Coding-assisted blind frame synchronization method and system based on maximum value detection

Abstract

The present invention discloses a coding-assisted blind frame synchronization method and system based on maximum value detection, which belongs to the field of communication signal processing. The present invention obtains a frame synchronization normalized confidence for frame synchronization position estimation based on the normalized satisfaction probability of the check nodes of the LDPC factor graph model, analyzes the relationship between the gain introduction point and the frame synchronization position according to the single peak characteristic of the frame synchronization normalized confidence, and completes the frame synchronization position estimation by determining the maximum value of the frame synchronization normalized confidence, thereby realizing frame synchronization and iterative decoding. The present invention does not need to insert a pilot to realize frame synchronization, significantly improving the throughput and transmission rate of the communication system; utilizing the structural characteristics of the coding itself to introduce coding gain through coding assistance, reducing the performance loss of the communication system; utilizing the single peak characteristic of the frame synchronization normalized confidence to find the maximum value, thereby estimating the frame synchronization position, realizing the frame synchronization process for the continuous communication system, and completing iterative decoding at the same time.

Description

Technical Field

[0001] The present invention relates to a maximum value detection-based coding-assisted blind frame synchronization method and system, and in particular to a maximum value detection-based LDPC coding-assisted blind frame synchronization method and system suitable for continuous communication systems, belonging to the field of communication signal processing. Background Art

[0002] Because channel coding and decoding technology can introduce coding gain and thus reduce the loss of communication system performance, it is often combined with frame synchronization, demodulation and other steps in the communication system receiver architecture to effectively improve the performance of the communication system. Code-assisted system architectures often iteratively update prior information based on external information to achieve maximum likelihood decoding, and use decoding iterative gain to improve the performance of processes such as frame synchronization and demodulation. Bit-Interleaved Coded Modulation-Iterative Decoding (BICM-ID), based on the combination of coding and modulation techniques, introduces channel coding and decoding gain into the demodulation process, effectively improving the demodulation threshold and system transmission efficiency.

[0003] Traditional communication receiver architectures typically achieve frame synchronization by inserting pilot symbols into the transmitted data time domain and combining them with binary hypothesis testing. The transmitter typically inserts a pilot symbol with a priori information at a fixed position in the data frame structure, and the receiver uses time-domain correlation or matched filtering to determine the frame start position. Due to the random nature of channel noise and data bits, to prevent missed detection or misestimation due to insufficient pilot information length, the pilot symbol inserted into the frame structure typically has a large time-domain footprint and signal energy, resulting in resource consumption and reduced transmission rate. Unlike traditional methods that add pilot symbols to the sequence, code-assisted blind frame synchronization leverages the inherent characteristics of the code structure to introduce coding gain. By eliminating the need for pilot symbol insertion for detection, transmission bandwidth efficiency is significantly improved. Furthermore, the coding gain is introduced into the frame synchronization process at the receiver, minimizing performance loss. However, current blind frame synchronization algorithms have high computational complexity. The iterative detection of sufficient compensation positions for each fixed frame increases system throughput and complexity. Furthermore, current receiver architectures based on blind frame synchronization mostly analyze LLR information in the digital domain, lacking probabilistic algorithm analysis and corresponding circuit implementation suitable for iterative message passing algorithms. Therefore, there is an urgent need to design low-power and low-complexity coding-assisted blind frame synchronization methods and systems. Summary of the Invention

[0004] In order to solve the following technical defects existing in the existing technology: (1) Traditional frame synchronization inserts a pilot with prior information in the time domain of the data frame structure, resulting in resource consumption and reduced transmission rate; (2) For continuous communication systems, the current receiver lacks the algorithm and architecture design to introduce coding gain into the frame synchronization technology, resulting in loss of communication system performance, and the current blind frame synchronization algorithm has large computational complexity and hardware complexity; The main purpose of the present invention is to provide a coding-assisted blind frame synchronization method and system based on maximum value detection, taking the LDPC decoding factor graph model as the research object, and determining the frame synchronization normalization confidence by the probability of normalization of the check nodes at each frame compensation position, analyzing the relationship between the gain introduction point and the frame synchronization position according to the single peak characteristic of the frame synchronization normalization confidence, and realizing frame synchronization position estimation by determining the maximum value of the frame synchronization normalization confidence, thereby realizing frame synchronization and iterative decoding of the continuous communication system. The present invention has the following advantages: (1) there is no need to insert pilot symbols at the transmitting end, and the coding gain is introduced through coding assistance by utilizing the structural characteristics of the coding itself to complete the frame synchronization process, thereby avoiding the problem of reduced spectrum utilization and significantly improving the system throughput and transmission rate; (2) the decoding factor graph is used to check the node normalization to meet the cost function of the probability determination of the frame synchronization position, thereby completing frame synchronization and iterative decoding, introducing coding gain compared with the traditional frame synchronization and decoding sequence processing architecture, and reducing the performance loss of the communication system; (3) the single peak characteristic of the frame synchronization normalization confidence is used to find the maximum value to realize the fast and accurate frame synchronization process of the continuous communication system.

[0005] The purpose of the present invention is achieved through the following technical solutions.

[0006] The present invention discloses a coding-assisted blind frame synchronization method based on maximum value detection. According to the normalized satisfaction probability of the check nodes of the LDPC factor graph model in the iterative decoding process, a frame synchronization normalized confidence for estimating frame synchronization is obtained. The frame synchronization normalized confidence is the cost function of the frame synchronization position. For a continuous communication system receiving sequence, the frame synchronization normalized confidence of each frame structure position is obtained by processing the data frames of each continuous time delay in the receiving sequence. According to the single peak characteristic of the frame synchronization normalized confidence, the relationship between the gain introduction point and the frame synchronization position is analyzed. The frame synchronization position estimation is realized by determining the maximum value of the frame synchronization normalized confidence, thereby realizing frame synchronization and iterative decoding. The present invention does not need to insert a pilot to achieve frame synchronization, solves the problem of reduced spectrum utilization, and significantly improves the throughput and transmission rate of the communication system; utilizes the structural characteristics of the coding itself to introduce coding gain through coding assistance, thereby reducing the performance loss of the communication system; utilizes the single peak characteristic of the normalized confidence level of frame synchronization to find the maximum value, and then estimates the frame synchronization position, realizing a fast and accurate frame synchronization process for continuous communication systems, and completing the iterative decoding process at the same time, compared with the traditional frame synchronization and decoding sequential processing architecture, it completes frame synchronization and decoding at the same time with low complexity.

[0007] The present invention discloses a coding-assisted blind frame synchronization method based on maximum value detection, comprising the following steps:

[0008] Step 1: Based on the LDPC decoding factor graph model, add check nodes to meet the probability detection node set GNs and connect them to the check nodes in a unidirectional manner; add check nodes to normalize the probability statistics node G ave , output the normalized satisfaction probability of the check node, and obtain the normalized confidence of frame synchronization for estimating frame synchronization. The normalized confidence of frame synchronization is the cost function of the frame synchronization position.

[0009] Establish the LDPC decoding factor graph model G = (VNs∪CNs,Ξ), where VNs represents the variable node set and the size is the row dimension N of the check matrix C ; CNs represents the check node set, the size of which is the column dimension N of the check matrix C -N b , where N C is the codeword sequence length, N b is the length of the information sequence; Ξ represents the edge set connecting the variable node and the check node. At the check node CNs=(C1, C2, ..., C k ) based on the addition of check nodes to meet the probability detection node GNs = (G1, G2, ..., G k ), output the check node to meet the probability Indicates that under the conditions of codeword sequence R and check matrix H, node G is detected after k iterations n Output check node C n The probability of meeting the check constraint, that is, the check node C n The probability that the sum of the symbolic information products modulo 2 of all variable nodes of the constraint is 0:

[0010]

[0011] Among them, C n Represents the check node receiving probability information, V m Represents the variable node that transmits probability information, g n Indicates the detection node symbol index, the value is 0 or 1, V m' ∈N(C n ) represents V m' Belongs to the check node C n The set of all connected variable nodes, {V m} represents the set of all variable nodes, the superscript k represents the decoding iteration cycle, I c (C n ) represents the check node C n Corresponding validation constraints Established, that is:

[0012]

[0013] Node G ave Output the normalized satisfaction probability P of the check node in the kth iteration cycle (k) (G(k)|R(τ)) is:

[0014]

[0015] Where ρ is the normalization factor, so that P (k) The value of (G(k)|R(τ)) ranges from 0 to 1. The summation symbol indicates that the satisfaction probabilities of all check nodes are summed, and the average value is finally obtained to represent the constraint satisfaction probability of all check nodes.

[0016] For the complete data frame received by the receiver, a complete codeword length of data in the received sequence is taken as a frame with different delays as the input of the factor graph model, that is, N in the sliding window Γ(τ). C Length information sequence R(τ:τ+N C -1), where τ is the frame synchronization transmission delay, defining the starting position of the sliding window. The factor graph output is bit B = (B1, B2, .... B Nb ), which represents the hard decision information of the approximate posterior probability information of the codeword.

[0017] Step 2: Determine the frame synchronization sliding window position Γ(τ), move the sliding window Γ(τ) to obtain the observation frame data R(τ: τ+N C -1), the information sequence R(τ:τ+N) in the sliding window C -1) is the input of the factor graph model.

[0018] A frame synchronization system model is established. Considering that the previous-stage timing synchronization and carrier synchronization have completed the accurate estimation of the timing error and carrier phase, the frame synchronization process is simplified to the estimation problem of the frame compensation amount τ0. r(t) represents the received signal input into the frame synchronization system after the AWNG signal, s(t) represents the transmitting end signal, n(t) is the noise superimposed at the symbol level, which is a complex Gaussian white noise with a mean of 0 and a unilateral power spectral density of N0, and τ0 represents the transmission delay. The frame synchronization system model is obtained as follows:

[0019] r(t)=s(t-τ0)+n(t)

[0020] Where s(t) is the symbol sent at time t, which is 0 or 1. Based on this system model, a frame synchronization sliding window Γ(τ) is set, and the sliding window length is selected as N C , that is, the sliding window always contains data of a complete coding sequence length.

[0021] By moving the frame synchronization sliding window in units of t on the received sequence, the frame structure of the relative position of the sliding window Γ(τ) is obtained, and the observation frame data R(τ:τ+N C -1). The correct frame synchronization probability corresponding to position τ is:

[0022] P(τ|r(t)),τ∈[0,n c -1]

[0023] τ represents the possible position of correct frame synchronization when the receiving sufficient statistics is r(t), n c is the length of the frame. By maximizing the above posterior probability by sliding the frame synchronization sliding window on r(t), the possible position of the correct frame synchronization when receiving r(t) is obtained.

[0024] Step 3: According to the observation frame data R(τ:τ+N C -1) to initialize the variable node.

[0025] The observation frame R(τ:τ+N C -1) Input the channel prior probability information into the factor graph model and initialize the variable nodes. According to the Gaussian white noise channel probability distribution, the initialization information of the factor graph variable nodes when the frame structure t = τ is:

[0026]

[0027] Among them, x i Indicates the symbolic index of the variable node, which takes a value of 0 or 1. Represents the sequence R(τ:τ+N) when the sliding window is Γ(τ) C -1) channel prior probability information, where φ(t) represents the amplitude of the received sequence, l∈[0,N C ) represents the sequence index in the frame synchronization sliding window, σ 2 is the channel noise variance.

[0028] Step 4: Update the global factor graph check nodes according to the variable node information and check constraint relationship.

[0029] In the known observation frame sequence R(τ:τ+N C -1) and the check constraint relationship H, the check node transmits the update probability information set to the variable node. That is, the check node passes the symbol x to the variable node i The probability of being 0 or 1:

[0030]

[0031] Among them, V m' ∈N(C n )\V m Indicates V m' Belongs to C n connected, excluding V m The set of all variable nodes, ~{V m} means except V m The set of all variable nodes except , the superscript k represents the decoding iteration cycle. Check node information Represents V m Other than check node C n The probability density function representation of all connected variable nodes modulo 2 sum to 0.

[0032] Step 5: According to R(τ:τ+N C -1) and the check node information are used to update the global factor graph variable nodes. When the maximum number of iterations is reached, the iteration process stops and jumps to step 6. Otherwise, jump to step 4 and repeat steps 4 to 5 until the maximum number of iterations is reached. This step ensures that the subsequent frame synchronization position estimation step is performed while fully incorporating coding gain.

[0033] The variable node transmits the updated probability information set to the check node That is, under the condition of known channel receiving sequence R and check matrix H, the variable node transmits symbol x to the check node i The probability of being 0 or 1, according to the verification constraint relationship and verification node information, the variable node update information set is:

[0034]

[0035] Among them, C n' ∈N(V m )\C n Indicates C n' Belongs to V m Connected, excluding C n The set of all check nodes, λ mn is the normalization parameter, so that the relationship satisfies

[0036] Step 6: Determine the frame synchronization normalization confidence Π(τ) based on the check node normalization satisfaction probability, and introduce the coding gain into the frame synchronization process.

[0037] When calculating the last iteration cycle, node G ave The normalized check node output satisfies the probability

[0038]

[0039] Where k0 is the stopping iteration period, i.e. the maximum number of iterations; Π(τ) represents the observation frame R(τ:τ+N c -1) Normalized confidence of frame synchronization after iterative update.

[0040] Step 7: Calculate the approximate posterior probability of the codeword and make a hard decision, output the decoding decision result of the frame synchronization position, and obtain the bit sequence after the frame synchronization and decoding process.

[0041] For the sequence R(τ0:τ0+N C -1) Update the approximate posterior probability and determine the output codeword. The approximate posterior probability of the codeword is:

[0042]

[0043] when When bit decision output B m =1, when When bit decision output B m =0, output decoding sequence

[0044] In step eight, the frame synchronization position is determined based on the normalized frame synchronization confidence obtained in step six. The relationship between the iterative gain introduction point and the frame synchronization position is analyzed using the single-peak characteristic of the normalized frame synchronization confidence. The frame synchronization position is estimated by determining the maximum value of the normalized frame synchronization confidence. By leveraging the inherent structural characteristics of the code to complete the frame synchronization process through the coding-assisted introduction of coding gain, the accuracy of frame synchronization is improved, eliminating the need to insert pilot symbols at the transmitter, avoiding the problem of reduced spectrum utilization, and significantly improving the throughput and transmission rate of the communication system.

[0045] The sliding window observation frame R(τ:τ+N C -1) frame synchronization normalized confidence Π(τ) and hard decision codeword sequence Store, and set τ = τ + 1, jump to step 2, repeat steps 2 to 7, iteratively update the information sequence of the next sliding window to calculate Π (τ). After the frame synchronization normalized confidence and hard decision codeword sequence of all sliding windows are calculated and stored, the frame synchronization normalized confidence Π (τ) of all sliding window observation frames is compared to find the maximum value. Corresponding sliding window but For the correct frame synchronization position, is the correct frame sequence, namely:

[0046]

[0047] From the changing trend of the normalized satisfaction probability of different code type check nodes in the iterative process, we can know that the non-frame synchronization sequence R(τ:τ+N c -1) The sequence of input factor graph is undecodable, and its check node normalization satisfies the probability P (k) (G(k)|R(τ)) converges or oscillates continuously in any interval less than 1, and finally converges to a value range less than the decoding success threshold. For the sequence R(τ0:τ0+N C -1), considering that the sequence of input factor graphs has the characteristics of encoding structure itself, it constitutes a decodable type, and the check node normalization satisfies the probability P (k) In most cases, (G(k)|R(τ)) can reach the decoding success threshold T, which is significantly greater than the normalized probability of satisfaction of the check nodes at the remaining non-frame synchronization positions. In low signal-to-noise ratio conditions, the sequence of the correct frame synchronization position, even if there are non-decodable types, still better meets the check constraint. Therefore, based on distribution statistics, it can be seen that Π(τ) at the correct frame synchronization position is generally greater than Π(τ) at the non-frame synchronization position. By finding the maximum Π(τ), the frame synchronization position can be determined.

[0048] Step nine: Output the decoding decision bit sequence for the correct frame synchronization position, completing the frame synchronization and decoding process. Compared to the conventional receiver frame synchronization and channel decoding sequential processing architecture, this invention introduces coding gain to achieve fast and accurate frame synchronization and channel decoding synchronization, minimizing communication system performance loss.

[0049] When determining the frame synchronization position Then, the sequence R(τ0:τ0+N c -1) hard decision codeword sequence As the decoded output. At this time, the frame synchronization position is determined according to the maximum value distribution of Π(τ) The corresponding decoding process is completed through an iterative message passing algorithm, and a decoding codeword sequence is output, thereby realizing frame synchronization and decoding process.

[0050] The present invention also discloses a maximum detection-based coding-assisted blind frame synchronization system for implementing the maximum detection-based coding-assisted blind frame synchronization method. The maximum detection-based coding-assisted blind frame synchronization system mainly comprises an observation frame sliding window module, a channel decoding module, a detection node module, a hard decision module, a buffer module, a confidence detection module, and a decoding output module.

[0051] The observation frame sliding window module is mainly used to implement the following functions: determine the observation frame structure on the received sequence according to the frame structure sliding window, and provide input for the channel decoding module.

[0052] The channel decoding module is mainly used to achieve the following functions: improve frame synchronization performance by introducing iterative gain through channel decoding, and complete the channel decoding process at the same time.

[0053] A detection node module is added on the basis of the traditional channel decoding module structure. The detection node module mainly functions to construct a cost function for estimating the frame synchronization position, namely the normalized confidence of the frame synchronization, and input the cost function into the cache module.

[0054] The hard decision module is mainly used to implement the following functions: through hard decision, the channel decoding soft information is judged as 0 or 1 bit, and the codeword sequence after frame synchronization and channel decoding is output to the cache module.

[0055] The cache module and confidence detection module are mainly used to implement the following functions: storing and comparing the frame synchronization normalized confidence, using the single peak characteristic of the frame synchronization normalized confidence to determine the correspondence between the iterative gain introduction point and the frame synchronization position, and estimating the frame synchronization position by performing maximum value detection on the cached frame synchronization normalized confidence.

[0056] The decoding output module is mainly used to realize the following functions: reading and outputting the hard decision cache data of the synchronization position to realize decoding output.

[0057] Beneficial effects:

[0058] 1. The present invention discloses a code-assisted blind frame synchronization method and system based on maximum value detection, which achieves frame synchronization without inserting pilot signals, solves the problem of reduced spectrum utilization, and significantly improves the throughput and transmission rate of the communication system;

[0059] 2. The present invention discloses a maximum detection-based coding-assisted blind frame synchronization method and system, which utilizes the structural characteristics of the coding itself to introduce channel coding gain into frame synchronization. Blind frame synchronization is achieved by analyzing the correspondence between the iterative gain introduction point and the frame synchronization position, thereby reducing communication system performance loss.

[0060] 3. The present invention discloses a coding-assisted blind frame synchronization method and system based on maximum value detection, which uses the single-peak characteristic of the normalized confidence of frame synchronization to find the maximum value, and then estimates the frame synchronization position, to achieve a fast and accurate frame synchronization process for continuous communication systems, and simultaneously complete the iterative decoding process, compared with the traditional frame synchronization and decoding sequential processing architecture, it achieves frame synchronization and decoding at the same time with low complexity. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 This is a flow chart of a coding-assisted blind frame synchronization method based on maximum value detection according to the present invention;

[0062] Figure 2It is a decoding factor graph model with added detection nodes in the coding-assisted blind frame synchronization method based on maximum value detection according to an embodiment of the present invention;

[0063] Figure 3 Schematic diagram of the relationship between the sliding window and the observation frame structure of the frame synchronization system model according to an embodiment of the present invention;

[0064] Figure 4 is a relationship diagram between the normalized confidence of frame synchronization and the frame synchronization position according to an embodiment of the present invention;

[0065] Figure 5 This is a simulation curve comparing the frame synchronization error rate (FSER) performance of the blind frame synchronization method described in this embodiment of the invention for the LDPC code type (480, 240);

[0066] Figure 6 This is a simulation curve comparing the bit error rate performance of the blind frame synchronization method described in the embodiment of the invention for the LDPC code type (480, 240);

[0067] Figure 7 This is a hardware architecture design of a coding-assisted blind frame synchronization system based on maximum value detection according to an embodiment of the present invention; DETAILED DESCRIPTION

[0068] To make the above-mentioned objects, features and advantages of the present invention easier to understand, the following is a further detailed description thereof with reference to the accompanying drawings and specific embodiments. This embodiment is directed to the LDPC (480, 240) code type, a continuous frame coding sequence is a data frame, and a code-assisted blind frame synchronization method based on maximum value detection is used to achieve frame synchronization and decoding. The system parameters in this embodiment are shown in the following table:

[0069] parameter Details Modulation method BPSK Bitrate 1 / 2 LDPC code type (480，240) Channel Model Gaussian white noise Maximum number of decoder iterations 50

[0070] like Figure 1 As shown, the present embodiment discloses a coding-assisted blind frame synchronization method based on maximum value detection, and the specific implementation steps are as follows:

[0071] Step 1, such as Figure 2 As shown, based on the LDPC decoding factor graph model, the check nodes are added to meet the probability detection node set GNs and are unidirectionally connected to the check nodes; the check nodes are added to normalize the probability statistics node G ave , output the normalized satisfaction probability of the check node, and obtain the normalized confidence of frame synchronization for estimating frame synchronization. The normalized confidence of frame synchronization is the cost function of the frame synchronization position.

[0072] Establish the LDPC decoding factor graph model G = (VNs∪CNs,Ξ), where VNs represents the variable node set and the size is the row dimension N of the check matrix C ; CNs represents the check node set, the size of which is the column dimension N of the check matrix C -N b , where N C is the codeword sequence length, N b is the length of the information sequence; Ξ represents the edge set connecting the variable node and the check node. At the check node CNs=(C1, C2, ..., C k ) based on the addition of check nodes to meet the probability detection node GNs = (G1, G2, ..., G k ), output the check node to meet the probability Indicates that under the conditions of codeword sequence R and check matrix H, node G is detected after k iterations n Output check node C n The probability of meeting the check constraint, that is, the check node C n The probability that the sum of the symbolic information products modulo 2 of all variable nodes of the constraint is 0:

[0073]

[0074] Among them, C n Represents the check node receiving probability information, V m Represents the variable node that transmits probability information, g n Indicates the detection node symbol index, the value is 0 or 1, V m' ∈N(C n ) represents V m' Belongs to the check node C n The set of all connected variable nodes, {V m} represents the set of all variable nodes, the superscript k represents the decoding iteration cycle, I c (C n ) represents the check node C n Corresponding validation constraints Established, that is:

[0075]

[0076] Node G ave Output the normalized satisfaction probability P of the check node in the kth iteration cycle (k) (G(k)|R(τ)) is:

[0077]

[0078] Where ρ is the normalization factor, so that P (k)The value of (G(k)|R(τ)) ranges from 0 to 1. The summation symbol indicates that the satisfaction probabilities of all check nodes are summed, and the average value is finally obtained to represent the constraint satisfaction probability of all check nodes.

[0079] For the complete data frame received by the receiver, a complete codeword length of data in the received sequence is taken as a frame with different delays as the input of the factor graph model, that is, N in the sliding window Γ(τ). C Length information sequence R(τ:τ+N C -1), where τ is the frame synchronization transmission delay, defining the starting position of the sliding window. The factor graph output is bit Hard decision information representing the approximate a posteriori probability information of the codeword.

[0080] Step 2: Determine the frame synchronization sliding window position Γ(τ), move the sliding window Γ(τ) to obtain the observation frame data R(τ: τ+N C -1), the information sequence R(τ:τ+N) in the sliding window C -1) is the input of the factor graph model, and the relationship between the sliding window and the observation frame structure is shown in the figure below. Figure 3 shown.

[0081] A frame synchronization system model is established. Considering that the previous-stage timing synchronization and carrier synchronization have completed the accurate estimation of the timing error and carrier phase, the frame synchronization process is simplified to the estimation problem of the frame compensation amount τ0. r(t) represents the received signal input into the frame synchronization system after the AWNG signal, s(t) represents the transmitting end signal, n(t) is the noise superimposed at the symbol level, which is a complex Gaussian white noise with a mean of 0 and a unilateral power spectral density of N0, and τ0 represents the transmission delay. The frame synchronization system model is obtained as follows:

[0082] r(t)=s(t-τ0)+n(t)

[0083] Where s(t) is the symbol sent at time t, which is 0 or 1. Based on this system model, a frame synchronization sliding window Γ(τ) is set, and the sliding window length is selected as N C , that is, the sliding window always contains data of a complete coding sequence length.

[0084] By moving the frame synchronization sliding window in units of t on the received sequence, the frame structure of the relative position of the sliding window Γ(τ) is obtained, and the observation frame data R(τ:τ+N C -1). The correct frame synchronization probability corresponding to position τ is:

[0085] P(τ|r(t)),τ∈[0,n c -1]

[0086] τ represents the possible position of correct frame synchronization when the receiving sufficient statistics is r(t), n c is the length of the frame. By maximizing the above posterior probability by sliding the frame synchronization sliding window on r(t), the possible position of the correct frame synchronization when receiving r(t) is obtained.

[0087] Step 3: According to the observation frame data R(τ:τ+N C -1) to initialize the variable node.

[0088] The observation frame R(τ:τ+N C -1) Input the channel prior probability information into the factor graph model and initialize the variable nodes. According to the Gaussian white noise channel probability distribution, the initialization information of the factor graph variable nodes when the frame structure t = τ is:

[0089]

[0090] Among them, x i Indicates the symbolic index of the variable node, which takes a value of 0 or 1. Represents the sequence R(τ:τ+N) when the sliding window is Γ(τ) C -1) channel prior probability information, where φ(t) represents the amplitude of the received sequence, l∈[0,N C ) represents the sequence index in the frame synchronization sliding window, σ 2 is the channel noise variance.

[0091] Step 4: Update the global factor graph check nodes according to the variable node information and check constraint relationship.

[0092] In the known observation frame sequence R(τ:τ+N C -1) and the check constraint relationship H, the check node transmits the update probability information set to the variable node. That is, the check node passes the symbol x to the variable node i The probability of being 0 or 1:

[0093]

[0094] Among them, V m' ∈N(C n )\V m Indicates V m' Belongs to C n connected, excluding V m The set of all variable nodes, ~{V m} means except V m The set of all variable nodes except , the superscript k represents the decoding iteration cycle. Check node information Represents Vm Other than check node C n The probability density function representation of all connected variable nodes modulo 2 sum to 0.

[0095] Step 5: According to R(τ:τ+N C -1) and the check node information are used to update the global factor graph variable nodes. When the maximum number of iterations is reached, the iteration process stops and jumps to step 6. Otherwise, jump to step 4 and repeat steps 4 to 5 until the maximum number of iterations is reached. This step ensures that the subsequent frame synchronization position estimation step is performed while fully incorporating coding gain.

[0096] The variable node transmits the updated probability information set to the check node That is, under the condition of known channel receiving sequence R and check matrix H, the variable node transmits symbol x to the check node i The probability of being 0 or 1, according to the verification constraint relationship and verification node information, the variable node update information set is:

[0097]

[0098] Among them, C n' ∈N(V m )\C n Indicates C n' Belongs to V m Connected, excluding C n The set of all check nodes, λ mn is the normalization parameter, so that the relationship satisfies

[0099] Step 6: Determine the frame synchronization normalization confidence Π(τ) based on the check node normalization satisfaction probability, and introduce the coding gain into the frame synchronization process.

[0100] When calculating the last iteration cycle, node G ave The normalized check node output satisfies the probability

[0101]

[0102] Where k0 is the stopping iteration period, i.e. the maximum number of iterations; Π(τ) represents the observation frame R(τ:τ+N c -1) Normalized confidence of frame synchronization after iterative update.

[0103] Step 7: Calculate the approximate posterior probability of the codeword and make a hard decision, output the decoding decision result of the frame synchronization position, and obtain the bit sequence after the frame synchronization and decoding process.

[0104] For the sequence R(τ0:τ0+N C -1) Update the approximate posterior probability and determine the output codeword. The approximate posterior probability of the codeword is:

[0105]

[0106] when When bit decision output B m =1, when When bit decision output B m =0, output decoding sequence

[0107] Step 8: Determine the frame synchronization position based on the frame synchronization normalization confidence obtained in step 6, and analyze the relationship between the iterative gain introduction point and the frame synchronization position using the single peak characteristic of the frame synchronization normalization confidence to determine the maximum value of the frame synchronization normalization confidence to achieve frame synchronization position estimation. The relationship between the frame synchronization normalization confidence and the frame synchronization position when Eb / N0 is 2dB is as follows: Figure 4 As shown in the figure, ITE represents the maximum number of iterations. The curves in the figure show the normalized confidence level of frame synchronization at each frame compensation position under different iteration settings. For a received sequence of length 480, the frame compensation positions are [-240, 240]. After iterative update, each frame compensation position outputs the normalized confidence level of frame synchronization. Due to the introduction of iteration gain, the frame offset corresponding to the maximum normalized confidence level of frame synchronization is the correct frame synchronization position. By leveraging the inherent structural characteristics of the code to achieve frame synchronization through coding-assisted coding gain, the accuracy of frame synchronization is improved, eliminating the need to insert pilot symbols at the transmitter, avoiding the problem of reduced spectrum utilization, and significantly improving the throughput and transmission rate of the communication system.

[0108] The sliding window observation frame R(τ:τ+N C -1) frame synchronization normalized confidence Π(τ) and hard decision codeword sequence Store, and set τ = τ + 1, jump to step 2, repeat steps 2 to 7, iteratively update the information sequence of the next sliding window to calculate Π (τ). After the frame synchronization normalized confidence and hard decision codeword sequence of all sliding windows are calculated and stored, the frame synchronization normalized confidence Π (τ) of all sliding window observation frames is compared to find the maximum value. Corresponding sliding window but For the correct frame synchronization position, is the correct frame sequence, namely:

[0109]

[0110] From the changing trend of the normalized satisfaction probability of different code type check nodes in the iterative process, we can know that the non-frame synchronization sequence R(τ:τ+N c -1) The sequence of input factor graph is undecodable, and its check node normalization satisfies the probability P (k) (G(k)|R(τ)) converges or oscillates continuously in any interval less than 1, and finally converges to a value range less than the decoding success threshold. For the sequence R(τ0:τ0+N C -1), considering that the sequence of input factor graphs has the characteristics of encoding structure itself, it constitutes a decodable type, and the check node normalization satisfies the probability P (k) In most cases, (G(k)|R(τ)) can reach the decoding success threshold T, which is significantly greater than the normalized probability of satisfaction of the check nodes at the remaining non-frame synchronization positions. In low signal-to-noise ratio conditions, the sequence of the correct frame synchronization position, even if there are non-decodable types, still better meets the check constraint. Therefore, based on distribution statistics, it can be seen that Π(τ) at the correct frame synchronization position is generally greater than Π(τ) at the non-frame synchronization position. By finding the maximum Π(τ), the frame synchronization position can be determined.

[0111] Step nine: Output the decoding decision bit sequence for the correct frame synchronization position, completing the frame synchronization and decoding process. Compared to the conventional receiver frame synchronization and channel decoding sequential processing architecture, this invention introduces coding gain to achieve fast and accurate frame synchronization and channel decoding synchronization, minimizing communication system performance loss.

[0112] When determining the frame synchronization position Then, the sequence R(τ0:τ0+N c -1) hard decision codeword sequence As the decoded output. At this time, the frame synchronization position is determined according to the maximum value distribution of Π(τ) The corresponding decoding process is completed through an iterative message passing algorithm, and a decoding codeword sequence is output, thereby realizing frame synchronization and decoding process.

[0113] The frame synchronization error rate (FSER) performance analysis was performed on the code type of this embodiment. The comparison algorithm selected the frame synchronization method mentioned in the journal "Optimum frame synchronization" published by J. Massey in 1972 in IEEE Transactions on Communications. The pilot hard decision, soft decision and optimal pilot methods were used to perform FSER performance simulation comparisons with the blind frame synchronization method described in this embodiment. The simulation results are shown in Figure 2. Figure 5As shown in the figure. Due to the coding gain introduced by this blind frame synchronization method, the FSER performance has a "waterfall region" and is close to and better than the pilot-based frame synchronization algorithm at high signal-to-noise ratios. It not only solves the transmission rate drop problem caused by the use of pilots, but also reduces system performance loss and ensures frame synchronization performance. The performance of the bit error rate under different average iteration number settings is analyzed, and the results are shown in the figure. Figure 6 As shown, it can be seen that it maintains the excellent decoding performance of LDPC and has a coding gain of several dB at medium and high signal-to-noise ratios.

[0114] This embodiment also discloses a maximum detection-based coding-assisted blind frame synchronization system for implementing the maximum detection-based coding-assisted blind frame synchronization method. The maximum detection-based coding-assisted blind frame synchronization system is designed and analyzed according to the factor graph model in step 1, and mainly consists of an observation frame sliding window module, a channel decoding module, a detection node module, a hard decision module, a buffer module, a confidence detection module, and a decoding output module. Figure 7 shown.

[0115] The observation frame sliding window module involves step two, which determines the observation frame structure on the received sequence based on the frame structure sliding window, and provides input for the channel decoding module. The channel decoding module involves steps three, four, and five, which introduces iterative gain through channel decoding to improve frame synchronization performance and complete the channel decoding process at the same time. The detection node module involves step six, which adds a detection node module based on the traditional channel decoding module structure, analyzes and detects the behavior of the check nodes in the system model, obtains the frame synchronization normalized confidence by normalizing the check nodes to meet the probability, and caches it. The hard decision module involves step seven, which implements the approximate posterior probability calculation and hard decision of the codeword, and judges the soft information as 0 or 1 bit and caches it. The cache module and confidence detection module involve step eight, which stores and compares the frame synchronization normalized confidence, uses the single peak characteristic of the frame synchronization normalized confidence to determine the correspondence between the iterative gain introduction point and the frame synchronization position, and estimates the frame synchronization position by performing maximum value detection on the cached frame synchronization normalized confidence. The decoding output module involves step nine, reading and outputting the hard decision cache data of the synchronization position to achieve decoding output.

[0116] The above specific description further illustrates the purpose, technical solutions and beneficial effects of the invention in detail. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A coding-assisted blind frame synchronization method based on maximum value detection, characterized in that: The following steps are included: Step 1: Based on the LDPC decoding factor graph model, add check nodes to satisfy the probability detection node set GNs and connect them unidirectionally. Add check nodes to normalize and satisfy the probability statistics node G ave , output the normalized satisfaction probability of the check node, and obtain the normalized confidence of frame synchronization for estimating frame synchronization, wherein the normalized confidence of frame synchronization is the cost function of the frame synchronization position; Step 2: Determine the position of the frame synchronization sliding window Γ(τ), move the sliding window Γ(τ) to obtain the observation frame data R(τ: τ+N C -1), the observation frame data R(τ:τ+N) in the sliding window C -1) is the input of the factor graph model; Step 3: According to the observation frame data R(τ:τ+N C -1)’s amplitude initialization variable node; Step 4: Update the global factor graph check nodes according to the variable node information and check constraint relationship; Step 5: According to the observation frame data R(τ:τ+N C -1) updates the global factor graph variable nodes; when the maximum number of iterations is reached, the iteration process is stopped and the process jumps to step 6; otherwise, the process jumps to step 4 and repeats steps 4 to 5 until the maximum number of iterations is reached and the iteration is stopped; this step ensures that the subsequent frame synchronization position estimation step is performed with sufficient coding gain; Step 6: Determine the frame synchronization normalization confidence Π(τ) based on the check node normalization satisfaction probability, and introduce the coding gain into the frame synchronization process; Step 7: Calculate the approximate posterior probability of the codeword and make a hard decision, output the decoding decision result of the frame synchronization position, and obtain the bit sequence after the frame synchronization and decoding process. Step 8: Determine the frame synchronization position based on the frame synchronization normalization confidence obtained in step 6, analyze the relationship between the iterative gain introduction point and the frame synchronization position using the single peak characteristic of the frame synchronization normalization confidence, and determine the maximum value of the frame synchronization normalization confidence to achieve frame synchronization position estimation; since the coding gain is introduced by coding assistance using the coding structure characteristics to complete the frame synchronization process, the frame synchronization accuracy is improved, and there is no need to insert pilot symbols at the transmitting end, avoiding the problem of reduced spectrum utilization, and significantly improving the throughput and transmission rate of the communication system; Step nine: Output the decoding decision bit sequence of the correct frame synchronization position to complete the frame synchronization and decoding process; compared with the traditional receiver frame synchronization and channel decoding sequential processing architecture, the introduction of coding gain achieves fast and accurate frame synchronization and channel decoding synchronization, reducing the loss of communication system performance.

2. The coding-assisted blind frame synchronization method based on maximum value detection according to claim 1, wherein: The implementation method of step one is: Establish the LDPC decoding factor graph model G = (VNsvCNs,Ξ), where VNs represents the variable node set and the size is the row dimension N of the check matrix C ; CNs represents the check node set, the size of which is the column dimension N of the check matrix C -N b , where N C is the codeword sequence length, N b is the length of the information sequence; Ξ represents the edge set connecting the variable node and the check node; at the check node CNs=(C1, C2, ..., C k ) based on the addition of check nodes to meet the probability detection node GNs = (G1, G2, ..., G k ), output the check node to meet the probability Indicates that under the conditions of codeword sequence R and check matrix H, node G is detected after k iterations n Output check node C n The probability of meeting the check constraint, that is, the check node C n The probability that the sum of the symbolic information products modulo 2 of all variable nodes of the constraint is 0: Among them, C n Represents the check node receiving probability information, V m Represents the variable node that transmits probability information, g n Indicates the detection node symbol index, the value is 0 or 1, V m' ∈N(C n ) represents V m' Belongs to the check node C n The set of all connected variable nodes, {V m } represents the set of all variable nodes, the superscript k represents the decoding iteration cycle, I c (C n ) represents the check node C n Corresponding validation constraints Established, that is: Node G ave Output the normalized satisfaction probability P of the check node in the kth iteration cycle (k) (G(k)|R(τ)) is: Where ρ is the normalization factor, so that P (k) The value of (G(k)|R(τ)) ranges from 0 to 1. The summation symbol indicates that the satisfaction probability of all check nodes is summed up, and the average value is finally obtained to represent the constraint satisfaction probability of all check nodes. For the complete data frame received by the receiver, a complete codeword length of data in the received sequence is taken as a frame with different delays as the input of the factor graph model, that is, N in the sliding window Γ(τ). C Length observation frame data R(τ:τ+N C -1), where τ is the frame synchronization transmission delay, defining the starting position of the sliding window; the factor graph output is bit Hard decision information representing the approximate a posteriori probability information of the codeword.

3. The coding-assisted blind frame synchronization method based on maximum value detection according to claim 2, wherein: The implementation method of step 2 is: A frame synchronization system model is established. Considering that the previous-stage timing synchronization and carrier synchronization have completed the accurate estimation of the timing error and carrier phase, the frame synchronization process is simplified to the estimation problem of the frame compensation amount τ0. r(t) represents the received signal input into the frame synchronization system after the AWNG signal, s(t) represents the transmitting end signal, n(t) is the noise superimposed at the symbol level, which is a complex Gaussian white noise with a mean of 0 and a unilateral power spectral density of N0, and τ0 represents the transmission delay. The frame synchronization system model is obtained as follows: r(t)=s(t-τ0)+n(t) Where s(t) is the symbol sent at time t, which is 0 or 1. Based on this system model, a frame synchronization sliding window Γ(τ) is set, and the sliding window length is selected as N. C , that is, the sliding window always contains data of a complete coding sequence length; By moving the frame synchronization sliding window in units of t on the received sequence, the frame structure of the relative position of the sliding window Γ(τ) is obtained, and the observation frame data R(τ:τ+N C -1); the probability of correct frame synchronization corresponding to position τ is: P(τ|r(t)),τ∈[0,n c -1] τ represents the possible position of correct frame synchronization when the receiving sufficient statistics is r(t), n c is the length of the frame; by maximizing the above posterior probability by sliding the frame synchronization sliding window on r(t), the possible position of the correct frame synchronization when receiving r(t) is obtained .

4. The coding-assisted blind frame synchronization method based on maximum value detection according to claim 3, wherein: The implementation method of step three is: The observation frame data R(τ:τ+N C -1) Input the channel prior probability information into the factor graph model and initialize the variable nodes. According to the Gaussian white noise channel probability distribution, the initialization information of the factor graph variable nodes when the frame structure t = τ is: Among them, x i Indicates the symbolic index of the variable node, with a value of 0 or 1; Represents the observation frame data R(τ:τ+N) when the sliding window is Γ(τ) C -1) channel prior probability information, where φ(t) represents the amplitude of the received sequence, l∈[0,N C ) represents the sequence index in the frame synchronization sliding window, σ 2 is the channel noise variance.

5. The coding-assisted blind frame synchronization method based on maximum value detection according to claim 4, wherein: The implementation method of step 4 is: In the known observation frame data R(τ:τ+N C -1) and the check constraint relationship H, the check node transmits the update probability information set to the variable node. That is, the check node passes the symbol x to the variable node i The probability of being 0 or 1: Among them, V m' ∈N(C n )\V m Indicates V m' Belongs to C n connected, excluding V m The set of all variable nodes, ~{V m } means except V m The set of all variable nodes except the one with superscript k indicating the decoding iteration cycle; the check node information Represents V m Other than check node C n The probability density function representation of all connected variable nodes modulo 2 sum to 0.

6. The coding-assisted blind frame synchronization method based on maximum value detection according to claim 5, wherein: The implementation method of step five is: The variable node transmits the updated probability information set to the check node That is, under the condition of known channel receiving sequence R and check matrix H, the variable node transmits symbol x to the check node i The probability of being 0 or 1, according to the verification constraint relationship and verification node information, the variable node update information set is: Among them, C n' ∈N(V m )\C n Indicates C n' Belongs to V m Connected, excluding C n The set of all check nodes, λ mn is the normalization parameter, so that the relationship satisfies When the maximum number of iterations is reached, the iteration process is stopped and the process jumps to step six. Otherwise, the process jumps to step four and repeats steps four to five until the maximum number of iterations is reached and the iteration is stopped. This step ensures that the subsequent frame synchronization position estimation step is performed with sufficient coding gain.

7. The coding-assisted blind frame synchronization method based on maximum value detection according to claim 6, wherein: The implementation method of step six is: When calculating the last iteration cycle, node G ave The normalized check node output satisfies the probability Where k0 is the stopping iteration period, i.e. the maximum number of iterations; Π(τ) represents the observation frame data R(τ:τ+N c -1) Normalized confidence of frame synchronization after iterative update; The implementation method of step seven is: For the sequence data R(τ0:τ0+N C -1) Update the approximate posterior probability and determine the output codeword. The approximate posterior probability of the codeword is: when When bit decision output B m =1, when When bit decision output B m =0, output decoding sequence .

8. The coding-assisted blind frame synchronization method based on maximum value detection according to claim 7, wherein: The implementation method of step eight is: The sliding window observation frame data R(τ:τ+N C -1) frame synchronization normalized confidence Π(τ) and hard decision codeword sequence Store, and set τ = τ + 1, jump to step 2, repeat steps 2 to 7, iteratively update the information sequence of the next sliding window to calculate Π(τ); when the frame synchronization normalized confidence and hard decision codeword sequence of all sliding windows are calculated and stored, the frame synchronization normalized confidence Π(τ) of all sliding window observation frames is compared to find the maximum value Corresponding sliding window but For the correct frame synchronization position, is the correct frame sequence, namely: From the changing trend of the normalized satisfaction probability of different code type check nodes in the iterative process, we can know that the non-frame synchronization sequence R(τ:τ+N c -1) The sequence of input factor graph is undecodable, and its check node normalization satisfies the probability P (k) (G(k)|R(τ)) converges or oscillates continuously in any interval less than 1, and finally converges to a value range less than the decoding success threshold; and for the sequence of the correct frame synchronization position, according to R(τ0:τ0+N C -1), considering that the sequence of input factor graphs has the characteristics of encoding structure itself, it constitutes a decodable type, and the check node normalization satisfies the probability P (k) In most cases, (G(k)|R(τ)) can reach the decoding success threshold T, which is much larger than the normalized satisfaction probability of the check nodes at the other non-frame synchronization positions. In the case of low signal-to-noise ratio, the sequence of the correct frame synchronization position still better conforms to the check constraint even if there is a non-decodable type. Therefore, according to distribution statistics, it can be seen that Π(τ) of the correct frame synchronization position is greater than Π(τ) of the non-frame synchronization position in most cases. The frame synchronization position can be determined by finding the maximum value of Π(τ).

9. The coding-assisted blind frame synchronization method based on maximum value detection according to claim 8, wherein: The implementation method of step nine is: When determining the frame synchronization position Then, the sequence R(τ0:τ0+N c -1) hard decision codeword sequence As the decoding output; at this time, the frame synchronization position is determined according to the maximum distribution of Π(τ) The corresponding decoding process is completed through an iterative message passing algorithm, and a decoding codeword sequence is output, thereby realizing frame synchronization and decoding process.

10. A maximum detection-based coding-assisted blind frame synchronization system, configured to implement a maximum detection-based coding-assisted blind frame synchronization method according to claim 1, 2, 3, 4, 5, 6, 7, 8, or 9, characterized in that: It mainly consists of observation frame sliding window module, channel decoding module, detection node module, hard decision module, cache module, confidence detection module and decoding output module; The observation frame sliding window module is mainly used to implement the following functions: determine the observation frame structure on the received sequence according to the frame structure sliding window, and provide input for the channel decoding module; The channel decoding module is mainly used to achieve the following functions: improve frame synchronization performance by introducing iterative gain through channel decoding, and complete the channel decoding process at the same time; A detection node module is added to the traditional channel decoding module structure. The detection node module is mainly used to construct a cost function for estimating the frame synchronization position, namely the normalized confidence of the frame synchronization, and input the cost function into the cache module. The hard decision module is mainly used to implement the following functions: through hard decision, the channel decoding soft information is judged as 0 or 1 bits, and the codeword sequence after frame synchronization and channel decoding is output to the buffer module; The cache module and confidence detection module are mainly used to implement the following functions: storing and comparing the normalized confidence of frame synchronization, determining the correspondence between the iterative gain introduction point and the frame synchronization position by using the single peak characteristic of the normalized confidence of frame synchronization, and estimating the frame synchronization position by performing maximum value detection on the cached normalized confidence of frame synchronization; The decoding output module is mainly used to realize the following functions: reading and outputting the hard decision cache data of the synchronization position to realize decoding output.

Images