High-efficiency analog-to-digital converter wireless receiver and power consumption optimization method

Abstract

This invention relates to the field of signal transmission, specifically to a high-efficiency analog-to-digital converter (ADC) wireless receiver and a power consumption optimization method. The method includes: dividing future time into several reference time periods, each of which sequentially includes a first time period and a second time period; adjusting the sampling frequency based on differences in several data points within the first time period; determining the sampling frequency for the second time period based on the information recognition capability of the first time period; extracting several reference data sequences from historical data and obtaining an information attention coefficient; and within the first time period of the next reference time period, combining the information recognition capability of the second time period with the differences in several data points using the information attention coefficient, and adjusting the sampling frequency for the next analog-to-digital conversion. This invention dynamically and recursively adjusts the sampling frequency across multiple consecutive time periods, ensuring that the data output across multiple time periods balances data accuracy and power consumption.

Description

Technical Field

[0001] This invention relates to the field of signal transmission, specifically to a high-efficiency analog-to-digital converter wireless receiver and a power consumption optimization method. Background Technology

[0002] Analog-to-digital converter (ADC) wireless receivers are important modules in radio frequency (RF) processors or RF chips. They use wireless communication technology to receive raw analog signals (such as audio signals) and convert them into digital signals that the processor can process.

[0003] In operation, an analog-to-digital (ADC) wireless receiver needs to receive the original signal via a wireless communication module and convert it using an analog-to-digital (ADC) converter. Therefore, the operation of an ADC wireless receiver consumes a significant amount of power. In many practical applications, the original signal being acquired does not always fill the entire bandwidth or maintain high dynamic range. For example, during voice pauses, stable phases of sensor monitoring, or when channel conditions are favorable, the effective information rate of the signal is low. If it continues to operate mechanically at the highest sampling frequency in these situations, it will sample and convert a large amount of redundant information or noise, resulting in a huge waste of energy. Conversely, operating at a lower sampling frequency cannot guarantee the accuracy of the output signal, which is detrimental to signal recognition. Summary of the Invention

[0004] To address the above problems, this invention provides a high-efficiency analog-to-digital converter wireless receiver and a power consumption optimization method.

[0005] The high-efficiency analog-to-digital converter wireless receiver and power consumption optimization method of the present invention adopt the following technical solution:

[0006] One embodiment of the present invention provides a power consumption optimization method for a high-efficiency analog-to-digital converter wireless receiver, the method comprising the following steps:

[0007] The data output by the analog-to-digital converter wireless receiver when operating at the maximum sampling frequency is recorded as historical data; when low-power mode is enabled, future time is divided into several reference time periods, each of which contains a first time period and a second time period in sequence;

[0008] After outputting a certain number of data points in the first time period, the sampling frequency for the next analog-to-digital conversion is adjusted based on the differences between the data points. The information recognition capability F1 of the first data sequence output in the first time period is extracted, the sampling frequency in the second time period is determined based on F1, and the information recognition capability F2 of the second data sequence output in the second time period is extracted.

[0009] Several reference data sequences are extracted from historical data. The time length of the reference data sequence is equal to the length of the reference time period, and the data sequence output in the second time period of the reference data sequence is most similar to the second data sequence. The difference between the information recognition ability of the data sequence in the first time period and the distribution of F1 among all reference data sequences is denoted as the information attention coefficient.

[0010] In the first time period of the next reference time period, after outputting a number of data, the difference between F2 and the data is combined using the information attention coefficient to obtain the frequency adjustment coefficient. The sampling frequency for the next analog-to-digital conversion is then adjusted using the frequency adjustment coefficient.

[0011] Preferably, the specific steps for extracting the information recognition capability of the first data sequence output in the first time period are as follows:

[0012] Obtain the target filter. The filtering result obtained by the target filter on the first data sequence is denoted as L1. Continue to use the target filter to filter L1, and the resulting filtering result is denoted as L2.

[0013] The difference between two data points at the same position in L1 and L2 is denoted as the third difference at each position. The average of the third differences at all positions in L1 is denoted as the effective noise interference. The difference between any two adjacent data points in L1 is denoted as the fourth difference. The average of the fourth differences at all adjacent data points in L1 is denoted as the average fluctuation amplitude of L1. The difference between the average fluctuation amplitude and the effective noise interference is denoted as the information recognition capability.

[0014] Preferably, the specific steps for obtaining the target filter are as follows:

[0015] Several filters are preset, and each filter is used to filter the first data sequence to obtain the filtered result of the first data sequence. The difference between two data at the same position in the first data sequence and the filtered result is denoted as the second difference at each position. The mean of the second differences at all positions in the first data sequence is obtained and denoted as the noise interference amount of the first data sequence extracted based on each filter.

[0016] For all the noise interference extracted by the filters, the filter whose noise interference is less than the preset threshold th1 and whose noise interference differs from th1 the least is denoted as the target filter.

[0017] Preferably, the method for determining the sampling frequency within the second time period based on F1 includes:

[0018] The average sampling frequency of all data output by the analog-to-digital converter wireless receiver during the first time period is obtained and denoted as f0; the sampling frequency during the second time period is set to (1+w2)×f0; where w2 represents the sampling frequency fine-tuning coefficient, and w2 is negatively correlated with F1.

[0019] Preferably, several reference data sequences are extracted from the historical data. The time length of the reference data sequence is equal to the length of the reference time period, and the data sequence output within the second time period in the reference data sequence is most similar to the second data sequence. The specific steps include the following:

[0020] Several sample sequences are sampled from historical data, and the time length of each sample sequence is equal to the length of the reference time period; each sample sequence also contains a first time period and a second time period in sequence;

[0021] When the current time is in the second time interval of the reference time interval, and the current time is greater than the average of all times in the second time interval: the second data sequence output in the next second time interval after the current time is denoted as L3; in each sample sequence, a data sequence is extracted from the beginning time of the second time interval of the sample sequence, which is denoted as the subsequent subsequence of each sample sequence. The time length of this subsequent subsequence is the same as the time length of L3; the similarity between L3 and the subsequent subsequence of each sample sequence is obtained; for all sample sequences, the sample sequence with the highest similarity is obtained and marked as the reference data sequence obtained at the current time;

[0022] When the current time is at the end of the second time period within the reference time period, several reference data sequences are marked.

[0023] Preferably, the difference between the information recognition ability of the data sequences within the first time period and the distribution of F1 among all the reference data sequences is denoted as the information attention coefficient, and the specific steps include the following:

[0024] Cluster the information recognition capabilities of all reference data sequences to obtain all first categories;

[0025] The mean of all information recognition capabilities in each first category is denoted as the category center. The category center with the smallest difference from F1 is obtained, and the first category corresponding to this category center is denoted as the target category. The difference between the category center of F1 and the target category is denoted as Y1. The mean of the differences between F1 and the category centers of all other first categories besides the target category is obtained and denoted as Y2. The ratio of the number of information recognition capabilities contained in the target category to the number of information recognition capabilities in all reference data sequences is denoted as the density of the target category. The information attention coefficient is positively correlated with the ratio of Y1 to Y2 and negatively correlated with the density of the target category.

[0026] Preferably, the frequency adjustment coefficient is obtained by combining F2 and the differences of several data points using the information attention coefficient, and the specific formula is as follows:

[0027] The difference between F2 and F1 is denoted as the information adjustment coefficient; the corrected result F3 of F2 is obtained, and F3 is positively correlated with both F2 and the information adjustment coefficient.

[0028] The frequency adjustment coefficient is obtained by combining F3 and the differences of several data using the information attention coefficient.

[0029] Preferably, the frequency adjustment coefficient is obtained by combining F3 and the differences of several data points using the information attention coefficient, and the specific formula is as follows:

[0030] Set the frequency adjustment coefficient to Y×F3+(1-Y)×W0; where Y represents the information attention coefficient and W0 represents the difference between several data.

[0031] Preferably, the specific steps for adjusting the sampling frequency during the next analog-to-digital conversion are as follows:

[0032] The difference or frequency adjustment coefficient of several data points is represented as w; the sampling frequency during the next analog-to-digital conversion is w×Fmax+(1-w)×Fmin; where Fmax represents the maximum sampling frequency and Fmin represents the minimum sampling frequency.

[0033] Another embodiment of the present invention provides a high-efficiency analog-to-digital converter (ADC) wireless receiver, including a WiFi module, an ADC module, and a data cable. The WiFi module receives the original signal via wireless communication technology, and the ADC module converts the received analog signal into a digital signal and outputs it. The data cable is connected to a processor via a USB interface, and the processor's power supply powers the ADC wireless receiver via the data cable. After the processor reads the signal output by the ADC wireless receiver, it executes all the steps of the above-described high-efficiency ADC wireless receiver power consumption optimization method when running a computer program.

[0034] The beneficial effects of the technical solution of the present invention are:

[0035] This invention divides future time into several reference time periods, each of which sequentially includes a first time period and a second time period. Within the first time period, after outputting a certain amount of data, the sampling frequency for the next analog-to-digital conversion is adjusted based on the differences between these data points. This process dynamically adjusts the sampling frequency based on the data change patterns over a short period, initially saving power consumption and avoiding the loss of significant information to some extent, thus initially ensuring the accuracy of the output data.

[0036] Furthermore, the information recognition capability F1 of the first data sequence output in the first time period is extracted, and the sampling frequency in the second time period is determined based on F1. In this process, the information recognition capability F1 describes whether effective information can be clearly identified in the output data sampled according to the above method in the first time period. By dynamically determining the sampling frequency in the second time period through F1, the problem of not being able to guarantee the accuracy of the output information in the second time period when it is difficult to accurately and reliably sample effective information in the first time period can be avoided. This ensures the overall accuracy of the output information in each reference time period, while reducing power consumption.

[0037] Furthermore, the information recognition capability of the data sequence within the first time period in all reference data sequences of this invention differs from the distribution of F1 as the information attention coefficient. Within the first time period of the next reference time period, after outputting several data points, the information attention coefficient is used to combine the differences between F2 and the several data points, and the sampling frequency for the next analog-to-digital conversion is adjusted. In this process, a larger information attention coefficient indicates that the interference introduced by the sampled output data within the first time period of the reference time period is not commonly present in specific signal segments of historical data. This either indicates that the original signal within the first time period of the reference time period has unique characteristics, or that the sampling process within the first time period has caused severe distortion. A smaller information attention coefficient indicates that the interference introduced by the sampled output data within the first time period of the reference time period is commonly present in specific signal segments of historical data. This indicates that the interference caused by the sampling process within the first time period cannot be further eliminated by increasing the sampling frequency. Based on this, this invention uses the information attention coefficient to combine the differences between F2 and the several data points and adjust the sampling frequency for the next analog-to-digital conversion to determine whether to continue focusing on or using the sampling method from the previous reference time period in the next first time period, in order to ensure the accuracy of the output signal over a long period or further reduce power consumption.

[0038] In summary, this invention dynamically and recursively adjusts the sampling frequency across multiple consecutive time periods (including the first and second time periods of the previous reference time period, and the first time period of the next reference time period), enabling the output data from multiple time periods to balance data accuracy and power consumption. Attached Figure Description

[0039] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0040] Figure 1This is a flowchart illustrating the steps of a high-efficiency analog-to-digital converter (ADC) wireless receiver power consumption optimization method provided in an embodiment of the present invention. Detailed Implementation

[0041] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of the high-efficiency analog-to-digital converter wireless receiver and power consumption optimization method proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0042] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0043] The following description, in conjunction with the accompanying drawings, details the specific scheme of the high-efficiency analog-to-digital converter wireless receiver and power consumption optimization method provided by this invention. Example

[0044] Please see Figure 1 The diagram illustrates a flowchart of a power consumption optimization method for a high-efficiency analog-to-digital converter wireless receiver according to an embodiment of the present invention. The method includes the following steps:

[0045] Step S101: When the analog-to-digital converter wireless receiver operates at the maximum sampling frequency, the data output is recorded as historical data. When the low-power mode is enabled, the future time is divided into several reference time periods, each of which includes a first time period and a second time period. After outputting several data points within the first time period, the sampling frequency for the next analog-to-digital conversion is adjusted based on the differences between the data points.

[0046] The analog-to-digital converter wireless receiver receives the original signal via wireless communication technology, then performs analog-to-digital conversion on the original signal and outputs it. In this embodiment, the original signal refers to the voice signal collected by the microphone.

[0047] In this embodiment, when the analog-to-digital converter wireless receiver has a high battery level (e.g., when the power supply battery level is greater than 30%), it operates in normal mode. The normal mode refers to the analog-to-digital converter wireless receiver operating at the maximum sampling frequency for analog-to-digital conversion. The data output in this mode is recorded as historical data.

[0048] When the analog-to-digital converter wireless receiver has low power (e.g., when the power supply is less than or equal to 30%), it operates in a low-power mode. The low-power mode refers to the sampling frequency of the analog-to-digital converter wireless receiver dynamically changing when performing analog-to-digital conversion.

[0049] It should be noted that when the analog-to-digital converter wireless receiver is in low-power mode, the sampling frequency of the analog-to-digital converter is equal to the frequency of the signal received by the wireless communication technology. That is, the wireless communication technology is only enabled to receive a signal once every time an analog-to-digital conversion is performed. This embodiment reduces the power consumption of analog-to-digital conversion and wireless communication by dynamically reducing the sampling frequency.

[0050] When low-power mode is enabled, the future time is divided into several reference time periods, each of which contains a first time period and a second time period in sequence.

[0051] As an example, each reference time period is 1 second long, with the first 0.3 seconds of each reference time period being the first time period and the last 0.7 seconds being the second time period.

[0052] After outputting a certain amount of data in the first time period of each reference time period, the sampling frequency for the next analog-to-digital conversion is adjusted based on the differences between the data. The sampling frequency is positively correlated with the differences in the data, meaning that when the differences in the data are relatively large, it indicates that the signal has significant changes, and the sampling frequency is larger in order to accurately capture the fluctuations in the signal; when the differences in the data are relatively small, it indicates that the signal fluctuations are small in a short period of time and may not contain effective information, and the sampling frequency is smaller in order to save power consumption.

[0053] As an example, after outputting a certain number of data points within the first time period, the sampling frequency for the next analog-to-digital conversion is adjusted based on the differences between these data points. The methods include the following:

[0054] Within the first time period, acquire the N0 most recently output data points and obtain the difference between the N0 data points. This difference is denoted as the data difference coefficient w1. The sampling frequency for the next analog-to-digital conversion is w1×Fmax+(1-w1)×Fmin.

[0055] Fmax represents the maximum sampling frequency, and Fmin represents the minimum sampling frequency. Specifically, when w1 is less than the preset threshold th1, w1 is set to 0. This is to ensure that when the differences between consecutively output data are small, sampling is performed at the minimum sampling frequency. When w1 is greater than 1, w1 is set to 1. This is to ensure that when the data differences are too large, sampling is performed at the maximum sampling frequency.

[0056] This embodiment uses N0=5 as an example; where Fmin is set to twice the original signal acquisition frequency (satisfying the Nyquist-Shannon sampling theorem). Specifically, in this embodiment, Fmin is 4kHz and Fmax is 8kHz. In other embodiments, to further save power consumption, Fmin can be set to 1.2 times the original signal acquisition frequency, i.e., 2.4kHz.

[0057] As an example, obtaining the differences among N0 data points involves the following steps:

[0058] Find the maximum value (max) and minimum value (min) among N0 data points, and denote the difference between the N0 data points as (max-min) / m.

[0059] The purpose of using m as the denominator is to remove dimensions and orders of magnitude. In this embodiment, m = max + 1 is set so that both dimensions and orders of magnitude can be removed, while avoiding a denominator of 0.

[0060] As an optional example, th1 can be set to 0.1.

[0061] As a preferred example, methods for obtaining th1 include:

[0062] The sequence of all historical data is denoted as the historical sequence. Gaussian filtering is applied to the historical sequence to obtain the filtered historical sequence, with the Gaussian filter kernel length set to 5. For two data points at the same position in the historical sequence and the filtered historical sequence, the difference between these two data points is calculated using the steps described above and recorded as the first difference at each position. The first differences at all positions in the historical sequence are obtained; these first differences represent the noise interference distribution of the data sampled at the maximum sampling frequency (i.e., the original signal). Mean-shift clustering is performed on all the first differences at all positions to obtain all categories. The category containing the most first differences is selected; this category represents the main noise interference amplitude of the original signal. The mean of all first differences within this category is used as th1, representing the main noise interference amplitude of the original signal (or historical data). Mean-shift clustering is a known technique; in this embodiment, the bandwidth of mean-shift clustering is set to 0.1.

[0063] In this example, it means that sampling is performed at the minimum sampling frequency when the difference between the continuously output data is less than the magnitude of the main noise interference of the original signal.

[0064] In other examples, to save computation, when low-power mode is enabled, historical data within a recent period (e.g., within 10 minutes) is extracted from all historical data and recorded as a historical sequence.

[0065] Step S102: Extract the information recognition capability F1 of the first data sequence output by the analog-to-digital converter wireless receiver within the first time period.

[0066] All data output within the first time period of each reference time period are arranged in chronological order to form the first data sequence. The information recognition capability of the first data sequence is extracted and denoted as F1. Information recognition capability describes whether valid information can be clearly identified in the output data sampled using the above method within the first time period. The greater the information recognition capability, the more valid information can be identified; the smaller the information recognition capability, the less valid information can be identified, or in other words, the valid signal is drowned out by useless noise fluctuations.

[0067] As an example, the methods for extracting information recognition capabilities from the first data sequence include:

[0068] Set up several filters, and use each filter to filter the first data sequence to obtain the filtered result of the first data sequence.

[0069] For two data points at the same position in the first data sequence and the filtering result, the difference between the two data points is calculated using the method in step S101 and recorded as the second difference at each position. The mean of the second differences at all positions in the first data sequence is obtained and recorded as the noise interference amount of the first data sequence extracted based on each filter.

[0070] For all the noise interference extracted by the filters, the filter whose noise interference is less than th1 and whose noise interference differs from th1 the smallest is denoted as the target filter. The filtering result obtained by filtering the first data sequence using the target filter is denoted as L1. Continuing to filter L1 using the target filter, the resulting filter is denoted as L2.

[0071] For two data points at the same location in L1 and L2, the difference between these two data points is calculated using the method in step S101, and recorded as the third difference at each location. The average of the third differences across all locations in L1 is then recorded as the effective noise interference. For any two adjacent data points in L1, the difference between these two data points is calculated using the method in step S101, and recorded as the fourth difference at each location. The average of the fourth differences across all adjacent data points in L1 is recorded as the average fluctuation amplitude of L1. The difference between the average fluctuation amplitude and the effective noise interference is recorded as the information recognition capability.

[0072] In this process, the filter whose noise interference is less than th1 and whose noise interference differs from th1 the least is called the target filter. Its purpose is to enable the target filter to remove the main noise from the original signal in the first data sequence. That is, the filtering result L1 represents the result after removing the main noise of the original signal. The above process can avoid noise interference from the original signal when extracting information recognition ability based on the filtering result L1, and only focuses on the noise interference caused by the change of sampling frequency.

[0073] Further, based on the filtered result L1, the target filter is used for further filtering, so that the obtained effective noise interference can describe the noise interference amplitude caused by the change in sampling frequency. The average fluctuation amplitude describes the fluctuation range of the filtered result (i.e., L1) after removing the noise interference from the original signal. The larger this value, the more obvious the fluctuation of L1. This embodiment uses the difference between the average fluctuation amplitude and the effective noise interference to obtain information recognition capability. The larger the difference between the average fluctuation amplitude and the effective noise interference, the more obvious the fluctuation trend of the filtered result L1 still is after subtracting the noise interference amplitude caused by the change in sampling frequency, indicating that it can identify more effective information, i.e., the greater the information recognition capability. Conversely, the smaller the difference between the average fluctuation amplitude and the effective noise interference, the less obvious the fluctuation trend of the filtered result L1 is after subtracting the noise interference amplitude caused by the change in sampling frequency, and it is overwhelmed by the noise interference caused by the change in sampling frequency, indicating that it cannot identify more effective information, i.e., the smaller the information recognition capability.

[0074] As an example, several filters are set up, including Gaussian filters with Gaussian kernel lengths of 3, 5, 7, 9, 11, and 13.

[0075] Specifically, in the above process, when there are multiple filters whose noise interference is less than th1 and whose noise interference differs from th1 the least, the filter with the smallest Gaussian kernel length is denoted as the target filter.

[0076] It should be noted that when using a Gaussian filter to filter the first data sequence, the step size is set to 1. Additionally, the length of the filtered result is different from that of the first data sequence; that is, some data at both ends of the first data sequence are missing at corresponding positions in the filtered result. In this embodiment, a linear interpolation algorithm is used to interpolate the missing data at these positions, ensuring that the length of the filtered result is the same as that of the first data sequence.

[0077] As an example, the difference between the average fluctuation amplitude and the effective noise interference is denoted as the information recognition capability, and includes the following formulas:

[0078] Let (Q1-Q2) / m1 be denoted as the information recognition capability, where Q1 represents the average fluctuation amplitude and Q2 represents the effective noise interference. The purpose of using m1 as the denominator is to remove dimensions and orders of magnitude. In this embodiment, m1 = Q1 + 1, ensuring that m1 removes both dimensions and orders of magnitude while avoiding a denominator of 0. Specifically, when the information recognition capability is less than 0, it indicates that no effective information content can be obtained from it; in this case, the information recognition capability is set to 0.

[0079] Step S103: Determine the sampling frequency within the second time period based on F1, and extract the information recognition capability F2 of the second data sequence output within the second time period.

[0080] (1) The smaller the information recognition capability F1, the more difficult it is to accurately and reliably sample effective information during the sampling process in the first time period. In order to further ensure the accuracy and reliability of the information contained in the output data of the analog-to-digital conversion wireless receiver, this embodiment determines the sampling frequency in the second time period based on F1. This sampling frequency is negatively correlated with F1.

[0081] As an example, determining the sampling frequency within the second time period based on F1 includes the following methods:

[0082] The average sampling frequency of all data output by the analog-to-digital converter wireless receiver within the first time period is obtained and denoted as f0.

[0083] The sampling frequency for the second time period is set to (1+w2)×f0. Specifically, when this sampling frequency is greater than 90% of the maximum sampling frequency, the sampling frequency is set to 90% of the maximum sampling frequency. The purpose of this is to avoid excessive power consumption caused by working at a high sampling frequency for a long time.

[0084] w2 represents the sampling frequency fine-tuning coefficient, w2=exp(-F1), where exp() represents an exponential function with the natural constant as the base.

[0085] During the above process, the analog-to-digital converter wireless receiver operates at this sampling frequency during the second time period. Specifically, the more severe the noise interference in the data sampled during the first time period, the higher the sampling frequency can be during the second time period. This avoids the problem of insufficient sampling frequency introducing noise into the output data throughout the entire reference time period, which would prevent the output data during the reference time period from completely failing to reflect the information contained in the original signal.

[0086] (2) Record the data output by the analog-to-digital converter wireless receiver during the second time period as the second data sequence, and extract the information recognition capability of the second data sequence, which is denoted as F2. See step S102 for the specific method.

[0087] Step S104: Extract several reference data sequences from historical data. The time length of the reference data sequence is equal to the length of the reference time period, and the data sequence output in the second time period of the reference data sequence is most similar to the second data sequence. The information recognition ability of the data sequence in the first time period of all reference series and the distribution difference of F1 are recorded as the information attention coefficient.

[0088] In this embodiment, when the analog-to-digital converter wireless receiver is working in normal mode, for the data output by the analog-to-digital converter wireless receiver, a data sequence with a time length of L0 is extracted every preset time period and recorded as a sample sequence. The length of L0 is equal to the length of the reference time period. Therefore, the sample sequence can also be divided into two time periods, namely, the first time period and the second time period in chronological order.

[0089] In this embodiment, the preset time period is 5 seconds. That is, when the analog-to-digital converter wireless receiver operates in normal mode, a sample sequence is captured after a 5-second interval, and then another sample sequence is captured after a 5-second interval. The purpose of this process is to collect samples of the original signal to aid in subsequent calculations in low-power mode. In other embodiments, to save storage space, the preset time period can be set to a larger value, such as 1 minute; or a portion of the sample sequences can be randomly removed. In some embodiments, all data from all sample sequences can be directly used as historical data in step S101. Specifically, if the normal mode operating time of the analog-to-digital converter wireless receiver is less than L0, then no more sample sequences are collected.

[0090] When the analog-to-digital converter wireless receiver enables low-power mode, and the current time falls within the latter half of the second time period of the reference time period (i.e., when the current time is greater than the average of all times within the second time period), the following processing is performed:

[0091] The second data sequence output within the second time period following the current time (that is, the sequence consisting of data output by the analog-to-digital converter wireless receiver from the start of the second time period to the current time) is denoted as L3.

[0092] In each sample sequence, a data sequence is extracted from the beginning of the second time period of the sample sequence, and is called the post-subsequence of each sample sequence. The time length of the post-subsequence is the same as the time length of L3.

[0093] Obtain the similarity between L3 and the subsequent subsequence of each sample sequence. For all sample sequences, obtain the sample sequence with the highest similarity and mark it as the reference data sequence.

[0094] As an example, the similarity between L3 and the subsequent subsequences of each sample sequence is obtained as follows:

[0095] The DTW algorithm is used to obtain the DTW distance between L3 and the subsequent subsequence of each sample sequence, denoted as x. exp(-x) is used as the similarity score, where exp represents an exponential function with the natural constant as the base. When L3 is identical or similar to the subsequent subsequence of each sample sequence, the value of x is smaller, and the similarity score is greater.

[0096] Thus, a reference data sequence has been obtained at the current time within the second time period of the reference time period. As the analog-to-digital converter wireless receiver continues to operate during the second time period, it outputs data and obtains reference data sequences simultaneously. Note that the already selected reference data sequences are not included in the above processing to ensure that the selected reference data sequences are different. When the second time period ends, multiple reference data sequences can be obtained using the above method.

[0097] These reference data sequences have the most similar signal information to the second data sequence output in the second time period within the reference time period. At the same time, the reference data sequences are output at the maximum sampling frequency, which can accurately describe the original signal.

[0098] For any reference data sequence, the ability to extract information from the data sequence within the first time period of the reference data sequence is referred to as the information recognition ability of any reference data sequence.

[0099] The difference between the information recognition capability of the reference data sequence and the distribution of F1 is denoted as the information attention coefficient. A larger information attention coefficient indicates a different distribution of information recognition capability between F1 and the reference data sequence, or that F1 is more anomalous compared to the information recognition capability of the reference data sequence. This further indicates that the interference introduced by the data sampled in step S101 within the first time period of the reference time period is not commonly found in specific signal segments of historical data. This either indicates that the original signal within the first time period of the reference time period has unique characteristics, or that the sampling process within the first time period has caused severe distortion. Conversely, a smaller information attention coefficient indicates that the distribution of information recognition capability between F1 and the reference data sequence is the same. This further indicates that the interference introduced by the data sampled in step S101 within the first time period of the reference time period is commonly found in specific signal segments of historical data. This means that the interference introduced by the sampling process within the first time period cannot be further eliminated by increasing the sampling frequency.

[0100] As an example, the difference between the information recognition ability of the reference data sequence and the distribution of F1 is denoted as the information attention coefficient. The methods include:

[0101] The mean-shift clustering algorithm is used to cluster the information identification capabilities of all reference data sequences, resulting in all first-order clusters. In this example, the bandwidth of the mean-shift clustering algorithm is set to 0.2. Specifically, each isolated information identification capability that is not clustered into a first-order cluster is also considered as a first-order cluster.

[0102] The mean of all information recognition capabilities in each first category is denoted as the category center. The category center with the smallest absolute difference from F1 is selected, and the first category corresponding to this category center is denoted as the target category. The absolute value of the difference between F1 and the category center of the target category is denoted as Y1. The mean of the absolute values ​​of the differences between F1 and the category centers of all other first categories besides the target category is denoted as Y2. The ratio of the number of information recognition capabilities contained in the target category to the number of information recognition capabilities in all reference data sequences is denoted as the density of the target category. The information attention coefficient is positively correlated with the ratio of Y1 to Y2 and negatively correlated with the density of the target category.

[0103] Each first category represents a distribution of the information recognition ability of all reference data sequences. The higher the density of the target category, the more representative the target category is of the distribution of the information recognition ability of all reference data sequences. Since F1 is the least different from the target category, F1 is more likely to be the same as the distribution of the information recognition ability of all reference data sequences, and the smaller the information attention coefficient is.

[0104] The smaller the ratio of Y1 to Y2, the greater the difference between F1 and the target category, and the greater the difference between F1 and other first categories. In other words, F1 more closely matches the distribution of information recognition ability contained in the target category, and the smaller the information attention coefficient. Conversely, the larger the ratio of Y1 to Y2, the less significant the difference between F1 and the target category and the difference between F1 and other first categories. This indicates that F1 does not conform to the distribution of information recognition ability contained in the target category, or that F1 is different from the distribution of information recognition ability represented by all first categories, and the larger the information attention coefficient.

[0105] As an example, the formula for calculating the information attention coefficient is: Where Y represents the information attention coefficient, The density of the target category is represented by exp(), which represents an exponential function with the natural constant as the base.

[0106] Step S105: In the first time period of the next reference time period, after outputting several data points, use the information attention coefficient to combine F2 and the difference between the several data points to obtain the frequency adjustment coefficient, and use the frequency adjustment coefficient to adjust the sampling frequency for the next analog-to-digital conversion.

[0107] The information attention coefficient was obtained at the end of the second time period of each reference time period. Next, the analog-to-digital converter wireless receiver samples and outputs data during the first time period of the next reference time period.

[0108] Before sampling and outputting data in the first time period of the next reference time period, it should be noted that for the information attention coefficient obtained in the above steps, the larger the information attention coefficient, the more likely that some of the data output in the previous reference time period may have unique characteristics different from historical data or have obvious errors (i.e., serious distortion). In this case, it is necessary to continue to pay attention to or continue to use the sampling method in the previous reference time period in the next reference time period to ensure the accuracy of the output signal over a long period of time. The smaller the information attention coefficient, the more likely that continuing to pay attention to or continue to use the sampling method in the previous reference time period in the next reference time period will not be able to further eliminate interference, and may instead lead to greater power consumption.

[0109] Based on this, in the first time period of the next reference time period, after outputting a certain number of data (for example, after outputting N0 data), the difference between F2 and the data is combined using the information attention coefficient to obtain the frequency adjustment coefficient, and the sampling frequency of the next analog-to-digital conversion is adjusted using the frequency adjustment coefficient.

[0110] The larger the information attention coefficient, the more attention is paid to F2 when combining F2 with the differences of several data points. This allows the sampling method from the previous reference time period to continue to be used in the first time period of the next reference time period to ensure the accuracy of the output signal over a long period. Conversely, the smaller the information attention coefficient, the more attention is paid to the differences of the data when combining F2 with the differences of several data points. This means that the sampling method from the previous reference time period can no longer be used to further eliminate interference. Instead, sampling is based on the changes in the data in the first time period of the next reference time period to minimize power consumption.

[0111] As an example, the frequency adjustment coefficient is obtained by combining F2 and the differences of several data points using the information attention coefficient. The sampling frequency for the next analog-to-digital conversion is then adjusted using the frequency adjustment coefficient. The steps include:

[0112] Let w3 = Y × F2 + (1 - Y) × W0.

[0113] Where w3 represents the frequency adjustment coefficient, Y represents the information attention coefficient, and W0 represents the difference of several data points. The method and steps for obtaining these data are the same as those in step S101.

[0114] Furthermore, the sampling frequency for the next analog-to-digital conversion is w3×Fmax+(1-w3)×Fmin.

[0115] This concludes this step.

[0116] The subsequent steps need to be repeated from step S102 to implement this embodiment.

[0117] This concludes the example. Example

[0118] In the example of step S103 of the above embodiment, by increasing the sampling frequency in the second time period, the problem of noise being introduced into all the data output in the entire reference time period due to insufficient sampling frequency is avoided, which would cause the data output in the reference time period to completely fail to reflect the information contained in the original signal. This improves the reliability of the output data and the effective information content that can be clearly identified. However, there may be a problem that the method of improving reliability and identifiable information content by increasing the sampling frequency is not effective. On the contrary, increasing the frequency will lead to wasted power consumption. This will not only increase the power consumption in the second time period in step S103, but also make the power consumption in the first time period of the next reference time period in step S105 too high.

[0119] Based on this, this embodiment provides another method for obtaining the frequency adjustment coefficient mentioned in step S105, specifically including:

[0120] At the end of step S103, the difference between F2 and F1 is recorded as the information adjustment coefficient. The larger the information adjustment coefficient, the more the reliability of the output data and the effective information content that can be clearly identified can be significantly improved by increasing the sampling frequency. The smaller the information adjustment coefficient, the less effective the method of improving reliability and identifiable information content by increasing the sampling frequency is, and the more power consumption is wasted due to the increase in frequency.

[0121] As an example, methods for obtaining information adjustment coefficients include:

[0122] Let the information adjustment coefficient equal to (F2-F1) / m2, where m2 is used as the denominator to remove dimensions and orders of magnitude. In this embodiment, m2=F2+1, so that m2 can both remove dimensions and orders of magnitude and avoid a denominator of 0. Specifically, when the information adjustment coefficient is less than 0, it indicates that increasing the sampling frequency cannot improve reliability or the amount of clearly identifiable effective information; in this case, the information adjustment coefficient is set to 0.

[0123] Furthermore, let the frequency adjustment coefficient w3 = Y × F3 + (1 - Y) × W0.

[0124] F3 represents the correction result of F2, and F3 is positively correlated with both F2 and the information adjustment coefficient.

[0125] As an example, F3 = F2 × g, where g represents the information adjustment coefficient.

[0126] Specifically, a larger information adjustment coefficient indicates that increasing the sampling frequency can significantly improve the reliability of the output data and the amount of clearly identifiable effective information. In this case, a larger F3 also indicates a greater tendency to determine the sampling frequency based on the information recognition capability F2 of the second time period to ensure the accuracy of the output data. Conversely, a smaller information adjustment coefficient indicates that increasing the sampling frequency is not effective in improving reliability and identifiable information content; instead, it leads to wasted power. In this case, it is less likely to determine the sampling frequency based on the information recognition capability F2 of the second time period to save power. Example

[0127] This embodiment provides a high-efficiency analog-to-digital converter (ADC) wireless receiver, which includes a WiFi module, an ADC module, and a data cable. The WiFi module receives the original signal via wireless communication technology. The ADC module converts the received analog signal into a digital signal and outputs it. The data cable powers the ADC wireless receiver and outputs the digital signal. In this embodiment, the data cable is connected to a laptop computer via a USB interface. After the processor in the laptop computer reads the signal output by the ADC wireless receiver, it executes the methods described in all the above embodiments, obtains the sampling frequency, and transmits the sampling frequency to the ADC wireless receiver, which then operates at this sampling frequency.

[0128] In this embodiment, the power supply of the laptop computer powers the analog-to-digital converter wireless receiver via a data cable. The power supply mentioned in step S101 of embodiment one refers to the power supply of the laptop computer.

[0129] In other embodiments, the WiFi module can be replaced with a Bluetooth module.

[0130] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for optimizing power consumption of a high-efficiency analog-to-digital converter wireless receiver, characterized in that, The method includes the following steps: The data output by the analog-to-digital converter wireless receiver when operating at the maximum sampling frequency is recorded as historical data; when low-power mode is enabled, future time is divided into several reference time periods, each of which contains a first time period and a second time period in sequence; After outputting a certain number of data points in the first time period, the sampling frequency for the next analog-to-digital conversion is adjusted based on the differences between the data points. The information recognition capability F1 of the first data sequence output in the first time period is extracted. Based on F1, the sampling frequency in the second time period is determined, and the information recognition capability F2 of the second data sequence output in the second time period is extracted. The sampling frequency in the second time period is negatively correlated with F1. Several reference data sequences are extracted from historical data. The time length of the reference data sequence is equal to the length of the reference time period, and the data sequence output in the second time period of the reference data sequence is most similar to the second data sequence. The difference between the information recognition ability of the data sequence in the first time period and the distribution of F1 among all reference data sequences is denoted as the information attention coefficient. In the first time period of the next reference time period, after outputting a number of data, the difference between F2 and the data is combined with the information attention coefficient to obtain the frequency adjustment coefficient. The sampling frequency during the next analog-to-digital conversion is adjusted using the frequency adjustment coefficient. The specific steps involved in extracting and identifying the information of the first data sequence output in the first time period are as follows: Obtain the target filter. The filtering result obtained by the target filter on the first data sequence is denoted as L1. Continue to use the target filter to filter L1, and the resulting filtering result is denoted as L2. The difference between two data points at the same position in L1 and L2 is denoted as the third difference at each position. The average of the third differences at all positions in L1 is denoted as the effective noise interference. The difference between any two adjacent data points in L1 is denoted as the fourth difference. The average of the fourth differences at all adjacent data points in L1 is denoted as the average fluctuation amplitude of L1. The difference between the average fluctuation amplitude and the effective noise interference is denoted as the information recognition capability. The difference between the information recognition ability of the data sequences within the first time period and the distribution of F1 among all reference data sequences is denoted as the information attention coefficient. The specific steps involved are as follows: Cluster the information recognition capabilities of all reference data sequences to obtain all first categories; The mean of all information recognition capabilities in each first category is denoted as the category center. The category center with the smallest difference from F1 is obtained, and the first category corresponding to this category center is denoted as the target category. The difference between the category center of F1 and the target category is denoted as Y1. The mean of the differences between F1 and the category centers of all other first categories besides the target category is obtained and denoted as Y2. The ratio of the number of information recognition capabilities contained in the target category to the number of information recognition capabilities in all reference data sequences is denoted as the density of the target category. The information attention coefficient is positively correlated with the ratio of Y1 to Y2 and negatively correlated with the density of the target category. The frequency adjustment coefficient is obtained by combining F2 and the differences of several data points using the information attention coefficient, and the specific formulas are as follows: The difference between F2 and F1 is denoted as the information adjustment coefficient; the corrected result F3 of F2 is obtained, and F3 is positively correlated with both F2 and the information adjustment coefficient. The frequency adjustment coefficient is obtained by combining F3 with the differences of several data using the information attention coefficient. The frequency adjustment coefficient is obtained by combining F3 and the differences of several data points using the information attention coefficient, and the specific formulas are as follows: Set the frequency adjustment coefficient to Y×F3+(1-Y)×W0; where Y represents the information attention coefficient and W0 represents the difference between several data.

2. The power consumption optimization method for a high-efficiency analog-to-digital converter wireless receiver according to claim 1, characterized in that, The specific steps for obtaining the target filter are as follows: Several filters are preset, and each filter is used to filter the first data sequence to obtain the filtered result of the first data sequence. The difference between two data at the same position in the first data sequence and the filtered result is denoted as the second difference at each position. The mean of the second differences at all positions in the first data sequence is obtained and denoted as the noise interference amount of the first data sequence extracted based on each filter. For all the noise interference extracted by the filters, the filter whose noise interference is less than the preset threshold th1 and whose noise interference differs from th1 the least is denoted as the target filter.

3. The power consumption optimization method for a high-efficiency analog-to-digital converter wireless receiver according to claim 1, characterized in that, The specific method for determining the sampling frequency within the second time period based on F1 is as follows: The average sampling frequency of all data output by the analog-to-digital converter wireless receiver during the first time period is obtained and denoted as f0; the sampling frequency during the second time period is set to (1+w2)×f0; where w2 represents the sampling frequency fine-tuning coefficient, and w2 is negatively correlated with F1.

4. The power consumption optimization method for a high-efficiency analog-to-digital converter wireless receiver according to claim 1, characterized in that, Several reference data sequences are extracted from the historical data. The time length of each reference data sequence is equal to the length of a reference time period, and the data sequence output within the second time period of the reference data sequence is most similar to the second data sequence. The specific steps include the following: Several sample sequences are sampled from historical data, and the time length of each sample sequence is equal to the length of the reference time period; each sample sequence also contains a first time period and a second time period in sequence; When the current time is in the second time interval of the reference time interval, and the current time is greater than the average of all times in the second time interval: the second data sequence output in the next second time interval after the current time is denoted as L3; in each sample sequence, a data sequence is extracted from the beginning time of the second time interval of the sample sequence, which is denoted as the subsequent subsequence of each sample sequence. The time length of this subsequent subsequence is the same as the time length of L3; the similarity between L3 and the subsequent subsequence of each sample sequence is obtained; for all sample sequences, the sample sequence with the highest similarity is obtained and marked as the reference data sequence obtained at the current time; When the current time is at the end of the second time period within the reference time period, several reference data sequences are marked.

5. The power consumption optimization method for a high-efficiency analog-to-digital converter wireless receiver according to claim 1, characterized in that, The specific steps involved in adjusting the sampling frequency during the next analog-to-digital conversion are as follows: The difference or frequency adjustment coefficient of several data points is represented as w; the sampling frequency during the next analog-to-digital conversion is w×Fmax+(1-w)×Fmin; where Fmax represents the maximum sampling frequency and Fmin represents the minimum sampling frequency.

6. A high-efficiency analog-to-digital converter (ADC) wireless receiver, comprising a WiFi module, an ADC module, and a data cable; the WiFi module receives raw signals via wireless communication technology, the ADC module converts the received analog raw signals into digital signals and outputs them; the data cable is connected to a processor via a USB interface, and the processor's power supply powers the ADC wireless receiver via the data cable, characterized in that... After the processor reads the signal output by the analog-to-digital converter wireless receiver, the processor executes all the steps of the power consumption optimization method for the high-efficiency analog-to-digital converter wireless receiver as described in any one of claims 1 to 5 when running a computer program.

Images