Vehicle tracking method, vehicle tracking system, and program

Abstract

To provide a vehicle tracking method, a vehicle tracking system, and a program.SOLUTION: A vehicle tracking method includes: receiving dispersion type optical fiber sensing (DFOS) data; identifying hit points in the DFOS data, each hit point corresponding to a position of a corresponding vehicle at a detection time; clustering the identified hit points to define one or more clusters; classifying each of the one or more clusters into a first classification or a second classification based on a quantity of hit points and a quantity of detection times in each of the one or more clusters; and estimating a vehicle parameter of a first vehicle corresponding to the hit point of the first cluster of one or more clusters based on the hit point of the first cluster having the first classification.SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] The present disclosure relates to a distributed fiber optic sensing (DFOS) system and method of use. [Background technology]

[0002] Optical fibers exist along many roadways. Distributed Acoustic Sensors (DAS) attached to these optical fibers can detect vibrations where the optical fibers are located. In some cases, these vibrations are the result of passing vehicles. The DAS can collect data on the number of vehicles, vehicle lane position, and vehicle speed.

[0003] The DAS generates waterfall data based on time and distance to determine traffic parameters. The ability of the DAS to detect individual vehicles is related to the amount of noise in the signal detected by the DAS. Summary of the Invention [Problem to be solved by the invention]

[0004] It is desirable to improve the accuracy of determining vehicle parameters. [Means for solving the problem]

[0005] A vehicle tracking method according to a first embodiment of the present disclosure includes receiving distributed optical fiber sensing (DFOS) data, identifying hit points in the DFOS data, each hit point corresponding to the position of a corresponding vehicle at a detection time, clustering the identified hit points to define one or more clusters, classifying each of the one or more clusters into a first classification or a second classification based on the number of hit points in each of the one or more clusters and the number of detection times, and estimating vehicle parameters of a first vehicle corresponding to the hit points of the first cluster of the one or more clusters based on the hit points of the first cluster having the first classification.

[0006] A vehicle tracking system according to a second embodiment of the present disclosure comprises a computer-readable medium configured to store instructions and a processor connected to the computer-readable medium, wherein the processor is configured to execute the instructions to receive distributed optical fiber sensing (DFOS) data, identify hit points in the DFOS data, each hit point corresponding to a corresponding vehicle position at a detection time, cluster the identified hit points to define one or more clusters, classify each of the one or more clusters into a first classification or a second classification based on the number of hit points in each of the one or more clusters and the number of detection times, and estimate vehicle parameters of a first vehicle corresponding to the hit points of the first cluster of the one or more clusters based on the hit points of the first cluster having the first classification.

[0007] A program according to a third embodiment of the present disclosure causes a computer to perform the following steps: receive distributed optical fiber sensing (DFOS) data; identify hit points in the DFOS data, each of which corresponds to the position of a corresponding vehicle at a detection time; cluster the identified hit points to define one or more clusters; classify each of the one or more clusters into a first classification or a second classification based on the number of hit points in each of the one or more clusters and the number of detection times; and estimate vehicle parameters of a first vehicle corresponding to the hit points of the first cluster of the one or more clusters based on the hit points of the first cluster having the first classification. [Brief explanation of the drawings]

[0008] Aspects of the present disclosure are best understood from the following detailed description when read in conjunction with the accompanying drawings. It should be noted that, according to standard industry practice, various features are not drawn to scale. In fact, the dimensions of various features may be arbitrarily increased or decreased for clarity of discussion. [Figure 1A] FIG. 1 is a schematic diagram of a distributed acoustic sensor (DAS) system along a roadway according to some embodiments. [Figure 1B] FIG. 1 is a schematic diagram of a distributed acoustic sensor (DAS) system along a roadway according to some embodiments. [Figure 2] 1 is a flowchart of a method for analyzing data from distributed fiber optic sensing (DFOS) data according to some embodiments. [Figure 3] FIG. 1 is a schematic diagram of a DAS system with waterfall data collected by the DAS system according to some embodiments. [Figure 4] FIG. 1 is a diagram of filtered DFOS data according to some embodiments. [Figure 5] FIG. 1 is a plot of detected vehicle positions based on DFOS data according to some embodiments. [Figure 6] FIG. 1 is a plot of detected vehicle positions based on DFOS data according to some embodiments. [Figure 7] FIG. 1 is a plot of denoised vehicle positions based on DFOS data according to some embodiments. [Figure 8A] FIG. 1 is a diagram of a plot for determining vehicle parameters according to some embodiments. [Figure 8B] FIG. 1 is a diagram of a plot for determining vehicle parameters according to some embodiments. [Figure 9] 1 is a flowchart of a method for denoising data from distributed optical fiber sensing (DFOS) data according to some embodiments. [Figure 10] FIG. 1 is a block diagram of a system for analyzing DFOS data according to some embodiments. DETAILED DESCRIPTION OF THE INVENTION

[0009] The following disclosure provides many different embodiments or examples for implementing different features of the given subject matter. To simplify the disclosure, certain examples of components, values, steps, materials, arrangements, etc. are described below. Of course, these are merely examples and are not intended to be limiting. Other components, values, steps, materials, arrangements, etc. are contemplated. For example, in the following description, forming a first feature above or on a second feature may include embodiments in which the first and second features are formed in direct contact, as well as embodiments in which an additional feature may be formed between the first and second features such that the first and second features are not in direct contact. Furthermore, the present disclosure may repeat reference numerals and / or letters in various examples. This repetition is for simplicity and clarity and does not, in itself, dictate a relationship between the various embodiments and / or configurations discussed.

[0010] Additionally, spatially relative terms such as "beneath," "below," "lower," "above," and "upper" may be used herein for ease of description to describe the relationship of one element or feature to another, as shown in the figures. Spatially relative terms are intended to encompass different orientations of the device during use or operation in addition to the orientation shown in the figures. The device may be otherwise oriented (rotated 90 degrees or at other orientations) and the spatially relative descriptors used herein may be similarly interpreted accordingly.

[0011] Using data from optical fibers along roadways is useful for determining traffic volume, traffic speed, accidents, and other events along roadways. To enhance the usefulness of traffic information obtained based on optical fiber data, determining vehicle parameters such as speed, acceleration, and lane position provides information useful for identifying traffic patterns and navigation information. The quality of collected distributed optical fiber sensing (DFOS) data is determined by many factors. DFOS data is based, at least in part, on vibration, vehicle size, traffic volume, and roadway type, which all affect data quality. For example, a single vehicle traveling along a roadway directly above ground is likely to provide a higher-quality signal than a large truck traveling on a busy bridge. Denoising DFOS data helps analyze DFOS data for use in applications such as traffic analysis, autonomous driving, and navigation directions.

[0012] Additionally, the accuracy of traffic information aids in the navigation of vehicles traveling along roadways. By providing drivers with more accurate traffic data, navigation systems and / or navigation applications become more useful to the driver. Improved navigation accuracy is also useful to autonomous drivers or driver assistance features in vehicles. By accurately determining where traffic congestion or accidents have occurred, autonomous vehicles or driver assistance systems can guide vehicles along more efficient routes.

[0013] This application describes a method and system for performing denoising of DFOS data. The method involves identifying hit points, also called seeds, at various times and locations based on the DFOS data. The DFOS data is analyzed to identify the location of a vehicle at a specific time based on detected DFOS data vibrations. A hit point indicates the location of a vehicle at a specific time. In situations where the DFOS data is of high quality, a vehicle will generate a single hit point at a specific time. In contrast, in situations where the DFOS data is noisy, i.e., of low quality, there is a high risk that a vehicle will generate multiple hit points simultaneously. That is, the DFOS data may indicate that the same vehicle is in two different locations at the same time. By denoising the DFOS data, multiple hit points for a single vehicle can be removed or combined to provide usable data.

[0014] In addition to situations where a single vehicle simultaneously generates multiple hitpoints, DFOS data also occasionally contains erroneous non-vehicle data. For example, construction adjacent to the roadway, bridge vibrations caused by strong winds, or overpass vibrations caused by vehicles passing underneath increase the likelihood of hitpoints in DFOS data. Filtering out these erroneous hitpoints helps improve the accuracy of DFOS data so that it can be used for traffic monitoring, autonomous driving, navigation instructions, or other suitable applications.

[0015] 1A is a schematic diagram of a distributed acoustic sensor (DAS) system 100A along a roadway 130A according to some embodiments. The DAS system 100A includes a traffic monitoring device 111 in communication with a DAS 112. The DAS system 100A further includes an optical fiber 121 connected to the DAS 112. The optical fiber 121 runs along the roadway 130A. The roadway 130A includes three lanes. A large number of vehicles are on the roadway 130A. Some vehicles 140 on the roadway 130A are larger than other vehicles 150 on the roadway 130A. While the description refers to the optical fiber 121, one skilled in the art will understand that in some embodiments, the optical fiber 121 includes a multi-fiber bundle.

[0016] As vehicles 140 and 150 pass along roadway 130A, they generate vibrations. These vibrations change the way light propagates along optical fiber 121. DAS 112 is connected to optical fiber 121, sends optical signals onto optical fiber 121, and detects return light from optical fiber 121. The resulting data is called waterfall data. The waterfall data provides information about the number of vehicles, the direction of travel by the vehicles, the speed and lane location of the vehicles on roadway 130A.

[0017] 1 is on solid ground. The ground does not vibrate at a sufficiently high amplitude to obscure the detection of vehicles 140 and 150 traveling along roadway 130A. As a result, DAS 112 can accurately detect vehicles 140, 150 traveling on roadway 130A. In some embodiments, roadway 130A includes at least one bridge, such as roadway 130B in FIG. 1B.

[0018] Unlike solid ground, bridges exhibit different vibration characteristics, such as damping. The vibration characteristics of a bridge are affected by the bridge length, the bridge construction materials, wind, and other factors. These differences in the vibration characteristics of bridges can be used to determine where the bridge should be located along the optical fiber 121.

[0019] 1A also includes exemplary measured DAS data, which is provided to aid in understanding the waterfall data collected by DAS 112.

[0020] FIG. 1B is a schematic diagram of a DAS system 100B along a roadway 130B according to some embodiments. Similar to the DAS system 100A of FIG. 1A, the DAS system 100B includes a DAS 112 and an optical fiber 121. In contrast to FIG. 1A, the roadway 130B of FIG. 1B includes a first bridge 160A and a second bridge 160B. Additionally, FIG. 1B includes an additional fiber portion 170.

[0021] Considering the first vehicle 150A and the second vehicle 150B helps understand the use of fixed reference points, such as the first bridge 160A and the second bridge 160B, in accurately determining the location corresponding to the traffic information. The distance between the first vehicle 150A and the second vehicle 150B along the roadway 130B is significantly different from the length of the optical fiber 121 between the positions of the first vehicle 150A and the second vehicle 150B. This difference is due to the presence of the additional fiber portion 170 and the fact that the optical fiber 121 is not exactly parallel to the roadway 130B. Determining the position of the first bridge 160A along the optical fiber 121 helps determine the precise location of the first vehicle 150A along the roadway 130B. The position of the first bridge 160A along the roadway 130B is known based on publicly available geographic data. By determining the position of the first bridge 160A relative to the optical fiber 121, the length of the optical fiber 121 from the DAS 112 to the end of the first bridge 160A closest to the first vehicle 150A is determined based on the waterfall data. Next, the length of the optical fiber 121 from the end of the first bridge 160A to the first vehicle 150A is determined based on the waterfall data. By limiting the distance of the optical fiber 121 from the fixed reference point on the first bridge 160A to the first vehicle 150A, errors in the length of the roadway 130B from the end of the first bridge 160A to the DAS 112 are eliminated from the position determination. As a result, the position of the first vehicle 150A along the roadway 130B can be more accurately determined using the fixed reference point on the first bridge 160A.

[0022] Similarly, the position of the second vehicle 150B is more accurately determined by using a fixed reference point on the second bridge 160B. Waterfall data from the DAS 112 can be used to determine the length of optical fiber 121 between the DAS 112 and the end of the second bridge 160B closest to the second vehicle 150B. Then, only the length of optical fiber 121 between the second vehicle 150B and the second bridge 160B is used to determine the position of the second vehicle 150B along the roadway 130B. By using this short optical fiber 121, the length of optical fiber 121 between the second vehicle 150B and the DAS 112, including the additional fiber portion 170, is excluded from the position determination. As a result, the position of the second vehicle 150B is more accurately determined by using a fixed reference point on the second bridge 160B.

[0023] 2 is a flowchart of a method for analyzing data from distributed optical fiber sensing (DFOS) data according to some embodiments. Method 200 can be used with DAS system 100A (FIG. 1A), DAS system 100B (FIG. 1B), or another suitable system that provides DFOS data.

[0024] In operation 205, raw DFOS data is received. In some embodiments, the raw DFOS data includes waterfall data, such as waterfall data detected by DAS 112 (FIG. 1). The waterfall data includes information about the time and location along the optical fiber at which the vibration data was detected. Further details of the waterfall data are described with respect to FIG. 3 below. In some embodiments, the waterfall data includes data along a roadway that includes both solid ground and at least one bridge.

[0025] In operation 210, the raw DFOS data is localized using structural information. The structural information includes information obtained from external sources to fix reference points along the roadway. For example, in some embodiments, the structural information includes the location of bridges, e.g., bridge 160A (FIG. 1B), the location of additional fibers, e.g., additional fiber 170 (FIG. 1B), or other suitable reference points along the roadway. Localizing the DFOS data using structural information helps improve the accuracy and precision of detected vehicle locations and accurate determination of vehicle parameters such as speed and acceleration. Localizing the DFOS data also helps improve the accuracy and precision of navigation instructions, autonomous driving functions, traffic monitoring, or other suitable applications of DFOS data.

[0026] In operation 215, the raw DFOS data is preprocessed to enhance the received data. Preprocessing the data includes normalizing the vibration amplitude of the data at each location along the roadway, e.g., roadway 130B (FIG. 1B), over a predetermined time period. Normalizing the vibration amplitude helps account for variations in optical fiber sensitivity. Variations in optical fiber sensitivity result from several causes, including, but not limited to, uneven roadway surfaces, inconsistent optical fiber installation, and optical fiber misalignment. Normalizing the vibration amplitude also helps account for variations in traffic volume. For example, as the number of vehicles on the roadway increases, the magnitude of vibrations detected by the DAS increases. Normalizing the vibration amplitude based on a predetermined duration reduces the impact of large vibrations detected during high traffic conditions on periods of low traffic conditions, producing more accurate data for estimating traffic flow characteristics.

[0027] In some embodiments, preprocessing the data also limits the maximum vibration amplitude at each location along the optical fiber over a predetermined period of time. Limiting the maximum vibration amplitude helps prevent vibrations from large vehicles, such as trucks or construction vehicles, from obscuring vibrations generated by smaller vehicles, such as passenger cars.

[0028] The preprocessed data is then filtered by a bandpass filter based on the estimated frequency range from operation 210. Filtering the data excludes portions of the roadway from the data that do not exhibit the vibration damping characteristics of the bridge. Figure 4 below provides an example of filtered DFOS data.

[0029] In operation 220, initial hit points are identified for each vehicle. Hit points are initially identified, for example, at time (t)=0. In some embodiments, the first time is the time when the vehicle first enters the detection area of ​​a DAS system, such as DAS system 100A (FIG. 1A), DAS system 100B (FIG. 1B), or another suitable DAS system. In some embodiments, the first time is a time selected by a user of the DAS system. Hit points are identified based on vibrations detected in the DFOS data. The detected vibrations are determined based on the width of a line in the DFOS data. In some cases, vibrations occur more widely, indicating higher vibration intensity. A wider line increases the risk of multiple hit points being identified for the same vehicle. Furthermore, in some embodiments, additional vibrations not caused by vehicles traveling along the roadway generate vibrations detected in the DFOS data. These additional vibrations can result in false hit points being identified where no vehicles are present.

[0030] In some embodiments, a trained neural network (NN) is utilized to identify hit points in the DFOS data. In some embodiments, hit points are identified or verified by a user of the DAS system. Identifying initial hit points assists in tracking a vehicle through at least a portion of the DAS system's detection area to determine vehicle parameters for use in automated driving, navigation instructions, traffic monitoring, or other suitable applications. Additional details regarding detecting initial hit points, according to some embodiments, are provided with respect to FIG. 5 below.

[0031] In operation 225, subsequent hit points are identified for each vehicle. The subsequent hit points are vehicle positions determined using DFOS data at a first time, e.g., t=1, t=2, and other later times. In some embodiments, the interval between the first time and the subsequent times is uniform across each time. In some embodiments, there are irregular intervals between the first time and the subsequent times. In some embodiments, the intervals are predetermined. In some embodiments, the intervals are set based on the speed limit along the roadway. In some embodiments, the intervals are set based on the expected traffic volume along the roadway, e.g., shorter intervals for areas with higher traffic congestion. In some embodiments, subsequent hit points are identified in the same manner as initial hit points. In some embodiments, in addition to detecting oscillations based on DFOS data, subsequent hit points are identified based on a number of initial hit points that can be used to predict the number of vehicles tracked through the detection area of ​​the DAS system. In some embodiments, in addition to detecting oscillations based on DFOS data, subsequent hit points are identified based on the expected position of the vehicle. In some embodiments, the trained NN can be used to identify subsequent hit points. Further details regarding initial hit point detection, according to some embodiments, are provided with respect to FIG. 6 below.

[0032] In operation 230, iterative denoising is applied to the hit points. The iterative denoising is applied to the initial and subsequent hit points of each vehicle detected using DFOS data. The iterative denoising attempts to remove hit points that are erroneous due to vibrations generated by sources other than the vehicle or due to multiple hit points generated simultaneously for the same vehicle.

[0033] The iterative denoising involves classifying clusters of hit points as either good or bad clusters. As previously mentioned, subsequent hit point identification in operation 225 can be performed based in part on the number of initial hit points. Thus, based on the initial number of hit points and the expected movement of the vehicle, hit points can be grouped into clusters for classification. Each cluster represents a vehicle traveling through the detection area of ​​the DAS system. In some embodiments, the clustering of hit points is performed using a trained neural network. In some embodiments, the clustering is performed using K-means clustering, density-based spatial clustering for applications with noise (DBSCAN) clustering, Gaussian mixture model clustering, balanced iterative reduction and clustering using hierarchies (BIRCH) clustering, affinity propagation clustering, mean-shift clustering, ordered points to identify clustering structures (OPTICS) clustering, agglomerative hierarchical clustering, or another suitable type of clustering.

[0034] In some embodiments, a good cluster is a set of hit points associated with the same vehicle with a single hit point for each time. In some embodiments, a bad cluster is a set of hit points associated with the same vehicle with at least one hit point or no hit points. In some embodiments, thresholding is used to classify clusters as either good or bad clusters. In some embodiments, thresholding is performed based on the number of detection times. For example, in a situation where there are six detection times, e.g., t=0 to t=5, thresholding is applied based on a selected percentage of vehicle hit points across the detection times. That is, in some embodiments, the threshold is the selected percentage multiplied by the number of detection times. For example, a selected percentage of 75% and a number of detection times of 6 results in a threshold of 4.5, i.e., (0.75)(6). Using this threshold, any cluster with 4.5 or more hit points is considered a good cluster, and any cluster with fewer than 4.5 hit points is considered a bad cluster. In some embodiments, the selected percentage is a predetermined value set by a user or based on empirical data. In some embodiments, the selected percentage is adjusted based on noise in the DFOS data. For example, during construction conditions or on windy days, the selected percentage is reduced to account for increased noise present in the DFOS data. In some embodiments, the selected percentage is determined based on a trained neural network. In some embodiments, the selected percentage ranges from about 65% to about 85%. If the selected percentage is too small, clusters containing noisy data are more likely to be included in the vehicle parameter estimation of operation 240, potentially reducing the reliability of the data utilized in operation 245. If the selected percentage is too large, the denoising will exclude too much data, and the data utilized in operation 245 will be less robust, potentially reducing the impact of method 200.

[0035] Once the cluster classification is complete, the good clusters are considered reliable indicators of the location of a single vehicle at various times within the DAS system's detection area. To improve the completeness of the analysis and maximize the value of the DFOS data, we analyze the bad clusters and adjust the hit points, attempting to aggregate one or more bad clusters into a good cluster.

[0036] For each bad cluster, an initial noisy hit point is identified. In some embodiments, the initial noisy hit point is identified as the hit point at the first time, or the earliest time, at which a cluster is detected. The initial noisy hit point is then removed from consideration, and clustering and classification are repeated. During clustering and classification, the number of detection times is reduced by one because an initial noisy hit point was removed for each bad cluster. In some circumstances, the result of the iterative clustering and classification results in the combination of one or more bad clusters into a good cluster. That is, because the noisy hit points caused the clustering algorithm to detect more vehicles than actually existed along the roadway, once the noisy hit points are removed, the clustering properly identifies the correct number of vehicles traveling along the roadway.

[0037] The initial noisy hit point removal, clustering, and classification are repeated iteratively until all remaining clusters are deemed good clusters. With each iteration of the process, additional detected hit points for each bad cluster are removed with each iteration, further reducing the detection time by 1. Further details regarding initial hit point detection, according to some embodiments, are provided below with respect to Figures 6 and 7.

[0038] In operation 235, the denoised individual vehicles are tracked. Individual vehicle tracking is performed using the good clusters remaining after operation 230. Each good cluster corresponds to the movement of a single vehicle through the detection area of ​​the DAS system. Individual vehicle tracking determines the position of the vehicle at each detection time.

[0039] At operation 240, individual vehicle parameters are estimated. Estimates of the vehicle parameters are determined by tracking individual vehicles within the detection area of ​​the DAS system. In some embodiments, the vehicle parameters include speed, acceleration, lane position, or another suitable parameter. In some embodiments, the vehicle parameters are estimated by generating a regression line of the hit points in the cluster. In some embodiments, the regression line is generated using linear regression, polynomial regression, Lasso regression, or another suitable regression model. The slope of the regression line can estimate the vehicle's speed. A flatter slope indicates a higher speed than a more vertical slope. Based on a change in the slope of the regression line, the vehicle's acceleration can be estimated. Vehicle stall can be estimated based on a failure to detect additional hit points at expected locations over one or more detection time intervals. Vehicle lane changes can be estimated by a shift in the regression line along the x-axis, which indicates whether the vehicle is closer to or farther from the optical fiber. As one skilled in the art will recognize, the above description of vehicle parameters and methods for estimating the vehicle parameters is not exhaustive and does not limit the scope of this description. Those skilled in the art may determine additional vehicle parameters or utilize other options for determining vehicle parameters based on individual tracked vehicles. Further details regarding initial hit point detection, according to some embodiments, are provided below with respect to Figures 8A and 8B.

[0040] In operation 245, the vehicle parameters are reported for use in one or more applications. In some embodiments, the applications include navigation instructions, autonomous driving, traffic monitoring, or other suitable applications. In some embodiments, the vehicle parameters are reported by transmitting the vehicle parameters to an external device, such as a server, a vehicle, a mobile device, or another suitable system. In some embodiments, the report is transmitted over a wired connection. In some embodiments, the report is transmitted wirelessly, for example, using a cellular network, a local area network, or another suitable network.

[0041] In some embodiments, method 200 includes additional operations. For example, in some embodiments, method 200 includes generating instructions for controlling the autonomous vehicle based on the estimated vehicle parameters. In some embodiments, at least one operation of method 200 is omitted. For example, in some embodiments, operation 235 is omitted, and vehicle tracking and vehicle parameter estimation are performed simultaneously. In some embodiments, the order of operations of method 200 is adjusted. For example, in some embodiments, operation 215 is performed before operation 210.

[0042] Using method 200, DFOS data can be used to determine vehicle performance along roadways with increased accuracy and precision compared to other approaches that cannot iteratively cluster and classify hit points. Improved determination of vehicle performance helps improve the accuracy of traffic monitoring, navigation instructions, autonomous driving instructions, and other applications.

[0043] 3 is a schematic diagram of a DAS system 100B along with waterfall data 300 collected by the DAS system according to some embodiments. The DAS system 100B is the same as the DAS system 100B of FIG. 1B. As in FIG. 1B, the roadway (not shown) in FIG. 3 includes two bridges, as indicated by the waterfall data 300. The waterfall data 300 is preprocessed waterfall data.

[0044] Waterfall data 300 includes regions 302, 304, 306, 308, and 310. Regions 302, 306, and 310 contain discernible lines indicative of vibrations generated by vehicles crossing the roadway. Regions 304 and 308 represent bridges. Compared to regions 302, 306, and 310, regions 304 and 308 do not contain discernible lines because the damped vibrations of the bridge obscure the detected vibrations of vehicles crossing the bridge.

[0045] 4 is a diagram of filtered DFOS data 400 according to some embodiments. The filtered DFOS data 400 includes a region 410 that exhibits higher vibration intensity within an estimated frequency range. Region 410 is likely to be a bridge along the roadway. The filtered DFOS data 400 also includes a region 420 that exhibits lower vibration intensity within an estimated frequency range. Region 420 is likely to be a non-bridge portion of the roadway.

[0046] 5 is an illustration of a plot 500 of detected vehicle positions based on DFOS data, according to some embodiments. In some embodiments, plot 500 is generated during operation 220 (FIG. 2). In some embodiments, plot 500 is generated by another method. In some embodiments, plot 500 is generated by system 1000 (FIG. 10), a trained NN, or another suitable system.

[0047] Plot 500 includes a first vibration line 510 and a second vibration line 520. Plot 500 further includes an initial time t=0, also referred to as the first time. Plot 500 further includes multiple hit points at time t=0. Hit point 530 is located on neither first vibration line 510 nor second vibration line 520. Hit point 540 is at the intersection of a first side of first vibration line 510 and the t=0 line. Hit point 550 is at the intersection of a second side of first vibration line 510 and the t=0 line. Hit point 560 is at the intersection of second vibration line 520 and the t=0 line.

[0048] The first vibration line 510 is thicker than the second vibration line 520. The increased thickness of the first vibration line 510 relative to the second vibration line 520 indicates that the vehicle generating the first vibration line 510 has a higher vibration amplitude than the vehicle generating the second vibration line 520. Potential reasons for the increased thickness include heavier vehicles, longer vehicles, and differences between different lanes on a roadway. As one skilled in the art will recognize, these potential reasons are not intended to be exhaustive or limiting of this description, but merely to provide a context for understanding this description. The increased thickness of the first vibration line 510 results in two hit points 540, 550 associated with this line. According to some embodiments, one of these hit points 540 or 550 is likely a false or noisy hit point that should be removed in a noise removal process, as previously described and discussed below with respect to FIG. 7.

[0049] Hit point 530 is not associated with a vibration line. Therefore, hit point 530 is likely a noisy hit point. Potential reasons for hit point 530 include construction near the roadway, high winds, or erroneous data. As one skilled in the art will recognize, these potential reasons do not make this description exhaustive or limiting, but merely provide a context for understanding this description.

[0050] Hit point 540 is on a first side of first vibration line 510, and hit point 550 is on a second side of first vibration line 510. The thickness of first vibration line 510 causes a system, such as system 1000 (FIG. 10), a trained NN, or another suitable system, to identify two different hit points associated with first vibration line 510.

[0051] The hit point 550 is on the second vibration line 520. The second vibration line 520 is thin enough that only a single hit point occurred in the system based on the second vibration line 520.

[0052] 6 is an illustration of a plot 600 of detected vehicle positions based on DFOS data, according to some embodiments. In some embodiments, plot 500 is generated during operations 225 and 230 (FIG. 2). In some embodiments, plot 500 is generated by another method. In some embodiments, plot 500 is generated by system 1000 (FIG. 10), a trained NN, or another suitable system.

[0053] In comparison to plot 500 (FIG. 5), plot 600 includes hit points at multiple times, i.e., t=0 through t=5. Plot 600 includes multiple clusters of hit points. First cluster 610 includes initial hit point 615 at time t=0. Second cluster 620 includes initial hit point 625 at time t=0. Third cluster 630 includes initial hit point 635 at time t=3. Fourth cluster 640 includes initial hit point 645 at time t=0. Each of clusters 610, 620, 630, and 640 is a group of hit points clustered together using a clustering algorithm such as those described above. The following description is based on an example implementation of operation 230 (FIG. 2) that includes a selected percentage of 75%. Those skilled in the art will understand that these selected percentages are merely exemplary and that other selected percentages are within the scope of this description. Furthermore, as previously mentioned, this discussion is not limited to the use of thresholds in classifying clusters as good and bad clusters.

[0054] The first cluster 610 contains three hit points, including the initial hit point 615. There are six detection times, and the selected percentage in this example is set to 75%, so the hit point threshold for a good cluster is 4.5. Because the hit points of the first cluster 610 are less than 4.5, the first cluster 610 is classified as a bad cluster.

[0055] The second cluster 620 includes four hit points, including the initial hit point 625. As noted above, the hit point threshold for a good cluster is 4.5. Because the hit points of the second cluster 620 are less than 4.5, the second cluster 620 is classified as a bad cluster.

[0056] The third cluster 630 includes three hit points, including the initial hit point 635. As noted above, the hit point threshold for a good cluster is 4.5. Because the hit points of the third cluster 630 are less than 4.5, the third cluster 630 is classified as a bad cluster.

[0057] The fourth cluster 640 includes six hit points, including the initial hit points 645. As noted above, the hit point threshold for a good cluster is 4.5. Because the fourth cluster 640 has more than 4.5 hit points, the fourth cluster 640 is classified as a good cluster.

[0058] Based on the initial classification of the clusters, the fourth cluster 640 is deemed to be a good cluster and analysis of the fourth cluster for denoising purposes ends. The remaining clusters 610, 620 and 630 continue the iterative denoising process of initial hit point removal, clustering and classification.

[0059] In the next iteration, initial hit point 615 is removed from first cluster 610, initial hit point 625 is removed from second cluster 620, and initial hit point 635 is removed from third cluster 630. The remaining hit points are then subjected to another clustering process using the clustering algorithm described above. As a result of the removal of the initial hit point, first cluster 610 is reduced to a cluster containing two hit points, i.e., those at times t=1 and t=2. However, upon removal of initial hit point 635, the remaining hit points in third cluster 630 are clustered with the hit points in second cluster 620 other than the removed initial hit point 625. As a result, after this clustering iteration, only two clusters remain: the reduced first cluster 610 and the combined cluster of the remaining hit points in second and third clusters 620 and 630. Classification of the remaining clusters involves comparing the hit points in each cluster with a threshold of 4 as the number of detection times is reduced from 6 to 5. The reduced number of detection times is multiplied by the selected 75% percentage. The reduced first cluster has two hit points, which is less than four, and continues to be classified as a bad cluster. However, the combined cluster of the remaining hit points of the second and third clusters has five hit points, which is greater than four. Therefore, the combined cluster of the remaining hit points of the second and third clusters is classified as a single good cluster. Noise removal now stops with respect to the combined cluster of the remaining hit points of the second and third clusters.

[0060] The reduced first cluster undergoes another iteration of initial hit point removal, clustering, and classification. As a result, the single remaining hit point in the first cluster is less than the threshold number of 3, which is 75% of the four detection times in this iteration. Again, the further reduced first cluster is still classified as a bad cluster. During the next iteration, all hit points from the first cluster 610 are removed. This causes the first cluster 610 to be discarded entirely.

[0061] Returning to the combined cluster of the remaining hit points of the second and third clusters, the good cluster resulting from the denoising process includes all hit points of the second cluster 620, including initial hit point 625, and all hit points of the third cluster 630, excluding initial hit point 635. The reason for including initial hit point 625 instead of initial hit point 635 is because each good cluster must have a single hit point at each detection time. Initial hit point 635 is at the same detection time as a hit point in second cluster 620. Therefore, initial hit point 635 is discarded from the good cluster resulting from the iterative denoising process.

[0062] 7 is an illustration of a plot 700 of denoised vehicle positions based on DFOS data, according to some embodiments. In some embodiments, plot 700 is generated during operation 230 (FIG. 2). In some embodiments, plot 700 is generated by another method. In some embodiments, plot 700 is generated by system 1000 (FIG. 10), a trained NN, or another suitable system. In some embodiments, plot 700 is a plot resulting from the iterative denoising process described above with respect to FIG. 6.

[0063] Similar to plot 600 (FIG. 6), plot 700 includes six detection times. Because plot 700 is obtained after an iterative denoising process, plot 700 includes only good clusters. Plot 700 includes a first good cluster 720, which corresponds to the combined cluster of second cluster 620 and third cluster 630 described in the description of plot 600 (FIG. 6). First good cluster 720 includes an initial hit point 725, which corresponds to initial hit point 625 (FIG. 6). Plot 700 includes a second good cluster 740, which corresponds to fourth cluster 640 described in the description of plot 600 (FIG. 6). Second good cluster 740 includes an initial hit point 745, which corresponds to initial hit point 645 (FIG. 6).

[0064] FIG. 8A is an illustration of a plot 800A for determining vehicle parameters according to some embodiments. Plot 800A is the result of an iterative denoising process of DFOS data. In some embodiments, plot 800A is generated during operation 240 (FIG. 2). In some embodiments, plot 800A is generated during another method. In some embodiments, plot 800A is generated by system 1000 (FIG. 1), a trained NN, or another suitable system. In some embodiments, plot 800A is the result of continuous tracking of vehicles associated with good clusters in plot 700 (FIG. 7).

[0065] Plot 800A includes a first regression line 810 that extends through the hit points 815 of a first cluster. First regression line 810 has a first slope 820. Plot 800A includes a second regression line 830 that extends through the hit points 835 of a second cluster. Second regression line 830 has a second slope 840.

[0066] As noted above in FIG. 4, the DFOS data has distance along the x-axis of the plot. Thus, the slope of the regression line can be used to determine the speed of the vehicles associated with each cluster. The first slope 820 can be used to determine a first speed of the vehicles associated with the first cluster. The second slope 840 can be used to determine a second speed of the vehicles associated with the second cluster. The second slope 640 is slightly steeper than the first slope 620, indicating that the second vehicle is traveling at a slower average speed than the first vehicle.

[0067] While plot 800A provides an example of determining speed as a vehicle parameter, those skilled in the art will appreciate that the present application is not limited to determining speed. Those skilled in the art will appreciate that changes in the slope of the regression line within the same cluster can be used to determine the corresponding change in speed, i.e., acceleration, of the vehicle. Similarly, a sudden shift along the x-axis indicates a sudden change in the distance between the vehicle and the detector, potentially indicating a lane change by the vehicle.

[0068] 8B is an illustration of a plot for determining vehicle parameters according to some embodiments. Plot 800B is the result of an iterative denoising process of DFOS data. In some embodiments, plot 800B is generated during operation 240 (FIG. 2). In some embodiments, plot 800B is generated during another method. In some embodiments, plot 800B is generated by system 1000 (FIG. 1), a trained NN, or another suitable system. In some embodiments, plot 800B is the result of continuous tracking of vehicles associated with good clusters in plot 700 (FIG. 7).

[0069] Plot 800B includes a first regression line 810 that extends through hit points 815 of the first cluster. Plot 800B includes a second regression line 830 that extends through hit points 835 of the second cluster. In contrast to plot 800A (FIG. 8A), plot 800B does not include hit points after hit point 835′. Failure to detect a hit point at expected location 850 indicates that an anomaly has occurred. In some cases, failure to detect a hit point after hit point 835′ indicates that the vehicle has stopped and is no longer moving along the roadway. For example, the vehicle is stuck in traffic. In some cases, failure to detect a hit point after hit point 835′ indicates that the vehicle has exited the roadway, for example, using an exit ramp. In some embodiments, a comparison between structural features of the roadway, such as in operation 210 (FIG. 2), can be used to determine whether an exit ramp is located near hit point 835′. The presence of an exit ramp near hit point 835′ suggests that the vehicle has exited the roadway. In contrast, the absence of an exit ramp near hit point 835' suggests that the vehicle was stopped along the roadway. Utilizing a combination of roadway structural information in the DFOS data combination allows a system such as system 1000 (FIG. 10), a trained NN, or another suitable system to accurately determine the cause of the anomaly.

[0070] 9 is a flowchart of a method 900 for denoising data from distributed optical fiber sensing (DFOS) data according to some embodiments. Method 200 can be used with DAS system 100A (FIG. 1A), DAS system 100B (FIG. 1B), or another suitable system that provides DFOS data. Some of the operations of method 900 are similar to those of method 200 (FIG. 2) and will not be described in detail for the sake of brevity.

[0071] Raw DFOS data is acquired in operation 905. In some embodiments, operation 905 is similar to operation 205 (FIG. 2).

[0072] Hit points are identified in the DFOS data in operation 910. In some embodiments, operation 910 is similar to operations 220 and 225 (FIG. 2).

[0073] The hit points are clustered in operation 915. In some embodiments, the hit points are clustered using K-means clustering, with the number of initial hit points determined in operation 910 being used as the value of K. In some embodiments, the clusters are performed using an algorithm other than K-means clustering, such as the clustering algorithms described above.

[0074] In operation 920, the clusters are classified. The clusters are classified into good clusters and bad clusters. In some embodiments, the classification of the clusters is similar to the classification described above with respect to operation 230 (FIG. 2).

[0075] At operation 925, a determination is made as to whether only good clusters remain. In response to a determination that only good clusters remain, method 900 proceeds to operation 935. In response to a determination that at least one bad cluster remains, method 900 proceeds to operation 930.

[0076] At operation 930, hit points are removed from each cluster. In the first iteration of the loop of operations 910, 915, 920, and 925, the initial hit points of each cluster are removed. In subsequent iterations of the loop of operations 910, 915, 920, and 925, the earliest remaining hit points are removed from each cluster. In some embodiments, the removal of hit points is similar to the removal described above with respect to operation 230 (FIG. 2) or with respect to plot 600 (FIG. 6).

[0077] The loop of operations 910, 915, 920, 925 and 930 is repeated until only good clusters remain, or the loop of operations 910, 915, 920, 925 and 930 is repeated until all hit points are removed.

[0078] Vehicles associated with good clusters are tracked in operation 935. In some embodiments, operation 935 is similar to operation 235 (FIG. 2).

[0079] Vehicle parameters are estimated in operation 940. In some embodiments, operation 940 is similar to operation 240 (FIG. 2).

[0080] The vehicle parameters are reported for use by the application in operation 945. In some embodiments, operation 945 is similar to operation 245 (FIG. 2).

[0081] In some embodiments, method 900 includes additional operations. For example, in some embodiments, method 900 includes generating instructions for controlling the autonomous vehicle based on the estimated vehicle parameters. In some embodiments, at least one operation of method 900 is omitted. For example, in some embodiments, operation 935 is omitted, and vehicle tracking and vehicle parameter estimation are performed simultaneously. In some embodiments, the order of operations of method 900 is adjusted. For example, in some embodiments, operation 945 is performed simultaneously with operation 940.

[0082] Using method 900, DFOS data can be used to determine vehicle performance along roadways with increased accuracy and precision compared to other approaches that cannot iteratively cluster and classify hit points. Improved determination of vehicle performance helps improve the accuracy of traffic monitoring, navigation instructions, automated driving instructions, and other applications.

[0083] 10 is a block diagram of a system 1000 for analyzing DFOS data according to some embodiments. The system 1000 includes a hardware processor 1002 and a non-transitory computer-readable storage medium 1004 encoded with, or storing, computer program code 1006, i.e., a set of executable instructions. The computer-readable storage medium 1004 is also encoded with instructions 1007 for interfacing with external devices. The processor 1002 is electrically coupled to the computer-readable storage medium 1004 via a bus 1008. The bus 1008 also electrically couples the processor 1002 to an I / O interface 1010. A network interface 1012 is also electrically connected to the processor 1002 via the bus 1008. The network interface 1012 is connected to a network 1014, thereby allowing the processor 1002 and the computer-readable storage medium 1004 to connect to external elements via the network 1014. Processor 1002 is configured to execute computer program code 1006 encoded on computer-readable storage medium 1004 to enable system 100 to perform some or all of the operations described with respect to DAS system 100A (FIG. 1A), DAS system 100B (FIG. 1B), method 200 (FIG. 2), method 900 (FIG. 9), or another suitable system for analyzing DFOS data.

[0084] In some embodiments, the processor 1002 is a central processing unit (CPU), a multiprocessor, a distributed processing system, an application specific integrated circuit (ASIC), and / or other suitable processing unit.

[0085] In some embodiments, computer-readable storage medium 1004 is an electronic, magnetic, optical, electromagnetic, infrared, and / or semiconductor system (or apparatus or device). For example, computer-readable storage medium 1004 includes semiconductor or solid-state memory, magnetic tape, a removable computer diskette, a random access memory (RAM), a read-only memory (ROM), a rigid magnetic disk, and / or an optical disk. In some embodiments using an optical disk, computer-readable storage medium 1004 includes a compact disk read-only memory (CD-ROM), a compact disk read / write (CD-R / W), and / or a digital video disk (DVD).

[0086] In some embodiments, storage medium 1004 stores computer program code 1006 configured to cause system 1000 to perform some or all of the operations described with respect to DAS system 100A (FIG. 1A), DAS system 100B (FIG. 1B), method 200 (FIG. 2), method 900 (FIG. 9), or another suitable system for analyzing DFOS data. In some embodiments, storage medium 1004 also stores information necessary to perform some or all of the operations described with respect to DAS system 100A ( FIG. 1A ), DAS system 100B ( FIG. 1B ), method 200 ( FIG. 2 ), method 900 ( FIG. 9 ), or another suitable system for analyzing DFOS data, as well as information generated while performing some or all of the operations described with respect to DAS system 100A ( FIG. 1A ), DAS system 100B ( FIG. 1B ), method 200 ( FIG. 2 ), method 900 ( FIG. 9 ), or another suitable system for analyzing DFOS data, such as sensor data parameters 1016, threshold parameters 1018, vehicle parameters 1020, hit point parameters 1022, and / or sets of executable instructions for performing some or all of the operations described with respect to DAS system 100A ( FIG. 1A ), DAS system 100B ( FIG. 1B ), method 200 ( FIG. 2 ), method 900 ( FIG. 9 ), or another suitable system for analyzing DFOS data.

[0087] In some embodiments, storage medium 1004 stores instructions 1007 for interfacing with an external device. Instructions 1007 enable processor 1002 to generate instructions readable by an external device to effectively perform some or all of the operations described with respect to DAS system 100A (FIG. 1A), DAS system 100B (FIG. 1B), method 200 (FIG. 2), method 900 (FIG. 9), or another suitable system for analyzing DFOS data.

[0088] System 1000 includes an I / O interface 1010. I / O interface 1010 is coupled to external circuitry. In some embodiments, I / O interface 1010 includes a keyboard, keypad, mouse, trackball, and / or cursor direction keys for communicating information and commands to processor 1002.

[0089] System 1000 also includes a network interface 1012 coupled to processor 1002. Network interface 1012 enables system 1000 to communicate with a network 1014 to which one or more other computer systems are connected. Network interface 1012 may include a wireless network interface, such as BLUETOOTH, WIFI, WIMAX, GPRS, or WCDMA, or a wired network interface, such as ETHERNET, USB, or IEEE-1394. In some embodiments, some or all of the operations described with respect to DAS system 100A (FIG. 1A), DAS system 100B (FIG. 1B), method 200 (FIG. 2), method 900 (FIG. 9), or another suitable system for analyzing DFOS data are performed by two or more systems 1000, and information such as sensor data, bridge locations, frequency ranges, and redundant fiber sections is exchanged between the different systems 1000 via network 1014.

[0090] The system 1000 is configured to receive information related to DFOS data via an I / O interface 1010 or a network interface 1012. The DFOS data is transferred to the processor 1002 via a bus 1008 for frequency range estimation and pre-processing and / or filtering. The frequency range is stored in a computer-readable medium 1004 as a frequency range parameter 1020. In some embodiments, the estimated frequency range parameter 1020 is received via the I / O 1010 or the network interface 1012. The pre-processed DFOS data is then stored in the computer-readable medium 1004 as a sensor data parameter 1016. The processor 1002 retrieves the sensor data parameter 1016 from the computer-readable medium 1004 and retrieves the hit points of the hit point parameter 1022. The processor 1002 performs iterative denoising of the hit point parameter 1022, such as using a threshold parameter 1018. Processor 1002 performs an analysis of the good clusters remaining after the iterative denoising to determine vehicle parameters of vehicle parameters 1020. In some embodiments, system 1000 is usable to implement a trained NN that can be used to effectively perform some or all of the operations described with respect to DAS system 100A (FIG. 1A), DAS system 100B (FIG. 1B), method 200 (FIG. 2), method 900 (FIG. 9), or another suitable process for analyzing DFOS data.

[0091] Some or all of the above embodiments can be described as, but are not limited to, the following supplementary notes.

[0092] (Appendix 1) receiving distributed optical fiber sensing (DFOS) data; identifying hit points in the DFOS data, each hit point corresponding to a corresponding vehicle position at a detection time; clustering the identified hit points to define one or more clusters; classifying each of the one or more clusters into a first classification or a second classification based on the number of hit points and the number of detection times in each of the one or more clusters; estimating vehicle parameters of a first vehicle corresponding to the hit points of a first cluster of the one or more clusters based on the hit points of the first cluster having the first classification; A vehicle tracking method comprising:

[0093] (Appendix 2) removing the earliest detected hit point of each of the one or more clusters having the second classification; repeating the clustering, classifying, and removing iteratively until only clusters of the one or more clusters having the first classification remain; 2. The vehicle tracking method of claim 1, further comprising:

[0094] (Appendix 3) 3. The vehicle tracking method of claim 2, further comprising reducing the detection time quantity by one in response to removing the earliest detected hit point.

[0095] (Appendix 4) 2. The vehicle tracking method of claim 1, wherein estimating the vehicle parameters includes estimating at least one of speed, acceleration, or lane changes.

[0096] (Appendix 5) The classification may include: determining a threshold value based on the quantity of detection times; comparing, for each of the one or more hit points, a quantity of the hit points in a corresponding one of the one or more clusters; classifying the corresponding cluster as the first classification in response to determining that the quantity of hit points in the corresponding cluster is greater than or equal to the threshold; classifying the corresponding cluster as the second classification in response to determining that the quantity of hit points in the corresponding cluster is less than the threshold; 2. The vehicle tracking method of claim 1, comprising:

[0097] (Appendix 6) 6. The vehicle tracking method of claim 5, wherein determining the threshold value includes multiplying the quantity of detection times by a selected percentage.

[0098] (Appendix 7) 2. The vehicle tracking method of claim 1, wherein the vehicle parameters include speed, acceleration, or lane changes.

[0099] (Appendix 8) 10. The vehicle tracking method of claim 1, further comprising transmitting the estimated vehicle parameters to an external device.

[0100] (Appendix 9) a computer-readable medium configured to store instructions; a processor coupled to the computer-readable medium; wherein the processor: receiving distributed optical fiber sensing (DFOS) data; identifying hit points in the DFOS data, each hit point corresponding to a corresponding vehicle position at a detection time; clustering the identified hit points to define one or more clusters; classifying each of the one or more clusters into a first classification or a second classification based on the number of hit points and the number of detection times in each of the one or more clusters; and estimating a vehicle parameter of a first vehicle corresponding to the hit point of the first cluster of the one or more clusters based on the hit point of the first cluster having the first classification. a vehicle tracking system configured to execute the instructions for:

[0101] (Appendix 10) The processor: removing the earliest detected hit point for each of the one or more clusters having the second classification; repeating the clustering, classifying, and removing iteratively until only clusters of the one or more clusters having the first classification remain. 10. The vehicle tracking system of claim 9, further configured to execute the instructions for:

[0102] (Appendix 11) 11. The vehicle tracking system of claim 10, wherein the processor is further configured to execute the instructions to decrease the detection time quantity by one in response to removing the earliest detected hit point.

[0103] (Appendix 12) 10. The vehicle tracking system of claim 9, wherein the processor is configured to execute the instructions to estimate the vehicle parameters by estimating at least one of speed, acceleration, or lane changes.

[0104] (Appendix 13) The processor: determining a threshold value based on the quantity of detection times; comparing, for each of the one or more hit points, the quantity of the hit points in a corresponding one of the one or more clusters; classifying the corresponding cluster as the first classification in response to determining that the quantity of hit points in the corresponding cluster is equal to or greater than the threshold; classifying the corresponding cluster as the second classification in response to determining that the quantity of hit points in the corresponding cluster is less than the threshold. 10. The vehicle tracking system of claim 9, configured to execute the instructions for the clustering by

[0105] (Appendix 14) 14. The vehicle tracking system of claim 13, wherein the processor is configured to execute the instructions to determine the threshold by multiplying the quantity of detection times by a selected percentage.

[0106] (Appendix 15) 10. The vehicle tracking system of claim 9, wherein the vehicle parameters include speed, acceleration, or lane changes.

[0107] (Appendix 16) 10. The vehicle tracking system of claim 9, wherein the processor is further configured to execute the instructions to instruct a transmitter to transmit the estimated vehicle parameters to an external device.

[0108] (Appendix 17) receiving distributed optical fiber sensing (DFOS) data; identifying hit points in the DFOS data, each hit point corresponding to a corresponding vehicle position at a detection time; clustering the identified hit points to define one or more clusters; classifying each of the one or more clusters into a first classification or a second classification based on the number of hit points and the number of detection times in each of the one or more clusters; estimating vehicle parameters of a first vehicle corresponding to the hit points of a first cluster of the one or more clusters based on the hit points of the first cluster having the first classification; A program that causes a computer to execute the following.

[0109] (Appendix 18) removing the earliest detected hit point of each of the one or more clusters having the second classification; iteratively repeating the clustering, the classification, and the removal until only clusters of the one or more clusters having the first classification remain; 18. The program according to claim 17, further causing the computer to execute the steps.

[0110] (Appendix 19) 19. The program of claim 18, further causing the computer to decrease the detection time quantity by one in response to removing the earliest detected hit point.

[0111] (Appendix 20) determining a threshold value based on the quantity of detection times; comparing, for each of the one or more hit points, a quantity of the hit points in a corresponding one of the one or more clusters; classifying the corresponding cluster as the first classification in response to determining that the quantity of hit points in the corresponding cluster is greater than or equal to the threshold; classifying the corresponding cluster as the second classification in response to determining that the quantity of hit points in the corresponding cluster is less than the threshold; 18. The program according to claim 17, further causing the computer to execute the steps.

[0112] The foregoing outlines features of several embodiments so that those skilled in the art may better understand aspects of the present disclosure. Those skilled in the art will readily appreciate that this disclosure may be used as a basis for designing or modifying other processes and structures to carry out the same purposes and / or obtain the same advantages of the embodiments presented herein. Those skilled in the art will also appreciate that such equivalent structures do not depart from the spirit and scope of the present disclosure, and that various changes, substitutions, and alterations can be made herein without departing from the spirit and scope of the present disclosure.

[0113] This application claims priority to U.S. Patent Application No. 18 / 358,945, filed July 26, 2023, the disclosure of which is incorporated herein in its entirety. [Explanation of symbols]

[0114] 100 systems 100A DAS system 100B DAS System 111 Traffic monitoring equipment 121 Optical Fiber 130 Roadway 130A Roadway 130B Roadway 140 vehicles 150 vehicles 150A First car 150B Second car 160A First Bridge 160B Second Bridge 170 Fiber part 300 Waterfall Data 302 areas 304 area 306 areas 308 areas 400 DFOS data 410 areas 420 areas 500 plots 510 First Vibration Line 520 Second Vibration Line 530 hit points 540 hit points 550 hit points 560 hit points 600 plots 610 First Cluster 615 starting hit points 620 Second cluster, first slope 625 starting hit points 630 Third Cluster 635 starting hit points 640 4th cluster, 2nd slope 645 starting hit points 700 plots 720 Cluster 725 starting hit points 740 cluster 745 starting hit points 800A Plot 800B Plot 810 First regression line 815 hit points 820 First Inclination 830 Second regression line 835 hit points 835' Hit Points 840 Second Inclination 850 predicted position 1000 systems 1002 processor 1004 Computer-readable storage medium 1006 Computer Program Code 1007 Instructions 1008 Bus 1010 I / O 1012 Network Interface 1014 Network 1016 Sensor Data Parameters 1018 Threshold Parameter 1020 Frequency range parameters, vehicle parameters 1022 Hit Point Parameter

Claims

1. receiving distributed fiber optic sensing (DFOS) data; identifying hit points within the DFOS data, each hit point corresponding to a corresponding vehicle position at a detection time; clustering the identified hit points to define one or more clusters; classifying each of the one or more clusters into a first classification or a second classification based on the number of hit points and the number of detection times in each of the one or more clusters; estimating vehicle parameters of a first vehicle corresponding to the hit points of a first cluster of the one or more clusters based on the hit points of a first cluster having the first classification; Including, The classification may include: determining a threshold value based on the quantity of detection times; comparing, for each of the one or more hit points, the quantity of the hit points in a corresponding one of the one or more clusters; classifying the corresponding cluster as the first classification in response to determining that the quantity of hit points in the corresponding cluster is equal to or greater than the threshold; classifying the corresponding cluster as the second classification in response to determining that the quantity of hit points in the corresponding cluster is less than the threshold; A vehicle tracking method comprising:

2. removing the earliest detected hit point of each of the one or more clusters having the second classification; repeating the clustering, classifying, and removing iteratively until only clusters of the one or more clusters having the first classification remain; The vehicle tracking method of claim 1 further comprising:

3. The vehicle tracking method of claim 2 , further comprising: decreasing the detection time quantity by one in response to removing the earliest detected hit point.

4. The method of claim 1 , wherein estimating the vehicle parameters includes estimating at least one of speed, acceleration, or lane changes.

5. The method of claim 1 , wherein determining the threshold value comprises multiplying the quantity of detection times by a selected percentage.

6. The vehicle tracking method of claim 1 , wherein the vehicle parameters include speed, acceleration, or lane changes.

7. The vehicle tracking method of claim 1 , further comprising transmitting the estimated vehicle parameters to an external device.

8. a computer-readable medium configured to store instructions; a processor coupled to the computer-readable medium; wherein the processor: receiving distributed fiber optic sensing (DFOS) data; identifying hit points within the DFOS data, each hit point corresponding to a corresponding vehicle position at a detection time; clustering the identified hit points to define one or more clusters; classifying each of the one or more clusters into a first classification or a second classification based on the number of hit points and the number of detection times in each of the one or more clusters; and estimating a vehicle parameter of a first vehicle corresponding to the hit point of the first cluster of the one or more clusters based on the hit point of a first cluster having the first classification. configured to execute the instructions for The classification may include: determining a threshold value based on the quantity of detection times; comparing, for each of the one or more hit points, the quantity of the hit points in a corresponding one of the one or more clusters; classifying the corresponding cluster as the first classification in response to determining that the quantity of hit points in the corresponding cluster is equal to or greater than the threshold; classifying the corresponding cluster as the second classification in response to determining that the quantity of hit points in the corresponding cluster is less than the threshold; Vehicle tracking systems, including:

9. receiving distributed fiber optic sensing (DFOS) data; identifying hit points within the DFOS data, each hit point corresponding to a corresponding vehicle position at a detection time; clustering the identified hit points to define one or more clusters; classifying each of the one or more clusters into a first classification or a second classification based on the number of hit points and the number of detection times in each of the one or more clusters; estimating vehicle parameters of a first vehicle corresponding to the hit points of a first cluster of the one or more clusters based on the hit points of a first cluster having the first classification; A program for causing a computer to execute the above, The classification may include: determining a threshold value based on the quantity of detection times; comparing, for each of the one or more hit points, the quantity of the hit points in a corresponding one of the one or more clusters; classifying the corresponding cluster as the first classification in response to determining that the quantity of hit points in the corresponding cluster is equal to or greater than the threshold; classifying the corresponding cluster as the second classification in response to determining that the quantity of hit points in the corresponding cluster is less than the threshold; Including, the program.

Images