An Iterative Receiving Processing Method for THP FTN System in the Presence of Phase Noise

Abstract

The present invention discloses an iterative receiving and processing method for a THP FTN system in the presence of phase noise, belonging to the field of communication technologies. Based on the statistical characteristics of the residual phase noise (RPHN) after pilot symbol assisted modulation (PSAM) in the THP-FTN system, the present invention realizes a receiving and processing method with a phase locked loop structure applicable to the THP-FTN system, namely a code-aided phase locked loop based on the extended a priori demapping (EAD) scheme. Through iteration, the RPHN can be more accurately estimated and eliminated to reduce the impact of PHN on the system performance. The present invention can effectively reduce the signal-to-noise ratio loss caused by RPHN in the THP-FTN system. Especially in the case of severe FTN-ISI and rapidly changing PHN, the reliability of data transmission can still be guaranteed.

Description

Technical Field

[0001] The present invention belongs to the field of communication technology, and in particular relates to an iterative receiving and processing method of a THP FTN system when phase noise exists. Background Art

[0002] In recent years, the Beyond Nyquist (FTN) technology, originally proposed by Mazo in the 1970s, has attracted much attention for its ability to improve spectral efficiency. In a Beyond Nyquist system, in order to obtain a higher symbol transmission rate, the FTN signal abandons the Nyquist criterion's time interval restriction on adjacent symbols, but this operation also introduces inter-symbol interference (ISI). At the same time, in scenarios with high spectral efficiency requirements such as microwave backhaul links, we combine high-order quadrature amplitude modulation (QAM) with FTN. However, high-order QAM signals are susceptible to noise, especially phase noise (PHN) caused by defects in the transmitter and receiver local oscillators. PHN will seriously affect the demodulation performance of the system and reduce spectral efficiency, especially in scenarios with high PHN levels such as millimeter wave communication links. Secondly, in the FTN system, the coupling of PHN and ISI (FTN-ISI) introduced by FTN causes the PHN compensation scheme used in the Nyquist transmission system to no longer be applicable. Therefore, it is necessary to study how to effectively compensate for PHN in the FTN system to improve system performance.

[0003] In traditional research, the research on PHN suppression schemes in FTN transmission is still limited. Among them, the iterative scheme based on factor graph has good signal detection performance, but the computational complexity is too high to be accepted; the adaptive decision feedback equalizer (ADFE) can eliminate FTN-ISI while compensating for PHN. However, due to the error propagation of DFE, it may lead to convergence failure and can only work effectively when there is a very small amount of FTN-ISI.

[0004] Recently, it has been proposed that FTN-ISI can be pre-eliminated at the transmitter to separate PHN from the state of coupling with FTN-ISI. The most common pre-elimination algorithm for FTN-ISI is linear equalization, but it will cause spectrum regrowth, resulting in SE loss; a more widely used method is to use Tomlinson-Harashima precoding (THP) to pre-eliminate FTN-ISI at the transmitter and perform pilot symbol assisted modulation (PSAM) at the receiver.

[0005] In high-order QAM systems, if PSAM is used to suppress PHN to a sufficiently low level, a higher pilot overhead is required, which will significantly reduce the spectral efficiency. In recent years, in Nyquist systems, some scholars have improved the PHN estimation method based on PSAM. This method first uses pilot-assisted linear interpolation to remove a large amount of PHN, and then uses a grid-based sequential search algorithm to accurately estimate RPHN (residual phase noise); in addition, a code-assisted phase-locked loop based on Wiener filter is proposed to estimate RPHN. However, these methods cannot be used in THP-FTN systems because the modulus processing of THP introduces an unknown random interference component. Summary of the invention

[0006] The purpose of the present invention is to provide an iterative receiving processing method of a code-assisted phase-locked loop (CA-PLL) structure combined with EAD (expanded apriori demapper) to solve the above-mentioned problems. The phase-locked loop uses an iterative method to estimate and eliminate RPHN. Compared with the traditional scheme, the phase-locked loop takes into account the random interference component introduced by THP, replaces the traditional QAM soft demodulation with EAD, and redesigns the input symbol of the phase detector to further improve the demodulation performance.

[0007] The technical solution adopted by the present invention is: a THP FTN system iterative receiving processing method when phase noise exists, the method comprising the following steps:

[0008] Step 1: Obtain the symbol sequence output by the pilot symbol assisted modulation (PSAM) module of the receiving end of the THP FTN system

[0009] Step 2: Perform QAM mapping on the codeword output by the low-density parity check LDPC decoder to obtain an estimated value of the transmission sequence.

[0010] Step 3, based on the current symbol sequence and estimated values Get Observables And send it to the phase detector;

[0011] in, M represents the number of bits per packet when the transmitter of the THP FTN system generates the transmission sequence a(n) by packet mapping, and round[·] represents rounding the number to the nearest integer;

[0012] Step 4: The phase detector obtains the phase detection result Input filter to obtain an estimate of the residual phase noise

[0013] in, arg[·] represents the argument of a complex number;

[0014] Step 5, based on the estimate For symbol sequence Perform phase noise compensation to obtain the symbol sequence after phase noise compensation And sent to the extended prior demapping module (EAD) to calculate the log-likelihood ratio LLR value;

[0015] Step 6, the LDPC decoder decodes the LLR value obtained in step 5 and outputs a codeword;

[0016] Step 7: If the codeword output by the LDPC decoder meets the valid codeword requirement or the recorded number of iterations reaches the preset maximum value, the receiving end outputs it; otherwise, update and record the number of iterations and return to step 2.

[0017] Furthermore, in step 4, the estimate of the residual phase noise is where ω e (n) represents the zero-mean Gaussian noise in the residual phase noise, and its variance E{} represents the mathematical expectation, ω(n) represents the zero-mean Gaussian noise introduced by the (AWGN) channel, v(n) represents the transmitted symbol, and ρ 0 represents the power normalization factor, the intermediate parameter λ 0 satisfy in l represents the sequence symbol number, ω(ω∈[-π,π]) represents the angle of integration, and L represents the sequence length. represents the response of the pulse shaping filter, T represents the Nyquist symbol period, τ represents the acceleration factor, and the superscript “*” represents conjugate.

[0018] The technical solution provided by the present invention brings at least the following beneficial effects:

[0019] The present invention is used to solve the problem that the high RPHN level of the PSAM scheme leads to a decrease in the bit error rate performance of the high-order modulation system. The signal reception is realized by an iterative method, and the CA-PLL scheme is adopted to effectively eliminate the RPHN introduced by the PSAM, and solve the problem that the THP modulus is difficult to determine the symbol during the iteration process. The invention can effectively reduce the signal-to-noise ratio loss caused by RPHN in the THP-FTN system, especially in the case of severe FTN-ISI and rapidly changing PHN, and can still ensure the reliability of data transmission. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0021] Figure 1 This is the structural block diagram of the THP-FTN system.

[0022] Figure 2 Schematic diagram of the structural block of CA-PLL.

[0023] Figure 3 This is the RPHN power spectrum.

[0024] Figure 4 To detect errors, a(n) and Schematic diagram of the location.

[0025] Figure 5 is the residual phase noise power spectral density after PSAM and CA-PLL when the modulation order is 1024-QAM, at this time Es / N0=33dB, T -1 =24*10 -6 , τ=0.8, PHN level is -90dBc / Hz@100kHz.

[0026] Figure 6 BER performance of the EAD-CA-PLL receiver when the modulation order is 256-QAM.

[0027] Figure 7 BER performance of the EAD-CA-PLL receiver when the modulation order is 1024-QAM. DETAILED DESCRIPTION

[0028] In order to make the objectives, technical solutions and advantages of the present invention more clear, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

[0029] The invention is an iterative receiving processing method suitable for a THP-FTN system. By adopting a CA-PLL scheme, the RPHN introduced by the PSAM is effectively eliminated, and the problem that the THP modulus is difficult to determine the sign during the iteration process is solved.

[0030] Therefore, when the present invention is used to compensate for phase noise, Figure 1 As shown, the specific processing process of the transmitter and the receiver is as follows:

[0031] Transmitter processing steps:

[0032] Channel coding and interleaving: Channel coding (such as convolutional coding, Turbo coding, LDPC coding, etc.) is performed on the binary information bit sequence to be transmitted to obtain a coded bit sequence;

[0033] Symbol mapping and pilot insertion: The coded bit sequence is grouped into M bits; each group containing M bits is mapped to obtain a symbol a ds [k], where k represents the kth symbol. Then, at every N-1 a ds A customized pilot symbol is inserted after the [k] symbol, where N is the pilot interval, to obtain the transmission sequence a[n].

[0034] Tomlinson-Harashima precoding: The THP module consists of a 2M modulo component and a feedback filter, the filter transfer function of which is F(z)-1, where is the coefficient sequence of FTN-ISI, and L represents the sequence length of FTN-ISI.

[0035] The output of the THP module is:

[0036]

[0037] in, represents the rounding down operation, p(n)=a(n)-I(n) is the input signal of THP, where is the output of the feedback filter. Definition Then the transmitted symbol v(n) = a(n) + 2Md(n). This modulo operation limits the amplitude of the output x(n) to (-M, M]. In order to keep the symbol energy normalized,

[0038] FTN modulation: After THP processing, the generated symbol x(n) is passed through the RRC filter Pulse shaping, with a rate of 1 / τT, generates the FTN signal s(t), which is sent into the channel for transmission through up-conversion.

[0039] Processing steps at the receiving end:

[0040] Matched filtering and whitening filtering: The RF front end receives a signal from a channel interfered with. In this specific implementation, an AWGN (Additive White Gaussian Noise) channel is considered, and a baseband signal r(t) is obtained by down-conversion. Then matched filtering and down-sampling are performed to extract statistical information for detection. Note that the noise component in the sample is not white noise. Therefore, whitening filtering is required, and the output after processing is:

[0041]

[0042] Among them, ω(n) is zero-mean Gaussian noise, and its variance is ρ 0 represents the power normalization factor, In this formula, T represents the Nyquist symbol period, and τ represents the acceleration factor.

[0043] In this system, the phase noise is formulated as a Wiener process, defined as

[0044] θ(n)=θ(n-1)+γη(n)

[0045] Among them, phase noise θ(n) = θ(nτT), η(n) ~ N(0,1), is a Gaussian random variable with unit distribution. γ is a constant coefficient that characterizes the speed of change of phase noise.

[0046] In summary, the received signal r(n) can be expressed as:

[0047]

[0048] PSAM: This module consists of the following three steps:

[0049] 1) Estimate the phase noise at the pilot position, denoted by φ(kN).

[0050] 2) The obtained φ(kN) is interpolated to obtain the phase noise at the data position.

[0051] 3) Compensate using the estimated value from step 2.

[0052] That is, in the THP-FTN system, the signal after the pilot is inserted is a(n), the transmitted symbol is v(n), and the output symbol after the PSAM module is Expressed as Among them, θ e (n) represents the residual phase noise, Represents the estimated value of zero-mean Gaussian noise. In order to estimate and compensate θ e (n), a phase detector (PD) is needed to extract The phase change relative to v(n). The estimation of a(n) can be obtained through QAM soft demodulation, LDPC decoder and QAM mapping. However, due to the 2M modulo operation of THP, an unknown random interference component 2Md(n) is introduced, so the observation from a(n) to v(n) There are still errors.

[0053] Next, the data after phase noise compensation is sent to the phase-locked loop for processing.

[0054] The principle block diagram of CA-PLL is as follows Figure 2 As shown. The phase-locked loop uses the EAD scheme to calculate the LLR (Log-Likelihood Ratio) value. Using the LLR value as input, the LDPC decoder performs soft decision decoding. If the decoder outputs an invalid codeword, the receiver performs the following iterative process:

[0055] 1) Initialize the iteration parameter i=1.

[0056] 2) Perform QAM mapping on the decoder output to obtain the sequence

[0057] 3) Utilize and a(n) to obtain the observed value of v(n)

[0058] in And send it to the PD module;

[0059] 4) PD module utilization Calculated And input it into the filter to obtain the estimate of residual phase noise

[0060] 5) The phase noise compensated sequence Input into EAD to calculate LLR.

[0061] 6) The LDPC decoder decodes the above LLR values ​​and outputs a codeword.

[0062] 7) If the codeword output by the decoder meets the valid codeword requirement or the number of iterations i reaches the preset maximum value, the receiving end outputs it; otherwise, i increases by 1 and returns to step 2.

[0063] In step 2, the sequence output by the QAM Mapper is an estimate of a(n) rather than the required v(n). Therefore, in order to ensure the correctness of the PD module output, the sequence obtained in step 3 should be kept as constant as possible. To this end, the following two problems need to be solved: first, how to accurately estimate the 2Md(n) component introduced by THP, and second, how to When an error occurs, reduce a(n) and In the THP-FTN system, when a(n) is located at the edge of the constellation and is detected incorrectly, There are two situations for the position, such as Figure 4 As shown: One is Located near a(n), this situation is acceptable in PLL. Figure 4 As shown in the upper left corner of , we can consider a(n) to be in the extended constellation and then take the modulus. This will result in a(n) and The distance between is too large (close to 2M), which is unacceptable. In order to improve performance, the phase-locked loop needs to compensate for the second type of error as much as possible. Therefore, according to the above analysis, v(n) and The distance between them is approximately an integer multiple of 2M.

[0064] In step 4, where ω e (n) is the zero-mean Gaussian noise in the residual phase noise, and its variance is E represents the mathematical expectation. Since the power spectrum (PSD) of RPHN is known as Figure 3 , so the Wiener filter can be used as the filter in the phase-locked loop.

[0065] Example

[0066] Figure 1 This is the block diagram of the TFP-FTN system structure of this embodiment. The source inputs the binary information bit stream to be sent, and the encoder performs channel coding (convolutional code, turbo code, LDPC code, etc.) to obtain the coding sequence; the coding sequence is pre-eliminated by FTN-ISI to obtain the transmission symbol sequence x(n); after filtering and shaping, s(t) is obtained and sent to the channel. The receiver RF front end receives the transmitted signal contaminated by the channel. Then, the FTN demodulator completes matched filtering and downsampling of the received signal, and sends it to the whitening filter to obtain the symbol sequence r(n). The sequence is processed by the PSAM module and the phase-locked loop, and finally the output sequence is obtained by decoding.

[0067] Traditional PSAM schemes usually use QPSK symbols as pilots, but when THP is used for ISI pre-elimination, its nonlinear operation will cause the pilot symbols to be deformed and fail to meet the requirements of PHN estimation. Therefore, this embodiment uses a pilot symbol energy reduction (PSER) generation method to solve this problem.

[0068] In the PSAM module, the estimated value of phase noise is:

[0069] φ(kN)=arg[r(kN)a * (kN)]

[0070] Wherein, N represents the pilot interval.

[0071] Based on this, the received symbol corresponding to the pilot position can be expressed as:

[0072]

[0073] In the above formula, the uncertainty of d(kN) leads to the error of φ(kN). To solve this problem, the PSER method will reduce the energy of some pilot symbols, that is, let d(kN) = 0. The pilot symbol at this time can be written as:

[0074]

[0075] in, is the constellation symbol per unit energy, and α(k) is the energy reduction factor. Based on the above information, the estimated value of phase noise can be written as:

[0076] φ(kN)=θ(kN)+ω θ (kN)

[0077] Among them, ω θ (kN) is zero-mean Gaussian noise with variance It can be seen that when the energy reduction coefficient α(k) is larger, the variance of the noise is smaller. In order to obtain the best reduction coefficient, the optimization problem can be listed as follows:

[0078]

[0079] After passing through the PSAM module, For the pilot interval N, PSAM can suppress the frequency less than f s / 2N PHN component in the frequency band, where f s =1 / τT is the signal rate. However, due to the pursuit of high transmission capacity in mobile communications, the pilot overhead required for PHN compensation is relatively small. In this case, the PSAM obtains Not accurate enough, resulting in The level is high, which seriously deteriorates the bit error performance of the THP-FTN system.

[0080] Taking the LDPC coding system with a code rate of 0.9 as an example, when 1024-QAM is used, the BER achieved by using the PSAM scheme is 10 compared with the case without PHN. -5 The required additional Es / N0 is 2.4dB when N = 50 and 4.3dB when N = 100. In order to reduce the RPHN level and improve the bit error rate performance, the phase-locked loop structure introduced above is required.

[0081] Other key parameters of the system are shown in Table 1.

[0082] Table 1 Simulation parameters

[0083] parameter Configuration Modulation 256-QAM, 1024-QAM Channel Coder DVB-S2 LDPC Coding Bitrate 0.9 Code length 64800 Shaping filter RRC (roll-off factor 0.3) Nyquist symbol rate 24Mbaud <![CDATA[PSAM filter order (N WF )]]> 21 <![CDATA[CA-PLL filter order (N WF )]]> 51 Maximum number of CA-PLL iterations 3 Channel AWGN

[0084] Figure 5 The PSD of RPHN after PSAM and CA-PLL is given for the modulation order of 1024-QAM and N=50. The results show that the RPHN is significantly reduced to -104dBc / Hz@100kHz by adding CA-PLL. Compared with the original PSAM, the low-frequency RPHN level is reduced by about 12dB. For 256-QAM, when ES / N0=33dB, CA-PLL can reduce the RPHN-PSD to about -97dBc / Hz@100kHz.

[0085] Figure 6 and Figure 7 The demodulation performance of the system is shown when the modulation order is 256-QAM and 1024-QAM. At this time, the acceleration factor is set to τ = 0.8, the PHN level is set to -90dBc / Hz@100kHz (γ≈0.2°), and the pilot interval is N = 50 or N = 100. The detailed description of these methods is shown in Table 2.

[0086] Table 2 Specific settings

[0087]

[0088]

[0089] Depend on Figure 6 and Figure 7 It can be seen that in the 256-QAM THP-FTN system, when the CA-PLL iteration scheme is not used, the PSAM-RPHN scheme has an advantage of about 0.2dB over the PSAM scheme when N=50, and the PSAM scheme has an advantage of about 0.55dB when N=100. Under 1024-QAM transmission conditions, the bit error rate performance improvements brought by PSAM-RPHN are 0.6dB and 1.6dB respectively. From the above results, it can be seen that the larger the modulation order and the pilot interval, the greater the benefit of optimizing the LLR calculation. At the same time, for 256-QAM transmission, the CA-PLL scheme achieves 10 -5 The bit error rate required Es / N0 is 0.85-1.55dB lower than that of the PSAM scheme; and at 1024-QAM, the bit error rate gain of CA-PLL can reach 1.6-3.3dB. It can be seen that the iterative scheme has better bit error rate performance than the non-iterative scheme.

[0090] In the iterative receiving processing method of the THP FTN system in the presence of phase noise disclosed in the present invention, based on the statistical characteristics of the residual phase noise (RPHN) after PSAM of the THP-FTN system, a receiving processing method for the phase-locked loop structure suitable for the THP-FTN system is provided. The method is based on the code-assisted phase-locked loop (CA-PLL) of the EAD scheme, and the RPHN can be estimated and eliminated more accurately through iteration to reduce the impact of PHN on system performance. Finally, we conducted simulations based on high-order QAM and low-density parity check (LDPC) codes. The simulation results verify that the above scheme effectively reduces the bit error rate of the THP-FTN system.

[0091] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

[0092] The above are only some embodiments of the present invention. For those skilled in the art, several modifications and improvements can be made without departing from the creative concept of the present invention, which all belong to the protection scope of the present invention.

Claims

1. An iterative receiving processing method for THP FTN system in the presence of phase noise, characterized in that, it includes the following steps: Step 1, obtain the symbol sequence output by the pilot symbol assisted modulation (PSAM) module at the receiving end of the THP FTN system Step 2, perform quadrature amplitude modulation (QAM) mapping on the codeword output by the low-density parity-check (LDPC) decoder to obtain an estimated value of the transmission sequence Step 3, based on the current symbol sequence and the estimated value obtain the observed quantity and send it to the phase detector; Among them, M represents the number of bits per packet when the transmitting end of the THP FTN system generates the transmission sequence a(n) through packet mapping, and round[·] represents rounding the number to the nearest integer; Step 4, the phase detector inputs the obtained phase detection result to an input filter to obtain an estimate of the residual phase noise wherein, arg[·] represents taking the argument of a complex number; Step 5, based on the estimator perform phase noise compensation on the symbol sequence to obtain the symbol sequence after phase noise compensation and send it to the extended prior demapping module to calculate the log-likelihood ratio (LLR) value; Step 6, the LDPC decoder decodes the LLR values obtained in Step 5 and outputs the codewords; Step 7, if the codewords output by the LDPC decoder meet the requirements of valid codewords or the recorded number of iterations reaches the preset maximum value, the receiving end outputs them; otherwise, update and record the number of iterations and return to Step 2; Among them, in step 4, the estimator of the residual phase noise where ω e (n) represents the zero-mean Gaussian noise in the residual phase noise, and its variance E{} represents the mathematical expectation, ω(n) represents the zero-mean Gaussian noise introduced by the (AWGN) channel, v(n) represents the transmitted symbol, ρ 0 represents the power normalization factor, and the intermediate parameter λ 0 satisfies where l represents the sequence symbol number, ω represents the integration angle, L represents the sequence length, represents the response of the pulse shaping filter, T represents the Nyquist symbol period, and τ represents the acceleration factor.

Images