has a level N2 lower than the level S2 at the end time of AS. In the example, both N1 and N2 have a voltage polarity opposite to the voltage polarity of PL. This makes the filtered IQ waveform at the time t ZC 912 and 914 are pushed out in the opposite direction (as will be seen later). Fig.12 This type of energy distribution results in benefits such as improved EVM compared to traditional filtered pulse schemes.

[0056] exist Fig. 9 In the specific example, the peak level PL is approximately 1 unit of amplitude; the trough zero value levels N1, N2 are both between -0.2 units and -0.4 units; and the start time level S1 and the end time level S2 are both between 0 units and -0.1 units.

[0057] Fig.10 1 is a flow chart of an example method 1000 for digital modulation performed by a polar transmitter 100 according to the present inventive concept. A bit stream is received (1002) and a stream of IQ symbol pairs representing a continuous n-bit sequence of the bit stream is calculated by the symbol mapper 110 (1004). The stream of IQ symbol pairs is low-pass filtered by the pulse shaper 120 to obtain a filtered IQ waveform (interchangeably referred to as a filtered IQ trace) (1006). Here, the symbol mapper 110 may use conventional low-pass digital filtering techniques to calculate and store a sequence of discrete samples representing a time-varying IQ trace, such as Figure 6Waveform F i and F q Evenly spaced samples.

[0058] The IQ filtered waveform is then modified by intermittently adding the zero crossing avoidance signal to the filtered IQ waveform using the avoidance signal generator 140, the delay element 130 and the adder 150 (1008). Figure 8 As explained, the zero crossing avoidance signal according to the present invention can have a spectrum with a zero frequency (DC) null value between the first peak and the second peak, and can be generated and added only near the time when the π phase shift transition is detected. Then, based on the modified IQ filtered waveform, a polarization signal (1010) is generated by the IQ to polarization conversion block 160. The RF carrier is then modulated by the polar modulator 170 and output to the antenna (1012) for wireless transmission, wherein the modulated output carries information of the bit stream.

[0059] Fig.11 is a flow chart of an example method for modifying a filtered IQ waveform by adding a zero crossing avoidance signal according to the present invention. Specifically, Fig.11 It is used to realize Fig.10 An example method of operation 1008 is provided. Fig.11 The operation will refer to Fig. 9 and Fig.12 An explanation is given, wherein the latter shows example traces of a filtered IQ waveform and a signal-avoided corrected (and filtered) IQ waveform.

[0060] The π transition is detected by the symbol mapper 110 (1102), whereupon the avoidance signal generator 140 initiates the generation of the avoidance signal. The avoidance signal may be based on a signal directly from the pulse shaper 120 (along the Figure 1 To this end, for a π transition, the avoidance signal generator 140 may select two intermediate transition samples (1104) to calculate the vector from the origin to the closest point of the trajectory (1106). This can be done assuming a linear trajectory through the two samples. For example, Fig.12 As shown, the filtered IQ waveform (trace) 1310 (represented as dotted line samples) in the IQ space can be obtained at the starting time t 2 π transition begins at time t ZC Close to the origin (0,0), and at the end time t 3 The two intermediate transition samples k1 and k2 can be used to calculate the vector (Notice, Fig.12The π transition of π starts near (1,0) corresponding to phase = 0 and ends near (-1,0) corresponding to phase = π. Figures 3 to 7 The π transformation is different).

[0061] As two intermediate transition samples (1104) are selected to calculate the closest vector This can be achieved in different ways. For highly oversampled systems this can be done from the nearest point, but it can also be calculated from knowledge of the pulse shape and the sign before and after the π transition.

[0062] Assume that the origin is centered with a value designated as zc limit The method can then calculate the magnitude of the avoidance signal (1108). ZC The desired avoidance vector at The amplitude of mag It can be calculated as:

[0063]

[0064] Among them, time t ZC Avoid vectors at Can be pointing and vector vectors with the same direction. Fig.12 In FIG. 1 , waveform 1312 represents the filtered IQ waveform with the avoidance signal modification (correction). It can be seen that the uncorrected waveform 1310 is along the origin. The direction of is pushed out near the origin, and at time t μ1 and t μ2 The vicinity is pushed out in the opposite direction. Fig. 9 As will be appreciated in , this is because the troughs 912 and 914 of the avoidance vector are arranged to have opposite polarity to the main peak.

[0065] For vector-based Calculate from the start time (for example, t 2 ) to the end time (e.g., t 3 ), the avoidance signal can then be calculated (1110) as:

[0066]

[0067] Among them, "tone_template" can be fixed for all correction signals. Generally, tone_template can be determined as follows: Assuming that the vector is avoided has a value at each of the n samples starting from the (-n / 2)th sample to the (n-1) / 2th sample (including the sample at n=0) (here, the number of samples "n" should not be confused with the "n-bit data" discussed above). The Tone_template for the entire range of n samples can then be determined as

[0068] tone_template(n)=(h(n) / (k1+k2))x(k1cos(w1 x 2πx n) / (OSR))

[0069] +(k2cos(w2 x 2πx n) / (OSR))

[0070] Where OSR is the oversampling rate = number of samples per symbol, i.e., IQ versus duration (e.g., from time t 2 To time t 3 duration); h(n) is the filter window function, k1, w1, w2 and k2 are constants.

[0071] In the above equation, the term (k1cos(w1 x 2πx n) / (OSR)) can be understood as producing Figure 8 The spectrum of the example tones T1 and T2. (k2cos(w2 x 2πx n) / (OSR)) can be understood as producing Figure 8 The variables w1 and w2 determine the shape of the trajectory. For example, consider Fig. 9 , if w1 and w2 are set to be greater than the values ​​used to generate waveform 901, multiple "trough zero values ​​and peaks" can be generated on each side of the main peak.

[0072] Pre-computed values ​​for each sample number n of tone_template may be stored in a lookup table (eg, resident in memory within avoidance signal generator 140), or alternatively, a tone_template value may be generated for each correction signal.

[0073] The filter window function h(n) may be a Hamming window function. Some examples of alternative window functions include a Hann window, a Blackman window, a Gaussian window, a Nutall window, a Blackman-Nutall window, and a Blackman-Harris window.

[0074] exist Figure 8 and Fig. 9 In the specific example, tone_template can be calculated from the following pseudo code: for n = -12:11

[0075]

[0076] End

[0077] For this pseudocode, the variable n represents the sample number, and h is a Hamming window function of length 24. Therefore, there are 24 samples, i.e., from n = -12 to +11 (where 0 is included as a sample). In this example, the sampling rate is 16 (e.g., for a 1 MHz symbol rate, 16 samples per symbol = 16 samples per MHz). If a different oversampling rate is desired, these variables can be changed appropriately.

[0078] The above example assumes the use of two tone sets, but in other embodiments more than two tone sets may be generated. Therefore, tone_template may generally be determined by the following equation:

[0079]

[0080] where h(n) is a suitable index window function, p is the number of tones, and w i Can be any number greater than OSR / 2.

[0081] A Hamming window function or other suitable window function may be used to prevent spectral leakage from the avoidance signal from increasing to undesirable levels that may degrade both EVM and ACP (adjacent channel power).

[0082] In avoiding vector After being calculated, it can be added to the corresponding portion of the IQ samples (1112) to obtain a corrected IQ waveform (e.g., Fig.12 As mentioned, since the desired shape for the avoidance vector includes troughs 912 and 914 ( Fig. 9 ), so by moving along direction is pushed out, and at the intermediate time t μ1 and t μ2 Pushed in the opposite direction (to a lesser extent), the corrected waveform 1312 may differ from the uncorrected waveform 1310. The simulations summarized below demonstrate the performance benefits by implementing this type of avoidance signal waveform.

[0083] Please note here, Fig.11 The method of may be modified in the following manner: Instead of the π transition detection in operation 1104 by the symbol mapper 110 (which is described as detecting the π transition and providing the π transition signal to the avoidance signal generator 140), the detection of the π transition can alternatively be performed by directly observing (by the avoidance signal generator 140) the amplitude of the IQ data output from the pulse shaper 120 along the path 122. If the IQ data represents <zclimit , we can determine at which point a zero crossing avoidance signal is needed and can construct it based on this.

[0084] Embodiments of the inventive concept as described above may advantageously allow the necessary frequency from the PLL 190 for a particular specification such as Bluetooth EDR 3 to be reduced without being replaced by standard filtered pulses (e.g., Figure 8 Spectrum 815 or Fig. 9 The following Table 1 compares the simulation results of the multi-tone avoidance signal embodiment described above (denoted as "out-of-band") with the simulation results of the standard filtered pulse implementation of zero crossing avoidance. In this comparison, Bluetooth EDR 3 parameters are assumed, which include a channel bandwidth of 1 MHz, a Bluetooth-compatible data rate, and an RF frequency in the range of 2.4 GHz-2.48 GHz. In Table 1, "Baseline" represents a polar transmitter architecture without any avoidance signal correction. "Pulse Size" represents the size of the radius of the circular hole H around the origin, where the avoidance signal is assumed to pass tangentially through the circumferential point of the hole H in all cases (excluding the baseline case). "Out-of-band" represents the use of a filter with a filter such as Figure 8 The spectrum of the multi-tone signal shown in the polarization avoidance transmitter.

[0085] The maximum frequency offset is the maximum deviation of the VCO output frequency from the nominal carrier frequency for a given sample. In a polar system, the VCO frequency is updated at the reference frequency of the PLL. Therefore, without zero crossing avoidance (baseline case) and π transition, the maximum offset is 7.45 MHz, which is close to ref_freq / 2 (ref_freq=16 MHz in Table 1). In this example, it is expected to reduce the maximum frequency offset to below 7.45 MHz of the baseline design. It can be seen in Table 1 that for various pulse sizes, both the out-of-band case and the filtered pulse case significantly reduce the maximum frequency offset to approximately the same level, but the out-of-band case exhibits significantly better EVM performance. Therefore, a reduction in the maximum frequency offset is obtained without excessive EVM degradation as in the filtered pulse case.

[0086] Table 1

[0087]

[0088]

[0089] It is noted here that an additional advantage of the out-of-band configuration is that it reduces the requirements for timing alignment between the AM path and the PM path of the polar modulator 170.

[0090] Fig.13A graph is shown depicting the maximum frequency offset for the baseline case ("raw IQ") relative to the out-of-band case ("final IQ") of Table 1. It is seen that in the baseline case, the frequency offset has a sharp peak over one symbol duration, while the zero-crossing avoidance implementation of the inventive concept greatly limits this peak.

[0091] Fig.14 1 is a graph showing simulation results of the power spectral density of the amplitude path versus frequency for the exemplary cases in Table 1. Curve 1402 representing a pulse of 0.0 corresponds to the baseline case. Pulses of 0.05, 0.10, and 0.15 (corresponding to curves 1404, 1406, and 1408, respectively) are candidates for out-of-band embodiments. It can be seen that each out-of-band embodiment exhibits a reduction in signal bandwidth on the amplitude path. As shown, the amount of signal at higher frequencies is reduced, which reduces the bandwidth requirements of the path and makes the configuration easier to manufacture.

[0092] Fig.15 1 is a functional block diagram of an example electronic device 1500 including the polar transmitter 100 described above. The electronic device 1500 may be any electronic device that may benefit from including a PSK modulator with reduced power consumption. Examples include any Bluetooth-compatible device, a smart phone, a tablet, a laptop, etc. The electronic device 1500 has an example configuration, wherein in the example configuration, the processor 1506 includes each of the symbol mapper 110, the pulse shaper 120, the delay element 130, the adder 150, and the avoidance signal generator 140. The IQ to polar conversion block 160 is illustrated as including a digital IQ to polar portion 160a and a D / A converter portion 160b as part of the processor 1506, wherein the D / A converter portion 160b converts a digital sample of a polar signal output by the IQ to polar portion 160a into an analog polar signal provided to the polar modulator 170. The bit stream source 1504 may provide the above bit stream to the symbol mapper 110. The processor 1506 is coupled to the memory 1502, wherein the memory 1502 may store instructions executed by the processor 1506 to perform the above-mentioned functions of the various components. The memory 1502 may be used for temporary storage of any component of the processor 1506, and may also store the above-mentioned lookup table (if included) for storing tone_template. The polar modulator 170 provides the modulated RF signal to the antenna 1510. Except for the bit stream source 1504 and the antenna 1510, all the elements shown in the electronic device 1500 may be part of the polar transmitter 100. The electronic device 1500 may include many other elements (not shown) to perform various functions, such as a PSK receiver for wirelessly receiving a PSK signal transmitted by a counterpart device and demodulating the PSK signal transmitted by the counterpart device.

[0093] The various elements described above have been described with labels consistent with their functions, but are implemented using hardware circuits. In this regard, elements such as the symbol mapper, pulse shaper, delay element, adder and avoidance signal generator, IQ to polarization conversion block, phase differentiator, divider, PFD and TDC block, and divider loop filter described above may be alternatively referred to as symbol mapper circuit, pulse shaper circuit, ..., divider loop filter circuit, respectively.

[0094] Exemplary embodiments of the inventive concept have been described herein with reference to signal arrows, block diagrams, and algorithmic representations. Each block of the block diagram and the combination of blocks in the block diagram and the operation according to the algorithmic representation can be implemented by hardware accompanied by computer program instructions. Such computer program instructions may be stored in a non-temporary computer-readable medium (such as a CDROM, RAM, floppy disk, hard disk, or magneto-optical disk), or a computer code downloaded over a network, so that the method described herein can be presented using such software stored on a recording medium using a general-purpose computer or a dedicated processor or stored in programmable or dedicated hardware (such as an ASIC or FPGA) that can guide a computer, other programmable data processing equipment, or other device to run in a specific manner, so that the instructions stored in the computer-readable medium produce an article of instruction including instructions for implementing the functions / actions specified in the flowchart and / or block / schematic diagram.

[0095] The term "processor" as used herein is intended to include any processing device, such as, for example, a processing device including a central processing unit (CPU) and / or other processing circuits (e.g., a digital signal processor (DSP), a microprocessor, etc.). In addition, a "processor" includes computing hardware and may refer to a multi-core processor including multiple processing cores in a computing device. Various elements associated with a processing device may be shared by other processing devices.

[0096] While the inventive concepts described herein have been particularly shown and described with reference to example embodiments of the inventive concepts, it will be understood by those skilled in the art that various changes in form and details may be made without departing from the spirit and scope of the claimed subject matter as defined by the appended claims and their equivalents.

Claims

1. A method for modulating a bit stream, comprising: generating, with a processor, a filtered IQ waveform representable in IQ space based on a stream of in-phase I and quadrature-phase Q symbol pairs, wherein each symbol pair represents a portion of a bit stream; modifying, using a processor, a filtered IQ waveform to avoid zero crossing regions in the IQ space by intermittently adding a zero crossing avoidance signal to the filtered IQ waveform, wherein the zero crossing avoidance signal has a frequency spectrum including at least a first tone and a second tone, wherein the first tone and the second tone define a first peak and a second peak on opposite sides of a center frequency trough; generating a polarization signal including polarization amplitude and phase based on the modified filtered IQ waveform; and The polarization signal is used to modulate the radio frequency (RF) carrier.

2. The method of claim 1, wherein: The steps to intermittently add a zero-crossing avoidance signal include: Each time the portion of the bit stream matches a bit sequence for generating a 180 degree phase shift of phase shift keying (PSK), a zero crossing avoidance signal is generated and added to the filtered IQ waveform.

3. The method of claim 1, wherein: The center frequency trough is at 0 Hz, the first tone is at frequency "f", and the second tone is at frequency -f.

4. The method of claim 3, wherein: A frequency bandwidth B is allocated to the modulated RF carrier; and The frequency f is greater than B / 2.

5. The method of claim 4, wherein: The spectrum includes a third tone between frequency B / 2 and frequency f, and a fourth tone between frequency -B / 2 and frequency -f.

6. The method of claim 5, wherein: B / 2 equals 0.5 MHz, f equals 1.0 MHz, and the third tone is at 2.0 MHz.

7. The method of claim 1, wherein: The RF carrier is a Bluetooth EDR 3 compatible RF carrier.

8. The method of claim 1, wherein: The zero-crossing avoidance signal is intermittently applied in a plurality of windows, wherein each of the plurality of windows has a duration within a range of (1 / 2-2) symbol periods centered around a symbol period containing a zero-crossing region in the IQ space.

9. The method of claim 1, wherein: The zero crossing avoidance signal is intermittently applied in a plurality of windows, wherein each window in the plurality of windows has a center at a symbol period containing a zero crossing region in the IQ space. The duration of a symbol period.

10. The method of claim 1, further comprising calculating the amplitude of the zero crossing avoidance signal by the following steps: Define a radius zc around the origin of the IQ space limit The circular hole; Calculate the vector from the origin of the IQ space to the closest point to the filtered IQ waveform And the amplitude s of the zero-crossing avoidance signal at the closest point is calculated as follows mag , 11. The method of claim 10, wherein: The zero-crossing avoidance signal is defined as an avoidance vector having a value at each of the n samples starting from the (-n / 2)th sample to the ((n-1) / 2)th sample. The n samples include the sample at n=0, wherein the vector is avoided is defined as in, tonetemplate=(h(n) / (k1+k2))x(k1cos(w1 x 2πx n) / (OSR))+(k2cos(w2 x 2πxn) / (OSR)) Where OSR is the oversampling rate, which is equal to the number of samples per symbol; h(n) is the filter window function, and k1, w1, w2 and k2 are constants.

12. The method of claim 11, wherein: h(n) is the Hamming window function.

13. The method of claim 10, wherein: The zero-crossing avoidance signal is defined as an avoidance vector having a value at each of the n samples starting from the (-n / 2)th sample to the ((n-1) / 2)th sample. The n samples include the sample at n=0, wherein the vector is avoided is defined as in, Wherein, h(n) is an index window function, p is the number of tones of the at least first tone and second tone, k i is a constant, OSR is the oversampling ratio, w i is a number greater than OSR / 2.

14. A polarization transmitter, comprising: a symbol mapper circuit configured to receive an input bit stream and generate a stream of in-phase I and quadrature-phase Q symbol pairs to obtain an IQ waveform, wherein each symbol pair represents a portion of the bit stream; a pulse shaper circuit configured to low pass filter the IQ waveform to obtain a filtered IQ waveform representable in an IQ space; a zero-crossing avoidance signal generator circuit configured to generate a zero-crossing avoidance signal, wherein the zero-crossing avoidance signal is applied during a window containing a π phase shift transition and has characteristics defined by a main peak at a zero-crossing time coinciding with a zero crossing of the filtered IQ waveform, a first trough at a time between a start time of the window and the zero-crossing time, and a second trough at a time between the zero-crossing time and an end time of the window; a delay circuit configured to delay the filtered IQ waveform to provide a delayed IQ waveform; an adder circuit configured to add the zero-crossing avoidance signal to the delayed IQ waveform to obtain a modified IQ signal; An IQ to polar conversion circuit configured to convert the modified IQ signal into a polar signal; and The polarization modulator circuit is configured to adjust a radio frequency (RF) carrier using a polarization signal.

15. The polar transmitter according to claim 14, wherein: The first wave valley has a first zero position value at a first zero position value level, the second wave valley has a second zero position value at a second zero position value level, the main wave peak represents a first voltage polarity, and each of the first zero position value level and the second zero position value level represents a second voltage polarity opposite to the first voltage polarity.

16. The polar transmitter according to claim 14, wherein: The RF carrier is a Bluetooth EDR 3 compliant RF carrier.

17. The polar transmitter according to claim 14, wherein: The polarization signal includes an amplitude signal and a phase signal, and the polarization modulator circuit includes: Power amplifier; a power modulator configured to modulate a power supply of the power amplifier using the amplitude signal; a phase differentiator configured to differentiate the phase signal into a frequency signal; A voltage controlled oscillator VCO is coupled to the input terminal of the power amplifier; an adder coupled between the phase differentiator and the VCO; and a phase-locked loop PLL configured to receive a phase signal as a first input, an output signal of the VCO as a second input, a PLL reference signal as a third input, and generate a PLL feedback signal therefrom; Therein, the adder adds the frequency signal and the PLL feedback signal to obtain an added signal which is applied as an input to the VCO.

18. The polar transmitter according to claim 14, wherein: The polar modulator circuit has an envelope cancellation and restoration modulator architecture.

19. The polar transmitter according to claim 14, wherein: The symbol mapper circuit, the pulse shaper circuit, the zero-crossing avoidance signal generator circuit, the delay circuit, the adder circuit, and a portion of the IQ to polarization conversion circuit are implemented by a general purpose processor that reads and executes instructions from a memory.

20. A non-transitory computer-readable recording medium storing instructions, the instructions, when executed by at least one processor, implementing a method for modulating a bit stream, the method comprising: generating a filtered IQ waveform in IQ space based on a stream of in-phase I and quadrature-phase Q symbol pairs, wherein each symbol pair represents a portion of a bit stream; modifying, using the at least one processor, the filtered IQ waveform to avoid zero crossing regions in the IQ space by intermittently adding a zero crossing avoidance signal to the filtered IQ waveform, wherein the zero crossing avoidance signal has a frequency spectrum including at least a first tone and a second tone, wherein the first tone and the second tone define a first peak and a second peak on opposite sides of a center frequency trough; generating a digital polarization signal including a digital polarization amplitude and phase based on the modified filtered IQ waveform; and The digital polarization signal is output to a digital-to-analog converter so as to subsequently modulate a radio frequency (RF) carrier using the analog polarization signal.

Images