UWB positioning method based on signal difference

Abstract

The invention provides a UWB positioning method based on signal difference, and belongs to the field of wireless signal positioning of the Internet of Things. In order to solve the problem that an existing UWB positioning system is limited by ToF estimation errors, signal conflicts and synchronization complexity, the method mainly adopts signal difference operation to be combined with PoA and ToF estimation, eliminates initial offset and antenna errors through single-difference and double-difference operation, calculates and eliminates clock offset residual errors in a double-difference operation result by using phase difference, and finally achieves the purpose of positioning the UWB positioning system. And distance estimation is optimized through a fusion filtering and frequency hopping method. According to the invention, high-efficiency UWB positioning with millimeter-level precision can be realized, and large-scale label deployment is supported.

Description

Technical Field

[0001] The present invention belongs to the field of Internet of Things wireless signal positioning, and specifically relates to a UWB positioning method based on signal difference. Background Art

[0002] In recent years, Ultra-Wideband (UWB) technology has received extensive attention due to its wide application in consumer devices such as smartphones. Thanks to the large bandwidth of 500 MHz, UWB signals can provide signal flight time estimates at the picosecond level. This enables commercially available UWB devices to provide centimeter-level positioning accuracy, which is better than other wireless signal modalities such as Wi-Fi, Bluetooth, and acoustic waves.

[0003] Existing UWB-based positioning systems can generally be divided into two categories: time-based systems [1-3] and angle-based systems [4-6]. Time-based systems usually include multiple positioning anchors with known positions, and each anchor estimates the flight time (ToF) of the signal sent from the positioning tag. Then, through trilateration combined with multiple ToF estimates, the position of the tag is determined. However, limited by the limited ToF estimation accuracy of commercially available UWB devices (the error is usually greater than 150 picoseconds), time-based systems cannot provide millimeter-level positioning accuracy. In contrast, angle-based methods use antenna arrays to measure the angle of arrival (AOA) of the signals sent by the tag. By fusing ToF and AOA estimates, single-anchor positioning can be achieved [6]. However, due to the limited number of antennas on commercial UWB devices (usually 2 or 3), such methods usually have relatively large errors exceeding 10 centimeters. Some new UWB positioning systems use the Time Difference of Arrival (TDOA) scheme to support the positioning of a large number of positioning tags [7-8]. However, in traditional TDOA schemes, the positioning tag listening (hereinafter referred to as TO) mechanism is used to reduce ranging signal conflicts. As Figure 1As shown, the anchors in the TDOA system are synchronized by a global clock and sequentially broadcast ranging signals. The positioning tag listens to these ranging signals and uses one-way ranging to calculate the time of flight (ToF) between itself and each anchor. To eliminate the clock bias in the time of flight, the positioning tag calculates the difference in ToF to different anchors and determines its own position through trilateration. Since the positioning tag listens passively and does not transmit signals, theoretically TDOA can support an infinite number of tags. However, this scheme has two fundamental limitations. First, the ToF estimation of commercially available UWB modules usually has a large deviation of more than 150 picoseconds, which results in the positioning accuracy of the current TDOA scheme being only centimeter-level. Although the phase of arrival (PoA) of UWB can provide more accurate distance estimation, due to the existence of time-varying offsets (such as carrier frequency offset) caused by transmitter-receiver separation, it cannot be directly used for ranging. In addition, TDOA requires tight synchronization between anchors (for example, through wired connections or GPS clocks), which may significantly increase the complexity and cost of deployment. It can be seen that due to the error in the ToF estimation of the device, the accuracy of this type of method is still limited to centimeter-level.

[0004] Some methods utilize the high-resolution characteristic of carrier phase for distance in positioning and propose a positioning system based on carrier phase to further improve positioning accuracy [9]. However, this type of method relies on the exchange of two-way ranging (TWR) signals between positioning anchors and tags, which will cause serious signal conflicts when the number of positioning tags is large (for example, in factory material positioning, the number of tags is usually greater than 100), reducing the positioning accuracy and position update rate of the system.

[0005] In summary, there is an urgent need for a positioning method that can support a large number of positioning tags while achieving millimeter-level UWB positioning accuracy.

[0006] References:

[0007] [1]Kempke B, Pannuto P, Campbell B, et al. Surepoint: Exploiting ultrawideband flooding and diversity to provide robust, scalable, high-fidelity indoor localization [C] / / Proceedings of the 14th ACM Conference on Embedded Network Sensor Systems CD-ROM. 2016: 137-149.

[0008] [2] Tiemann J, Wietfeld C. Scalable and precise multi-UAV indoor navigation using TDOA-based UWB localization[C] / / 2017 international conference on indoor positioning and indoor navigation(IPIN). IEEE, 2017:1-7.

[0009] [3] Yang J, Dong B S, Wang J. VULoc: Accurate UWB localization for countless targets without synchronization[J]. Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies, 2022, 6(3):1-25.

[0010] [4] Dotlic I, Connell A, Ma H, et al. Angle of arrival estimation using decawave DW1000 integrated circuits[C] / / 2017 14th Workshop on Positioning, Navigation and Communications(WPNC). IEEE, 2017:1-6.

[0011] [5] Heydariaan M, Dabirian H, Gnawali O. Anguloc: Concurrent angle of arrival estimation for indoor localization with uwb radios[C] / / 2020 16th International Conference on Distributed Computing in Sensor Systems(DCOSS). IEEE, 2020:112-119.

[0012] [6]Zhao M, Chang T, Arun A, et al. Uloc: Low-power, scalable and cm-accurate uwb-tag localization and tracking for indoor applications[J]. Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies, 2021, 5(3): 1-31.

[0013] [7]Groβwindhager B, Stocker M, Rath M, et al. SnapLoc: An ultra-fast UWB-based indoor localization system for an unlimited number of tags[C] / / Proceedings of the 18th International Conference on Information Processing in Sensor Networks. 2019: 61-72.

[0014] [8]Corbalán P, Picco G P, Palipana S. Chorus: UWB concurrent transmissions for GPS-like passive localization of countless targets[C] / / Proceedings of the 18th International Conference on Information Processing in Sensor Networks. 2019: 133-144.

[0015] [9]Jiang J,Wang J,Chen Y,et al.LocRa:Enable practical long-rangebackscatter localization for low-cost tags[C] / / Proceedings of the 21st AnnualInternational Conference on Mobile Systems,Applications and Services.2023:317-329. Summary of the Invention

[0016] The objective of the present invention is to solve the problems that the existing UWB positioning system is limited by the ToF estimation error, signal collision, and synchronization complexity. A UWB positioning method based on signal difference is provided, which eliminates the initial offset and time-varying deviation through signal difference operation, combines PoA and ToF estimation, realizes efficient UWB positioning with millimeter-level accuracy, and supports large-scale tag deployment at the same time.

[0017] To achieve the above objective, the technical solution adopted by the present invention is as follows:

[0018] A UWB positioning method based on signal difference includes the following steps:

[0019] 1) The positioning tag and the reference anchor receive the UWB ranging signal broadcast by the anchor, and respectively extract the time of flight ToF of the signal and the phase of arrival PoA of the signal;

[0020] 2) Error modeling is performed on the ToF and PoA extracted by the positioning tag and the reference anchor. A single-difference operation is performed on the estimated values of ToF and PoA to eliminate the errors related to the transmitting anchor, and then a double-difference operation is performed to eliminate the initial error and antenna error;

[0021] 3) Calculate the PoA phase difference between two adjacent reported signals of the reference anchor, eliminate the phase ambiguity interference in the PoA phase difference, calculate the estimated value of the clock offset residual, and subtract the estimated value of the clock offset residual from the result of the double-difference operation to obtain the fine-grained ToF and PoA estimated values;

[0022] 4) Calculate the estimated values of the ToF and PoA distance differences from each anchor to the positioning tag based on the fine-grained ToF and PoA estimated values, solve the integer ambiguity in the estimated value of the PoA distance difference according to the estimated value of the ToF distance difference, then suppress the additive random noise in the estimated value of the ToF distance difference through fusion filtering, and then eliminate the position distance deviation of the positioning tag through the frequency hopping method to obtain the fine PoA distance difference estimated value;

[0023] 5) Calculate the theoretical distance difference between the positioning tag and two anchor points based on the Time Difference of Arrival (TDOA) multilateration method. Construct an optimization function based on the refined PoA distance difference estimate and the theoretical distance difference, and solve this optimization function to obtain the position of the positioning tag.

[0024] Further, the steps of extracting ToF and PoA in step 1) include:

[0025] The anchor points send ranging signals in sequence and enter the listening state after sending.

[0026] The positioning tag receives the ranging signals from each anchor point and extracts the ToF and PoA parameters.

[0027] The reference anchor point receives the ranging signals from each anchor point and extracts the ToF and PoA parameters.

[0028] Further, the steps of error modeling for the ToF and PoA extracted from the positioning tag and the reference anchor point in step 2) include:

[0029] Perform error modeling on the ToF extracted from the positioning tag. Calculate the ToF estimate through propagation delay, clock offset, device startup initial offset, and antenna delay.

[0030] Perform error modeling on the PoA extracted from the positioning tag. Calculate the PoA estimate through modulo operation on propagation delay phase offset, carrier frequency offset, starting phase offset, and antenna phase shift.

[0031] Perform error modeling on the ToF extracted from the reference anchor point. Calculate the ToF estimate through propagation delay, clock offset, device startup initial offset, and antenna delay.

[0032] Perform error modeling on the PoA extracted from the reference anchor point. Calculate the PoA estimate through modulo operation on propagation delay phase offset, carrier frequency offset, starting phase offset, and antenna phase shift.

[0033] Further, the steps of performing single-difference operation in step 2) include:

[0034] Perform single-difference operation on ToF. Subtract the ToF estimate of the positioning tag from the ToF estimate of the reference anchor point and compensate for the path delay to obtain the result of the ToF single-difference operation.

[0035] Perform single-difference operation on PoA. Subtract the PoA estimate of the positioning tag from the PoA estimate of the reference anchor point and compensate for the path delay to obtain the result of the PoA single-difference operation.

[0036] Further, the steps of performing double-difference operation in step 2) include:

[0037] Calculate the difference between the ToF single-difference operation results of two different anchor points to obtain the ToF double-difference operation result;

[0038] Calculate the difference between the PoA single-difference operation results of two different anchor points to obtain the PoA double-difference operation result.

[0039] Further, the step of calculating the PoA phase difference between two adjacent reports of the reference anchor point in step 3) includes:

[0040] Extract the PoA data in the adjacent report signals of the reference anchor point to obtain two PoA estimated values;

[0041] Calculate the PoA phase difference between the adjacent report signals based on these two PoA estimated values.

[0042] Further, the step of eliminating the phase ambiguity interference in the PoA phase difference in step 3) includes:

[0043] Based on the PoA phase difference, known carrier frequency, and time interval, calculate the ambiguous solution of the clock offset difference;

[0044] Extract the clock offset difference estimated value provided by the commercially available UWB module, calculate the integer offset, eliminate the phase ambiguity interference, and obtain an accurate clock offset difference estimated value;

[0045] Based on the clock offset difference estimated value and the transmission time interval, calculate the clock offset residual estimated value.

[0046] Further, the step of suppressing the additive random noise in the ToF distance difference estimated value by fusion filtering in step 4) includes:

[0047] Calculate the relative displacement of the positioning tag according to the fine-grained PoA estimated values of two adjacent signals;

[0048] Fuse and filter the relative displacement of the positioning tag and the ToF distance difference estimated value to suppress the additive random noise in the ToF distance difference estimated value, and calculate the filtered distance difference estimated value.

[0049] Further, the step of eliminating the position distance deviation of the positioning tag by the frequency hopping method in step 4) includes:

[0050] Calculate the fine-grained PoA estimated values of at least two center frequencies of the UWB signal to form an ambiguity resolution equation system;

[0051] Calculate the equivalent wavelength according to the phase difference between the above two center frequencies, and round the quotient of the ToF distance difference estimated value and the equivalent wavelength to calculate the new integer ambiguity;

[0052] Solve the above equations based on the new integer ambiguity and equivalent wavelength, and calculate the refined PoA distance difference estimate value.

[0053] Further, in step 5), the least squares method or an optimization method based on search is used to solve the optimization function.

[0054] The beneficial effects achieved by the present invention are as follows:

[0055] 1. By keeping the listening state after the anchor point sends a signal, this method enables the positioning tag and the reference anchor point to extract the ToF and PoA parameters of the same ranging signal, ensuring the effectiveness of the differential operation and avoiding the need for clock synchronization between anchor points.

[0056] 2. Through single-difference and double-difference operations, this method eliminates the errors related to the transmitting anchor point and significantly reduces the influence of time-varying errors on the ToF and PoA estimates.

[0057] 3. By calculating the phase difference and resolving the integer ambiguity, this method realizes high-precision clock offset estimation, eliminates the influence of clock offset on the ToF and PoA estimates, and completely cancels its time-varying component.

[0058] 4. This method fuses the ToF and PoA estimates to reduce the random noise of the distance difference estimate, and uses frequency hopping to extend the PoA ambiguity period, eliminating the integer ambiguity problem of the PoA estimate and improving the distance difference estimate accuracy, providing high-precision input data for TDOA positioning.

[0059] 5. Based on the TDOA multilateration positioning of the PoA estimate, this method calculates the optimal tag position, improving the positioning accuracy and reducing the error compared with the traditional TDOA method. Description of the Drawings

[0060] Figure 1 is a schematic diagram of the UWB signal transmission process in the traditional TDOA positioning mode.

[0061] Figure 2 is a flowchart of the UWB positioning method based on signal difference of the present invention.

[0062] Figure 3 is a schematic diagram of the UWB signal transmission process based on the anchor point listening mode of the present invention.

[0063] Figure 4 is a schematic diagram of the original ToF estimate and the original ToA estimate extracted from the positioning tag.

[0064] Figure 5 is a schematic diagram of the original single-difference calculation.

[0065] Figure 6 is a schematic diagram of the original double-difference calculation.

[0066] Figure 7 It is a ToF and PoA estimation diagram processed by double difference calculation.

[0067] Figure 8 It is a fine-grained ToF and PoA estimation diagram that eliminates clock offset residuals.

[0068] Figure 9 It is a schematic diagram of the basic principle of TDOA positioning.

[0069] Figure 10 It is a ToF-based distance difference estimation diagram before and after filtering.

[0070] Figure 11 It is a schematic diagram of the positioning error of the present invention in a typical indoor scenario. Detailed implementation manners

[0071] To make the technical features, advantages or technical effects in the above technical solutions of the present invention more obvious and understandable, the following will be described in detail through embodiments.

[0072] As Figure 2 shown, an embodiment of the present invention provides a UWB positioning method based on signal difference, and its steps mainly include: (1) using a positioning tag and a reference anchor to receive UWB ranging signals broadcast by the anchor; (2) the positioning tag performs signal difference operation to eliminate the initial offset, antenna delay and most of the time-varying clock offset in the time of flight and arrival phase estimation; (3) using the UWB notification signal broadcast by the reference anchor received by the positioning tag to estimate and eliminate the time-varying clock offset residuals in the time of flight and arrival phase; (4) using ToF to assist in solving the PoA integer ambiguity to obtain an unambiguous distance difference estimate; (5) using the obtained unambiguous distance difference estimate to solve the position of the positioning tag. The specific implementation steps of this method are as follows:

[0073] (1) Broadcast and reception of UWB ranging signals

[0074] The present invention provides a ranging signal broadcast and reception mechanism called anchor listening (hereinafter referred to as AO) to improve positioning accuracy. Compared with the traditional TDOA scheme, in the AO scheme, the anchor remains in the listening state after sending data so as to be able to listen to signals from other anchors. As Figure 3 shown, assuming there are N anchors in the system, anchor 1 to anchor N broadcast in turn at t1 to t NSend ranging signals at all times. Consider that after one of the anchors (e.g., anchor N) sends a signal, it enters the listening state and is denoted as the reference anchor. These signals are received by the positioning tag and the reference anchor through TO and AO respectively. The positioning tag and the reference anchor extract ToF and PoA from the received ranging signals respectively. Among them, the ToF and PoA broadcast from anchor i (i ∈ [1, N]) extracted by the positioning tag are denoted as r i,tag and φ i,tag . Similarly, the ToF and PoA broadcast from anchor i extracted by the reference anchor are denoted as r i,ref and φ i,ref .

[0075] The advantage of the AO mode is that the UWB ranging signals (including ToF and PoA) extracted from the positioning tag and the reference anchor have similar error components. Therefore, the differences between them can be used for differential operations to eliminate these errors and improve the positioning accuracy. At the same time, the signal differential-based method does not require clock synchronization between the positioning anchors, reducing the deployment cost of the UWB positioning system. The following combines formula derivation to illustrate the specific steps of the signal differential operation.

[0076] (2) The positioning tag performs signal differential operation

[0077] First, error modeling is performed on the ToF (r i,tag ) and PoA (φ i,tag ) extracted by the positioning tag. According to the IEEE802.15.4 standard, commercial ultra-wideband (UWB) devices such as DW1000 use short pulses of only 2 nanoseconds as their baseband waveforms, with an absolute bandwidth of 500 MHz. This enables UWB receivers to distinguish direct-path signals and multipath reflections in the channel impulse response (CIR) profile and estimate the flight time and phase of the direct-path signal.

[0078] Let d i,tag represent the distance between anchor i and the positioning tag. The UWB signal from anchor i arrives at the tag after a propagation delay τ i,tag = d i,tag / c, where c represents the speed of light. Considering the inherent hardware imperfections of commercially available UWB devices, including clock offset (δ), initial offset (ΔT) after device startup, and antenna delay (ψ), these components will introduce additional time delays in the ToF estimation. Considering these errors, the ToF estimated value r i,tag of the ranging signal received by the positioning tag from anchor i is modeled as:

[0079]

[0080] where t i represents the time when the UWB signal is sent from anchor i, and δ tagand δ i represent the clock offsets of the positioning tag and the anchor i respectively, ΔT tag and ΔT i represent the initial offsets of the positioning tag and the anchor i respectively, ψ tag and ψ i represent the antenna delays of the positioning tag and the anchor i respectively.

[0081] The PoA estimate value φ of the signal received by the tag from the anchor i i,tag is modeled as:

[0082]

[0083] where f c represents the carrier frequency of the UWB signal, 2πf c τ i,tag represents the phase offset corresponding to the propagation delay. and represent the PLL start offsets of the positioning tag and the anchor i respectively, and represent the phase shifts caused by the antenna delays of the positioning tag and the anchor i respectively. mod(·) represents the modulo operation. Equations (1) and (2) show that the error sources of the ToF and PoA estimates extracted by the positioning tag mainly include three aspects: time-varying errors, initial errors after device startup, and antenna-related errors. As Figure 4 shown, there is a large deviation of more than 2 milliseconds in the ToF, while the PoA is affected by random noise within the range of [0, 2π].

[0084] Similarly, the ToF estimate value r of the signal received by the reference anchor from the anchor i i.ref is modeled as:

[0085]

[0086] where τ i,ref represents the path delay of the signal received by the reference anchor from the anchor i. δ ref , ΔT ref and ψ ref represent the clock offset, initial offset and antenna delay of the reference anchor respectively.

[0087] The PoA estimate value φ of the signal received by the reference anchor from the anchor i i,ref is modeled as:

[0088]

[0089] where, and represent the starting phase offset and antenna phase shift of the reference anchor respectively.

[0090] Comparing the ToF and PoA estimates of the comparison positioning tags (Equations (1) and (2)) and the reference anchor points (Equations (3) and (4)), it can be seen that both have the same error components related to anchor point i, such as δ i , ΔT i , ψ i , and as Figure 5 shown. Therefore, these common errors can be eliminated through the differential operation between the two.

[0091] Single difference calculation: Given the ToF estimates r i,tag and r i,ref , the ToF single difference operation can be expressed as:

[0092] r i,tag -r i,ref =τ i.tag -τ i,ref +(δ tag -δ ref )t i +ΔT tag -ΔT ref +ψ tag -ψ ref (5)

[0093] Assuming that the distance d i,tag between anchor point i and the reference anchor point is known, then τ i,trf can be compensated according to τ i,ref =d i,ref / c. The compensated ToF single difference operation result is expressed as:

[0094]

[0095] Similarly, the PoA single difference operation can be expressed as:

[0096]

[0097] After compensating τ i,ref , the single difference operation result of PoA can be expressed as:

[0098]

[0099] Note that in Equations (6) and (8), the errors related to anchor point i have been eliminated. Although there are still errors related to the reference anchor point and the positioning tag. However, it is not difficult to see that in the single difference results calculated using different transmitting anchor points, these errors are consistent, as Figure 6As shown. Therefore, this consistent error can be utilized to perform double-difference operations on these single-difference results to further reduce the error.

[0100] Double-difference calculation: Considering the UWB ranging signal sent from another anchor point j (j≠i≠ref), the positioning tag and the reference anchor point respectively listen to this data packet, and calculate the single-difference estimates of ToF and PoA according to formula (6) and formula (8) respectively. The results are denoted as and The double-difference operation result of ToF can be obtained by subtracting the ToF single-difference operation result and :

[0101]

[0102] where t j represents the time when the UWB signal is sent from anchor point j, and τ j,tag represents the path delay experienced by the UWB signal from the time it is sent from anchor point j to reaching the tag.

[0103] Similarly, the double-difference operation result of PoA can be obtained by subtracting the PoA single-difference operation result and :

[0104]

[0105] Observing formula (9) and formula (10), it can be seen that the initial error after the device starts and the error related to the antenna have been completely eliminated. In addition, the time-varying errors (time-varying bias and carrier frequency offset) are also significantly reduced, and only the residual part remains. Figure 7 shows the double-difference results of ToF and PoA. Compared with the estimated values in Figure 4 , the large time bias in ToF and the random offset in PoA estimation are effectively eliminated. However, the existence of the clock offset residual term results in a slow-time drift still existing in the results, which needs to be further eliminated to ensure the stability of the positioning results.

[0106] Analyzing the composition of the clock offset residual, it can be found that it is the product of the clock offset difference δ tag -δ ref between the positioning tag and the reference anchor point and the time interval t i -t j . Among them, the time interval t i -t j can be obtained by reading the UWB module transmission time register, while δ tag -δ refIt is even more difficult to calculate. This is because the clock offset estimates provided by most commercially available UWB modules usually have an estimation error of more than 0.1 parts per million (ppm), and their accuracy is not sufficient to eliminate the clock offset residuals in the PoA double-difference results.

[0107] (3) Clock offset difference estimation and clock offset residual elimination

[0108] To obtain an accurate clock offset estimate, the present invention proposes a clock offset estimation and residual elimination method based on PoA estimation. Note that in order for the positioning tag to know the reference anchor estimate values (including ToF and PoA estimate values) used for performing signal difference calculations, the reference anchor needs to periodically send report signals carrying these estimate values, and the positioning tag can also extract the PoA of these report signals. It is worth noting that the phase difference between these adjacent report signals contains fine clock offset information. Assume that the reference anchor sends two report data packets at times t ref and t ref +Δt respectively, from which the PoA estimate values φ ref,tag and φ′ ref,tag can be obtained respectively. The derivation of the phase difference Δφ between the two is as follows:

[0109] Δφ = φ ref,tag -φ′ ref,tag = [-2πf c (τ ref,tag -τ′ ref,tag ) + 2πf c (δ tag -δ ref )Δt] mod 2π

[0110] ≈ [2πf c (δ tag -δ ref )Δt] mod 2π (11)

[0111] Since f c , Δφ and Δt are both known, the clock offset difference δ tag -δ ref can be calculated as:

[0112]

[0113] Due to phase wrapping, the result in formula (12) is ambiguous and contains an integer l to be solved, which represents the number of complete tag -δ ref parts contained in δ . Note that commercially available UWB modules can also provide a clock offset difference estimate, denoted as Although this estimate has a large error, it is not ambiguous. Therefore, using Solve for the integer l as follows:

[0114]

[0115] where, represents the floor operation. After obtaining l, the accurate and unambiguous clock offset difference estimate δ can be solved using formula (12) tag -δ ref . The typical variance magnitude of the solution result is only 0.005 ppm, which is much smaller than the estimation error (0.1 ppm) of the built-in method of commercially available UWB modules. Note that this result can be further processed in combination with filtering algorithms (such as SG filtering, Kalman filtering) to reduce the error.

[0116] After obtaining the clock offset difference estimate, multiply it by the transmission time interval to obtain the clock offset residual estimate. Finally, the fine-grained ToF and PoA estimates and can be obtained by subtracting the clock offset residual estimate from the double-difference results represented by formula (9) and formula (10):

[0117]

[0118] Figure 8 Shows the ToF and PoA estimation results after eliminating the clock offset residual, and it can be seen that the time-varying components therein have been completely eliminated.

[0119] (4) PoA ambiguity resolution

[0120] Note that in formula (14) and formula (15), and The remaining terms represent the UWB signal propagation delay differences caused by the different distances from each anchor to the positioning tag. Therefore, a multilateration algorithm similar to time difference of arrival (hereinafter referred to as TDOA) can be used to determine the position of the positioning tag. The core idea of TDOA positioning is as Figure 9 shown. The difference in distances from the positioning tag to two different anchors (in other words, the difference in propagation delays) defines a hyperbola on a two-dimensional plane, characterizing the possible positions of the positioning tag. When the number of anchors is greater than 2, multiple hyperbolas can be obtained using the distance differences. At this time, the position of the positioning tag can be obtained by finding the common intersection of multiple hyperbolas.

[0121] Obviously, in order to obtain an accurate label position estimate, high-precision distance difference information is crucial. Note that the fine-grained ToF and PoA estimates in formula (14) and formula (15) can both provide distance difference estimates. Using the fine-grained ToF estimate the distance difference between anchor i and anchor j to the positioning tag can be calculated (referred to as the ToF distance difference estimate value), and its calculation formula is:

[0122]

[0123] Although the calculation of formula (16) is relatively simple, due to its own inaccuracy, its result usually has a relatively large error of more than 10 centimeters.

[0124] In contrast, the distance difference calculated using the fine-grained PoA estimate value (referred to as the PoA distance difference estimate value) has higher accuracy, but there is a phase ambiguity problem, and its calculation formula is:

[0125]

[0126] where N i,j is the integer ambiguity, which represents the number of signal wavelength λ length parts included in the PoA distance difference estimate value . Since and λ are both known (λ can be determined according to the UWB signal frequency), only by determining N i.j can an unambiguous estimate be obtained. Considering that commercially available UWB devices usually have a phase noise of about 0.3 radians and a signal center frequency of 3.5 GHz, according to formula (17), it can provide a distance difference estimate accuracy of about 0.4 millimeters, which is much better than Therefore, is used as the input of the TDOA positioning algorithm.

[0127] To solve the integer ambiguity N i,j , note that is unambiguous. Ideally, assuming that the error of is less than half of the UWB signal wavelength (for example, when using a 3.5 GHz center frequency, the signal half-wavelength is 4.29 centimeters), the integer ambiguity can be solved in the following way:

[0128]

[0129] However, in actual situations there are often errors greater than half the wavelength. Its error components are mainly composed of two parts: additive random noise and position-related distance deviation.

[0130] To reduce the influence of the additive random noise in , a fusion filtering method combining and is proposed. Let and respectively represent the fine-grained PoA estimation values at the k-th and (k + 1)-th times, and the relative displacement of the positioning tag can be calculated by the following formula:

[0131]

[0132] Formula (19) holds when the movement speed of the positioning tag is less than where ΔT represents the time interval between two adjacent rounds of positioning measurements. The typical values of ΔT and λ are 2.5 milliseconds and 8.57 centimeters respectively, and the corresponding maximum movement speed of the positioning tag is 17.14 m / s, which is sufficient to support most positioning and tracking applications (such as indoor personnel navigation, UAV positioning). After obtaining the relative displacement of the positioning tag, it is fused with for filtering to reduce the noise of the ToF distance difference estimation value:

[0133]

[0134] where represents the filtered distance difference estimation value, and F is the smoothing coefficient. The typical value range of F is 0.6 - 0.8 to balance the initial value error and the noise reduction effect. Figure 10 compares the before filtering with the after filtering, and it can be seen that the additive random noise is significantly suppressed.

[0135] Although the additive random noise is reduced, there is still a distance deviation related to the position of the positioning tag. For this, it is proposed to use the frequency hopping of UWB signals to expand the cycle of the integer ambiguity. Commercially available UWB devices usually support multiple center frequencies (such as Qorvo DW1000 supports 3.5 GHz, 4 GHz, 4.5 GHz, 6.5 GHz). Therefore, it is proposed to implement frequency hopping by switching between channel 1 (3.5 GHz) and channel 3 (4.5 GHz) for every two rounds of positioning of the anchor point and the positioning tag. For the two channels (center frequencies), the above signal difference and clock offset estimation methods are used to obtain the fine-grained PoA estimation values, denoted as and respectively. Using these PoA estimation values, the following equations can be listed:

[0136]

[0137] where and are the center frequencies of UWB channel 1 and channel 3 respectively. By solving formula (21), can be calculated as follows:

[0138]

[0139] Among them, centimeter is the new ambiguity period (in other words, the equivalent wavelength), and is the new integer ambiguity. Compared with the single-frequency case represented by formula (17), the ambiguity period increases significantly from 8.57 cm (using a 3.5 GHz center frequency) to 30 cm. At this time, even if there is an error at the centimeter level related to the position of the positioning tag in the filtered , the new integer ambiguity can still be determined using it as follows:

[0140]

[0141] After determining , the fine PoA distance difference estimate can be determined using formula (22)

[0142] Note that the system of equations obtained using two UWB center frequencies described by formula (21) can be further expanded with the number of frequency hops. In the case of complex indoor multipath reflections, the number of frequency hop points can be increased to obtain better ambiguity resolution performance.

[0143] (5) Solving the position of the positioning tag

[0144] After obtaining the fine PoA distance difference estimate , the UWB tag position is determined using the TDOA multilateration method. The goal of TDOA multilateration is to find an optimal position that closely matches the estimated value of the theoretical distance difference. Given a candidate tag position P and the positions p i , p j of anchor point i and anchor point j, the theoretical distance difference from the positioning tag to these two anchor points can be calculated as follows:

[0145]

[0146] The position of the positioning tag can be obtained by solving the following optimization function:

[0147]

[0148] Formula (25) can be solved using the Chan method based on the least squares method, or other search-based optimization methods (such as the Newton method).

[0149] Figure 11The positioning test results at 30 points in a 3*3-meter indoor space based on commercially available DW1000 UWB modules are given. It can be seen that the median positioning error of this method is only 0.47 cm, and the 90th percentile error is only 1.02 cm, which is better than the traditional TDOA method.

[0150] Although the present invention has been disclosed as above by way of examples, it is not intended to limit the present invention. Appropriate modifications or equivalent replacements made by those of ordinary skill in the art to the technical solutions of the present invention shall be covered within the protection scope of the present invention, and the protection scope of the present invention shall be subject to what is defined by the claims.

Claims

1. A UWB positioning method based on signal difference, characterized in that It includes the following steps: 1) The positioning tag and the reference anchor receive the UWB ranging signals broadcast by the anchor, and respectively extract the time of flight (ToF) of the signal and the phase of arrival (PoA) of the signal; 2) Error modeling is performed on the ToF and PoA extracted by the positioning tag and the reference anchor. The estimated values of ToF and PoA are subjected to a single-difference operation to eliminate the errors related to the transmitting anchor, and then a double-difference operation is performed to eliminate the initial errors and antenna errors; 3) Calculate the PoA phase difference between two adjacent reporting signals of the reference anchor, eliminate the phase ambiguity interference in the PoA phase difference, calculate the estimated value of the clock offset residual, and deduct the estimated value of the clock offset residual from the double-difference operation result to obtain the fine-grained ToF and PoA estimated values; 4) Based on the fine-grained ToF and PoA estimated values, calculate the estimated values of the ToF and PoA distance differences from each anchor to the positioning tag. Solve the integer ambiguity in the PoA distance difference estimated value according to the ToF distance difference estimated value, then suppress the additive random noise in the ToF distance difference estimated value through fusion filtering, and then eliminate the position distance deviation of the positioning tag through the frequency hopping method to obtain the fine PoA distance difference estimated value; 5) Calculate the theoretical distance difference from the positioning tag to two anchors based on the time difference of arrival (TDOA) multilateration method. Construct an optimization function according to the fine PoA distance difference estimated value and the theoretical distance difference, and solve the optimization function to obtain the position of the positioning tag.

2. The UWB positioning method based on signal difference as described in claim 1, wherein The steps of extracting ToF and PoA in step 1) include: The anchor sends the ranging signal in sequence and enters the listening state after sending; The positioning tag receives the ranging signals from each anchor and extracts the ToF and PoA parameters; The reference anchor receives the ranging signals from each anchor and extracts the ToF and PoA parameters.

3. The UWB positioning method based on signal difference as claimed in claim 1, wherein The steps of performing error modeling on the ToF and PoA extracted by the positioning tag and the reference anchor in step 2) include: Perform error modeling on the ToF extracted by the positioning tag, and calculate the ToF estimated value through propagation delay, clock offset, device startup initial offset, and antenna delay; Perform error modeling on the PoA extracted by the positioning tag, and calculate the PoA estimated value through modulo operation of propagation delay phase shift, carrier frequency offset, starting phase offset, and antenna phase shift; Perform error modeling on the ToF extracted by the reference anchor, and calculate the ToF estimated value through propagation delay, clock offset, device startup initial offset, and antenna delay; Perform error modeling on the PoA extracted by the reference anchor, and calculate the PoA estimated value through modulo operation of propagation delay phase shift, carrier frequency offset, starting phase offset, and antenna phase shift.

4. The UWB positioning method based on signal difference as described in claim 1, wherein The steps of performing the single-difference operation in step 2) include: Perform a single-difference operation on ToF, subtract the ToF estimated value of the positioning tag from the ToF estimated value of the reference anchor, and compensate for the path delay to obtain the result of the ToF single-difference operation; Perform a single-difference operation on PoA, subtract the PoA estimated value of the positioning tag from the PoA estimated value of the reference anchor, and compensate for the path delay to obtain the result of the PoA single-difference operation.

5. The UWB positioning method based on signal difference as claimed in claim 1 or 4, characterized in that, The steps of performing the double-difference operation in step 2) include: Calculate the difference between the results of the ToF single-difference operations of two different anchors to obtain the result of the ToF double-difference operation; Calculate the difference between the PoA single-difference operation results of two different anchor points to obtain the PoA double-difference operation result.

6. The UWB positioning method based on signal difference as claimed in claim 1, wherein Step 3) The steps of calculating the PoA phase difference between two adjacent reports of the reference anchor point include: Extract the PoA data in the adjacent report signals of the reference anchor point to obtain two PoA estimated values; Calculate the PoA phase difference between the adjacent report signals based on these two PoA estimated values.

7. The UWB positioning method based on signal difference as claimed in claim 1 or 6, characterized in that The steps of eliminating the phase ambiguity interference in the PoA phase difference in Step 3) include: Based on the PoA phase difference, the known carrier frequency, and the time interval, calculate the fuzzy solution of the clock offset difference; Extract the clock offset difference estimated value provided by the commercially available UWB module, calculate the integer offset, eliminate the phase ambiguity interference, and obtain an accurate clock offset difference estimated value; Based on the clock offset difference estimated value and the transmission time interval, calculate the clock offset residual estimated value.

8. The UWB positioning method based on signal difference as described in claim 1, wherein The steps of suppressing the additive random noise in the ToF distance difference estimated value by fusion filtering in Step 4) include: Calculate the relative displacement of the positioning tag according to the fine-grained PoA estimated values of two adjacent signals; Fuse and filter the relative displacement of the positioning tag and the ToF distance difference estimated value to suppress the additive random noise in the ToF distance difference estimated value, and calculate the filtered distance difference estimated value.

9. The UWB positioning method based on signal difference as claimed in claim 1, wherein The steps of eliminating the position distance deviation of the positioning tag by the frequency hopping method in Step 4) include: Calculate the fine-grained PoA estimated values of at least two center frequencies of the UWB signal to form an ambiguity resolution equation set; Calculate the equivalent wavelength according to the phase difference between the above two center frequencies, and take the integer of the quotient of the ToF distance difference estimated value and the equivalent wavelength to calculate a new integer ambiguity; Solve the above equation set based on the new integer ambiguity and the equivalent wavelength to calculate the fine PoA distance difference estimated value.

10. The UWB positioning method based on signal difference as described in claim 1, characterized in that, In Step 5), the least squares method or an optimization method based on search is used to solve the optimization function.

Images