A method for high-quality reconstruction of a reference signal of a terrestrial digital television multimedia broadcast

Abstract

The application discloses a reconstruction method, in particular to a ground digital television multimedia broadcast reference signal high-quality reconstruction method, which comprises the following steps: step 1, high-precision synchronization and parameter tracking; step 2, robust channel estimation and equalization; step 3, soft information demapping and iterative decoding; step 4, bit stream re-modulation. The two-stage synchronization mechanism combining coarse synchronization and fine synchronization, combined with the improved windowing peak detection algorithm and dynamic threshold, can effectively resist noise and multipath interference, and significantly reduce the false alarm and missed detection probability. The joint tracking loop continuously compensates the carrier frequency deviation and the sampling clock deviation, effectively suppresses the phase noise and the inter-subcarrier interference, and ensures stable timing and carrier synchronization under high-speed movement and severe channel conditions.

Description

Technical Field

[0001] This invention relates to the field of signal processing equipment technology, and in particular to a method for high-quality reconstruction of a reference signal for terrestrial digital television multimedia broadcasting. Background Technology

[0002] During transmission, terrestrial digital television multimedia broadcasting (DTMB) systems are affected by various complex channel factors such as multipath fading, Doppler shift, and noise interference, leading to a decline in reception quality. To achieve reliable reception, the receiver must complete key steps such as high-precision synchronization, channel estimation and equalization, and channel decoding. In existing technologies, synchronization often employs correlation peak detection based on a fixed threshold, which is prone to synchronization deviations under low signal-to-noise ratio or multipath conditions. Channel estimation often uses simple linear interpolation, which is insufficient for tracking rapidly time-varying or highly frequency-selective channels. Decoding typically involves single hard or soft decisions, failing to fully utilize the iterative gain of soft information. These factors limit further improvements in reception performance, especially in harsh environments such as mobile reception and dense urban multipath environments, where the system error rate is high, making it difficult to obtain a "clean" reference signal suitable for accurate channel analysis, interference troubleshooting, or receiver performance verification. Therefore, there is an urgent need for a method that can achieve high-precision synchronization, robust channel estimation, and efficient decoding under complex channel conditions, thereby reconstructing the transmitter reference signal with high quality to support applications such as advanced receiver processing, network optimization, and system testing. Summary of the Invention

[0003] In order to solve the problems existing in the prior art, the present invention provides a high-quality reconstruction method for terrestrial digital television multimedia broadcasting reference signals, thereby solving the current technical problems.

[0004] The technical solution adopted by this invention to solve its technical problem is:

[0005] This invention provides a method for high-quality reconstruction of terrestrial digital television multimedia broadcasting reference signals, comprising the following steps:

[0006] Step 1: High-precision synchronization and parameter tracking;

[0007] Step 2: Robust channel estimation and equalization;

[0008] Step 3: Soft information demapping and iterative decoding;

[0009] Step 4: Bitstream remodulation.

[0010] Preferably, step 1: high-precision synchronization and parameter tracking includes: step 1.1: coarse synchronization; step 1.2: fine synchronization and frequency offset estimation; and step 1.3: joint tracking loop.

[0011] Preferably, step 1.3: the joint tracking loop includes: step 1.3.1: frequency deviation tracking and step 1.3.2: sampling clock deviation tracking.

[0012] Preferably, step 2: robust channel estimation and equalization includes: step 2.1: initial channel estimation at the pilot; step 2.2: two-dimensional adaptive interpolation; and step 2.3: frequency domain equalization.

[0013] Preferably, step 2.2: two-dimensional adaptive interpolation includes: step 2.2.1: interpolation based on two-dimensional Wiener filtering and step 2.2.2: channel estimation based on two-dimensional discrete Fourier transform.

[0014] Preferably, step 3: soft information demapping and iterative decoding includes: step 3.1: soft demodulation and step 3.2: cascaded channel decoding.

[0015] Preferably, step 4: bitstream remodulation includes: step 4.1: channel coding; step 4.2: constellation mapping; step 4.3: OFDM modulation; and step 4.4: final output.

[0016] The beneficial effects of this invention are:

[0017] High synchronization accuracy and robustness: Employing a two-stage synchronization mechanism combining coarse and fine synchronization, along with an improved windowed peak detection algorithm and dynamic threshold, it effectively resists noise and multipath interference, significantly reducing the probability of false alarms and missed detections. The joint tracking loop (PLL+DLL) continuously compensates for carrier frequency deviation and sampling clock deviation, effectively suppressing phase noise and inter-carrier interference (ICI), ensuring stable timing and carrier synchronization even under high-speed movement and harsh channel conditions.

[0018] Accurate channel estimation and strong adaptability: By combining multiple channel estimation methods such as pilot averaging, two-dimensional Wiener filtering, and 2D-DFT interpolation, the optimal estimation strategy can be adaptively selected based on the time-varying and frequency-selective characteristics of the channel. This method significantly improves the accuracy of channel estimation while maintaining low computational complexity, and is particularly suitable for fast-changing channels and deep fading scenarios, laying a reliable foundation for subsequent equalization and demodulation.

[0019] Superior decoding performance and significant error correction capability: Employing a soft information demapping and iterative decoding mechanism, the decoding gain is significantly improved by calculating the bit-level log-likelihood ratio (LLR) and combining LDPC decoding with the Min-Sum algorithm and RS concatenated decoding. The introduction of an external iterative feedback mechanism further reduces the error floor, enabling the system to achieve reliable decoding even under extremely low signal-to-noise ratio conditions, providing a near-error-free bitstream for high-quality signal reconstruction.

[0020] The reconstructed signal boasts high purity and adherence to standards: Bitstream remodulation strictly follows the DTMB transmitter's coding, interleaving, modulation, and framing specifications, ensuring the generated reference signal is completely identical to the original transmitted signal in the time domain, frequency domain, and modulation format. An optional closed-loop correction mechanism further eliminates residual synchronization and estimation errors, ensuring the reconstructed signal achieves instrument-grade quality, making it suitable for high-precision applications such as channel diagnostics, interference analysis, and network optimization.

[0021] Flexible and widely applicable: The proposed method features a clear modular design and highly configurable parameters. Key parameters such as synchronization loop bandwidth, interpolation method, and number of iterations can be adjusted according to the actual receiving environment, balancing performance and complexity. It is not only suitable for fixed reception but can also be extended to mobile scenarios such as vehicle-mounted and portable systems, providing an effective tool for reception optimization, performance testing, and key technology verification in terrestrial digital television broadcasting systems. Attached Figure Description

[0022] The above-described aspects and advantages of the present invention will become apparent and readily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0023] Figure 1 This is a schematic diagram of a high-quality reconstruction method for terrestrial digital television multimedia broadcasting reference signals according to an embodiment of the present invention. Detailed Implementation

[0024] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0025] This invention proposes a high-quality reconstruction method for terrestrial digital television multimedia broadcasting reference signals, the specific steps of which are as follows:

[0026] Step 1: High-precision synchronization and parameter tracking

[0027] The goal of this stage is to accurately determine the frame start time of the signal and to estimate and compensate for carrier frequency offset (CFO) and sampling clock offset (SCO) in real time. The implementation is as follows:

[0028] Step 1.1: Coarse Synchronization

[0029] The receiver down-converts the signal to an intermediate frequency (IF), which is then sampled by an analog-to-digital converter (ADC) to obtain a discrete-time domain signal s(n). To combat noise, the signal is first digitally down-converted to baseband and then matched-filtered. Sliding correlation calculations are performed using two identical synchronization signals (PN sequences) from the terrestrial digital television multimedia broadcasting (DTMB) signal time slots.

[0030] R(n) = | sum_(i=n)^(n+Nsync-1) s(i) ×conj(s(i+Nsync)) |, where 1 <= n<= 2N;

[0031] Where: Nsync is the length of the synchronization sequence, which is 420 sampling points (corresponding to PN420 mode) or 945 sampling points (corresponding to PN945 mode) in the DTMB standard; N is the total number of sampling points in a time slot, which is typically 3780 (PN420 mode) or 4375 (PN945 mode) for an 8MHz bandwidth, 7.56MHz symbol rate system; to avoid missed detections, the sliding window length is set to 2N. To suppress the influence of noise spikes, an improved windowed peak detection algorithm is adopted:

[0032] Calculate the moving average of the sliding correlation sequence R(n): R_avg(n) = (1 / L) × sum_(i=nL / 2)^(n+L / 2) R(i), where L is the moving average window length, which is recommended to be an integer multiple of Nsync / 4, such as 105 points; set the dynamic threshold eta(n) = alpha × R_avg(n) + beta, where: alpha is the scaling factor, with a value ranging from 1.5 to 2.5, and a typical value of 2.0; beta is the fixed offset, with a value ranging from 0.1 to 0.5, used to prevent false triggering under extremely low signal-to-noise ratio.

[0033] When the value of R(n) first exceeds eta(n) and remains above 0.8×eta(n) within the subsequent Nsync points, the point n0 is determined to be the coarse synchronization position of the frame;

[0034] To improve accuracy, parabolic interpolation can be used to precisely locate the peak value near n0: Let n_max be the maximum value of R(n) near n0, then the precise location is:

[0035] n0_prime = n_max + [R(n_max-1) - R(n_max+1)] / [2 × (R(n_max-1) - 2×R(n_max) + R(n_max+1))].

[0036] Step 1.2: Fine synchronization and frequency offset estimation

[0037] Near the coarse synchronization position n0, two complete synchronization signal segments are extracted. The initial carrier frequency deviation Delta_f_init can be obtained by calculating the phase difference between these two sequences:

[0038] Delta_f_init = [1 / (2×pi×Nsync×Ts)] × arg[ sum_(i=0)^(Nsync-1) s(n0+i) × conj(s(n0+i+Nsync)) ];

[0039] Where: Ts is the sampling period, for an 8MHz bandwidth system, Ts = 1 / 32.5us ~= 30.769ns; arg[ ] represents the complex phase angle, in the range of (-pi, pi).

[0040] Use Delta_f_init to perform preliminary frequency offset compensation on the current frame signal:

[0041] s_corr(n) = s(n) × exp(-j×2×pi×Delta_f_init×n×Ts), where n is the index of the sampling point in the current frame.

[0042] Step 1.3: Joint Tracking Loop (used for continuous tracking of CFO and SCO)

[0043] After initial synchronization, the remaining CFO and SCO values ​​need to be continuously tracked and compensated in subsequent signal processing. This invention employs a pilot-assisted joint tracking loop.

[0044] Step 1.3.1: Frequency Deviation Tracking

[0045] The phase rotation information of the continuous pilot (or virtual continuous pilot composed of discrete pilots) in each OFDM symbol is utilized. Let the pilot value on the l-th OFDM symbol and the k-th subcarrier be Y(l,k), and the transmitted pilot value be P(l,k). Then the channel response at the pilot is: H(l,k) = Y(l,k) / P(l,k).

[0046] For continuous pilots, calculate the phase difference on the same subcarrier between adjacent symbols: Delta_phi(l,k) = arg[ H(l,k) × conj(H(l-1,k)) ]. Average the phase differences of all continuous pilots to obtain the average phase drift: Delta_phi_avg(l) = (1 / N_cp) × sum_(k in CP) Delta_phi(l,k). Where N_cp is the number of continuous pilots, and CP represents the continuous pilot index set.

[0047] The residual frequency offset is estimated as follows: Delta_f_res(l) = Delta_phi_avg(l) / (2×pi×T_u×Delta_l); where: T_u is the useful symbol duration of OFDM, and for the C=1 mode of DTMB, T_u = 500us; Delta_l is the symbol interval, which is usually 1.

[0048] The estimated value is fed into a second-order digital phase-locked loop (PLL) for smoothing and filtering. The PLL parameters are designed as follows: loop bandwidth B_L = 0.01×Rs, where Rs is the symbol rate, and B_L = 75.6kHz when the typical value is 7.56MHz; damping coefficient zeta = 0.707 (critical damping);

[0049] Loop filter coefficients: K_p = (4×zeta×B_L×T_s) / [1 + (2×zeta×B_L×T_s)^2] (proportional coefficients); K_i = (4×B_L^2×T_s^2) / [1 + (2×zeta×B_L×T_s)^2] (integral coefficients); The PLL update equation is: theta_e(l) = arg[ H(l,k) × exp(-j×theta_hat(l-1)) ] (phase error); theta_hat(l) = theta_hat(l-1) + K_p×theta_e(l) + K_i×sum_(i=0)^ltheta_e(i); The final frequency offset correction is: Delta_f_final = theta_hat(l) / (2×pi×T_u).

[0050] Step 1.3.2: Sampling Clock Deviation Tracking

[0051] SCO can cause inter-carrier interference (ICI) and phase drift. SCO is estimated by comparing the phase changes of pilots at different locations within the same OFDM symbol.

[0052] For the l-th OFDM symbol, select two pilot sets located at opposite ends of the frequency domain: low-frequency pilot set: k in[K_min, K_min + M - 1], where M is usually 8-16; high-frequency pilot set: k in [K_max - M + 1, K_max].

[0053] Calculate the average phase of the two sets: phi_low(l) = (1 / M) × sum_(k in low_set) arg[H(l,k)]; phi_high(l) = (1 / M) × sum_(k in high_set) arg[H(l,k)].

[0054] Since the phase caused by SCO changes linearly with frequency, the estimated value of SCO is: epsilon(l) = [phi_high(l) - phi_low(l)] / [2×pi×(f_high - f_low)×T_u]; where: f_high and f_low are the frequencies corresponding to the high-frequency and low-frequency pilots, respectively; epsilon is the normalized sampling clock deviation, that is, the relative deviation between the actual sampling frequency and the ideal sampling frequency; this estimated value is fed into a digital delay-locked loop (DLL). The loop filter design of the DLL is similar to that of the PLL, but the loop bandwidth is usually narrower, and it is recommended that B_L_DLL = 0.001×Rs = 7.56kHz.

[0055] The DLL controls a piecewise polynomial interpolation filter of a Farrow structure for resampling. The filter coefficients are: for the fractional delay mu in [0,1], the output sample y(n) = sum_(k=-2)^1 x(n+k) × h_k(mu); where: h_-2(mu) = (-mu^3 + 3×mu^2 - 3×mu + 1) / 6; h_-1(mu) = (3×mu^3 - 6×mu^2 + 4) / 6; h_0(mu) = (-3×mu^3 + 3×mu^2 + 3×mu + 1) / 6; h_1(mu) = mu^3 / 6.

[0056] Step 2: Robust Channel Estimation and Equalization

[0057] The synchronized and compensated frequency domain signal is Y(l,k). The goal of channel estimation is to obtain the accurate channel frequency response H_hat(l,k).

[0058] Step 2.1: Initial channel estimation at the pilot.

[0059] At a known pilot position (l_p, k_p), the channel response is: H_hat(l_p,k_p) = Y(l_p,k_p) / P(l_p,k_p); where P(l_p,k_p) is the known transmitted pilot value. To suppress pilot position noise, multiple adjacent identical pilots can be averaged.

[0060] For the DTMB system, the discrete pilots are periodically distributed on the time-frequency grid. Let the time-domain interval be D_t symbols and the frequency-domain interval be D_f subcarriers, then: for mode C=3780, D_t=1, D_f=12; for mode C=1, D_t=1, D_f=3; perform a 3-point average in both the time and frequency domains: H_hat_prime(l_p,k_p) = (1 / 9) × sum_(i=-1)^1 sum_(j=-1)^1 H_hat(l_p+i×D_t, k_p+j×D_f).

[0061] Step 2.2: Two-dimensional adaptive interpolation

[0062] An interpolation strategy is adaptively selected based on the time-varying characteristics and frequency selectivity of the channel.

[0063] Step 2.2.1: Interpolation based on two-dimensional Wiener filtering

[0064] Let the channel response vector at the pilot be H_p, and the response vector at the data subcarrier be H_d, which is estimated by the following formula: H_d = R_dp × inv(R_pp) × H_p; where: R_dp is the cross-correlation matrix between the data subcarrier and the pilot subcarrier; R_pp is the autocorrelation matrix between the pilot subcarriers; for a typical urban multipath channel, the correlation function can be modeled as: R(Delta_t,Delta_f) = J_0(2×pi×f_d×Delta_t) × exp(-abs(Delta_f) / f_c); where: J_0() is the zero-order Bessel function; f_d is the maximum Doppler shift, f_d = v×f_c / c, v is the moving speed, f_c is the carrier frequency, and c is the speed of light; f_c is the coherence bandwidth of the channel, f_c ~= 1 / (5×sigma_tau), where sigma_tau is the root mean square delay spread; in practical implementation, a simplified linear interpolation combined with frequency domain filtering can be used: Time-domain linear interpolation: For the k-th subcarrier, linear interpolation is performed between two pilot symbols; H_hat(l,k) = H_hat(l1,k) + (l-l1) / (l2-l1)× (H_hat(l2,k) - H_hat(l1,k)); where l1 and l2 are the positions of the two nearest pilot symbols; Frequency-domain MMSE filtering: For each symbol, frequency domain interpolation is performed using an MMSE filter of length L_f; H_hat(l,k) = sum_(i=-L_f / 2)^(L_f / 2) w_i × H_hat(l,k_p+i); the filter coefficients w_i are obtained by solving the Wiener-Hough equation: R_hh× w = r_hd; where R_hh is the autocorrelation matrix of the channel response at the pilot, and r_hd is the cross-correlation vector between the pilot and the data subcarrier.

[0065] Step 2.2.2: Channel estimation based on two-dimensional discrete Fourier transform (2D-DFT)

[0066] For fast-changing or highly frequency-selective channels, channel estimation based on 2D-DFT is adopted: H_hat(l_p,k_p) at the pilot position is placed on a time-frequency grid, and non-pilot positions are filled with zeros to obtain a sparse matrix H_grid; a 2D-IDFT transform is performed on this grid to the time-delay-Doppler domain: h_grid(tau, nu) = IDFT2D{ H_grid(l, k)}; the effective support region is determined based on the maximum channel delay spread tau_max and the maximum Doppler frequency shift nu_max: Time delay dimension: 0 <= tau <= tau_max is retained, and the rest are set to zero. A typical value of tau_max is 20µs (corresponding to 60 sampling points at a 3MHz sampling rate); Doppler dimension: -nu_max <= nu <= nu_max is retained, and the rest are set to zero. nu_max = f_d, typical value: when the vehicle speed is 120km / h and the carrier frequency is 800MHz, f_d ~= 89Hz; perform 2D-DFT transformation on the filtered result back to the time-frequency domain to obtain the channel response estimate H_hat(l,k) at all (l,k) positions;

[0067] Step 2.3: Frequency Domain Equalization

[0068] An MMSE equalizer is used to balance noise amplification and interference cancellation: X_hat(l,k) = [ Y(l,k) × conj(H_hat(l,k)) ] / [ |H_hat(l,k)|^2 + sigma_n^2 / sigma_s^2 ]; where: sigma_n^2 is the noise variance, which can be estimated from the pilot: sigma_n^2 = (1 / N_p) × sum_(p in pilot_set) |Y_p - H_hat_p×P_p|^2; sigma_s^2 is the average signal power, usually normalized to 1;

[0069] For signal-to-noise ratio (SNR) estimation, the following method can be used: SNR_est = [ (1 / N_p) × sum_(p in pilot_set) |H_hat_p×P_p|^2 ] / sigma_n^2; then the noise term in the equalizer coefficients can be expressed as: sigma_n^2 / sigma_s^2 = 1 / SNR_est; for low SNR cases, a regularization factor lambda can be introduced: X_hat(l,k) = [Y(l,k) × conj(H_hat(l,k)) ] / [ |H_hat(l,k)|^2 + lambda / SNR_est ]; where lambda ranges from 0.5 to 2.0, with a typical value of 1.0;

[0070] Step 3: Soft information demapping and iterative decoding

[0071] The goal of this stage is to convert the equalized complex symbols X_hat(l,k) into a highly reliable binary bit stream.

[0072] Step 3.1: Soft demodulation (soft decision demapping)

[0073] For M-QAM modulation, calculate the log-likelihood ratio (LLR) for each bit b_j. A Max-Log-MAP approximation is used to reduce computational complexity: for 64QAM, each symbol carries 6 bits (b_5 b_4 b_3 b_2 b_1 b_0). Let the equalized symbol be X_hat = I + j×Q, and the channel response amplitude be |H_hat|.

[0074] First, calculate the minimum Euclidean distance from the symbol to each bit that is 0 and 1: For the bit corresponding to the in-phase component I: d_I0_min = min_(s_I in S_I0) |I - |H_hat|×s_I|^2; d_I1_min = min_(s_I in S_I1)|I - |H_hat|×s_I|^2; where S_I0 and S_I1 are the sets of symbols in the in-phase component where the bit is 0 and 1, respectively. Then LLR is approximately: LLR(b_j) ~= (1 / (2×sigma_eff^2)) × (d_I0_min - d_I1_min) for the same bit; or LLR(b_j) ~= (1 / (2×sigma_eff^2)) × (d_Q0_min - d_Q1_min) for the orthogonal bit; where sigma_eff^2 is the equivalent noise variance: sigma_eff^2 = sigma_n^2 × [ |H_hat|^2 / (|H_hat|^2 + sigma_n^2 / sigma_s^2) ]^2; for 64QAM, the symbol sets S_I and S_Q are: {-7, -5, -3,-1, 1, 3, 5, 7}, and the normalization factor is 1 / sqrt(42). In practice, a lookup table method can be used to speed up the calculation: pre-calculate the LLR corresponding to all possible (I / |H_hat|, Q / |H_hat|) values ​​and store them as a lookup table.

[0075] Step 3.2: Cascaded Channel Decoding

[0076] The DTMB standard uses a concatenation of LDPC(7493,3048) or LDPC(7493,4572) codes and RS(208,188) codes. The decoding process is as follows:

[0077] Step 3.2.1: Deinterleave bits

[0078] The received LLR sequence is deinterleaved according to the DTMB specification. The interleaving mode is block interleaving, with the following parameters: number of rows: R = 384 (for 64QAM) or R = 192 (for 16QAM); number of columns: C = LDPC code length / R; deinterleaving algorithm: write by column, read by row.

[0079] Step 3.2.2: LDPC Decoding

[0080] Iterative decoding is performed using the Min-Sum algorithm of Soft-Input Soft-Output (SISO).

[0081] Let the parity-check matrix H of the LDPC code be M x N, where M = 7493 - 3048 = 4445 (for a code rate of 0.406) or M = 7493 - 4572 = 2921 (for a code rate of 0.61).

[0082] Initialization: Message from variable node to check node: Q(m,n) = LLR_n, where LLR_n is the initial LLR of the nth bit; Message from check node to variable node: R(m,n) = 0; Iteration process (maximum number of iterations I_max = 20~50):

[0083] Verification node update: For each verification node m, the set of variable nodes connected to it is N(m); R(m,n) =alpha × [ prod_(n_prime in N(m)-{n}) sign(Q(m,n_prime)) ] × min_(n_prime inN(m)-{n}) |Q(m,n_prime)|; where alpha is a correction factor, typically 0.75~0.9;

[0084] Variable node update: For each variable node n, the set of check nodes connected to it is M(n); Q(m,n) = LLR_n + sum_(m_prime in M(n)-{m}) R(m_prime,n); Total posterior information: L_n = LLR_n + sum_(m in M(n)) R(m,n); Hard decision: b_hat_n = 1 if L_n < 0 else 0; Check if all check equations are satisfied: H×b_hat = 0; If satisfied or the maximum number of iterations is reached, stop iterating; otherwise, return to step 1;

[0085] Step 3.2.3: RS Decoding

[0086] The hard decision byte stream output from the LDPC decoder is fed into the RS(208,188) decoder to correct any remaining burst errors.

[0087] RS decoding employs either the Berlekamp-Massey algorithm or the Euclidean algorithm: Calculate the adjoint polynomial S(x) = sum_(i=0)^(2×t-1) S_i×x^i, where S_i = sum_(j=0)^(n-1) r_j×alpha^(i×j), t = 10; Solve the key equation: Omega(x) = Lambda(x)×S(x) mod x^(2×t); Use Qian search to find the error location; Use the Furni algorithm to calculate the error value; Correct the error: c(x) = r(x) - e(x).

[0088] Step 3.2.4: Soft information iteration (external iteration)

[0089] To further improve performance, an external iteration can be introduced: re-encode the byte stream after RS ​​decoding into RS codewords; convert the re-encoded bytes into bits and calculate the confidence of each bit: for correctly decoded bytes, set the confidence of the corresponding bit to a high value (e.g., + / -10); for bytes with errors corrected, set the confidence of the corresponding bit to a medium value (e.g., + / -5); feed this prior information back to the LDPC decoder as a new initial LLR; perform LDPC decoding again, this time using: LLR_n_prime = LLR_n + gamma×L_prior_n; where gamma is the feedback factor, typically 0.3~0.7; repeat the above process 1~2 times.

[0090] Step 4: Bitstream Remodulation

[0091] After the above steps, a binary bit stream b with an extremely low bit error rate is obtained. The process of reconstructing the reference signal is completely deterministic:

[0092] Step 4.1: Channel Coding

[0093] For b, strictly re-encode according to the DTMB transmitter specification: RS encoding: Add 20 bytes of checksum to each 188-byte data block to generate a 208-byte RS codeword;

[0094] Byte interleaving: 12x17 convolutional interleaving is used, with interleaving depth B = 52 and interleaving mode M = 17;

[0095] LDPC encoding: Using the same generator matrix G as the transmitter, calculate c = u×G, where u is the information bit;

[0096] Bit interleaving: Block interleaving is used, and the number of rows R is determined according to the modulation method.

[0097] Step 4.2: Zodiac Mapping

[0098] Constellation mapping is performed according to the transmission mode: For 64QAM, every 6 bits are mapped to a complex symbol: In-phase component: I = (2×b_5 - 1)×(4×b_3 + 2×b_1 + 1); Quadrature component: Q = (2×b_4 - 1)×(4×b_2 + 2×b_0 + 1); Normalization: the symbol is multiplied by the normalization factor 1 / sqrt(42).

[0099] Step 4.3: OFDM Modulation

[0100] Map the complex symbol C(l,k) to the corresponding time-frequency resource grid and insert pilots: continuous pilots have fixed positions and known values; discrete pilots have positions that change according to a certain pattern and known values. Perform IFFT on each OFDM symbol: s_l(n) = (1 / sqrt(N_FFT)) × sum_(k=0)^(N_FFT-1) C(l,k)×exp(j×2×pi×k×n / N_FFT), n =0,...,N_FFT-1; where N_FFT is the number of FFT points, 4096 for C=3780 mode and 8192 for C=1 mode; add a cyclic prefix (CP): copy N_CP samples from the end of the symbol to the beginning; s_l_CP(n) = s_l(n+N_FFT-N_CP), n =0,...,N_CP-1 (cyclic prefix part); s_l_CP(n+N_CP) = s_l(n), n = 0,...,N_FFT-1 (data part); where N_CP is the cyclic prefix length, typically N_FFT / 8 or N_FFT / 16; Frame according to frame structure, add synchronization header (PN sequence): Frame header: PN sequence; Frame body: 3780 or 4375 OFDM symbols.

[0101] Step 4.4: Final Output

[0102] The generated time-domain discrete sequence s_ref(n) is the reference signal for high-quality reconstruction. To simulate actual transmission, an appropriate digital up-conversion can be added: s_RF(n) = Re{ s_ref(n) × exp(j×2×pi×f_IF×n×Ts)}; where f_IF is the intermediate frequency, which can be selected as 0 (baseband output) or a certain intermediate frequency value.

[0103] Step 5: Performance Enhancement Closed Loop (Optional Advanced Feature)

[0104] To achieve the highest reconstruction quality, a closed-loop correction mechanism can be introduced: the reconstructed clean reference signal s_ref(n) is compared with the received signal s_rx(n) after time delay alignment, and the error is calculated as: e(n) = s_rx(n) - s_ref(n)×h_est(n); where h_est(n) is the currently estimated channel impulse response;

[0105] Calculate the mean squared error: MSE = (1 / N) × sum_(n=0)^(N-1) |e(n)|^2; If the MSE exceeds the threshold (e.g., 10^-4), initiate closed-loop correction: a. Re-estimate the synchronization parameters using a longer training sequence; b. Update the channel estimate using the error signal e(n): H_hat_new = H_hat_old + mu×E×conj(S_ref) / |S_ref|^2; where mu is the step size, typically 0.01~0.1; c. Feed the updated parameters back to the synchronization and channel estimation modules; d. Reconstruct the signal; Repeat steps 1-3 until the MSE is below the threshold or the maximum number of iterations (e.g., 5 times) is reached.

[0106] Summary of implementation parameters:

[0107] For ease of implementation, key parameters are summarized as follows: Synchronization parameters: Sliding correlation window length: 2N (N = 3780 or 4375); Peak detection threshold: alpha = 2.0, beta = 0.2; PLL loop bandwidth: 0.01×Rs (Rs = 7.56MHz); DLL loop bandwidth: 0.001×Rs. Channel estimation parameters: Pilot averaging window: 3x3 (time domain x frequency domain); 2D-DFT filtering range: delay 0-20us, Doppler -100Hz~100Hz; Equalization regularization factor: lambda = 1.0. Decoding parameters: LDPC iterations: 20-50; Min-Sum correction factor: alpha = 0.8; Outer iterations: 1-2; Feedback factor: gamma = 0.5. Performance thresholds: Synchronization success: The relevant peak value exceeds the threshold by more than 3 times; Decoding success: LDPC verification passes and there are no errors after RS ​​error correction; Closed-loop trigger: MSE > 10^-4.

[0108] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for high-quality reconstruction of a terrestrial digital television multimedia broadcasting reference signal, characterized in that: Includes the following steps: Step 1: High-precision synchronization and parameter tracking; Step 2: Robust channel estimation and equalization; Step 3: Soft information demapping and iterative decoding; Step 4: Bitstream remodulation.

2. The method for high-quality reconstruction of terrestrial digital television multimedia broadcasting reference signals according to claim 1, characterized in that: Step 1: High-precision synchronization and parameter tracking includes: Step 1.1: Coarse synchronization; Step 1.2: Fine synchronization and frequency offset estimation; and Step 1.3: Joint tracking loop.

3. The method for high-quality reconstruction of terrestrial digital television multimedia broadcasting reference signals according to claim 2, characterized in that: Step 1.3: The joint tracking loop includes: Step 1.3.1: Frequency deviation tracking and Step 1.3.2: Sampling clock deviation tracking.

4. The method for high-quality reconstruction of terrestrial digital television multimedia broadcasting reference signals according to claim 1, characterized in that: Step 2: Robust channel estimation and equalization includes: Step 2.1: Initial channel estimation at the pilot; Step 2.2: Two-dimensional adaptive interpolation; and Step 2.3: Frequency domain equalization.

5. The method for high-quality reconstruction of terrestrial digital television multimedia broadcasting reference signals according to claim 4, characterized in that: Step 2.2: Two-dimensional adaptive interpolation includes: Step 2.2.1: Interpolation based on two-dimensional Wiener filtering and Step 2.2.2: Channel estimation based on two-dimensional discrete Fourier transform.

6. The method for high-quality reconstruction of terrestrial digital television multimedia broadcasting reference signals according to claim 1, characterized in that: Step 3: Soft information demapping and iterative decoding includes: Step 3.1: Soft demodulation Step 3.2: Cascaded channel decoding.

7. The method for high-quality reconstruction of terrestrial digital television multimedia broadcasting reference signals according to claim 1, characterized in that: Step 4: Bitstream remodulation includes: Step 4.1: Channel coding; Step 4.2: Constellation mapping; Step 4.3: OFDM modulation; and Step 4.4: Final output.

Images