Wireless communication frequency offset estimation method and device

Abstract

The invention relates to the technical field of industrial communication networks, in particular to a wireless communication frequency offset estimation method and device, and the method comprises the steps: receiving a wireless communication signal to obtain a baseband digital sampling sequence, and enabling a preamble to be composed of a short training field symbol and a long training field symbol; identifying preamble symbol configuration, and judging whether the frame structure is extended configuration or standard configuration according to the number of symbols; during extended configuration, executing multi-stage coarse frequency offset estimation, performing first-stage estimation by using the first part of short training field symbols and compensating subsequent symbols, performing second-stage residual estimation by using the second part of short training field symbols, and cascading to obtain a total coarse frequency offset estimation value; performing multiple times of fine frequency offset measurement, performing two times of independent measurement by using part of long training field symbols, and averaging to obtain a final fine frequency offset estimation value; using a plurality of long training field symbols to joint channel estimation to obtain an average channel response; and the combined frequency offset value compensates and demodulates the data part, so that the problem of low frequency offset estimation precision caused by insufficient preamble symbols is solved, and the reliability is improved.

Description

Technical Field

[0001] This invention relates to the field of industrial communication network technology, and in particular to a method and apparatus for estimating frequency offset in wireless communication. Background Technology

[0002] In wireless communication, frequency offset is mainly caused by the Doppler effect or differences in crystal oscillators at the transmitting and receiving ends, which can lead to signal phase rotation and demodulation performance degradation. To overcome this challenge, frequency offset estimation technology uses pre-inserted pilot symbols or inherent signal structures such as cyclic prefixes to extract frequency offset information by calculating the cross-correlation or phase difference between the received sequence and the local copy. Based on statistical signal processing theory, the frequency offset estimation process in wireless communication can derive the offset, thereby guiding the frequency correction module to achieve accurate compensation and ensure the stability of the communication link.

[0003] Existing wireless communication frequency offset estimation technologies suffer from the following technical challenges: In industrial wireless communication systems such as WIA-FA factory automation networks or outdoor private networks, the scarcity of physical layer preamble symbols limits their ability to support only single coarse and fine frequency offset estimations. Due to the insufficient number of preamble symbols, receivers in industrial scenarios with low signal-to-noise ratios and strong multipath interference cannot suppress estimation errors through multiple measurements or cascaded correction mechanisms, resulting in significant residual frequency offsets. This, in turn, leads to carrier synchronization inaccuracies and demodulation phase rotation, significantly increasing the bit error rate. For example, in automated production line control systems, inaccurate frequency offset estimation can cause sensor data packet demodulation errors, triggering abnormal robotic arm control commands or production interruptions; in outdoor communication, it can lead to the failure of critical command transmission or the loss of delay-sensitive data packets. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a wireless communication frequency offset estimation method and apparatus, solving the technical pain point of low frequency offset estimation accuracy and high bit error rate caused by insufficient number of preamble symbols.

[0005] To solve the above-mentioned technical problems, the specific contents of the present invention are as follows:

[0006] In a first aspect, the present invention provides a wireless communication frequency offset estimation method, comprising:

[0007] Step 1: Receive wireless communication signal and acquire baseband digital sampling sequence. The sequence includes a preamble symbol part and a data part. The preamble symbol part consists of short training field symbols and long training field symbols.

[0008] Step 2: Identify the configuration of the preamble symbol portion, and determine whether the frame structure is an extended configuration or a standard configuration based on the number of short training field symbols and long training field symbols;

[0009] Step 3: When the extended configuration is determined, perform multi-level coarse frequency offset estimation: use the first part of the short training field symbols to perform the first level coarse frequency offset estimation, obtain the first estimate value, and perform frequency offset compensation on the subsequent symbols; use the second part of the short training field symbols to perform the second level residual estimation on the compensated signal, obtain the second estimate value; combine the multi-level estimates to obtain the total coarse frequency offset estimate value.

[0010] Step 4: Perform multiple fine frequency offset measurements: Use the symbols of the first part of the long training field to perform the first fine frequency offset measurement to obtain the third estimate; use the symbols of the second part of the long training field to perform the second fine frequency offset measurement to obtain the fourth estimate; average the multiple measurements to obtain the final fine frequency offset estimate.

[0011] Step 5: Perform joint channel estimation using multiple long training field symbols to obtain the average channel response;

[0012] Step 6: Combine the total coarse frequency offset estimate with the final fine frequency offset estimate to obtain the total frequency offset value, perform frequency offset compensation on the data portion of the baseband digital sampling sequence, and apply the average channel response for frequency domain equalization and demodulation to output demodulated data.

[0013] Furthermore, in the wireless communication frequency offset estimation method of the present invention, step 1 includes: performing down-conversion and analog-to-digital conversion on the wireless communication signal received by the antenna, and outputting a baseband digital sampling sequence;

[0014] Step 3 includes: when the configuration is determined to be extended, extracting the first two short training field symbols from the baseband digital sampling sequence, calculating the first time-domain autocorrelation value, and deriving the first coarse frequency offset estimate; using the first coarse frequency offset estimate to generate a first compensation signal, and performing frequency offset compensation on the symbols in the baseband digital sampling sequence located after the first two short training field symbols to obtain the first compensated sequence; extracting the subsequent short training field symbols from the first compensated sequence, calculating the second time-domain autocorrelation value, and deriving the second coarse frequency offset estimate; adding the first coarse frequency offset estimate and the second coarse frequency offset estimate to obtain the total coarse frequency offset estimate.

[0015] Step 4 includes: extracting the first two long training field symbols from the coarse frequency offset compensated sequence, calculating the third time-domain autocorrelation value, and deriving the first fine frequency offset estimate; extracting the subsequent two long training field symbols from the coarse frequency offset compensated sequence, calculating the fourth time-domain autocorrelation value, and deriving the second fine frequency offset estimate; adding the first fine frequency offset estimate and the second fine frequency offset estimate and dividing by two to obtain the final fine frequency offset estimate.

[0016] Step 5 includes: performing a Fast Fourier Transform on each long training field symbol in the baseband digital sampling sequence, dividing the transformed frequency domain symbol by a known long training frequency domain sequence stored locally to obtain multiple initial channel frequency responses; adding the multiple initial channel frequency responses and dividing by the total number of long training field symbols to output the average channel response.

[0017] Step 6 includes: adding the total coarse frequency offset estimate to the final fine frequency offset estimate to obtain the total frequency offset value; using the total frequency offset value to generate a second compensation signal to perform frequency offset compensation on the data portion of the baseband digital sampling sequence to obtain a second compensated sequence; performing a fast Fourier transform on the second compensated sequence, dividing the transformed frequency domain data symbols by the average channel response to complete frequency domain equalization; and demapping the equalized frequency domain symbols to output the demodulated data bit stream.

[0018] Furthermore, in the wireless communication frequency offset estimation method of the present invention, the configuration for identifying the preamble portion includes:

[0019] Perform sliding correlation calculation on the baseband digital sampling sequence and output the correlation detection results, which include the repetition period information of short training field symbols and the sequence pattern information of long training field symbols;

[0020] Based on the relevant detection results, calculate the number of symbols in the short training field and the number of symbols in the long training field;

[0021] If the number of short training field symbols is six and the number of long training field symbols is four, then the frame structure is determined to be an extended configuration, and the first control signal is generated.

[0022] If the number of short training field symbols is two or three and the number of long training field symbols is two, then the frame structure is determined to be the standard configuration, and a second control signal is generated.

[0023] Furthermore, in the wireless communication frequency offset estimation method of the present invention, the step of extracting the first two short training field symbols from the baseband digital sampling sequence, calculating the first time-domain autocorrelation value, and deriving the first coarse frequency offset estimate from the first time-domain autocorrelation value includes:

[0024] Select the first two short training field symbols in the baseband digital sampling sequence and calculate the time-domain autocorrelation value between the sampling points corresponding to the two short training field symbols.

[0025] Phase information is extracted from the first time-domain autocorrelation value obtained from the calculation.

[0026] Calculate the first coarse frequency offset estimate based on the phase information;

[0027] The step of generating a first compensation signal using a first coarse frequency offset estimate, and compensating for the frequency offset of symbols in the baseband digital sampling sequence located after the first two short training field symbols, to obtain the first compensated sequence includes:

[0028] A first compensation signal is generated based on the first coarse frequency offset estimate;

[0029] Multiply the symbol in the baseband digital sampling sequence that is located after the first two short training field symbols with the first compensation signal to complete the frequency offset compensation and output the first compensated sequence.

[0030] The steps of extracting subsequent short training field symbols from the first compensated sequence, calculating the second time-domain autocorrelation value, and deriving the second coarse frequency offset estimate from the second time-domain autocorrelation value include:

[0031] The four short training field symbols are extracted from the first compensated sequence. The time-domain autocorrelation values ​​between adjacent short training field symbol pairs are calculated sequentially. The phase difference is extracted from each time-domain autocorrelation value. The arithmetic mean of the extracted phase differences is calculated. The second coarse frequency offset estimate is calculated based on the averaged phase difference.

[0032] Furthermore, in the wireless communication frequency offset estimation method of the present invention, the step of extracting the first two long training field symbols from the sequence after coarse frequency offset compensation, calculating the third time-domain autocorrelation value, and deriving the first fine frequency offset estimate from the third time-domain autocorrelation value includes:

[0033] Extract the first two long training field symbols from the sequence after coarse frequency offset compensation, calculate the time domain autocorrelation value between the two long training field symbols as the third time domain autocorrelation value, extract the phase difference from the third time domain autocorrelation value, and calculate the first fine frequency offset estimate based on the phase difference;

[0034] The steps of extracting the subsequent two long training field symbols from the coarse frequency offset compensated sequence, calculating the fourth time-domain autocorrelation value, and deriving the second fine frequency offset estimate from the fourth time-domain autocorrelation value include:

[0035] Extract the next two long training field symbols from the sequence after coarse frequency offset compensation, calculate the time-domain autocorrelation value between the two long training field symbols as the fourth time-domain autocorrelation value, extract the phase difference from the fourth time-domain autocorrelation value, and calculate the second fine frequency offset estimate based on the phase difference.

[0036] Furthermore, in the wireless communication frequency offset estimation method of the present invention, the step of performing a Fast Fourier Transform on each long training field symbol in the baseband digital sampling sequence, and performing a dot division operation between the transformed frequency domain symbol and the locally stored known long training frequency domain sequence to obtain multiple initial channel frequency responses includes:

[0037] Perform a Fast Fourier Transform on each long training field symbol in the baseband digital sampling sequence and output the frequency domain symbol;

[0038] Each frequency domain symbol is divided by a known long training frequency domain sequence stored locally to obtain multiple initial channel frequency responses;

[0039] The step of summing multiple initial channel frequency responses and dividing by the total number of symbols in the long training field to output the average channel response includes:

[0040] The average channel response is output by summing the frequency responses of multiple initial channels and dividing the sum by the total number of symbols in the long training field.

[0041] Furthermore, in the wireless communication frequency offset estimation method of the present invention, the step of generating a second compensation signal using the total frequency offset value and compensating for the frequency offset of the data portion in the baseband digital sampling sequence to obtain the second compensated sequence includes:

[0042] The total frequency offset is used to generate a complex compensation signal. The data part of the baseband digital sampling sequence is multiplied with the complex compensation signal to complete the frequency offset compensation and output the second compensated sequence.

[0043] The step of performing a Fast Fourier Transform on the second compensated sequence and dividing the transformed frequency domain data symbols by the average channel response to complete frequency domain equalization includes:

[0044] Perform a Fast Fourier Transform on the second compensated sequence to output frequency domain data symbols. Divide the frequency domain data symbols by the average channel response to complete frequency domain equalization. The step of demapping the equalized frequency domain symbols and outputting the demodulated data bit stream includes: demapping the equalized frequency domain symbols and outputting the demodulated data bit stream.

[0045] Furthermore, the wireless communication frequency offset estimation method of the present invention further includes:

[0046] When the second control signal indicates that the frame structure is in the standard configuration, according to the second control signal, the operation of extracting the subsequent short training field symbols from the first compensated sequence to calculate the second time-domain autocorrelation value is disabled, and the operation of extracting the subsequent two long training field symbols from the sequence after coarse frequency offset compensation to calculate the fourth time-domain autocorrelation value is also disabled.

[0047] Only the operation of extracting the first two short training field symbols from the baseband digital sampling sequence, calculating the first time-domain autocorrelation value, and deriving the first coarse frequency offset estimate is performed to obtain a single coarse frequency offset estimate.

[0048] Only the operation of extracting the first two long training field symbols from the coarse frequency offset compensated sequence, calculating the third time domain autocorrelation value, and deriving the first fine frequency offset estimate is performed to obtain a single fine frequency offset estimate.

[0049] The single coarse frequency offset estimate is used as the total coarse frequency offset estimate, and the single fine frequency offset estimate is used as the final fine frequency offset estimate.

[0050] Continue performing the operation of outputting the average channel response by performing Fast Fourier Transform on each long training field symbol in the baseband digital sampling sequence, as well as subsequent frequency offset compensation and demodulation operations.

[0051] Furthermore, the wireless communication frequency offset estimation method of the present invention further includes:

[0052] When the number of symbols in the short training field is greater than six and the number of symbols in the long training field is greater than four, the multi-level coarse frequency offset estimation also includes a third-level estimation, and the multiple fine frequency offset measurements also include a third measurement.

[0053] After completing the second-level residual estimation, the third-level coarse frequency bias estimation is performed using the symbols of the third part of the short training field to obtain the fifth estimate. The first estimate, the second estimate and the fifth estimate are combined to obtain the extended total coarse frequency bias estimate.

[0054] After completing the second fine frequency offset measurement, the third fine frequency offset measurement is performed using the symbols of the third part of the long training field to obtain the sixth estimate. The third estimate, the fourth estimate and the sixth estimate are averaged to obtain the extended final fine frequency offset estimate.

[0055] Joint channel estimation is performed using all long training field symbols to obtain the extended average channel response.

[0056] Secondly, the present invention provides a wireless communication frequency offset estimation device, applied to the aforementioned wireless communication frequency offset estimation method, comprising:

[0057] The signal receiving module is used to receive wireless communication signals and acquire baseband digital sampling sequences. The baseband digital sampling sequences include a preamble symbol part and a data part. The preamble symbol part is composed of short training field symbols and long training field symbols.

[0058] The configuration identification module is used to perform sliding correlation calculation on the baseband digital sampling sequence, identify the number of short training field symbols and the number of long training field symbols, and determine whether the frame structure is an extended configuration or a standard configuration based on the identified number.

[0059] The coarse frequency offset estimation module is used to extract the first two short training field symbols from the baseband digital sampling sequence when the extended configuration is determined, calculate the first time-domain autocorrelation value, derive the first coarse frequency offset estimate value from the first time-domain autocorrelation value, generate the first compensation signal using the first coarse frequency offset estimate value, perform frequency offset compensation on the symbols in the baseband digital sampling sequence located after the first two short training field symbols to obtain the first compensated sequence, extract the subsequent short training field symbols from the first compensated sequence, calculate the second time-domain autocorrelation value, derive the second coarse frequency offset estimate value from the second time-domain autocorrelation value, and add the first coarse frequency offset estimate value and the second coarse frequency offset estimate value to obtain the total coarse frequency offset estimate value.

[0060] The fine frequency offset measurement module is used to extract the first two long training field symbols from the coarse frequency offset compensated sequence, calculate the third time domain autocorrelation value, derive the first fine frequency offset estimate from the third time domain autocorrelation value, extract the subsequent two long training field symbols from the coarse frequency offset compensated sequence, calculate the fourth time domain autocorrelation value, derive the second fine frequency offset estimate from the fourth time domain autocorrelation value, add the first fine frequency offset estimate and the second fine frequency offset estimate and divide by two to obtain the final fine frequency offset estimate.

[0061] The channel estimation module performs a fast Fourier transform on each long training field symbol in the baseband digital sampling sequence, divides the transformed frequency domain symbol by a known long training frequency domain sequence stored locally to obtain multiple initial channel frequency responses, adds the multiple initial channel frequency responses and divides them by the total number of long training field symbols to output the average channel response.

[0062] The compensation demodulation module is used to add the total coarse frequency offset estimate to the final fine frequency offset estimate to obtain the total frequency offset value. The total frequency offset value is used to generate a second compensation signal to perform frequency offset compensation on the data part of the baseband digital sampling sequence to obtain the second compensated sequence. A fast Fourier transform is performed on the second compensated sequence, and the transformed frequency domain data symbols are divided by the average channel response to complete the frequency domain equalization. The equalized frequency domain symbols are demapped to output the demodulated data bit stream.

[0063] Beneficial effects of this invention;

[0064] This invention expands the preamble symbol configuration, increasing the number of short training field symbols to six and the number of long training field symbols to four, overcoming the data foundation limitation of insufficient preamble symbols in existing technologies. This provides hardware support for solving the problems of low frequency offset estimation accuracy and high bit error rate. Based on the expanded configuration, a multi-level coarse frequency offset estimation mechanism is adopted. First, the first two short training field symbols are used for the first-level estimation and immediate compensation to initially reduce the impact of frequency offset. Then, additional short training field symbols are used for the second-level residual estimation. The total coarse frequency offset estimate is obtained by cascading and adding these values. This iterative estimation, compensation, and re-estimation process promotes error correction in low signal-to-noise ratio environments. Fast convergence; the fine frequency offset measurement stage performs two independent operations, using different long training field symbols to calculate the time-domain autocorrelation value, and the results are arithmetically averaged to reduce noise variance and improve estimation accuracy by utilizing statistical independence; the channel estimation part performs joint processing on multiple long training field symbols, and reduces mean square error and improves robustness through data fusion; the system automatically identifies frame structure through sliding correlation, enables advanced functions in extended configuration, and disables redundant operations in standard configuration to maintain backward compatibility. The overall architecture improves communication reliability in industrial multipath interference scenarios through refined processing, while balancing performance and power consumption through adaptive design. Attached Figure Description

[0065] To more clearly illustrate the technical solution of the present invention, the drawings used in the embodiments will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on the drawings without creative effort.

[0066] Figure 1 This is a schematic flowchart of the overall process of a wireless communication frequency offset estimation method according to the present invention.

[0067] Figure 2 This is a schematic diagram illustrating the principle of a two-stage cascaded coarse frequency offset estimation method for wireless communication according to the present invention.

[0068] Figure 3 This is a schematic diagram illustrating the principle of the average fine frequency offset estimation based on two measurements in a wireless communication frequency offset estimation method according to the present invention.

[0069] Figure 4 This is a flowchart of the total frequency offset calculation and multi-LTF channel estimation of a wireless communication frequency offset estimation method according to the present invention. Detailed Implementation

[0070] To make the technical solution of the present invention clearer, the present invention will be clearly and completely described below with reference to specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. 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. The present invention provided by various embodiments will be described in detail below with reference to the accompanying drawings. To better understand the purpose of the present invention, the present invention will be described in further detail below.

[0071] Firstly, please refer to Figures 1 to 4 The present invention provides a wireless communication frequency offset estimation method, comprising:

[0072] Step 1: Receive wireless communication signal and acquire baseband digital sampling sequence. The sequence includes a preamble symbol part and a data part. The preamble symbol part consists of short training field symbols and long training field symbols.

[0073] Step 2: Identify the configuration of the preamble symbol portion, and determine whether the frame structure is an extended configuration or a standard configuration based on the number of short training field symbols and long training field symbols;

[0074] Step 3: When the extended configuration is determined, perform multi-level coarse frequency offset estimation: use the first part of the short training field symbols to perform the first level coarse frequency offset estimation, obtain the first estimate value, and perform frequency offset compensation on the subsequent symbols; use the second part of the short training field symbols to perform the second level residual estimation on the compensated signal, obtain the second estimate value; combine the multi-level estimates to obtain the total coarse frequency offset estimate value.

[0075] Step 4: Perform multiple fine frequency offset measurements: Use the symbols of the first part of the long training field to perform the first fine frequency offset measurement to obtain the third estimate; use the symbols of the second part of the long training field to perform the second fine frequency offset measurement to obtain the fourth estimate; average the multiple measurements to obtain the final fine frequency offset estimate.

[0076] Step 5: Perform joint channel estimation using multiple long training field symbols to obtain the average channel response;

[0077] Step 6: Combine the total coarse frequency offset estimate with the final fine frequency offset estimate to obtain the total frequency offset value, perform frequency offset compensation on the data portion of the baseband digital sampling sequence, and apply the average channel response for frequency domain equalization and demodulation to output demodulated data.

[0078] Step 1 includes: performing down-conversion and analog-to-digital conversion on the wireless communication signal received by the antenna, and outputting a baseband digital sampling sequence;

[0079] Step 3 includes: when the configuration is determined to be extended, extracting the first two short training field symbols from the baseband digital sampling sequence, calculating the first time-domain autocorrelation value, and deriving the first coarse frequency offset estimate; using the first coarse frequency offset estimate to generate a first compensation signal, and performing frequency offset compensation on the symbols in the baseband digital sampling sequence located after the first two short training field symbols to obtain the first compensated sequence; extracting the subsequent short training field symbols from the first compensated sequence, calculating the second time-domain autocorrelation value, and deriving the second coarse frequency offset estimate; adding the first coarse frequency offset estimate and the second coarse frequency offset estimate to obtain the total coarse frequency offset estimate.

[0080] Step 4 includes: extracting the first two long training field symbols from the coarse frequency offset compensated sequence, calculating the third time-domain autocorrelation value, and deriving the first fine frequency offset estimate; extracting the subsequent two long training field symbols from the coarse frequency offset compensated sequence, calculating the fourth time-domain autocorrelation value, and deriving the second fine frequency offset estimate; adding the first fine frequency offset estimate and the second fine frequency offset estimate and dividing by two to obtain the final fine frequency offset estimate.

[0081] Step 5 includes: performing a Fast Fourier Transform on each long training field symbol in the baseband digital sampling sequence, dividing the transformed frequency domain symbol by a known long training frequency domain sequence stored locally to obtain multiple initial channel frequency responses; adding the multiple initial channel frequency responses and dividing by the total number of long training field symbols to output the average channel response.

[0082] Step 6 includes: adding the total coarse frequency offset estimate to the final fine frequency offset estimate to obtain the total frequency offset value; using the total frequency offset value to generate a second compensation signal to perform frequency offset compensation on the data portion of the baseband digital sampling sequence to obtain a second compensated sequence; performing a fast Fourier transform on the second compensated sequence, dividing the transformed frequency domain data symbols by the average channel response to complete frequency domain equalization; and demapping the equalized frequency domain symbols to output the demodulated data bit stream.

[0083] The wireless communication frequency offset estimation method provided by this invention addresses the problem of insufficient frequency offset estimation accuracy in industrial wireless communication scenarios. It improves communication reliability by expanding the preamble symbol configuration and employing a multi-level estimation and averaging fusion mechanism. The method first receives the wireless communication signal, performs down-conversion and analog-to-digital conversion, and outputs a baseband digital sampling sequence. This sequence includes a preamble symbol portion and a data portion. The preamble symbol portion consists of short training field symbols and long training field symbols, providing the raw data foundation for subsequent processing.

[0084] When identifying the preamble symbol configuration, the method performs sliding correlation calculations on the baseband digital sampling sequence, analyzing the quantity and temporal characteristics of short and long training field symbols. Based on the correlation detection results, it automatically determines whether the frame structure is an extended or standard configuration. For example, in industrial automation networks, an extended configuration may include more preamble symbols to support advanced estimation mechanisms, while a standard configuration maintains backward compatibility. The decision logic outputs control signals to guide subsequent processing units to enable or disable specific functions.

[0085] When an extended configuration is identified, the method performs multi-level coarse frequency offset estimation. The first level of coarse frequency offset estimation uses the first two short training field symbols to calculate the time-domain autocorrelation value, extract phase difference information, and derive the initial coarse frequency offset estimate. Immediately afterwards, a compensation signal is generated to compensate for the frequency offset of subsequent symbols, initially reducing the impact of the frequency offset. The second level of residual estimation uses the additional short training field symbols in the compensated signal to calculate multiple pairs of autocorrelation values ​​and estimate the residual frequency offset. The two levels of estimates are cascaded and added to obtain the total coarse frequency offset estimate. This iterative estimation, compensation, and re-estimation mechanism can effectively converge errors in low signal-to-noise ratio environments.

[0086] The fine frequency offset measurement stage employs a multi-independent measurement averaging and fusion approach. The first fine frequency offset measurement uses the first two long training field symbols to calculate the time-domain autocorrelation value, obtaining a fine frequency offset estimate. The second measurement uses an additional long training field symbol, repeating the same operation on the coarse frequency offset compensated signal to obtain another independent estimate. The two measurement results are then arithmetically averaged to generate the final fine frequency offset estimate. This averaging and fusion leverages statistical independence to reduce noise variance and improve estimation accuracy, making it particularly suitable for industrial scenarios with strong multipath interference.

[0087] The channel estimation part utilizes multiple long training field symbols for joint processing. A Fast Fourier Transform is performed on each long training field symbol, and after transformation to the frequency domain, a point division operation is performed with the locally stored known sequence to obtain multiple initial channel frequency responses. These responses are then summed and averaged to output the average channel response. Joint estimation reduces the mean square error through data fusion, improving the robustness of channel estimation.

[0088] Finally, the method adds the total coarse frequency offset estimate to the final fine frequency offset estimate to obtain the total frequency offset value. This value is used to generate a compensation signal to compensate for the frequency offset of the data portion. The compensated sequence is converted into frequency domain data symbols through a Fast Fourier Transform, and frequency domain equalization is completed by dividing by the average channel response. The equalized symbols are then demapped to output a demodulated data bitstream.

[0089] This invention analyzes the baseband digital sampling sequence using sliding correlation calculations, outputting the repetition period of short training field symbols and the sequence pattern information of long training field symbols. Based on the correlation detection results, the system automatically calculates the number of symbols. If there are six short training field symbols and four long training field symbols, it is determined to be an extended configuration and a first control signal is generated; if the number of symbols is two or three short training field symbols and two long training field symbols, it is determined to be a standard configuration and a second control signal is generated. This identification logic enables the receiver to adapt to different frame structures, providing mode selection signals for subsequent processing modules, ensuring compatibility and flexibility.

[0090] This invention extracts the first two short training field symbols from the baseband digital sampling sequence, calculates the time-domain autocorrelation value between them, and extracts phase information to derive a first coarse frequency offset estimate. Subsequently, a first compensation signal is generated based on this estimate, and subsequent symbols are multiplied by the compensation signal to complete frequency offset compensation, outputting a first compensated sequence. In the compensated sequence, the next four short training field symbols are extracted, and the time-domain autocorrelation value between adjacent symbol pairs is calculated sequentially. Multiple phase differences are extracted and arithmetically averaged. A second coarse frequency offset estimate is calculated based on the average phase difference. This averaging process across multiple symbol pairs effectively suppresses random noise and improves the accuracy of residual estimation.

[0091] This invention extracts the first two long training field symbols from a sequence that has undergone coarse frequency offset compensation, calculates the temporal autocorrelation value between them as the third temporal autocorrelation value, and extracts the phase difference to derive the first fine frequency offset estimate. Similarly, it extracts the next two long training field symbols to calculate the fourth temporal autocorrelation value, extracts the phase difference to derive the second fine frequency offset estimate. The two measurements are performed independently to ensure the unbiasedness of the results, providing a basis for subsequent averaging and fusion.

[0092] This invention performs a Fast Fourier Transform on each long training field symbol in the baseband digital sampling sequence, outputting a frequency domain symbol. This symbol is then divided by a locally stored known long training frequency domain sequence to obtain multiple initial channel frequency responses. Subsequently, the responses are summed and divided by the total number of long training field symbols to output the average channel response. This joint estimation mechanism reduces mean square error through data fusion, making it suitable for industrial scenarios with severe multipath interference.

[0093] This invention utilizes the total frequency offset to generate a complex compensation signal. Multiplying the data portion by the compensation signal completes frequency offset compensation, outputting a second compensated sequence. A Fast Fourier Transform is performed on this sequence to obtain frequency domain data symbols. Dividing by the average channel response completes frequency domain equalization. Finally, the equalized frequency domain symbols are demapped to output a demodulated data bitstream.

[0094] In the standard configuration of this invention, when the generated second control signal indicates that the frame structure is in the standard configuration, the system disables redundant operations, such as extracting subsequent short training field symbols from the compensated sequence for second-level estimation, and extracting subsequent long training field symbols for a second measurement. Only a single coarse frequency offset estimation and a single fine frequency offset estimation are performed, and the results are directly used as the total estimate to continue performing channel estimation and demodulation operations. This design achieves backward compatibility and reduces power consumption and complexity.

[0095] This invention adds a third stage to the coarse frequency offset estimation when the number of short training field symbols is greater than six and the number of long training field symbols is greater than four. The third stage uses the short training field symbols to perform residual estimation, yielding a fifth estimate, which is then combined with the first two stages to obtain an extended total coarse frequency offset estimate. A third stage is added to the fine frequency offset measurement, using the third part of the long training field symbols to obtain a sixth estimate. This sixth estimate is then averaged with the first two stages to obtain the extended final fine frequency offset estimate. Simultaneously, all long training field symbols are used for joint channel estimation, outputting an extended average channel response. This scalable architecture is suitable for high-performance scenarios.

[0096] Secondly, the present invention provides a wireless communication frequency offset estimation device, applied to the aforementioned wireless communication frequency offset estimation method, comprising:

[0097] The signal receiving module is used to receive wireless communication signals and acquire baseband digital sampling sequences. The baseband digital sampling sequences include a preamble symbol part and a data part. The preamble symbol part is composed of short training field symbols and long training field symbols.

[0098] The configuration identification module is used to perform sliding correlation calculation on the baseband digital sampling sequence, identify the number of short training field symbols and the number of long training field symbols, and determine whether the frame structure is an extended configuration or a standard configuration based on the identified number.

[0099] The coarse frequency offset estimation module is used to extract the first two short training field symbols from the baseband digital sampling sequence when the extended configuration is determined, calculate the first time-domain autocorrelation value, derive the first coarse frequency offset estimate value from the first time-domain autocorrelation value, generate the first compensation signal using the first coarse frequency offset estimate value, perform frequency offset compensation on the symbols in the baseband digital sampling sequence located after the first two short training field symbols to obtain the first compensated sequence, extract the subsequent short training field symbols from the first compensated sequence, calculate the second time-domain autocorrelation value, derive the second coarse frequency offset estimate value from the second time-domain autocorrelation value, and add the first coarse frequency offset estimate value and the second coarse frequency offset estimate value to obtain the total coarse frequency offset estimate value.

[0100] The fine frequency offset measurement module is used to extract the first two long training field symbols from the coarse frequency offset compensated sequence, calculate the third time domain autocorrelation value, derive the first fine frequency offset estimate from the third time domain autocorrelation value, extract the subsequent two long training field symbols from the coarse frequency offset compensated sequence, calculate the fourth time domain autocorrelation value, derive the second fine frequency offset estimate from the fourth time domain autocorrelation value, add the first fine frequency offset estimate and the second fine frequency offset estimate and divide by two to obtain the final fine frequency offset estimate.

[0101] The channel estimation module performs a fast Fourier transform on each long training field symbol in the baseband digital sampling sequence, divides the transformed frequency domain symbol by a known long training frequency domain sequence stored locally to obtain multiple initial channel frequency responses, adds the multiple initial channel frequency responses and divides them by the total number of long training field symbols to output the average channel response.

[0102] The compensation demodulation module is used to add the total coarse frequency offset estimate to the final fine frequency offset estimate to obtain the total frequency offset value. The total frequency offset value is used to generate a second compensation signal to perform frequency offset compensation on the data part of the baseband digital sampling sequence to obtain the second compensated sequence. A fast Fourier transform is performed on the second compensated sequence, and the transformed frequency domain data symbols are divided by the average channel response to complete the frequency domain equalization. The equalized frequency domain symbols are demapped to output the demodulated data bit stream.

[0103] Wireless communication frequency offset estimation devices are used in industrial wireless communication networks, such as factory automation systems or outdoor private networks. They achieve high-precision frequency offset estimation and compensation through modular design. The device includes a signal receiving module, a configuration identification module, a coarse frequency offset estimation module, a fine frequency offset measurement module, a channel estimation module, and a compensation demodulation module. These modules cooperate sequentially according to the data stream to process the baseband digital sampling sequence to improve communication reliability.

[0104] The signal receiving module is responsible for capturing wireless communication signals, performing down-conversion and analog-to-digital conversion, and outputting a baseband digital sampling sequence including a preamble symbol portion and a data portion. The preamble symbol portion consists of short training field symbols and long training field symbols, providing the raw data foundation for subsequent modules. In industrial scenarios, signals may be affected by multipath effects and crystal oscillator deviations; therefore, the module must have anti-interference capabilities to ensure the integrity of the sampling sequence.

[0105] The configuration recognition module performs sliding correlation calculations on the baseband digital sampling sequence, analyzes the repetition period of short training field symbols and the sequence pattern of long training field symbols, identifies the number of symbols, and determines whether the frame structure is an extended configuration or a standard configuration. For example, an extended configuration may include six short training field symbols and four long training field symbols, while a standard configuration may only have two or three short training field symbols and two long training field symbols. The module outputs a mode selection signal to control whether subsequent modules enable or disable advanced functions, achieving backward compatibility.

[0106] The coarse frequency offset estimation module activates a multi-level estimation mechanism when the extended configuration is determined. The module first extracts the first two short training field symbols from the baseband digital sampling sequence, calculates the time-domain autocorrelation value, extracts phase information, and derives the first coarse frequency offset estimate. Then, it immediately generates a first compensation signal to compensate for the frequency offset of subsequent symbols, outputting the first compensated sequence to initially reduce the impact of the frequency offset. After compensation, the module extracts subsequent short training field symbols from the first compensated sequence, calculates multiple pairs of time-domain autocorrelation values, averages the phase difference, and derives the second coarse frequency offset estimate. The two estimates are added together to obtain the total coarse frequency offset estimate. This iterative estimation-compensation-re-estimation mechanism promotes error convergence in low signal-to-noise ratio environments. Figure 2 The timing and data flow of this cascading process are shown.

[0107] The fine frequency offset measurement module performs multiple measurements based on coarse frequency offset compensation. The module extracts the first two long training field symbols from the compensated sequence, calculates the time-domain autocorrelation value, and derives the first fine frequency offset estimate. Then, it extracts the subsequent two long training field symbols, calculates the second time-domain autocorrelation value, and derives the second fine frequency offset estimate. The arithmetic mean of the two measurements yields the final fine frequency offset estimate. This averaging and fusion utilizes statistical independence to reduce noise variance and improve estimation accuracy. Figure 3 The complete procedure for the two measurements is explained.

[0108] The channel estimation module performs a Fast Fourier Transform on each long training field symbol, and then performs a point division operation between the frequency domain symbol and the locally stored known long training frequency domain sequence to obtain multiple initial channel frequency responses. The sum of these responses is divided by the total number of long training field symbols to output the average channel response. Joint estimation reduces mean square error through data fusion and is suitable for industrial environments with severe multipath interference.

[0109] The compensation and demodulation module adds the total coarse frequency offset estimate to the final fine frequency offset estimate to obtain the total frequency offset value. The module uses this total frequency offset value to generate a complex compensation signal, performs frequency offset compensation on the data portion, and outputs the second compensated sequence. Subsequently, a Fast Fourier Transform is performed, dividing the frequency domain data symbols by the average channel response to complete frequency domain equalization. Finally, the equalized symbols are demapped to output the data bitstream. This invention prioritizes computational efficiency, maintaining a low bit error rate in real-time communication. Figure 4 The integrated process of total frequency offset calculation and multi-symbol channel estimation is demonstrated.

[0110] The logical relationships between the modules of the device are adaptively adjusted based on the preamble symbol configuration. The output of the identification module drives the estimation strategy. Coarse and fine frequency offset estimations work together to reduce residual errors. Channel estimation provides the basis for equalization. Finally, reliable data transmission is achieved through compensation and demodulation. In the standard configuration, the modules are simplified to reduce power consumption, reflecting design flexibility.

[0111] In industrial wireless communication scenarios, such as WIA-FA (Wireless Integrated Automation) factory networks or private networks, after the receiver antenna captures the wireless communication signal, it performs down-conversion and analog-to-digital conversion to output a baseband digital sampling sequence. This sequence includes a preamble symbol portion and a data portion. The preamble symbol portion consists of short training field symbols (STF) and long training field symbols (LTF). To address the low frequency offset estimation accuracy caused by insufficient preamble symbols in existing technologies, this invention employs an extended preamble configuration, for example, increasing the number of STF symbols to six and the number of LTF symbols to four, providing a data foundation for multi-level estimation.

[0112] The receiver first performs sliding correlation calculations on the baseband digital sampling sequence to automatically identify the preamble symbol configuration. If it detects that the short training field has six symbols and the long training field has four symbols, it determines that the frame structure is an extended configuration and enables the advanced estimation mechanism; if the number of symbols is a standard configuration, such as two or three STF symbols and two LTF symbols, it switches to compatibility mode. It identifies the logic output mode selection signal to control subsequent processing units, ensuring backward compatibility.

[0113] In the extended configuration, a two-stage cascaded coarse frequency offset estimation is performed. The first stage uses the first two short training field symbols (STF symbols) to calculate the time-domain autocorrelation value, extract phase difference information, and derive the initial coarse frequency offset estimate. Immediately afterwards, a compensation signal is generated to compensate for the frequency offset of subsequent symbols, initially reducing the impact of the frequency offset. The second stage residual estimation utilizes additional short training field symbols (STF symbols), such as the third to sixth STF symbols, to calculate multiple pairs of time-domain autocorrelation values ​​on the compensated signal, estimating the residual frequency offset. The two-stage estimates are cascaded and added together to obtain the total coarse frequency offset estimate. This iterative estimation, compensation, and re-estimation mechanism effectively converges the error in low signal-to-noise ratio environments.

[0114] The fine frequency offset measurement stage employs a two-step averaging and fusion approach. The first fine frequency offset measurement uses the first two long training field symbols (LTF symbols) to calculate the time-domain autocorrelation value, obtaining a fine frequency offset estimate. The second measurement uses additional long training field symbols (LTF symbols), such as the third and fourth LTF symbols, repeating the same operation on the coarse frequency offset compensated signal to obtain another independent estimate. The two measurement results are then arithmetically averaged to generate the final fine frequency offset estimate. This averaging and fusion utilizes statistical independence to reduce noise variance.

[0115] Joint channel estimation utilizes multiple Long Training Field (LTF) symbols. A Fast Fourier Transform (FFT) is performed on each LTF symbol, and after conversion to the frequency domain, a dot-matrix division operation is performed with a locally stored known long training frequency domain sequence to obtain multiple initial channel frequency responses. These responses are summed and averaged to output the average channel response, which improves estimation robustness through data fusion. The total frequency offset is obtained by adding the total coarse frequency offset estimate and the final fine frequency offset estimate, and is used to generate a compensation signal to compensate for the frequency offset of the data portion. The compensated sequence is converted to frequency domain data symbols using an FFT, divided by the average channel response to achieve frequency domain equalization, and finally demapped to output the demodulated data bitstream.

[0116] In industrial control applications, the carrier frequency is 2.4 GHz, the bandwidth is 20 MHz, the crystal oscillator accuracy is inconsistent, and the factory environment suffers from strong multipath interference and low signal-to-noise ratio. An extended preamble configuration enables the receiver to automatically identify the mode, activating a two-stage cascading and double-averaging mechanism. Coarse frequency offset estimation completes multi-stage correction within microseconds, fine frequency offset measurement improves accuracy through averaging and fusion, and channel estimation utilizes multiple symbols to reduce mean square error. This design, in outdoor communication or automated production lines, can reduce sensor data packet demodulation errors and prevent abnormal robotic arm control commands.

[0117] For high-performance scenarios such as outdoor communication, the preamble symbols can be further extended to include more STF and LTF symbols, supporting three-level coarse frequency offset estimation or four-level fine frequency offset measurement to improve estimation accuracy. However, this increases preamble overhead, requiring a trade-off between performance and efficiency. In standard frame compatibility mode, when insufficient symbol quantity is detected, the advanced estimation unit is automatically disabled, performing only a single coarse and fine frequency offset estimation. Clock gating reduces power consumption and ensures compatibility with existing standards such as WiFi. The solution utilizes efficient hardware deployment to adapt to different reliability requirements, highlighting flexibility and addressing constraints in practical applications.

[0118] The accompanying drawings of this invention address the technical problem of insufficient frequency offset estimation accuracy in industrial wireless communication scenarios. They visually present the core process and key mechanisms of the method, aiming to support the scope of the claims and demonstrate the specific implementation of the subordinate technical solutions. Background technology indicates that existing wireless communication systems such as WiFi or WIA-FA are limited by the scarcity of preamble symbols (e.g., standard configuration only has 2-3 short training field symbols and 2 long training field symbols). In industrial environments with low signal-to-noise ratios and strong multipath interference, they cannot suppress estimation errors through multiple measurements, resulting in large residual frequency offsets and increased demodulation bit error rates. This invention expands the preamble symbol configuration (e.g., increasing to 6 STFs and 4 LTFs) and introduces a multi-level estimation and averaging fusion architecture. The accompanying drawings detail how to overcome this bottleneck in practical applications.

[0119] Figure 1 As an overall flowchart, it depicts the complete sequence from signal reception to data demodulation, corresponding to steps 1 to 6 of claim 1. In industrial automation scenarios, such as a factory workshop with a carrier frequency of 2.4 GHz and a bandwidth of 20 MHz, the receiver antenna captures the signal and outputs a baseband digital sampling sequence after down-conversion and analog-to-digital conversion. The arrows in the diagram clearly indicate the direction of data flow: the preamble symbol portion is separated into STF and LTF groups, and the configuration identification module automatically determines the frame structure (extended or standard) through sliding correlation calculation. When identified as an extended configuration, the process branches to multi-level coarse frequency offset estimation and multiple fine frequency offset measurements, reflecting an adaptive switching mechanism to avoid resource waste. In actual deployment, if the environmental signal-to-noise ratio is less than 10 dB, this process can reduce the impact of frequency offset through iterative compensation, but it should be noted that the extended preamble will slightly increase timing overhead, and performance and efficiency need to be balanced in delay-sensitive applications.

[0120] Figure 2 The diagram highlights the principle of a two-stage cascaded coarse frequency offset estimation. Key nodes are marked on the timeline. For example, the first-stage estimation uses the first two STF symbols to perform time-domain autocorrelation calculations within microseconds, deriving the initial coarse frequency offset value and immediately compensating subsequent symbols, initially reducing the frequency offset from hundreds of kHz to tens of kHz. The second-stage residual estimation utilizes additional STF symbols (such as the 3rd to 6th STFs in the extended configuration) to calculate multiple pairs of autocorrelation values ​​on the compensated signal. After averaging the phase difference, the residual estimate is obtained, and the cascaded summation further reduces the total residual frequency offset. This architecture performs exceptionally well in high-reliability scenarios such as outdoor communication, but it relies on a sufficient number of preamble symbols. If a standard frame is detected, the system automatically disables the second-stage operation, performing only a single estimation to maintain compatibility.

[0121] Figure 3The process of averaging fine frequency offset estimation between two measurements is explained. All LTF symbols are first coarsely compensated for frequency offset to ensure consistent measurement baselines. The first measurement uses the first two LTF symbols to calculate the time-domain autocorrelation value, outputting an independent estimate. The second measurement repeats the same operation using an additional LTF symbol. The two results are fused by arithmetic averaging, utilizing statistical independence to reduce noise variance by approximately half, thereby improving estimation accuracy. In outdoor communication examples, frequency offset fluctuations caused by the Doppler effect can be smoothed using this averaging mechanism. However, if the channel time-varying is too rapid, the measurement interval may lead to error accumulation; therefore, this scheme is more suitable for medium-speed mobile environments. The measured data (e.g., frequency offset values ​​of approximately several kHz) marked in the figure verify the effectiveness of the averaging fusion.

[0122] Figure 4 This method integrates total frequency offset calculation with multi-LTF joint channel estimation, corresponding to the compensation and demodulation steps. The flowchart is divided into four parts: coarse frequency offset concatenated addition, fine frequency offset averaging and fusion, total frequency offset summation, and multi-symbol channel estimation. For example, in an industrial control system, the total frequency offset value is used to generate a complex compensation signal to correct the frequency offset of the data portion; simultaneously, FFT and point division operations are performed on the four LTF symbols respectively, and the averaged channel responses significantly reduce the mean square error. This joint processing can improve the bit error rate by several times in strong multipath environments, but the computational complexity is high, requiring dedicated integrated circuits to optimize power consumption.

[0123] Embodiment 1 of this invention: In industrial automation networks such as WIA-FA systems, the carrier frequency is 2.4 GHz, the bandwidth is 20 MHz, and the factory environment suffers from strong multipath interference and low signal-to-noise ratio. Crystal oscillator accuracy deviations can lead to frequency offsets of hundreds of kHz. This embodiment employs an extended preamble configuration with six short training field symbols and four long training field symbols. The time-domain preamble structure is [STF×6+LTF×4+SIG+DATA]. The receiver automatically identifies this extended frame mode through sliding correlation detection and enables an advanced estimation mechanism. In the coarse frequency offset estimation stage, the first stage uses the first two short training field symbols to calculate the time-domain autocorrelation value, completing the initial estimation within 1.6 microseconds. The measured frequency offset value is approximately 234 kHz, and subsequent symbols are immediately compensated, reducing the residual frequency offset to approximately 48 kHz. The second stage uses four additional short training field symbols to calculate multiple pairs of autocorrelation values ​​on the compensated signal, estimating a residual of approximately 12 kHz. After cascading and summing, the total coarse frequency offset estimate is approximately 246 kHz, further reducing the residual frequency offset to approximately 5 kHz. In the fine frequency offset measurement phase, the first measurement uses the first two long training field symbols, with an estimated value of approximately 3.8 kHz. The second measurement uses the subsequent two long training field symbols, with an estimated value of approximately 4.2 kHz. After averaging and fusing, the fine frequency offset estimate is approximately 4.0 kHz, improving accuracy by approximately 1.4 times. Channel estimation utilizes frequency domain point division operations on each of the four long training field symbols. After averaging and fusing, the mean square error is improved from -15.1 dB in the standard configuration to -21.2 dB. The total frequency offset value of approximately 250 kHz is used for data compensation. After demodulation, the bit error rate is reduced to 1.8 × 10⁻⁶ at a signal-to-noise ratio of 15 dB. -4 This reduces errors by approximately 11.7 times compared to standard methods. This design can reduce sensor data packet errors in automated production lines, but it's important to note that extending the preamble increases timing overhead, requiring a performance trade-off in latency-sensitive applications.

[0124] Embodiment 2 of this invention: For high-reliability scenarios such as outdoor communication, this embodiment further expands the preamble symbol configuration, increasing the number of short training field symbols to twelve and the number of long training field symbols to eight, to support a more advanced estimation architecture. Coarse frequency offset estimation employs a three-stage cascade mechanism: the first stage uses the first two short training field symbols to complete initial estimation and compensation; the second stage uses the middle four symbols for residual estimation; and the third stage uses the subsequent four symbols for further refinement. After cascading, the overall estimation accuracy is improved by approximately 3.5 times. Fine frequency offset measurement involves four independent operations, each using different pairs of long training field symbols, resulting in an average accuracy improvement of approximately 2 times after fusion. Channel estimation is jointly processed using eight long training field symbols, improving the mean square error by approximately 9.6 dB. Under a signal-to-noise ratio of 20 dB, the bit error rate can be reduced to 4.8 × 10⁻⁶. -5The performance improvement is over 40 times, but the preamble overhead increases to 1.59%, making it suitable for command transmission with extremely high reliability requirements, such as outdoor tactical communications, where the Doppler effect and crystal oscillator deviation are more significant. This solution demonstrates scalability, but relies on application-specific integrated circuits (ASICs) to optimize computational complexity.

[0125] In standard frame compatible mode, when the receiver detects two or three short training field symbols and two long training field symbols, it automatically disables multi-level estimation and average fusion functions, performing only single coarse and fine frequency offset estimation. This design ensures backward compatibility with existing standards such as WiFi, with a switching latency of less than 1 microsecond and power consumption comparable to standard receivers. It is suitable for mixed network environments, but its performance is on par with standard methods, highlighting the adaptability and practicality of the solution.

[0126] This invention addresses the problems of low frequency offset estimation accuracy and high bit error rate caused by insufficient preamble symbols in industrial wireless communication. It improves communication reliability by expanding the preamble symbol configuration and introducing a multi-level estimation and averaging fusion mechanism. Existing wireless communication systems, such as standard WiFi or WIA-FA, only configure a small number of short and long training field symbols in the time domain, for example, two to three short training field symbols and two long training field symbols. Each limited number of symbols is only sufficient to complete a single coarse frequency offset estimation and a single fine frequency offset estimation. In industrial environments with low signal-to-noise ratios and strong multipath interference, a single estimation cannot suppress errors through iterative correction, resulting in a large residual frequency offset. This causes carrier synchronization misalignment and demodulation phase rotation, significantly increasing the bit error rate. For example, in factory automation networks, inaccurate frequency offset estimation may lead to sensor data packet demodulation errors, triggering abnormal robotic arm control.

[0127] The core solution of this invention is to expand the number of preamble symbols, for example, increasing the number of short training field symbols to six and the number of long training field symbols to four, providing a data foundation for advanced estimation architectures. The receiver first performs sliding correlation calculations on the baseband digital sampling sequence to automatically identify the preamble symbol configuration. If it detects that the number of short training field symbols is six and the number of long training field symbols is four, it determines that the frame structure is an expanded configuration and activates a multi-level estimation mechanism; if the number of symbols is the standard configuration, it switches to a compatibility mode to maintain backward compatibility. This identification logic guides subsequent processing units through output control signals, ensuring flexible adaptation to different scenarios.

[0128] In the extended configuration, multi-stage coarse frequency offset estimation is performed to improve convergence accuracy. The first stage of coarse frequency offset estimation uses the first two short training field symbols to calculate the time-domain autocorrelation value, extract phase difference information, and derive the initial coarse frequency offset estimate. A compensation signal is then immediately generated to compensate for the frequency offset of subsequent symbols, initially reducing the impact of the frequency offset. The second stage of residual estimation utilizes additional short training field symbols, such as the third to sixth symbols, to calculate multiple pairs of time-domain autocorrelation values ​​on the compensated signal, estimating the residual frequency offset. The two-stage estimates are cascaded and added to obtain the total coarse frequency offset estimate. This iterative estimation-compensation-re-estimation mechanism promotes rapid error convergence in low signal-to-noise ratio environments. In industrial scenarios, strong multipath interference may cause large initial estimation deviations, but the second stage, utilizing the improved signal-to-noise ratio of the compensated symbols, effectively refines the results.

[0129] Fine frequency offset measurement employs a two-step averaging and fusion approach to reduce noise variance. The first measurement uses the first two long training field symbols to calculate the time-domain autocorrelation value, obtaining a fine frequency offset estimate. The second measurement uses additional long training field symbols, such as the third and fourth symbols, repeating the same operation on the coarse frequency offset compensated signal to obtain an independent estimate. The arithmetic mean of the two results generates the final fine frequency offset estimate; this averaging and fusion leverages statistical independence to halve the error variance. In outdoor communication scenarios, frequency offset fluctuations caused by the Doppler effect can be smoothed using this mechanism. However, if the channel time-varying is too rapid, the measurement interval may lead to error accumulation; therefore, this approach is more suitable for medium-speed mobile environments.

[0130] The channel estimation section utilizes multiple long training field symbols for joint processing to improve robustness. A Fast Fourier Transform is performed on each long training field symbol, and after conversion to the frequency domain, a dot-matrix division operation is performed with the locally stored known sequence to obtain multiple initial channel frequency responses. These responses are summed and averaged to output the average channel response, reducing the mean square error through data fusion. In industrial environments with strong multipath propagation, joint estimation can suppress the influence of random noise. The total frequency offset is obtained by adding the coarse and fine frequency offset estimates, used to generate a compensation signal for frequency offset correction of the data portion. The compensated sequence undergoes frequency domain equalization and demodulation to output a reliable data bitstream.

[0131] In industrial control applications such as automated production lines, the carrier frequency is 2.4 GHz, the bandwidth is 20 MHz, and the environment is subject to crystal oscillator deviation and multipath interference. Extended preamble configuration enables the receiver to adaptively enable advanced estimation, reducing the residual frequency offset from hundreds of kHz to several kHz through multi-stage correction, significantly improving the bit error rate. However, it should be noted that extended symbols increase timing overhead, requiring a trade-off between performance and efficiency in delay-sensitive applications. For standard frames, the system automatically disables advanced functions, performing only a single estimation, with power consumption comparable to existing equipment, ensuring compatibility. This design meets high reliability requirements through efficient hardware deployment while retaining the constraints of practical deployment conditions.

Claims

1. A method for estimating frequency offset in wireless communication, characterized in that, include: Step 1: Receive wireless communication signal and acquire baseband digital sampling sequence. The sequence includes a preamble symbol part and a data part. The preamble symbol part consists of short training field symbols and long training field symbols. Step 2: Identify the configuration of the preamble symbol portion, and determine whether the frame structure is an extended configuration or a standard configuration based on the number of short training field symbols and long training field symbols; Step 3: When the extended configuration is determined, perform multi-level coarse frequency offset estimation: use the first part of the short training field symbols to perform the first level coarse frequency offset estimation, obtain the first estimate value, and perform frequency offset compensation on the subsequent symbols; use the second part of the short training field symbols to perform the second level residual estimation on the compensated signal, obtain the second estimate value; combine the multi-level estimates to obtain the total coarse frequency offset estimate value. Step 4, perform multiple fine frequency offset measurements: use the symbols of the first part of the long training field to perform the first fine frequency offset measurement to obtain the third estimate; The second fine frequency offset measurement is performed using the symbols of the second part of the long training field to obtain the fourth estimate; the final fine frequency offset estimate is obtained by averaging the multiple measurements. Step 5: Perform joint channel estimation using multiple long training field symbols to obtain the average channel response; Step 6: Combine the total coarse frequency offset estimate with the final fine frequency offset estimate to obtain the total frequency offset value, perform frequency offset compensation on the data portion of the baseband digital sampling sequence, and apply the average channel response for frequency domain equalization and demodulation to output demodulated data.

2. The wireless communication frequency offset estimation method as described in claim 1, characterized in that, Step 1 includes: performing down-conversion and analog-to-digital conversion on the wireless communication signal received by the antenna, and outputting a baseband digital sampling sequence; Step 3 includes: when the configuration is determined to be extended, extracting the first two short training field symbols from the baseband digital sampling sequence, calculating the first time-domain autocorrelation value, and deriving the first coarse frequency offset estimate; using the first coarse frequency offset estimate to generate a first compensation signal, and performing frequency offset compensation on the symbols in the baseband digital sampling sequence located after the first two short training field symbols to obtain the first compensated sequence; extracting the subsequent short training field symbols from the first compensated sequence, calculating the second time-domain autocorrelation value, and deriving the second coarse frequency offset estimate; adding the first coarse frequency offset estimate and the second coarse frequency offset estimate to obtain the total coarse frequency offset estimate. Step 4 includes: extracting the first two long training field symbols from the coarse frequency offset compensated sequence, calculating the third time-domain autocorrelation value, and deriving the first fine frequency offset estimate; extracting the subsequent two long training field symbols from the coarse frequency offset compensated sequence, calculating the fourth time-domain autocorrelation value, and deriving the second fine frequency offset estimate; adding the first fine frequency offset estimate and the second fine frequency offset estimate and dividing by two to obtain the final fine frequency offset estimate. Step 5 includes: performing a Fast Fourier Transform on each long training field symbol in the baseband digital sampling sequence, dividing the transformed frequency domain symbol by a known long training frequency domain sequence stored locally to obtain multiple initial channel frequency responses; adding the multiple initial channel frequency responses and dividing by the total number of long training field symbols to output the average channel response. Step 6 includes: adding the total coarse frequency offset estimate to the final fine frequency offset estimate to obtain the total frequency offset value; using the total frequency offset value to generate a second compensation signal to perform frequency offset compensation on the data portion of the baseband digital sampling sequence to obtain a second compensated sequence; performing a fast Fourier transform on the second compensated sequence, dividing the transformed frequency domain data symbols by the average channel response to complete frequency domain equalization; and demapping the equalized frequency domain symbols to output the demodulated data bit stream.

3. The wireless communication frequency offset estimation method as described in claim 1, characterized in that, The configuration for identifying the leading symbol portion includes: Perform sliding correlation calculation on the baseband digital sampling sequence and output the correlation detection results, which include the repetition period information of short training field symbols and the sequence pattern information of long training field symbols; Based on the relevant detection results, calculate the number of symbols in the short training field and the number of symbols in the long training field; If the number of short training field symbols is six and the number of long training field symbols is four, then the frame structure is determined to be an extended configuration, and the first control signal is generated. If the number of short training field symbols is two or three and the number of long training field symbols is two, then the frame structure is determined to be the standard configuration, and a second control signal is generated.

4. The wireless communication frequency offset estimation method as described in claim 2, characterized in that, The step of extracting the first two short training field symbols from the baseband digital sampling sequence, calculating the first time-domain autocorrelation value, and deriving the first coarse frequency offset estimate from the first time-domain autocorrelation value includes: Select the first two short training field symbols in the baseband digital sampling sequence and calculate the time-domain autocorrelation value between the sampling points corresponding to the two short training field symbols. Phase information is extracted from the first time-domain autocorrelation value obtained from the calculation. Calculate the first coarse frequency offset estimate based on the phase information; The step of generating a first compensation signal using a first coarse frequency offset estimate, and compensating for the frequency offset of symbols in the baseband digital sampling sequence located after the first two short training field symbols, to obtain the first compensated sequence includes: A first compensation signal is generated based on the first coarse frequency offset estimate; Multiply the symbol in the baseband digital sampling sequence that is located after the first two short training field symbols with the first compensation signal to complete the frequency offset compensation and output the first compensated sequence. The steps of extracting subsequent short training field symbols from the first compensated sequence, calculating the second time-domain autocorrelation value, and deriving the second coarse frequency offset estimate from the second time-domain autocorrelation value include: The four short training field symbols are extracted from the first compensated sequence. The time-domain autocorrelation values ​​between adjacent short training field symbol pairs are calculated sequentially. The phase difference is extracted from each time-domain autocorrelation value. The arithmetic mean of the extracted phase differences is calculated. The second coarse frequency offset estimate is calculated based on the averaged phase difference.

5. The wireless communication frequency offset estimation method as described in claim 2, characterized in that, The steps of extracting the first two long training field symbols from the coarse frequency offset compensated sequence, calculating the third time-domain autocorrelation value, and deriving the first fine frequency offset estimate from the third time-domain autocorrelation value include: Extract the first two long training field symbols from the sequence after coarse frequency offset compensation, calculate the time domain autocorrelation value between the two long training field symbols as the third time domain autocorrelation value, extract the phase difference from the third time domain autocorrelation value, and calculate the first fine frequency offset estimate based on the phase difference; The steps of extracting the subsequent two long training field symbols from the coarse frequency offset compensated sequence, calculating the fourth time-domain autocorrelation value, and deriving the second fine frequency offset estimate from the fourth time-domain autocorrelation value include: Extract the next two long training field symbols from the sequence after coarse frequency offset compensation, calculate the time-domain autocorrelation value between the two long training field symbols as the fourth time-domain autocorrelation value, extract the phase difference from the fourth time-domain autocorrelation value, and calculate the second fine frequency offset estimate based on the phase difference.

6. The wireless communication frequency offset estimation method as described in claim 2, characterized in that, The process involves performing a Fast Fourier Transform on each long training field symbol in the baseband digital sampling sequence, and then performing a dot division operation between the transformed frequency domain symbol and the locally stored known long training frequency domain sequence to obtain multiple initial channel frequency responses, including: Perform a Fast Fourier Transform on each long training field symbol in the baseband digital sampling sequence and output the frequency domain symbol; Each frequency domain symbol is divided by a known long training frequency domain sequence stored locally to obtain multiple initial channel frequency responses; The step of summing multiple initial channel frequency responses and dividing by the total number of symbols in the long training field to output the average channel response includes: The average channel response is output by summing the frequency responses of multiple initial channels and dividing the sum by the total number of symbols in the long training field.

7. The wireless communication frequency offset estimation method as described in claim 2, characterized in that, The step of generating a second compensation signal using the total frequency offset value and compensating for the frequency offset of the data portion in the baseband digital sampling sequence to obtain the second compensated sequence includes: The total frequency offset is used to generate a complex compensation signal. The data part of the baseband digital sampling sequence is multiplied with the complex compensation signal to complete the frequency offset compensation and output the second compensated sequence. The step of performing a Fast Fourier Transform on the second compensated sequence and dividing the transformed frequency domain data symbols by the average channel response to complete frequency domain equalization includes: Perform a Fast Fourier Transform on the second compensated sequence to output frequency domain data symbols. Divide the frequency domain data symbols by the average channel response to complete frequency domain equalization. The step of demapping the equalized frequency domain symbols and outputting the demodulated data bit stream includes: demapping the equalized frequency domain symbols and outputting the demodulated data bit stream.

8. The wireless communication frequency offset estimation method as described in claim 3, characterized in that, Also includes: When the second control signal indicates that the frame structure is in the standard configuration, according to the second control signal, the operation of extracting the subsequent short training field symbols from the first compensated sequence to calculate the second time-domain autocorrelation value is disabled, and the operation of extracting the subsequent two long training field symbols from the sequence after coarse frequency offset compensation to calculate the fourth time-domain autocorrelation value is also disabled. Only the operation of extracting the first two short training field symbols from the baseband digital sampling sequence, calculating the first time-domain autocorrelation value, and deriving the first coarse frequency offset estimate is performed to obtain a single coarse frequency offset estimate. Only the operation of extracting the first two long training field symbols from the coarse frequency offset compensated sequence, calculating the third time domain autocorrelation value, and deriving the first fine frequency offset estimate is performed to obtain a single fine frequency offset estimate. The single coarse frequency offset estimate is used as the total coarse frequency offset estimate, and the single fine frequency offset estimate is used as the final fine frequency offset estimate. Continue performing the operation of outputting the average channel response by performing Fast Fourier Transform on each long training field symbol in the baseband digital sampling sequence, as well as subsequent frequency offset compensation and demodulation operations.

9. The wireless communication frequency offset estimation method as described in claim 2, characterized in that, Also includes: When the number of symbols in the short training field is greater than six and the number of symbols in the long training field is greater than four, the multi-level coarse frequency offset estimation also includes a third-level estimation, and the multiple fine frequency offset measurements also include a third measurement. After completing the second-level residual estimation, the third-level coarse frequency bias estimation is performed using the symbols of the third part of the short training field to obtain the fifth estimate. The first estimate, the second estimate and the fifth estimate are combined to obtain the extended total coarse frequency bias estimate. After completing the second fine frequency offset measurement, the third fine frequency offset measurement is performed using the symbols of the third part of the long training field to obtain the sixth estimate. The third estimate, the fourth estimate and the sixth estimate are averaged to obtain the extended final fine frequency offset estimate. Joint channel estimation is performed using all long training field symbols to obtain the extended average channel response.

10. A wireless communication frequency offset estimation device, applied to the wireless communication frequency offset estimation method as described in any one of claims 1 to 9, characterized in that, include: The signal receiving module is used to receive wireless communication signals and acquire baseband digital sampling sequences. The baseband digital sampling sequences include a preamble symbol part and a data part. The preamble symbol part is composed of short training field symbols and long training field symbols. The configuration identification module is used to perform sliding correlation calculation on the baseband digital sampling sequence, identify the number of short training field symbols and the number of long training field symbols, and determine whether the frame structure is an extended configuration or a standard configuration based on the identified number. The coarse frequency offset estimation module is used to extract the first two short training field symbols from the baseband digital sampling sequence when the extended configuration is determined, calculate the first time-domain autocorrelation value, derive the first coarse frequency offset estimate value from the first time-domain autocorrelation value, generate the first compensation signal using the first coarse frequency offset estimate value, perform frequency offset compensation on the symbols in the baseband digital sampling sequence located after the first two short training field symbols to obtain the first compensated sequence, extract the subsequent short training field symbols from the first compensated sequence, calculate the second time-domain autocorrelation value, derive the second coarse frequency offset estimate value from the second time-domain autocorrelation value, and add the first coarse frequency offset estimate value and the second coarse frequency offset estimate value to obtain the total coarse frequency offset estimate value. The fine frequency offset measurement module is used to extract the first two long training field symbols from the coarse frequency offset compensated sequence, calculate the third time domain autocorrelation value, derive the first fine frequency offset estimate from the third time domain autocorrelation value, extract the subsequent two long training field symbols from the coarse frequency offset compensated sequence, calculate the fourth time domain autocorrelation value, derive the second fine frequency offset estimate from the fourth time domain autocorrelation value, add the first fine frequency offset estimate and the second fine frequency offset estimate and divide by two to obtain the final fine frequency offset estimate. The channel estimation module performs a fast Fourier transform on each long training field symbol in the baseband digital sampling sequence, divides the transformed frequency domain symbol by a known long training frequency domain sequence stored locally to obtain multiple initial channel frequency responses, adds the multiple initial channel frequency responses and divides them by the total number of long training field symbols to output the average channel response. The compensation demodulation module is used to add the total coarse frequency offset estimate to the final fine frequency offset estimate to obtain the total frequency offset value. The total frequency offset value is used to generate a second compensation signal to perform frequency offset compensation on the data part of the baseband digital sampling sequence to obtain the second compensated sequence. A fast Fourier transform is performed on the second compensated sequence, and the transformed frequency domain data symbols are divided by the average channel response to complete the frequency domain equalization. The equalized frequency domain symbols are demapped to output the demodulated data bit stream.

Images