UWB adaptive ternary sampling module based on zero proportion feedback and UWB receiving system

Abstract

The application discloses a UWB adaptive ternary sampling module based on zero proportion feedback and a UWB receiving system, and belongs to the technical field of wireless communication.The UWB receiving system comprises a radio frequency front end module, a digital processing module and a UWB adaptive ternary sampling module, wherein the UWB adaptive ternary sampling module comprises a first comparator, a second comparator, a logic encoding unit and a threshold generating unit.The UWB baseband energy signal is quantized into three levels, the average proportion of zero symbols in the quantized output is counted in real time, and the error of the average proportion and a preset target proportion is calculated;based on the Bayes theory, the basic step is optimized online by using historical error and threshold data, and different stages are dynamically switched according to the size and convergence state of the zero proportion error, so that the threshold center parameter and the width parameter are independently and adaptively updated, and closed-loop control is formed.The application significantly improves the signal detection and time stamp extraction precision of the UWB receiver in a dynamic environment, and has the advantages of low power consumption and easy implementation.

Description

Technical Field

[0001] This invention relates to the field of wireless communication technology, specifically to a UWB adaptive ternary sampling module and a UWB receiving system based on zero-proportional feedback. Background Technology

[0002] Ultra-wideband (UWB) technology is widely used in the Internet of Things (IoT), smart homes, and precise positioning due to its high bandwidth, high-precision positioning capabilities, and resistance to multipath interference. When a UWB receiver quantizes analog energy signals into digital signals, a high-speed, high-precision analog-to-digital converter (ADC) is required, with sampling rates typically reaching several GHz. Furthermore, ADCs with high bit depths consume extremely high power, making them unsuitable for battery-powered IoT devices. Consequently, UWB receivers require complex modules for matched filtering, carrier recovery, and timing synchronization, resulting in complex and costly hardware implementations.

[0003] Existing solutions typically employ 1-bit quantization (binary sampling), setting only a single threshold and outputting two states: "+1" and "-1". While this method boasts extremely low power consumption, it suffers significant information loss, losing both signal phase and amplitude information. Performance degrades severely under low signal-to-noise ratio or dynamic channel conditions. Ternary sampling, by introducing two thresholds (high and low) and outputting three states ("+1", "0", and "-1"), filters out noise while retaining more signal information, resulting in superior performance.

[0004] However, most existing ternary samplers employ fixed thresholds or threshold setting strategies based on simple statistics (such as signal mean and variance), which cannot consistently maintain optimal quantization performance in complex and variable real-world scenarios (such as distance variations, occlusion, and multipath interference). Improper threshold settings can lead to an excessively high or low proportion of "0" symbols: too high a proportion results in insufficient quantization information, while too low a proportion degenerates into binary quantization, losing its noise immunity advantage. Therefore, dynamically and intelligently adjusting the quantization threshold to ensure the system always operates in an optimal or near-optimal quantization state is crucial for improving the robustness and accuracy of UWB receivers. Summary of the Invention

[0005] In view of this, the present invention provides a UWB adaptive ternary sampling module and a UWB receiving system based on zero-proportional feedback. The present invention optimizes parameter updates through Bayesian theory and, combined with closed-loop control, controls the proportion of "zero" signs in the quantization output to achieve adaptive ternary sampling.

[0006] The technical solution of the present invention is as follows:

[0007] A UWB adaptive ternary sampling module based on zero-proportional feedback includes:

[0008] The first comparator is used to compare the energy signal with a high threshold V. HComparison: If the energy signal is below the high threshold V H If the condition is met, output 0; otherwise, output 1.

[0009] The second comparator is used to compare the energy signal with a low threshold V. L Comparison: If the energy signal is higher than the low threshold V L If the condition is met, output 0; otherwise, output 1.

[0010] The logic encoding unit is used to generate a corresponding three-valued digital signal D[n] based on the output results of the first comparator and the second comparator, where D[n] ∈ {+1, 0, -1} and [n] represents the timing number;

[0011] The threshold generation unit is used to statistically analyze the ternary digital signal generated by the logic encoding unit and dynamically update the high threshold V. H and low threshold V L .

[0012] Furthermore, the logic encoding unit generates the ternary digital signal in the following way:

[0013] When Cmp H =1, Cmp L When =0, D[n]=+1;

[0014] When Cmp H =0, Cmp L When = 0, D[n] = 0;

[0015] When Cmp H =0, Cmp L When =1, D[n]=-1.

[0016] Furthermore, the UWB adaptive ternary sampling module operates as follows:

[0017] Step 1: Initialize the target zero-proportion P0 = 0.6 and the basic step size α. base =0.02, β base =0.015, collect sampling points from the first 32 symbol periods, and calculate the sample mean μ. s and sample standard deviation σ s Set a high threshold low threshold Where h is a constant factor, ranging from 1.5 to 3.0; calculate the initial values ​​of the threshold center C and the threshold width W:

[0018]

[0019] Step 2, compare the energy signal with the current high threshold V. H and low threshold V LThe comparison is performed, and a three-valued digital signal D[n] is generated based on the comparison result. The continuously output three-valued digital signals D[n] constitute a three-valued sequence {D[n]}.

[0020] Step 3: Within an observation window of length M symbol periods, calculate the proportion P of zero values ​​in the three-valued sequence. 0_avg And calculate the zero proportional error e0 = P0 - P 0_avg In addition, the proportion P of positive values ​​in the three-valued sequence is statistically analyzed. +1_avg The proportion of negative values ​​P -1_avg And calculate the symmetry error e C = P +1_avg -P -1_avg ;

[0021] Step 4: Construct a posterior probability model for the step size parameter, and use Bayesian estimation to update the base step size α in real time. base and β base ;

[0022] Step 5, based on the zero proportional error e0 and the symmetry error e C Dynamically select adaptive strategies for different stages and update the threshold center C and threshold width W;

[0023] Step 6: Calculate the new high threshold V based on the updated threshold center C and threshold width W. H and low threshold V L :

[0024]

[0025] Return to step 2.

[0026] Furthermore, step 4 is specifically implemented as follows:

[0027] Step 401: Periodically collect the historical zero-proportional error sequence {e0} and symmetric error sequence {e... C} and the corresponding threshold center sequence {C} and threshold width sequence {W};

[0028] Step 402: Within an observation window of length M symbol periods, calculate the average values ​​of the zero-proportional error sequence, the symmetric error sequence, the threshold center sequence, and the threshold width sequence. And the historical maximum and minimum values ​​e of the zero-proportional error sequence and the symmetric error sequence 0,max e 0,min e C,max e C,min ;

[0029] Step 403: Based on Bayesian estimation theory, construct a posterior probability model and calculate α. base and β baseMaximum a posteriori estimate:

[0030]

[0031] Where Δ is a small constant to prevent division by zero. The variance of the error estimated from the extreme values ​​is calculated as follows:

[0032]

[0033] Step 404: Update the base step size α using a smoothing strategy. base and β base :

[0034] .

[0035] Furthermore, in step 5, the threshold center C and threshold width W are updated as follows:

[0036]

[0037] Where A and B are adjustment coefficients, and F is a scaling factor, the values ​​of which are determined in different stages as follows:

[0038] (1) When |e0|≥0.3, it is the coarse adjustment stage; at this time, the adjustment coefficients A=2, B=2, and the scaling factor F=G are used; G is the momentum acceleration factor based on the consistency of historical error direction, and the calculation method is as follows:

[0039]

[0040] Wherein, S(e0,e C This is a measure of consistency in the error direction.

[0041]

[0042] When S(e0,e C When λ > 0.8, C = 0.5, otherwise λ C = 0; sign is the sign function, which takes values ​​of -1, 0, and +1;

[0043] (2) When |e0|<0.3 and |e0|≥0.15, it is the fine-tuning stage; at this time, the adjustment coefficients A=1, B=1, and the scaling factor F=R are used; R is the damping factor based on error oscillation counting, and the calculation method is as follows:

[0044]

[0045] Where ρ = 0.7 is the damping coefficient; N osc This is the oscillation count, i.e., the number of times the sign is reversed due to the zero proportional error e0;

[0046] (3) When |e0|<0.15 and the system is in a stable state, it is in the steady state stage; at this time, the adjustment coefficients A =0.3, B =0.3, and scaling factor F =1 are used.

[0047] (4) When threshold saturation or signal loss is detected, it is the recovery phase; at this time, the adjustment coefficients A=3, B=3, and scaling factor F=1 are used.

[0048] Furthermore, the present invention also provides a UWB receiving system, including a radio frequency front-end module, a digital processing module, and the UWB adaptive ternary sampling module as described above, wherein:

[0049] The radio frequency front-end module is used to receive UWB radio frequency signals, perform low-noise amplification and down-conversion processing, and output energy signals to the UWB adaptive ternary sampling module.

[0050] The UWB adaptive ternary sampling module performs three-level quantization on the energy signal based on the high and low thresholds to obtain a three-valued digital signal, which is then output to the digital processing module. In addition, the high and low thresholds are adaptively adjusted based on the statistical results of the three-valued digital signal.

[0051] The digital processing module performs pulse detection, timestamp extraction, and position calculation based on the three-valued digital signal output by the UWB adaptive ternary sampling module.

[0052] The beneficial effects of this invention are as follows:

[0053] 1. Strong adaptability: By monitoring zero proportional error in real time and adjusting the threshold through closed-loop feedback, it can automatically adapt to channel changes, significantly improving the receiver's dynamic range and robustness.

[0054] 2. Fast convergence speed: It adopts a four-stage step-by-step update strategy. In the coarse adjustment stage, it uses large momentum to quickly approach the target. In the fine adjustment stage, it uses damping to suppress oscillations. In the steady state stage, it makes fine adjustments.

[0055] 3. High accuracy: By adjusting the basic step size online through Bayesian optimization, the system can maintain the optimal quantization state under different signal-to-noise ratio environments. Attached Figure Description

[0056] Figure 1 This is a block diagram of a UWB receiving system.

[0057] Figure 2 This is a flowchart of the UWB adaptive ternary sampling module.

[0058] Figure 3 This is a flowchart of Bayesian step size optimization.

[0059] Figure 4This is a flowchart of the four-stage threshold update. Detailed Implementation

[0060] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0061] A UWB receiving system, such as Figure 1 As shown, it includes an RF front-end module, a digital processing module, and a UWB adaptive ternary sampling module, wherein:

[0062] The radio frequency front-end module is used to receive UWB radio frequency signals, perform low-noise amplification and down-conversion processing, and output energy signals to the UWB adaptive ternary sampling module.

[0063] The UWB adaptive ternary sampling module performs three-level quantization on the energy signal based on the high and low thresholds to obtain a three-valued digital signal, which is then output to the digital processing module. In addition, the high and low thresholds are adaptively adjusted based on the statistical results of the three-valued digital signal.

[0064] The digital processing module performs pulse detection, timestamp extraction, and position calculation based on the three-valued digital signal output by the UWB adaptive ternary sampling module.

[0065] The UWB adaptive ternary sampling module includes:

[0066] The first comparator is used to compare the input energy signal with a high threshold.

[0067] The second comparator is used to compare the input energy signal with a low threshold.

[0068] The threshold generation unit dynamically generates high and low thresholds based on the statistical analysis of the sampled output.

[0069] The logic encoding unit encodes the outputs of the two comparators into three-valued digital signals (+1, 0, -1).

[0070] like Figure 2 As shown, the UWB adaptive ternary sampling module works as follows:

[0071] S1: Initialization parameters

[0072] The UWB radio frequency signal is received, and after low-noise amplification, downconversion and filtering, the baseband signal r(t) is obtained.

[0073] Set the initial threshold parameter: High threshold Low threshold ; Target zero proportion P0∈[0.4, 0.8]; Basic step size α base =0.02, β base=0.015. Where, μ s and σ s Here, represents the estimated mean and standard deviation of the signal, and h is a constant factor (typically 1.5-3.0).

[0074] The offset can be represented by two independent parameters: the threshold center C and the threshold width W. The initial values ​​of the threshold center C and the threshold width W are:

[0075]

[0076] S2: Ternary Quantization Sampling

[0077] For each sampling time n, the signal r[n] is subjected to three-level quantization:

[0078]

[0079] S3: Average Zero Scale Calculation

[0080] Suppose that the UWB signal frame structure contains M (M=128) consecutive symbols, and each symbol period contains N. s Each sampling point is used to generate real-time statistical quantization output and calculate the average distribution ratio of the sampled values ​​for each symbol period.

[0081] The starting index of the kth symbol is:

[0082]

[0083] The ending index of the kth symbol is:

[0084]

[0085] The number of samples with the kth symbol value of 0 and ±1 are respectively:

[0086]

[0087] Where δ(·) is the Kronecker function, δ(x) = 1 when x=0; and δ(x) = 0 when x≠0.

[0088] Calculate the average distribution proportions of M symbol period sample values ​​of 0, +1, and -1 respectively:

[0089]

[0090] Calculate the zero proportional error:

[0091]

[0092] Calculate the symmetry error:

[0093]

[0094] S4: Bayesian Basic Step Size Optimization

[0095] Construct a posterior probability model for the step size parameter, and update α using Bayesian estimation. base and β base The optimal value. Specifically, such as Figure 3 As shown, using historical zero-proportional error and symmetric error sequences, along with their corresponding historical thresholds, as observed data, and assuming the prior distribution of the adjusted step size is Gaussian, the update formula for the optimal basic step size is derived by maximizing the posterior probability, calculating the historical mean and variance:

[0096]

[0097] in, This represents the historical average of the symmetry error and the zero proportionality error; The threshold center and threshold width are the historical average values; Δ is a small constant to prevent division by zero (e.g., 10). -6 ); Simplified estimation of the variance for zero proportionality error and symmetry error:

[0098]

[0099] Among them, e C,max e C,min e 0,max e 0,min These are the historical maximum and minimum values ​​for symmetric error and zero proportional error, respectively.

[0100] Finally, update the base step size using the following smoothing strategy:

[0101]

[0102] S5: Four-stage threshold update

[0103] The system dynamically switches between four stages based on the magnitude of the zero-proportional error and the convergence status, implementing staged adaptive updates to the threshold center C and threshold width W. The specific process is as follows: Figure 4 As shown. The various stages are as follows:

[0104] The first stage is the coarse adjustment stage, which quickly reduces the zero proportional error from a large value (|e0|≥0.3) to a medium level, avoiding staying in the initial stage for too long.

[0105] Threshold update equation:

[0106]

[0107] Among them, A l= 2 is the center adjustment magnification factor; B l = 2 is the width adjustment magnification factor; G is the acceleration momentum factor:

[0108]

[0109] Wherein, S(e0,e C This is a measure of consistency in the error direction.

[0110]

[0111] When S(e0,e C When λ > 0.8, C = 0.5, otherwise λ C = 0. `sign` is the sign function, which takes values ​​of -1, 0, and +1.

[0112] The second stage, the fine-tuning stage, further reduces the zero proportional error from a medium level (0.15≤|e0|<0.3) to a small error (|e0|<0.15), achieving fine optimization.

[0113] Threshold update equation:

[0114]

[0115] Among them, A m =1 is the standard center adjustment factor; B m =1 is the standard width adjustment coefficient; R is the damping factor:

[0116]

[0117] Where ρ = 0.7 is the damping coefficient; N osc This is an oscillation counter, which increases when e0 reverses its sign.

[0118] The third stage: steady state stage, maintaining stability near the target value (|e0|<0.15), handling minor fluctuations and environmental changes.

[0119] Fine-tuning the update equation:

[0120]

[0121] Among them, A s =0.3 is the small center adjustment factor; B s =0.3 is a small width adjustment factor.

[0122] The fourth stage is the recovery stage, which involves recovering from severe anomalies such as threshold saturation, signal loss, and noise spikes.

[0123] Employs ultra-large adjustment step size:

[0124]

[0125] Among them, A r =3 is the adjustment factor for the ultra-large step size; B r =3 is the adjustment factor for the ultra-large step size.

[0126] S6: Threshold conversion output

[0127] Convert the updated threshold center C and threshold width W into the new high threshold V. H and low threshold V L Repeat step S2 to perform ternary quantization.

[0128] The system can be implemented using application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or high-performance embedded microcontrollers (MCUs).

[0129] The following is a more specific example, which uses an FPGA architecture as an example for illustration. However, those skilled in the art should understand that this does not constitute a limitation on the present invention.

[0130] This embodiment uses a single-chip FPGA to implement all functions. The specific hardware platform configuration is as follows: Components Model / Specification Function Description FPGA Xilinx Artix-7 A100T The main control chip implements all digital logic. Simulated front end UWB RF front-end module Low-noise amplification, downconversion ADC 8 channels, 249.6MHz / channel, 8 bits 8-channel parallel sampling, equivalent sampling rate 1996.8MHz Master clock 38.4MHz temperature-compensated crystal oscillator System reference clock power supply 3.3V LDO powered by

[0131] Overall architecture description:

[0132] like Figure 1 As shown, the system comprises three parts: a radio frequency (RF) front-end module, a UWB adaptive ternary sampling module, and a digital processing module. The RF front-end module and energy detection module are implemented using analog circuits, while the UWB adaptive ternary sampling module and digital processing module are implemented within an FPGA. The system operates as follows:

[0133] Step S1: System parameter initialization

[0134] The FPGA controls the ADC to continuously acquire samples from the first 32 symbol cycles at a sampling rate of 1996.8MHz, and calculates the sample mean μ. s and sample standard deviation σ s High threshold low threshold Where h is a constant factor, and in this embodiment, h = 2.0.

[0135] Control parameter initialization: parameter symbol Value illustrate Target zero ratio <![CDATA[P0]]> 0.6 Between 0.4 and 0.8 base step size <![CDATA[α base , b base ]]> 0.02, 0.015 initial value Number of observation window symbols M 32 Sliding window length Number of sampling points per symbol <![CDATA[N s ]]> 2176

[0136] Step S2: Ternary Quantization Sampling

[0137] Inside the FPGA, two 8-bit digital comparators are used to implement threshold comparison, and then the comparator output is encoded to obtain a three-value sequence {-1, 0, +1}.

[0138]

[0139] Step S3: Calculation of the zero-proportion average of multiple symbols

[0140] Instantiate a depth of N in the FPGA's internal Block RAM. s A FIFO with a bit width of 3 bits (containing 2^176 / 8 = 2^72 bits, storing the statistical results of 8 sampling points per cycle) is used to buffer the statistical information of each symbol. A statistical calculation is triggered every 2^72 clock cycles (i.e., one symbol cycle).

[0141] Count the number of 0s and ±1s in M ​​symbol periods and calculate their proportion.

[0142]

[0143] Calculate the zero proportional error:

[0144]

[0145] Calculate the symmetry error:

[0146]

[0147] Step S4: Bayesian basic step size optimization

[0148] like Figure 3 As shown, Bayesian step-size optimization is performed once per frame, triggered by an internal FPGA flag. A historical data buffer with depth L = 100 is maintained in the FPGA's internal Block RAM, storing e0, e1, and e2 of the most recent 100 frames. C C, W values, and historical extreme values ​​of statistical symmetry error and zero proportionality error e C,max e C,min e 0,max e 0,min Calculate the historical average of the symmetry error and the zero proportionality error. Historical average of threshold center and threshold width The variance is estimated using the range method:

[0149]

[0150] The optimal step size is calculated using the formula derived from Bayesian posterior maximization:

[0151]

[0152] Finally, update the base step size using the following smoothing strategy:

[0153]

[0154] S5: Four-stage threshold update

[0155] like Figure 4 As shown, this embodiment implements a four-state state machine within the FPGA, performing a stage judgment and threshold update once per symbol cycle.

[0156] The general update equations for each stage are as follows:

[0157]

[0158] Where A and B are adjustment coefficients for each stage; and F is the scaling factor for each stage.

[0159] The first stage, the coarse strip stage: A=2 and B=2, introduces a momentum acceleration factor G based on the consistency of historical error direction, i.e., F=G;

[0160] The second stage, fine-tuning stage: A=1 and B=1, introduce a damping factor R based on error oscillation counting, i.e., F=R;

[0161] The third stage, the steady-state stage: A = 0.3 and B = 0.3, the scaling factor F = 1 in this stage;

[0162] Phase 4, Recovery Phase: Use ultra-large adjustment coefficients A=3 and B=3, scaling factor F=1.

[0163] Step S6: Threshold conversion output

[0164]

[0165] Updated V H and V L The output is sent to the comparator reference voltage input for ternary quantization in the next symbol cycle.

[0166] This system achieves ternary sampling by controlling the proportion of "zero" signs in the quantization output through closed-loop control, and combines Bayesian theory to adaptively optimize the basic step size, thereby realizing rapid, stable and adaptive adjustment of the threshold in stages.

[0167] In summary, existing UWB receivers using fixed or simple adaptive ternary sampling methods cannot maintain optimal quantization performance under dynamic channel conditions, leading to inaccurate zero-symbol ratios and affecting receiver robustness and accuracy. This invention performs three-level (+1, 0, -1) quantization on the UWB baseband energy signal; it calculates the average proportion of zero symbols in the quantized output in real time and its error compared to a preset target proportion; based on Bayesian theory, it optimizes the base step size online using historical error and threshold data; and it dynamically switches between four stages—coarse adjustment, fine adjustment, steady state, and recovery—according to the magnitude and convergence state of the zero-proportion error, independently and adaptively updating the threshold center parameter and width parameter to form a closed-loop control. This invention, by combining zero-proportion closed-loop control with Bayesian step size optimization, achieves rapid, stable, and adaptive adjustment of the quantization threshold, enabling the system to maintain optimal quantization state under dynamic channel conditions. This allows the UWB receiver to maintain high-precision signal detection and timestamp extraction capabilities under various channel and rapidly changing environmental conditions, while maintaining low power consumption and ease of implementation.

Claims

1. A UWB adaptive ternary sampling module based on zero proportional feedback, characterized in that, include: The first comparator is used to compare the energy signal with a high threshold V. H Comparison: If the energy signal is below the high threshold V H If the condition is met, output 0; otherwise, output 1. The second comparator is used to compare the energy signal with a low threshold V. L Comparison: If the energy signal is higher than the low threshold V L If the condition is met, output 0; otherwise, output 1. The logic encoding unit is used to generate a corresponding three-valued digital signal D[n] based on the output results of the first comparator and the second comparator, where D[n] ∈ {+1, 0, -1} and [n] represents the timing number; The threshold generation unit is used to statistically analyze the ternary digital signal generated by the logic encoding unit and dynamically update the high threshold V. H and low threshold V L .

2. The UWB adaptive ternary sampling module based on zero proportional feedback according to claim 1, characterized in that, The logic encoding unit generates a ternary digital signal in the following way: When Cmp H =1, Cmp L When =0, D[n]=+1; When Cmp H =0, Cmp L When n = 0, D[n] = 0; When Cmp H =0, Cmp L When =1, D[n]=-1.

3. The UWB adaptive ternary sampling module based on zero proportional feedback according to claim 1, characterized in that, The working mode of the UWB adaptive ternary sampling module is as follows: Step 1: Initialize the target zero-proportion P0 = 0.6 and the basic step size α. base =0.02, β base =0.015, collect sampling points from the first 32 symbol periods, and calculate the sample mean μ. s and sample standard deviation σ s Set a high threshold low threshold Where h is a constant factor, ranging from 1.5 to 3.0; calculate the initial values ​​of the threshold center C and the threshold width W: Step 2, compare the energy signal with the current high threshold V. H and low threshold V L The comparison is performed, and a three-valued digital signal D[n] is generated based on the comparison result. The continuously output three-valued digital signals D[n] constitute a three-valued sequence {D[n]}. Step 3: Within an observation window of length M symbol periods, calculate the proportion P of zero values ​​in the three-valued sequence. 0_avg And calculate the zero proportional error e0 = P0 - P 0_avg In addition, the proportion P of positive values ​​in the three-valued sequence is statistically analyzed. +1_avg The ratio of negative values ​​P -1_avg And calculate the symmetry error e C = P +1_avg -P -1_avg ; Step 4: Construct a posterior probability model for the step size parameter, and use Bayesian estimation to update the base step size α in real time. base and β base ; Step 5, based on the zero proportional error e0 and the symmetry error e C Dynamically select adaptive strategies for different stages and update the threshold center C and threshold width W; Step 6: Calculate the new high threshold V based on the updated threshold center C and threshold width W. H and low threshold V L : Return to step 2.

4. The UWB adaptive ternary sampling module based on zero proportional feedback according to claim 3, characterized in that, The specific method for step 4 is as follows: Step 401: Periodically collect the historical zero-proportional error sequence {e0} and symmetric error sequence {e... C } and the corresponding threshold center sequence {C} and threshold width sequence {W}; Step 402: Within an observation window of length M symbol periods, calculate the average values ​​of the zero-proportional error sequence, the symmetric error sequence, the threshold center sequence, and the threshold width sequence. And the historical maximum and minimum values ​​e of the zero-proportional error sequence and the symmetric error sequence 0,max e 0,min e C,max e C,min ; Step 403: Based on Bayesian estimation theory, construct a posterior probability model and calculate α. base and β base Maximum a posteriori estimate: Where Δ is a small constant to prevent division by zero. The variance of the error estimated from the extreme values ​​is calculated as follows: Step 404: Update the base step size α using a smoothing strategy. base and β base : 。 5. The UWB adaptive ternary sampling module based on zero proportional feedback according to claim 3, characterized in that, In step 5, the threshold center C and threshold width W are updated as follows: Where A and B are adjustment coefficients, and F is a scaling factor, the values ​​of which are determined in different stages as follows: (1) When |e0|≥0.3, it is the coarse adjustment stage; at this time, the adjustment coefficients A=2, B=2, and the scaling factor F=G are used; G is the momentum acceleration factor based on the consistency of historical error direction, and the calculation method is as follows: Where S(e0,e C This is a measure of consistency in the error direction. When S(e0,e C When λ > 0.8, C = 0.5, otherwise λ C = 0; sign is the sign function, which takes values ​​of -1, 0, and +1; (2) When |e0|<0.3 and |e0|≥0.15, it is the fine-tuning stage; at this time, the adjustment coefficients A=1, B=1, and the scaling factor F=R are used; R is the damping factor based on error oscillation counting, and the calculation method is as follows: Where ρ = 0.7 is the damping coefficient; N osc This is the oscillation count, i.e., the number of times the sign is reversed due to the zero proportional error e0; (3) When |e0|<0.15 and the system is in a stable state, it is in the steady state stage; at this time, the adjustment coefficients A = 0.3, B = 0.3, and the scaling factor F = 1 are used. (4) When threshold saturation or signal loss is detected, it is the recovery phase; at this time, the adjustment coefficients A=3, B=3, and scaling factor F=1 are used.

6. A UWB receiving system, characterized in that, It includes an RF front-end module, a digital processing module, and a UWB adaptive ternary sampling module as described in any one of claims 1-5, wherein: The radio frequency front-end module is used to receive UWB radio frequency signals, perform low-noise amplification and down-conversion processing, and output energy signals to the UWB adaptive ternary sampling module. The UWB adaptive ternary sampling module performs three-level quantization on the energy signal based on the high and low thresholds to obtain a three-valued digital signal, which is then output to the digital processing module. In addition, the high and low thresholds are adaptively adjusted based on the statistical results of the three-valued digital signal. The digital processing module performs pulse detection, timestamp extraction, and position calculation based on the three-valued digital signal output by the UWB adaptive ternary sampling module.

Images