Differential Phase Shift Keying (DPSK) Receiver for Adapting to Transmitter Defects and Methods Performed by the DPSK Receiver

Abstract

This invention discloses a DPSK receiver for adapting to the defect of undesirable carrier frequency variation over time in the presence of a transmitter. The aim of providing an apparatus and method for adapting the receiver to transmitter defects is thus achieved. The frequency offset estimation unit includes a first phase estimation unit, a second phase estimation unit, and a clean / dirty selector, wherein a first phase error estimate e is output. dirty (k) The first phase estimation unit includes a first frequency offset estimation low-pass filter, the bandwidth of which is wider than the bandwidth of the second frequency offset estimation low-pass filter in the second phase estimation unit. The second phase estimation unit outputs a second phase error estimate e. clean (k), the dirty / clean selector is configured to compare the ratio of the second phase error estimate to the first phase error estimate with a threshold T and output a binary comparison result c(k). The demodulation reference estimation unit includes a first filter and a second filter, wherein the filter bandwidth of the first filter, which is part of the “dirty” reception path of the receiver, is wider than the filter bandwidth of the second filter, which is part of the “clean” reception path of the receiver. If the ratio of the second phase error estimate to the first phase error estimate is lower than the threshold T, the “clean” reception path is used to output the received information bits; otherwise, the “dirty” reception path is used.

Description

Technical Field

[0001] This invention relates to a differential phase shift keying (DPSK) receiver for adapting to the defect of undesirable carrier frequency variation over time in the presence of a transmitter.

[0002] The present invention also relates to a method executed by a DSPK receiver. Background Technology

[0003] Overall, the receiver of a modem utilizes a single parameter estimation and demodulation path with fixed parameter settings to determine the received data. Compared to theoretical limits, by selecting the architecture and parameterization, the receiver performs well when receiving signaling packets from so-called "dirty" transmitters, while its performance is inherently limited when receiving signals from "clean" transmitters.

[0004] Transmitters used for wireless data transmission in modems according to the Bluetooth (BT) standard (see Bluetooth Core Specification, Revision 5.2, Bluetooth SIG, December 31, 2019) can behave "dirty," implying that their carrier frequency may oscillate during packet transmission. The aforementioned standard defines the limits of the nominal carrier frequency's deviation over time, and the worst-case transmitter model with sinusoidal frequency modulation of a given amplitude and period. Transmitter implementations may behave as badly as defined, or even better. This provides room for low-complexity / low-cost transmitter implementations that reduce throughput performance. However, advanced transmitter implementations may behave completely "clean," meaning their carrier frequency may be considered stable over the duration of a packet, limited only by the phase noise of the local oscillator, which does not impair transmission. The Bluetooth Test Specification defines receiver performance tests using the aforementioned worst-case "dirty" transmitter; see "RF Bluetooth Test Specification, Revision RF.TS, page 30, Version 2, Bluetooth SIG, January 27, 2020."

[0005] The requirement to support "dirty" transmitters influences design choices in receiver implementations. A receiver optimized for a "clean" transmitter may fail to receive any packets from a "dirty" transmitter. On the other hand, a receiver optimized for a "dirty" transmitter can still receive from a "clean" transmitter, but with an increased probability of error compared to the theoretical limit.

[0006] The lack of error rate optimization when using a receiver optimized for a "dirty" transmitter to receive data packets from a "clean" transmitter is particularly evident in Bluetooth Enhanced Data Rate operating modes, which utilize the Serial Differential Phase Shift Keying (DPSK) transmission scheme.

[0007] The standard defines 2 Mbits / s and 3 MBits / s modes using 4-DPSK and 8-DPSK alphabets, respectively. Figure 1 The packet structure for these patterns is illustrated. The first part of the packet is a preamble modulated using Gaussian Frequency Shift Keying (GFSK), and the second part utilizes Differential Phase Shift Keying (DPSK) modulation. The second part begins with a Sync field, which is a sequence of DPSK symbols known at the receiver, followed by an unknown symbol selected at the transmitter based on the payload information, and a tail. The tail is irrelevant to this invention.

[0008] Figure 2 The general structure of a packet receiver using decision feedback for differential phase shift keying is shown. During the receive sync field, parameter estimators are fed known symbols, and during payload reception, they are fed previously determined symbols. Figure 3 This illustrates how a received noisy 8-DPSK symbol is determined and identified by its index. Parameter estimation includes time offset estimation, frequency offset estimation, and demodulation reference estimation. In the simplest DPSK receiver, the demodulation reference for symbol k would be symbol k-1, i.e., the previous symbol. In more advanced receivers, multiple previous symbols are appropriately combined to obtain a demodulation reference with less noise, thereby reducing the error probability. Figure 2 In this receiver, matched filtering, time offset estimation, and timing symbol extraction all operate on the complex baseband signal received from the transmitter. After converting the complex values ​​to phase values, the remainder of the receiver processes the phase values. Alternatively, the remainder of the receiver can be implemented using a complex signal, or a mixture of phase and complex signal values ​​can be processed after timing symbol extraction.

[0009] Typically, the receiver operates as follows: the received complex baseband signal 1 appears at an oversampling rate relative to the symbol rate defined in the receiver and passes through the matched filter 2. The purpose of time offset estimation, performed by the time offset estimation unit 3, is to obtain the sampling time point with minimal interference from previous and past symbols. The timing symbol extraction, performed by the timing symbol extraction unit 4, downsamples according to the symbol rate. The phase value p(k)16 of the obtained complex symbol y(k)5 is fed to the frequency offset estimation unit 6, where k is the symbol index. This frequency offset estimation unit 6 transmits a phase sequence s(k)8 representing the slope of the phase relative to time caused by frequency offset and frequency drift.

[0010] Figure 4 A possible implementation of the frequency offset estimation unit 6 is illustrated in detail. Assuming error-free reception, the known symbol index 48 transmitted in the sync field, or the determined symbol index 49 transmitted in the payload field, is mapped to their corresponding phase values ​​and accumulated in the data packets received from the transmitter, thereby replicating the sequence of transmitted phase values. The accumulation therein simulates differential modulation in a manner similar to that occurring at the transmitter. The difference between the received phase value p(k)16 and the... Based on the variations in noise, frequency offset, and frequency drift over time, the expanded phase u(k)17 is fed to a differentiating filter 18, which transmits the estimated instantaneous frequency offset. The frequency offset estimation low-pass filter 20 reduces noise-induced fluctuations in the instantaneous frequency offset 19 and delivers a more stable frequency offset estimate. Integrating the frequency offset estimate 21 produces a time-varying phase sequence s(k)8, the goal of which is to replicate the phase changes caused by frequency offset and frequency drift.

[0011] The frequency offset compensation unit 7 transmits the phase q(k) = p(k) - s(k)9 to the demodulation reference estimation unit 10. Figure 5 The demodulation reference symbol estimation unit 10 using a modified forgetting factor infinite impulse response (IIR) filter is illustrated in detail. The modification involves adding 23 of the phase of the newly determined (or known) symbol i(k) 14 to the filter memory 50 so that the filter state phase is always aligned with the filter state phase of the previous symbol, which is used in differential demodulation. The symbol demodulation unit 11 calculates the phase difference d(k) = q(k) - r(k) 45; symbol determination and bit mapping 12 are performed by quantizing d(k) according to the transmitted symbol alphabet, for example... Figure 3 An example of 8-DPSK is given. F represents the forgetting factor of 24.

[0012] In receivers of known technologies, there is always only a fixed receiving path to process data packets received from both clean and dirty transmitters. However, use cases with no line of sight between two communication BT modems, such as transmission between a mobile phone in a pocket and an earphone plugged into the ear, require as little link loss as possible. Developers can achieve this by implementing a clean transmitter on one side and a receiver on the other side that fully utilizes the clean signal. However, the receiver must still support the "dirty" transmitter well. Therefore, a receiver that performs optimally on either "dirty" or "clean" transmitters is needed.

[0013] The growing demand for battery-powered devices in mass-market applications is for low cost and low power consumption. This translates to a need for small hardware architectures and low implementation complexity.

[0014] Therefore, the object of this invention is to provide an apparatus and method for adapting a receiver to transmitter defects. This means that, assuming operation at a certain signal-to-noise ratio, where the noise is thermal noise from the radio front end, the receiver performance should be adjusted by reducing the error rate when receiving data packets from the net transmitter. Summary of the Invention

[0015] To address the aforementioned problems, a first aspect of the present invention provides a differential phase shift keying (DPSK) receiver for adapting to transmitter defects.

[0016] A differential phase shift keying (DPSK) receiver for adapting to transmitter defects according to the present invention includes: means for receiving signaling data packets from a transmitter, a time offset estimation unit, and a timing symbol extraction unit. The timing symbol extraction unit is configured to output a complex symbol y(k) of the signaling data packet, where k is a symbol index, and feed y(k) to a frequency offset estimation unit and a frequency offset compensation unit. The frequency offset estimation unit is configured to transmit the phase value s(k) of the symbol y(k). The frequency offset compensation unit is configured to perform frequency offset and frequency drift compensation on the received symbol and feed the compensation result q(k) to a demodulation reference estimation unit and a symbol demodulation unit. The demodulation reference estimation unit is configured to support symbol demodulation for symbol determination and bit mapping. According to the present invention, the frequency offset estimation unit includes a first phase estimation unit, a second phase estimation unit, and a dirty / clean selector, wherein a first phase error estimate e is output. dirty The first phase estimation unit of (k) includes a first frequency offset estimation low-pass filter, the bandwidth of which is wider than the bandwidth of the second frequency offset estimation low-pass filter of the second phase estimation unit. The second phase estimation unit outputs a second phase error estimate e. clean (k), the dirty / clean selector is configured to select the second phase error estimate e clean(k) and the first phase error estimate e dirty The ratio of (k) is compared with a threshold T, and a binary comparison result c(k) is output. The demodulation reference estimation unit includes a first filter and a second filter, wherein the filter bandwidth of the first filter, which is part of the "dirty" reception path of the receiver, is wider than the filter bandwidth of the second filter, which is part of the "clean" reception path of the receiver, wherein if the ratio e of the second phase error estimate to the first phase error estimate is... clean (k) / e dirty (k) If the threshold T is met, the “clean” receiving path is used to output the received information bits; otherwise, the “dirty” receiving path of the receiver is used.

[0017] The key idea of ​​this new receiver technology is to operate in parallel the differentiating elements of a receiver optimized for both "dirty" and "clean" transmitters, so as to evaluate incoming data packets relative to a more suitable reception path during operation and utilize this more suitable path. In the DPSK receiver, the differentiating elements are a frequency offset estimation low-pass filter and a demodulation reference filter; the receiver operates in parallel a wide filter and a narrow frequency offset estimation low-pass filter, as well as a wide filter and a narrow filter for demodulation reference filtering. The outputs of the two frequency offset estimation low-pass filters are fed into a corresponding sliding phase offset estimation filter and subsequently a corresponding sliding phase error averaging filter. The evaluation criterion is the ratio of the average phase error between the outputs of the narrow and wide frequency offset estimation low-pass filters. Below a threshold T, the receiver uses a "clean" reception path. A wide filter is a filter with a wider bandwidth than a narrow filter, and a narrow filter has a narrower bandwidth than a wide filter.

[0018] Assuming operation at a given signal-to-noise ratio, where the noise is thermal noise from the radio front end, the receiver of this invention reduces the error rate when receiving data packets from a "clean" transmitter. Therefore, it is possible to achieve the same error rate and user experience even with significant link loss, possibly due to a large distance between the transmitter and receiver, or due to strong signal attenuation caused by an obstructed radio wave propagation path between them.

[0019] The threshold T was determined through statistical simulation, with the goal of minimizing the packet error rate in a mix of clean and dirty transmitter scenarios.

[0020] In a variant of the DPSK receiver of the present invention, the frequency offset estimation unit includes a phase selection switch controlled by the output c(k) of the dirty selector, to select the first phase sequence s output by the first phase estimation unit and the second phase estimation unit, respectively. dirty(k) and the second phase sequence s clean Choose between (k).

[0021] First phase sequence s dirty (k) and the second phase sequence s clean (k) is used as input to the frequency offset compensation unit to compensate for the phase slope with respect to time caused by frequency offset and frequency drift. The frequency offset estimator of the new technology outputs the selected phase sequence s(k) for frequency offset compensation, that is, the frequency offset estimator and the phase reference estimator are switched.

[0022] In another variant of the DPSK receiver of the present invention, the first phase estimation unit and the second phase estimation unit include means for: estimating a low-pass filter by sliding phase offset to convert the fluctuating reference phase u(k) and the low-noise phase sequence s inst The difference between (k) is averaged, and the output of the sliding phase offset estimation low-pass filter is compared with the low-noise phase sequence s. inst (k) are added together to obtain the mean value of the fluctuation reference phase u. inst (k) low noise sequence and absolute difference The phase error estimate e is transmitted through a corresponding sliding phase error averaging low-pass filter to output the phase error estimate e. dirty (k) and e clean (k).

[0023] The index "inst" represents either "dirty" or "clean".

[0024] In another variant of the DPSK receiver of the present invention, the respective phase error estimates e of the phase estimation units are... dirty (k) and e clean (k) is input to the dirty / clean selector.

[0025] The basic idea is that the first "dirty" phase estimation unit uses a first frequency offset estimation low-pass filter with a wider bandwidth for frequency offset estimation, which better follows the frequency drift. Similarly, the second "clean" phase estimation unit uses a frequency offset estimation low-pass filter with a narrower filter bandwidth for frequency offset estimation, which better suppresses noise. Therefore, typically, a phase estimation unit transmits the estimated value of phase fluctuation e in the following manner. inst (k): Estimation of the low-pass filter's response to the fluctuating reference phase u(k) and the low-noise phase sequence s via sliding phase offset. inst The difference between (k) is averaged, and the output signal of the filter is compared with the low-noise phase sequence s. inst Adding (k) together yields a low-noise phase sequence with a mean of the fluctuating reference phase u(k). and difference The absolute value is passed through a sliding phase error averaging low-pass filter. The dirty-clean selector will select the "clean" receive path (i.e., the second phase error estimate e). clean (k) and the “dirty” receiving path (i.e., the first phase error estimate e shrunk by threshold T). dirty (k) is compared, and if the "net" error estimate is less than the "dirty" estimate scaled by the threshold T, the packet is declared "net".

[0026] In a variant of the DPSK receiver of the present invention, the demodulation reference estimation unit includes a demodulation reference selector configured to transmit a reference symbol r(k) determined / controlled by the output c(k) of the clean / dirty selector. dirty (k) or r clean (k).

[0027] The demodulation reference estimation unit of the present invention includes a function for determining a reference symbol r. dirty (k) and / or r clean Two paths of (k). Each path contains an infinite impulse response (IIR) low-pass filter; the filter bandwidth of the first path is wider than that of the filter of the second path. The phase of the newly determined (or known) symbol is added to the filter memory 50 to align the filter state phase with the filter state phase of the previous symbol, which is used in differential demodulation. The outputs of the two filter paths are fed into a demodulation reference selector, which is controlled by the output c(k) of the clean / dirty selector. The resulting output r(k) of the demodulation reference estimation unit is used by the symbol demodulation unit. The symbol demodulation unit calculates the phase difference d(k) = q(k) - r(k) and performs symbol determination by quantizing d(k) according to the transmitted symbol alphabet.

[0028] In another variant of the DPSK receiver of the present invention, the demodulation reference estimation unit includes a filter coefficient selector for selecting filter coefficients for a wide or narrow filter bandwidth, the filter coefficient selector being controlled by the output c(k) of the dirty selector.

[0029] The output c(k) of the dirty / clean selector is the filter coefficient of the wide or narrow filter bandwidth selector.

[0030] A second aspect of the invention provides a method, performed by a DPSK receiver, for adapting the receiving performance of a Bluetooth receiver to transmitter defects. This method can be performed by a differential phase-shift keying DPSK receiver as described in any embodiment of the first aspect.

[0031] A method for adapting the receiving performance of a Bluetooth receiver to transmitter defects, performed by a differential phase shift keying (DPSK) receiver as described in any embodiment of the first aspect, includes the following steps:

[0032] - Receive and match filtered signaling data packets from the transmitter.

[0033] - Obtain the sampling time point with minimal interference from previous and past symbols using the time offset estimation unit.

[0034] - The received signaling data packet's mapped symbols are downsampled to the symbol rate using a timing symbol extraction unit, and the complex symbol y(k) is obtained.

[0035] - Obtain the phase values ​​p(k) of the complex symbol y(k) and feed them to the frequency offset estimation unit and the frequency offset compensation unit.

[0036] - Map the symbol indices i(k) to their corresponding phase values ​​p(k) and accumulate them into the received signaling data packets to generate a replication sequence.

[0037] - Determine the difference The expanded phase u(k) is obtained and fed to a differentiating filter, which transmits the estimated instantaneous frequency offset value. in,

[0038] -The instantaneous frequency offset estimate The input is fed into the first phase estimation unit and the second phase estimation unit, and the first phase estimation unit outputs a first phase error estimate e. dirty (k) and the first phase sequence s dirty (k), the second phase estimation unit outputs the second phase error estimate e clean (k) and the second phase sequence s clean (k), and

[0039] - The second phase error estimate e is obtained through the dirty / clean selector. clean (k) and the first phase error estimate e dirty The ratio of (k) is compared with the threshold T, and

[0040] -If the second phase error estimate is e clean (k) and the first phase error estimate e dirty If the ratio of (k) is lower than the threshold T, then the "net" receive path of the demodulation reference estimation unit of the receiver is used to output the received information bits controlled by the output c(k) of the dirty-net selector; otherwise, the "dirty" receive path is used.

[0041] First phase sequence s dirty (k) and the second phase sequence s clean (k) is used as the input to the frequency offset compensation unit so that the phase q(k) = p(k) - s(k) is transmitted to the demodulation reference estimation unit and the symbol demodulation unit.

[0042] The basic idea is that the first phase estimation unit, also known as the "dirty" phase estimation unit, uses a filter with a wider filter bandwidth than the filter of the second or "net" phase estimation unit, so that it can better follow the frequency drift, while the "net" phase estimation unit uses a narrower filter bandwidth, so that it can better suppress noise.

[0043] According to the method of the present invention, the instantaneous frequency offset estimate The frequency offset is low-pass filtered by two frequency offset estimation low-pass filters with different bandwidths to obtain the frequency offset estimate. Integrate it to obtain the phase value sequence s inst (k), the fluctuating reference phase u(k) and the low-noise phase sequence s inst The difference between (k) is averaged using a sliding phase shift estimation low-pass filter, and the output signal of the sliding phase shift estimation low-pass filter is compared with the low-noise phase sequence s. inst (k) are added together to obtain a low-noise phase sequence with a mean of the fluctuating reference phase u(k). and difference The absolute value is transmitted through a sliding phase error averaging low-pass filter to convey the phase error estimate e. inst (k). Phase error estimate e inst (k) is e when output by the first phase estimation unit. dirty (k), which is e when output by the second phase estimation unit. clean (k).

[0044] In a variation of the method of the present invention, the output c(k) of the dirty / clean selector controls the demodulation reference selector of the demodulation reference estimation unit to transmit the reference symbol r. dirty (k) and r clean (k), which is input to the symbol demodulation unit.

[0045] In an alternative variation of the method of the present invention, the output c(k) of the dirty / clean selector controls the filter coefficient selector of the demodulation reference estimation unit to select filter coefficients for a wide or narrow filter bandwidth of the filter F(k) of the demodulation parameter estimation unit. An advantage is that only one receive path is needed for the "dirty" and "clean" transmitters in the demodulation reference unit.

[0046] The invention will be explained in more detail using exemplary embodiments. Attached Figure Description

[0047] The attached diagram shows

[0048] Figure 1 Bluetooth Enhanced Data Rate (BDR) packet format;

[0049] Figure 2 This application provides a DPSK data packet receiver;

[0050] Figure 3 8-DPSK determination scheme;

[0051] Figure 4 This application provides a frequency offset estimation unit in its embodiments;

[0052] Figure 5 This application provides a demodulation reference estimation unit in its embodiments;

[0053] Figure 6 The frequency offset estimation unit according to the present invention;

[0054] Figure 7a a) A general phase estimation unit according to the present invention; b) A "dirty" phase estimation unit; c) A "clean" phase estimation unit;

[0055] Figure 8 The dirty / clean selector according to the present invention;

[0056] Figure 9 Demodulation reference estimation unit according to a first embodiment of the present invention;

[0057] Figure 10 An alternative demodulation reference estimation unit according to a second embodiment of the present invention. Detailed Implementation

[0058] The receiver in this application embodiment may include: a means for receiving signaling data packets from a transmitter, a time offset estimation unit, and a timing symbol extraction unit. The timing symbol extraction unit is configured to output a complex symbol y(k) of the signaling data packet, where k is a symbol index, and feed y(k) to a frequency offset estimation unit and a frequency offset compensation unit. The frequency offset estimation unit is configured to transmit the phase value s(k) of the symbol y(k). The frequency offset compensation unit is configured to perform frequency offset and frequency drift compensation on the received symbol and feed the compensation result q(k) to a demodulation reference estimation unit and a symbol demodulation unit. The demodulation reference estimation unit is configured to support symbol demodulation for symbol determination and bit mapping.

[0059] Figure 6The frequency offset estimation unit of the present invention is shown, which differs from known technologies in that it generates an instantaneous frequency offset estimate after the differential filter 18. Based on the comparison result of the dirty / clean selector 32, the phase selection switch 33 selects between two different phase sequences 8, namely the first phase sequence s. dirty (k)29 and the second phase sequence s clean (k), and selects from among them, with the dirty / clean selector 32 feeding a first phase error estimate e. dirty (k)28 and the second phase error estimate e clean (k)30, and these two are obtained from the first “dirty” phase estimation unit or instance 26 and the second “net” phase estimation unit or instance 27, respectively. The basic idea is that the “dirty” phase estimation unit 26 uses a first frequency offset estimation low-pass filter 20a, the filter bandwidth of which is wider than that of the second frequency offset estimation low-pass filter 20b of the second “net” phase estimation unit 27. The first frequency offset estimation low-pass filter 20a can better follow the frequency drift, while the second “net” phase estimation unit 27 uses a second frequency offset estimation low-pass filter 20b, the filter bandwidth of which is narrower than that of the first phase estimation unit 26. The second frequency offset estimation low-pass filter 20b can better suppress noise.

[0060] Figure 7 shows the detailed structure of phase estimation units 26 and 27, used for "dirty" ( Figure 7b ) or "clean" Figure 7c Phase error estimation.

[0061] Similar to the frequency offset estimation unit 6 in the aforementioned embodiments, the instantaneous frequency offset value The obtained frequency offset estimate is transmitted through low-pass filters 20, 20a, and 20b. Integrating 22, 22a, 22b yields the phase value sequence s. inst (k), which is s dirty (k)29 or s dirty (k)31. Furthermore, without considering a constant phase offset, phase estimation units 26 and 27 transmit the estimated value of the phase fluctuation, also known as the phase error estimate e, in the following manner. inst (k), i.e., e dirty (k),e clean (k)28, 30: Estimate the fluctuating reference phase u(k)17 in the first phase estimation unit 26 or the phase equal to s in the second phase estimation unit 27 by sliding phase offset for low-pass filters 39, 39a, 39b. dirty The oscillation reference phase u(k) and the low-noise phase sequence s instThe difference between (k) is averaged to obtain s in the first phase estimation unit 26. dirty (k) and s in the second phase estimation unit 27 clean (k)31, the input signal of the filter is compared with the low-noise phase sequence s inst Adding (k)29 and 31, we get the mean value as the fluctuation reference phase u(k)17 or s. dirty (k)29 low-noise phase sequence 37a, 37b, and the difference The absolute value of 38b is passed through another sliding phase error averaging low-pass filter 35, 35a, 35b respectively.

[0062] Figure 8 The function of the dirty / clean selector 32 is shown. It estimates the phase error e of the second "clean" path. clean (k)30 and the first “dirty” path phase error estimate scaled by threshold T40 e dirty (k)28 is compared. If the "net" path phase error estimate is e clean (k)30 is less than the scaled dirty path error estimate T·e dirty If (k), then the dirty / clean selector 8 declares the received signal as "clean," where 0 < T < 1. In other words, if the ratio of the error estimates is less than the threshold, The received signal is considered clean.

[0063] If the transmitter's behavior is merely "dirty"—that is, exhibiting less frequency modulation than the specified worst-case scenario—the "dirty" / "clean" judgment may change back and forth during packet reception. The described method of weighting one of these estimates with a threshold and then comparing them is a low-complexity approach for accurately implementing this ratio. The division operations required to calculate the ratio are typically more complex than multiplication.

[0064] like Figure 9 As shown, the output c(k)34 of the dirty / clean selector controls the demodulation reference estimation unit of the new technology. The demodulation reference estimation unit includes two instances of the demodulation reference estimation unit in the aforementioned embodiments, one with a wide filter bandwidth 46 for transmitting r dirty (k)42 is the demodulation reference filter for the dirty receiving path, and another one with a narrow filter bandwidth 47 is used to transmit r. clean The demodulation reference filter for the net receive path of (k)43. The demodulation reference estimation unit also includes a demodulation reference selector 41 for transmitting r(k)25, which is determined by the output c(k)34 of the dirty / clean selector.

[0065] Figure 10A further simplified version of the demodulation reference estimation unit is shown, wherein the output c(k)34 of the dirty selector triggers the filter coefficient selector 44 to select the filter coefficients for wide or narrow filter bandwidths. Figure 9 In contrast, when switching back and forth between "dirty" and "clean", some filter ringing may occur in corner situations. In practice, this will only reduce the error rate by a very small amount, so much so that it is undetectable.

[0066] The key to this patent is the frequency offset estimation low-pass filter 20a, 20b. Figure 7a -c) and Figure 9 and Figure 10 The demodulation reference filters 52 and 53 switch between two bandwidths. To control this switching, evaluation functions have been built in for the sliding phase offset estimation low-pass filters 39, 39a, and 39b, the low-noise phase sequences 37, 37a, and 37b, the absolute value calculation blocks 38, 38a, and 38b, and the sliding phase error averaging low-pass filters 35, 35a, and 35b.

[0067] List of reference numerals

[0068] 1. Received baseband signal

[0069] 2. Matched Filter

[0070] 3 Time Offset Estimation Unit

[0071] 4. Timing Symbol Extraction Unit

[0072] 5. Complex number symbols

[0073] 6 Frequency Offset Estimation Unit

[0074] 7 Frequency Offset Compensation Unit

[0075] 8 Phase Sequence

[0076] 9. Compensation Results

[0077] 10 Demodulation Reference Estimation Unit

[0078] 11 Symbol Demodulation Unit

[0079] 12. Sign determination and bit mapping unit

[0080] 13. Results of generating the demodulation reference estimation unit

[0081] 14 Symbol Index

[0082] 15 received information bits

[0083] 16 phase values

[0084] 17. Phase expansion

[0085] 18 Differential Filter

[0086] 19. Estimated instantaneous frequency offset

[0087] 20 Frequency Offset Estimation Low-Pass Filter

[0088] The 20a first frequency offset estimation low-pass filter is "dirty".

[0089] 20b Second frequency offset estimation low-pass filter "net"

[0090] 21. Frequency offset estimate

[0091] 21a Frequency offset estimate "dirty"

[0092] 21b Frequency Shift Estimation "net"

[0093] 22 Integral Filter

[0094] The 22a integrator filter is "dirty".

[0095] 22b Integral Filter "Net"

[0096] 23. Addition

[0097] 24 Forgetting Factors

[0098] 25. Results of generating the demodulation reference estimation unit

[0099] 26 First Phase Estimation Unit

[0100] 27 Second Phase Estimation Unit

[0101] 28 First phase error estimate e dirty (k)

[0102] 29 First phase sequence s dirty (k)

[0103] 30 Second phase error estimate e clean (k)

[0104] 31 Second phase sequence s clean (k)

[0105] 32 Dirty / Clean Selector

[0106] 33 Phase Selection Switch

[0107] 34. Output c(k) of the dirty / clean selector

[0108] 35 Sliding phase error averaging low-pass filter

[0109] 35a Sliding Phase Error Averaging Low-Pass Filter "Dirty" Path

[0110] 35b Sliding Phase Error Averaged Low-Pass Filter "Net" Path

[0111] 37 Low-noise phase sequences

[0112] 38 Absolute value calculation block

[0113] 39. Sliding Phase Shift Estimation Low-Pass Filter

[0114] 39a Sliding Phase Shift Estimation Low-Pass Filter "Dirty" Path

[0115] 39b Sliding phase offset estimation of the "net" path of a low-pass filter

[0116] 40 Threshold T

[0117] 41 Demodulation Reference Selector

[0118] 42 First reference symbol r dirty (k)

[0119] 43 Second reference symbol r clean (k)

[0120] 44 Filter coefficient selector

[0121] 45 phase difference

[0122] 46. ​​"Dirty" demodulation reference filter for the "dirty" receiver path

[0123] 47. "Net" demodulation reference filter for the "net" receiver path

[0124] 48. Known Symbol Index

[0125] 49 Determined symbol index

[0126] 50 Filtering Memory

[0127] 51 Demodulation Reference Filter

[0128] 52. Filter coefficients with wide filter bandwidth

[0129] 53. Filter coefficients for narrow filter bandwidth

Claims

1. A differential phase shift keying, DPSK, receiver for accommodating transmitter imperfections, the imperfections being in that there is an undesired variation over time of the carrier frequency, the receiver comprising: An apparatus for receiving a signaling data packet from a transmitter, a time offset estimation unit (3) and a timing symbol extraction unit (4), the timing symbol extraction unit (4) being configured to output complex symbols of the signaling data packet y(k) (5), k for a symbol index and to feed y(k) to a frequency offset estimation unit (6) and a frequency offset compensation unit (7), wherein the frequency offset estimation unit (6) is configured to deliver a phase value y(k) (5) s(k) (8), the frequency offset compensation unit (7) is configured to perform a frequency offset and frequency drift compensation on the received symbols and to feed a compensation result q(k) (9) to a demodulation reference estimation unit (10) and a symbol demodulation unit (11), wherein the demodulation reference estimation unit (10) is configured to support a symbol demodulation for symbol decision and bit mapping, characterized in that the frequency offset estimation unit (6) comprises a first phase estimation unit (26), a second phase estimation unit (27) and a dirty-clean selector (32), wherein the first phase estimation unit (26) outputting a first phase error estimation value e dirty (k) (28) comprises a first frequency offset estimation low pass filter (20a) having a filter bandwidth wider than a filter bandwidth of a second frequency offset estimation low pass filter (20b) of the second phase estimation unit (27) outputting a second phase error estimation value e clean (k) (30), the dirty-clean selector (32) is configured to compare a ratio of the second phase error estimation value e clean (k) (30) to the first phase error estimation value e dirty (k) (28) to a threshold value T and to output a binary comparison result c(k) (34), the demodulation reference estimation unit (10) comprises a first filter (46) and a second filter (47), wherein a filter bandwidth of the first filter (46) being part of a "dirty" reception path of the receiver is wider than a filter bandwidth of the second filter (47) being part of a "clean" reception path of the receiver, wherein the ratio of the second phase error estimation value to the first phase error estimation value e clean (k) (30) e dirty (k) (28) below the threshold T then the "clean" receive path is used to output the received information bits, otherwise the "dirty" receive path is used.

2. The DPSK receiver of claim 1, wherein, The frequency offset estimation unit (6) comprises a phase selection switch (33) controlled by the output (34) of the dirty-clean selector (32) to select between a first phase sequence (28) output by the first phase estimation unit (26) and a second phase sequence (30) output by the second phase estimation unit (27). c(k) s dirty (k) s dirty (k) ​​​ 3. The DPSK receiver of claim 1, wherein, The first phase estimation unit (26) and the second phase estimation unit (27) comprise means for averaging the difference between the fluctuating reference phase u(k) (17, 29) and the low-noise phase sequence s inst (k) (29, 31), adding the output of the low-pass filter to the low-noise phase sequence s inst (k) (29, 31) and thus obtaining a low-noise sequence u(k) (37, 37a, 37b) having the fluctuating reference phase (17, 29) as mean value and passing the absolute difference (38, 38a, 38b) through a sliding phase error averaging low-pass filter (35, 35a, 35b) to output the phase error estimate e dirty (k) (28) and e clean (k) (30), respectively.

4. The DPSK receiver of claim 3, wherein, the respective phase error estimate of the phase estimation unit (26, 27) e dirty (k) (28) and e clean (k) (30) are input to the dirty selecter (32).

5. The DPSK receiver of claim 1, wherein, The demodulation reference estimation unit (10) includes a demodulation reference selector (41) configured to transmit the output of the dirty / clean selector (32). c(k) (34) Reference symbols for determination / control r dirty (k) (42) and r clean (k) (43).

6. The DPSK receiver of claim 1, wherein, The demodulation reference estimation unit (10) comprises a filter coefficient selector (44) for selecting filter coefficients for a wide filter bandwidth or a narrow filter bandwidth, the filter coefficient selector (44) being controlled by the output of the dirty clean selector (32) c(k) (34) control.

7. A method of adapting the reception performance of a Bluetooth receiver to transmitter imperfections, performed by a Differential Phase Shift Keying, DPSK, receiver, the method comprising the steps of: - receiving (1) and matched filtering (2) a signaling data packet from a transmitter, - obtaining, by a time offset estimation unit (3), a sampling time point that is least disturbed from previous and past symbols, - down-sampling the mapped symbols of the received signaling data packets to a symbol rate by a timing symbol extraction unit (4) and obtaining complex symbols y(k) (5), - obtaining the complex symbols y(k) the phase values of (5) p(k) (16) and feeding them to the frequency offset estimation unit (6) and the frequency offset compensation unit (7), - the symbol index i(k) (14) mapped to their corresponding phase values p(k) (16), and accumulate them into the received signaling data packet, resulting in a replica sequence , - determining the difference of the unwrapped phase u(k) (17) and feeding u(k) (17) to a differential filter (18) which delivers an estimated instantaneous frequency offset estimate (19), wherein, - said instantaneous frequency offset estimate (19) to a first phase estimation unit (26) and a second phase estimation unit (27), said first phase estimation unit (26) outputting a first phase error estimate e dirty (k) (28) and a first phase sequence s dirty (k) (29), said second phase estimation unit (27) outputting a second phase error estimate e clean (k) (30) and a second phase sequence s clean (k) (31), and -The second phase error estimate is obtained by the dirty / clean selector (32). e clean (k) (30) and the first phase error estimate e dirty (k) (28) ratio and threshold T (40) Comparison, and -If the second phase error estimate e clean (k) (30) and the first phase error estimate e dirty (k) (28) The ratio is lower than the threshold. T (40) The "clean" receiving path (47) of the demodulation reference estimation unit (10) of the receiver is used to output the received output from the dirty / clean selector (32). c(k) (34) Control information bits, otherwise use the "dirty" receive path (46).

8. The method of claim 7, wherein, to the instantaneous frequency offset estimate in the first phase estimation unit (26) (19) is low-pass filtered (20a) to obtain a frequency offset estimate (21a) which is integrated (22a) to obtain a phase value sequence s dirty (k) (29), the difference between the fluctuating reference phase u(k) (17) and the low-noise phase sequence s dirty (k) (29) is averaged by a sliding phase offset estimation low-pass filter (39a), the output signal of which is added to the low-noise phase sequence s dirty (k) (29) to obtain a low-noise phase sequence u (k) with the mean value of the fluctuating reference phase (37a) and the difference is passed through a sliding phase error average low-pass filter (35a) to deliver the phase error estimate e dirty (k) (28).

9. The method of claim 7, wherein, to the instantaneous frequency offset estimate in the second phase estimation unit (27) (19) is low-pass filtered (20b) to obtain a frequency offset estimate (21b), which is integrated (22b) to obtain a phase value sequence s clean (k) (31), the difference between the fluctuating reference phase u(k) (29) and the low-noise phase sequence s clean (k) (31) is averaged by a sliding phase offset estimation low-pass filter (39b), the output signal of which is added to the low-noise phase sequence s clean (k) (31) to obtain a low-noise phase sequence u (k) (29) with the mean of the fluctuating reference phase u clean (k) (37b), and the absolute value of the difference is passed through a sliding phase error averaging low-pass filter (35b) to deliver the phase error estimate e clean (k) (30).

10. The method of claim 7, wherein, The output of the dirty selecter (32) c(k) (34) controlling a demodulation reference selector (41) of the demodulation reference estimation unit (10) to deliver reference symbols r dirty (k) (42) and r clean (k) (43) which are input to the symbol demodulation unit (11).

11. The method of claim 7, wherein, The output of the dirty selecter (32) c(k) (34) controlling a filter coefficient selector (44) of the demodulation reference estimation unit (10) to select filter coefficients (52, 53) for a wide filter bandwidth or a narrow filter bandwidth of the filter F(k) of the demodulation reference estimation unit (10) to select filter coefficients (52, 53) for a wide filter bandwidth or a narrow filter bandwidth of the filter

Images