CN115022964A - Indoor positioning radio map reconstruction method and system based on map signals - Google Patents

Abstract

The invention provides an indoor positioning radio map reconstruction method and system based on map signals. By modeling each reference point as a graph vertex and its radio data as graph signals, a graph signal model is first developed in which the virtual reference points and their radio data can be interpolated into the radio map as missing vertices. Then, the potential spatial structure between all real and virtual reference points is explored to find the graph laplacian, which is used to reconstruct the radiomap by semi-supervised graph interpolation. Simulation experiments prove that the wireless map reconstruction method provided by the invention has good performance in the aspect of positioning accuracy, and shows the potential of the wireless map reconstruction based on the graph in the indoor positioning based on deep learning.

Description

Indoor positioning radio map reconstruction method and system based on map signals

Technical Field

The invention belongs to the field of graph signal processing, and particularly relates to an indoor positioning radio map reconstruction method and system based on graph signals.

Background

Indoor positioning is a key technology for Location Based Services (LBS) in many other areas where indoor navigation, building emergency rescue, and Global Navigation Satellite System (GNSS) are difficult to access. For many indoor positioning scenarios, traditional distance-based positioning techniques, including time of arrival (TOA), time difference of arrival (TDOA), angle of arrival (AOA), and Received Signal Strength (RSS) still apply. However, distance-based positioning depends to a large extent on the accuracy of the signal propagation model, which means that line-of-sight (LOS) propagation is often required for precise positioning. In contrast, indoor environments are generally more diverse and complex, so non line of sight (NLOS) is a more common channel model. Under the condition of NLOS, the radio fingerprint positioning method can adapt to irregular structures of indoor environments and obtain higher positioning accuracy.

Fingerprint-based positioning is generally divided into two phases, off-line training and on-line estimation. In the off-line training phase, radio fingerprints are collected at reference points of known locations, forming a radio map of the environment. In general, a Received Signal Strength Indicator (RSSI) and Channel State Information (CSI) may be used as fingerprints. In this field, many machine learning algorithms, such as K-nearest neighbor (KNN), Artificial Neural Network (ANN), Support Vector Machine (SVM), and constrained boltzmann machine, are widely used. Furthermore, since the relationship between radio fingerprints and positioning is substantially non-linear, deep learning based positioning is receiving increasing attention in this area. For example, Deep Neural Networks (DNNs), Convolutional Neural Networks (CNNs), and Recurrent Neural Networks (RNNs) have been proposed for indoor localization of radio fingerprints. Recently, Graphical Neural Networks (GNNs) have also been considered in this area. Compared with the traditional machine learning method, the indoor positioning based on deep learning has higher positioning precision.

The quality of the radio map at the off-line stage has a great influence on the accuracy of the indoor positioning. Intuitively, a more dense radiomap contains more details of the electromagnetic properties of the environment, and therefore a more accurate positioning can be achieved. Furthermore, radio samples should be collected multiple times per reference point to obtain finer grained and reliable data. However, in practice, the number of reference points is very limited due to the cost of hardware, time and labor. In some large buildings, the cost of intensive off-site surveys is even too high to be practical. At the same time, it is difficult to deploy as many reference points as possible to build a dense radio map due to privacy/security issues or deployment limitations. Second, since the indoor wireless environment is naturally time-varying, the radiomap may already be outdated, and the cost of real-time data calibration over a large number of reference points is unrealistic. In general, it is very expensive to repeatedly collect data at a reference point, and thus the collected data is generally not fine grained. Furthermore, despite the above difficulties, high resolution radio maps are still not possible due to unexpected hardware failures or transmission losses, which makes radio maps incomplete in space or time. In this regard, incomplete radio maps lacking radio data are one of the major challenges for deep learning based indoor positioning.

The wireless map building or rebuilding technology mainly comprises calibration-free data acquisition, a dynamic self-adaptive method, machine learning, deep learning and crowdsourcing. Wherein calibration-free data collection such as QR-Loc systems are difficult to achieve intensive scene monitoring. The kernel parameters of the improved bayesian regression method can adapt to dynamic changes, but a wireless monitor needs to be installed beside each AP, which increases the equipment cost. Based on deep learning methods, generation of a countermeasure network (GAN) is mostly adopted to extend real data, and the network needs to be trained greatly, and such methods mainly focus on fingerprint recovery and data enhancement of time samples, and do not consider spatial enhancement. Although enhancement can be performed spatially based on machine learning methods such as manifold learning techniques, gaussian process regression, etc., which depend largely on the number and arrangement of reference points and have an over-smoothing problem, they are not sparse, they use complete sample information for prediction, and not all sample information contributes to prediction. RGWR based on geographical weighting requires that a logarithmic distance model be measured in advance, which is difficult to obtain accurately throughout the environment, and the logarithmic path loss model still cannot accurately describe complex RSS distributions. Crowd-sourced based approaches require additional user intervention to fully exploit the full sensing capabilities of the mobile device, while continuous Inertial Measurement Unit (IMU) monitoring consumes a significant amount of the mobile device battery.

Disclosure of Invention

The present invention is directed to overcome the above-mentioned shortcomings in the prior art, and to provide a method and a system for reconstructing an indoor positioning radio map based on map signals, so as to overcome the disadvantage of spatial enhancement of radio map reconstruction in the prior art, thereby improving the performance of indoor positioning.

In order to achieve the purpose, the invention adopts the following technical scheme to realize the purpose:

an indoor positioning radio map reconstruction method based on map signals comprises the following steps:

constructing an indoor positioning frame, and obtaining an indoor initial radio map, wherein part of reference nodes are missing in the indoor initial radio map;

filling virtual reference nodes at reference nodes of the missing part of the initial radio map to obtain a radio map to be reconstructed, regarding the reference nodes in the radio map to be reconstructed as nodes of a graph topology, and regarding fingerprint information on each reference node as information on the nodes of the graph topology;

fingerprint information on each virtual node in the graph topology is recovered through a graph Laplace semi-supervised interpolation method; a reconstructed radiomap is obtained.

The invention is further improved in that:

preferably, a plurality of virtual nodes are filled in other positions in the initial radio map, and a new reconstructed radio map is obtained by a graph Laplacian semi-supervised interpolation method; the other positions are positions other than the reference node in the previously reconstructed radio map.

Preferably, the reconstructed radiomap is evaluated by RSS recovery accuracy analysis; the newly reconstructed radiomap is evaluated by a DNN neural network.

Preferably, the process of recovering the information on each virtual node in the graph topology by the graph laplacian semi-supervised interpolation method is as follows: establishing a weighted adjacency matrix reflecting K neighbors of all nodes in the graph topology, wherein weights in the weighted adjacency matrix are based on the position relation among the nodes in the graph topology; establishing a Laplace matrix by weighting the adjacency matrix; and establishing a smoothness equation based on the Laplace matrix, converting the smoothness equation into a problem function for solving missing values in the topology of the graph, and after the problem function is solved, completing information recovery on the virtual nodes.

Preferably, the weighted adjacency matrix is W _i,j In the graph signal, the closer the relationship between the node i and the node j is, the more W _i,j The greater the value of (A);

wherein m is _i,j A binary (0,1) adjacency matrix being a K neighbor, if node j is a K neighbor node of node i, then m _i,j 1, otherwise 0; g _i,j Is the Euclidean distance between node pair (i, j).

Preferably, the problem function of the missing value is:

where y is the signal to be recovered and x _i For the remaining reference nodes, S _K L is the laplacian matrix for the remaining set of reference nodes.

Preferably, the problem function of missing values is solved by the matlab's CVX toolkit.

Preferably, the smoothness equation is:

TV(x):＝x ^T Lx＝∑ _i≠j W _ij (x _i -x _j ) ² (2)

an indoor positioning radio map reconstruction system based on map signals, comprising:

the frame construction unit is used for constructing an indoor positioning frame and obtaining an indoor initial radio map;

the map topology establishing unit is used for filling virtual nodes in the initial radio map to obtain a radio map to be reconstructed, regarding reference nodes in the radio map to be reconstructed as nodes of a map topology, and regarding fingerprint information on each reference node as information on the nodes of the map topology;

the recovery unit is used for recovering information on each virtual node in the graph topology by a graph Laplace semi-supervised interpolation method; a reconstructed radiomap is obtained.

Preferably, the method further comprises the following steps:

an RSS recovery accuracy evaluation unit for evaluating the reconstructed radio map by RSS recovery accuracy analysis;

and the DNN neural network evaluation unit is used for evaluating the newly reconstructed radio map through the DNN neural network.

Compared with the prior art, the invention has the following beneficial effects:

the invention provides a map-based radio map reconstruction method, which is used for indoor positioning based on deep learning. By modeling each reference point as a graph vertex and its radio data as a graph signal, a graph signal model is first developed in which a virtual reference point and its radio data can be interpolated into a radio map as missing vertices. Then, the potential spatial structure between all real and virtual reference points is explored to find the graph laplacian, which is used to reconstruct the radiomap by semi-supervised graph interpolation. According to the invention, under the established graph signal processing model, the reconstruction of the radio map is taken as the problem of sampling and recovering the graph signals, and the prior spatial smooth knowledge of the graph signals is used for carrying out semi-supervised interpolation on the existing radio map by adopting a proper graph signal algorithm, so that the reconstruction of the radio map is finally completed, and the improvement of the spatial resolution is realized. And then evaluating the RSS recovery precision of the reconstructed radio map, or using the reconstructed radio map for deep neural network training to verify whether the indoor positioning accuracy is improved.

Simulation experiments prove that the wireless map reconstruction method provided by the invention has good performance in the aspect of positioning accuracy, and shows the potential of the wireless map reconstruction based on the graph in the indoor positioning based on deep learning. Compared with the prior art, the method utilizes the spatial smoothness of the graphic signal, does not need early training, and can be directly used for reconstructing the signal of the unknown region. This approach can exploit potential spatial correlation between data, not just relationships of data surfaces. Therefore, the method has the advantage of fingerprint information recovery precision corresponding to most APs.

Furthermore, the indoor positioning scene has wide NLOS shielding, which causes unsmooth space of some areas, thereby possibly influencing the positioning performance of the radio map reconstructed based on the map signal method. .

Drawings

Fig. 1 is a schematic diagram of radio fingerprint collection for indoor positioning.

Fig. 2 is an indoor radiomap reconstruction positioning system architecture based on a graph signal method.

Fig. 3 is a uciindor radiomap.

Fig. 4 is a radio map of a fourth floor of the first building selected by ucindoor.

Fig. 5 is a diagram of RSS recovery accuracy error for AP 41.

Fig. 6 is an RSS recovery accuracy error map of the AP 253.

Fig. 7 is a schematic radio map with 40 missing reference nodes.

Figure 8 is a schematic diagram of a carefully designed interpolation of a virtual node between two neighboring nodes.

Fig. 9 is a graph of the cumulative error distribution function CDF for different algorithms.

Fig. 10 is a schematic diagram after removing interference points.

FIG. 11 is a graph of the average positioning error after various algorithms have deleted the interference points.

Fig. 12 is a diagram of a variation of positioning error after three times of padding after 50 reference nodes are lost in a UWB ranging 3D map.

Fig. 13 is a diagram of a change in positioning error of a UWB ranging 3D map after losing 20 reference nodes and being filled three times.

Detailed Description

The invention is described in further detail below with reference to the accompanying drawings:

for the original radio map, the existing reference nodes may be damaged or lost due to certain factors, resulting in an incomplete radio map. Either the layout of existing reference nodes is adjusted, some are discarded, and a new radiomap is reconstructed, or it is filled directly from the existing reference points. The present invention refers collectively to these situations as the radiomap reconstruction problem. The aim is to obtain better positioning performance. These incomplete radio maps, or radio maps after the artificial rejection of nodes, are referred to as initial radio maps.

Based on the above problem, an embodiment of the present invention discloses an indoor positioning radio map reconstruction method based on map signals, which includes the following steps:

step 1, establishing a system model: the invention firstly selects the information given in the data set in the indoor positioning data set, so as to be convenient for constructing a reference node area (such as a building, a layer or the whole space) of a space map topology, thereby selecting and constructing the original radio map based on the invention. The invention considers most of the original radio map reference nodes as missing, obtains an initial radio map containing only a small number of partial reference nodes, and takes the initial radio map as the original radio map before reconstruction.

Based on the selected reference node area, an indoor positioning frame based on fingerprints is established, the reference nodes are regarded as nodes in Graph topology, the fingerprint data are regarded as Graph data, and the whole system is abstracted to a Graph Signal Processing (Graph Signal Processing) model, namely a Graph topology structure.

Fig. 1 illustrates the collection of radio fingerprints for indoor positioning. Suppose there are four AP nodes indoors and the initial radio map has N _r A reference point. Take the first reference point as an example. The reference point receives a plurality of wireless fingerprint information from different AP nodes, wherein

Fingerprint information indicating that the nth AP transmits to the reference point. And a total of N were collected at different times _s And (4) sampling. Thus, the size of the fingerprint data set is N _r ×N _s ×N _AP ，N _s Is the number of samples at a reference point, N _AP Is the number of AP nodes, i.e., the size of the feature dimension of the fingerprint data.

Assuming that the number of reference points after the wireless map is reconstructed is N, the size of the final fingerprint data set is N multiplied by N _s ×N _AP . The radiomap reconstruction problem can be described as using the original N _r ×N _s ×N _AP Data set, expansion or adaptation N _r To obtain a reconstructed NxN _s ×N _AP And (4) data.

The method provided by the invention aims to reduce the cost of manually reconstructing the off-line radio map by utilizing the fingerprint information and the spatial topology information of the coverage reference point area. Fingerprints of the non-covered reference point areas are generated to achieve more accurate indoor positioning. The whole positioning system architecture is shown in fig. 2. The architecture consists essentially of radiomap reconstruction, node data recovery using graphical signals, and DNN positioning model.

Step 2: and (5) reconstructing a radio map. The radio map reconstruction of the invention is based on this initial radio map, followed by filling in virtual nodes (either randomly or at selected specific locations) and reconstruction, all reference nodes (actual as well as virtual) being considered as map topology nodes, the fingerprint information being the map signal. And then, fingerprint information on the virtual nodes is generated by using a graph signal method (graph Laplacian semi-supervised interpolation) of a formula (3), so that different indoor positioning fingerprint training sets are obtained.

The radio map after filling has N reference nodes and K actual reference nodes which exist. The data of the other (N-K) nodes is considered lost. In addition to populating virtual nodes at the original lost location to verify the accuracy of RSS recovery, the present invention elaborates a new reconstruction method to overcome the NLOS situation by inserting a virtual node between each pair of real nodes that are already very close to each other in a certain direction (longitude, latitude or x, y or z direction) for the original radiomap. This way of constructing the invention can overcome most NLOS cases.

The invention uses graph signal processing means to recover or generate the fingerprint data on the reconstructed virtual reference node, and the specific graph Laplace semi-supervised interpolation process is as follows:

(1) considering those signals associated with undirected, weighted, connected graphs, it is noted that

And weighted undirected graph notation corresponding to the signal

Wherein

Is a set of N vertices that are,

is a collection of edges between nodes. Can use x _i To be represented at a node

The magnitude of the signal value at (a). The adjacent matrix W is a weighted symmetrical matrix reflecting the connection strength relation between edges, if there is an edge (i, j) epsilon, then W is _i,j ≥0，Otherwise W _i,j 0. The tighter the association (correlation) of node i and node j, the more W _i,j The larger the value of (c). The laplacian matrix L ═ D-W is introduced, where D is the degree matrix.

(2) And then constructing a space map topology by using a K-nearest neighbor method. Specifically, the distance g between all node pairs (i, j) is calculated based on the geographic locations of all nodes _i,j A distance matrix G is formed. It is then necessary to eliminate some of the connections between nodes that are less dependent on physical distance. Therefore, an all-zero matrix M is created, and then the first k elements with the minimum distance need to be found in the ith row of the matrix G and recorded as

Let M's corresponding element be 1, record as

The weight of each edge is inversely proportional to the square of the distance between the two nodes. Finally, the following weighted adjacency matrix W is constructed:

when W is constructed, the corresponding degree matrix D and laplace matrix L — D-W may also be constructed. In the scene of the invention, nodes are taken as vertexes of the graph, distance relations among the nodes are taken as weights of edges, and fingerprint data is taken as information on the graph nodes.

(3) Many recovery methods are based only on the surface relationships of the data, and do not extract the underlying relationships of the data. In the embodiment, potential spatial features exist between the reconstructed actual reference node and the virtual reference node. The wireless fingerprint information should be similar between nodes that are closer, and the data should be more different between nodes that are farther, a property called spatial smoothness. Spatial smoothness is an important feature of graphics signals. On the connected graph, assuming that the graph information is continuously diffused, the signal values of different nodes will gradually become similar as time goes by. In the limit, the signal values of all nodes on the graph will become consistent. To measure the smoothness or signal correlation of the signal y, the laplace quadratic of the following graph is used to describe this relationship. The smaller the value of the equation, the smoother the signal on the graph.

TV(y):＝y ^T Ly＝∑ _i≠j W _ij (y _i -y _j ) ² (2)

The present invention addresses this problem by a graph semi-supervised learning framework when data for one or more nodes is lost. The framework assumes that the data is located in the graph nodes. The data between spatially more adjacent nodes should be smoother, i.e. the graph signal should be smooth, and the values of the smoothness equation shown in equation (2) should be smaller and smoother. The invention considers that the value of the missing position has the smooth relation with the values of the known nodes adjacent to the missing position, and obtains a spatial smoothness expression y ^T Ly. The smoothness equation is then converted into a problem function that solves for missing values in the map topology, as shown in equation (3), and the global quantity represented by spatial smoothness is optimized by equation (3) to find the missing values, i.e., the information on the virtual nodes is recovered, and the radiomap can be reconstructed by solving equation (3). In this process the constraints limit that the value generated at a known position should be equal to a known value.

Specifically, if y is the signal to be recovered, then the signals on the K nodes are known to be x. The purpose of graph semi-supervised interpolation is to estimate y and guarantee x on known nodes _i ＝y _i ,i∈S _K . Similar to low rank matrix recovery, this method is semi-supervised. These unobserved values are calculated at each time step for a given observation. The method does not need a training stage, and when a node S is observed _K Then, the missing N-K node data can be reconstructed directly, or when a virtual sensor containing unknown values is desired. This problem can be expressed as:

by solving the above problem through the CVX toolbox of matlab, the signal y to be recovered can be obtained, that is, the fingerprint information in the reconstructed radio map is obtained, and the reconstructed radio map is correspondingly obtained, where the radio map has N reference nodes, each reference node has its own longitude and latitude, and each reference node has its corresponding fingerprint information.

Further, although the nodes have potential graph structure characteristics, a common NLOS problem exists in an indoor environment due to the particularity of an indoor positioning scene. Leading to the fact that: the distance between the generated virtual node and its neighboring nodes may be relatively far, so their fingerprint data may not be smooth due to signal occlusion and interruption. This problem may limit the application of graph smoothness to some extent. In the radiomap reconstruction problem of the present invention, it is desirable to maximize the potential smoothness of the map, although NLOS can obscure the map. Therefore, the embodiment also provides a well-designed radio map filling mode, and the test set positioning accuracy is used as an indirect evaluation index. Filling a plurality of virtual nodes in other positions in the reconstructed radio map, repeating the steps (1) to (3), recovering the fingerprint information on the newly added reference node, and obtaining the newly reconstructed radio map by a graph Laplacian semi-supervised interpolation method;

further, considering that there are areas inside the building that are walls or other non-free spaces where targets cannot be found, the present invention will also remove these extra interfering location reference nodes that are newly added in a well-designed filling manner. The indoor positioning accuracy of the radio map newly reconstructed by the method is superior to that of the initial radio map and also superior to that of the newly reconstructed by the baseline method.

And 4, step 4: training set training phase. Inputting the fingerprint information in the reconstructed radio map generated in the step 2 into the neural network of the invention for training, and obtaining each fingerprint data corresponding to the neural networkComparing and calculating the longitude and latitude obtained by calculation with the actual longitude and latitude of the fingerprint data to obtain a loss function, finishing the neural network training when the loss function meets the set requirement, and finally storing the obtained network model parameters. The indoor positioning model used in the present invention is a three-layer DNN model. The input size of the network is (NXN) _s )×N _AP I.e. having NxNxN _s Samples, each sample having a feature dimension of N _AP . The number of neurons in layers 1, 2 and 3 is different for different data sets. The fixed deprivation can be carried out according to actual conditions. The neural network positioning test set selects the area selected by the invention to be corresponding to the original test set for carrying out positioning performance test.

Using the MSE function as a loss function for DNN, the network will directly output the coordinates of the object. Using the Adam optimizer, the initial learning rate of the network was 0.01. Using L ₂ Loss as weight attenuation and set the value to 5 × 10 ^-4 In addition, to avoid death of network neurons to some extent, the Leakyrelu function is used for the activation function of each FC layer.

And 5: and a performance verification stage. Inputting the test set data into the network model stored in the step 3, and verifying the positioning performance of the method.

To facilitate the analysis of the RSS recovery accuracy. Assuming that the original radio map has N reference nodes, N remains after the nodes are lost randomly _r If the reference nodes are reconstructed into the radio map which is the same as the previous radio map, since the lost points are randomly distributed, the method can be regarded as a random reconstruction mode, and the fingerprint information in the reconstructed radio map and the information of the known original N reference nodes are subjected to RSS recovery accuracy analysis.

Evaluating the difference between a radio map reconstructed at the original lost node position by utilizing the map signal method and a baseline method through RSS recovery precision analysis; further, for the new layout mode well designed by the invention, because the positions of the original N reference points are not known, the DNN neural network of the step 4 evaluates the positioning accuracy performance gap used as indoor positioning between the radio map newly reconstructed at the well designed position by using the map signal method and the baseline method, and uses the gap as an indirect evaluation standard for radio map reconstruction. The existing, conventional method of recovering the map signal is set to the baseline method. It was found that the RSS information recovery accuracy at the lost node location of the radio map reconstructed using the map signal method is better than that of the baseline method.

The invention also discloses an indoor positioning radio map reconstruction system based on the map signal, which comprises the following steps:

the frame construction unit is used for constructing an indoor positioning frame and obtaining an indoor initial radio map;

the map topology establishing unit is used for filling virtual nodes in the initial radio map to obtain a radio map to be reconstructed, regarding reference nodes in the radio map to be reconstructed as nodes of a map topology, and regarding fingerprint information on each reference node as information on the nodes of the map topology;

the recovery unit is used for recovering information on each virtual node in the graph topology by a graph Laplace semi-supervised interpolation method; a reconstructed radiomap is obtained.

Further, the method also comprises the following steps:

an RSS recovery accuracy evaluation unit for evaluating the reconstructed radio map through RSS recovery accuracy analysis;

and a DNN neural network evaluation unit for evaluating the newly reconstructed radiomap through the DNN neural network.

Example 1

The invention is described in further detail below with reference to the figures and specific examples.

In the experiment, numerical results are displayed on the algorithm performance under a 2D scene of the fourth layer of the first building and a UWB ranging 3D scene data set under the UCIIndoor data set respectively. Meanwhile, Gaussian Process Regression (GPR) and Linear interpolation (Linear) are selected as the comparison method of the invention.

Fig. 3 is a ucindoor original radiomap, X-axis is longitude, Y-axis is latitude, and Z-axis is floor, and it is difficult to establish a connectivity map at all reference points because the three buildings are far away from each other. Furthermore, the raw data set does not give the floor height and is therefore not suitable for forming a three-dimensional distance-based graphical topology. For this purpose, fig. 4 is a radio map of the fourth floor of the first building selected, with 68 reference nodes in total. Table 1 and fig. 5, fig. 6 shows RSS recovery accuracy of recovering an originally lost node (random reconstruction mode) by using the remaining 28 reference nodes after 40 reference nodes are lost randomly, which cannot be shown in detail because there are 520 APs in total, and here, a part of representative results are shown, where the abscissa is 40 lost nodes and the ordinate is RSS bias. The fingerprint data recovery accuracy of the method of the graph signal can be found to be the highest, which is superior to that of the conventional method of comparison.

Fig. 7 is a schematic view of a radio map after 40 reference nodes are lost, and fig. 8 is a schematic view of a virtual node interpolated between two neighboring nodes with elaborate design. Table 2 shows that 40 nodes, referred to as Case1 and Case2, are randomly lost twice in the original data set, and according to the average positioning error of the graph laplacian semi-supervised interpolation, gaussian process regression and linear interpolation test set in the radio map reconstruction in this embodiment, it can be found that the positioning error of the graph signal method is significantly better than that of the gaussian process regression and linear interpolation in the scenario of this embodiment. Fig. 9 is a function of the cumulative error distribution of the different algorithms, and it can be seen that the method of the present embodiment is superior to the comparative scheme, especially for the control of large errors.

TABLE 1

RSS estimation error performance comparison (DBM) of different algorithms to reconstruct raw position data

Method	AP 7	AP15	AP23	AP69	AP184	AP185	AP220	……	Average
										Graph	0.1438	0	2.1633	0	1.0890	1.0858	0	……	0.4864
GPR	0.2045	0	2.2292	0	1.2045	1.3392	0	……	0.5192
										Linear	0.4561	0	5.5733	0	1.4225	1.3939	0	……	0.8981

TABLE 2

Comparison of different algorithm performances to reconstruct well-designed new position data

Scenario	No lost	Lost but not recovery	Graph	GPR	Linear interp
						case1	10.3687m	12.1040m	11.1908m	11.6164m	13.2037m
case2	10.3687m	10.6638m	10.4294m	10.9735m	12.4048m

Furthermore, since the building is "X" shaped, it can be found that the middle part of the "X" may be a wall or other part that cannot become free space. Therefore, these new nodes in the middle part are not possible in practice, so they have become the source of interference data for the interference localization result and should be eliminated. Fig. 10 is a schematic diagram after the interference points are removed. The present invention uses the suffix "-r" to indicate the deleted case. Fig. 11 is a graph of the average positioning error after various algorithms remove the interference points. It can be found that after the interference points are removed, the positioning accuracy of either method is further improved, and the method of the graph signal is still the optimal method.

Also, the above method can be applied to 3D scenes. Fig. 12 and fig. 13 show that after 50 and 20 reference nodes are randomly lost from a UWB ranging 3D data set (with 324 reference nodes), respectively, the UWB ranging 3D data set is first filled into 324 symmetrical nodes as before. More virtual nodes are then populated in a carefully designed manner, first according to the x-direction and then the y-direction. Finally, radio maps containing 324, 612 and 1156 points are respectively established to obtain the average positioning error trend change chart. It can be found that the graph signal method provided by the invention not only improves the performance in a 2D scene, but also obviously improves the positioning performance of the system in a 3D scene.

The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.

Claims (10)

1. An indoor positioning radio map reconstruction method based on map signals is characterized by comprising the following steps:

constructing an indoor positioning frame, and obtaining an indoor initial radio map, wherein part of reference nodes are missing in the indoor initial radio map;

filling virtual reference nodes at reference nodes of the missing part of the initial radio map to obtain a radio map to be reconstructed, regarding the reference nodes in the radio map to be reconstructed as nodes of a graph topology, and regarding fingerprint information on each reference node as information on the nodes of the graph topology;

fingerprint information on each virtual node in the graph topology is recovered through a graph Laplace semi-supervised interpolation method; a reconstructed radiomap is obtained.

2. The indoor positioning radio map reconstruction method based on map signals as claimed in claim 1, wherein a plurality of virtual nodes are filled in other positions in the initial radio map, and a new reconstructed radio map is obtained by a graph laplacian semi-supervised interpolation method; the other locations are locations other than the reference node in the previously reconstructed radio map.

3. The method of claim 2, wherein the reconstructed radiomap is evaluated by RSS recovery accuracy analysis; the newly reconstructed radiomap is evaluated by a DNN neural network.

4. The indoor positioning radio map reconstruction method based on graph signals as claimed in any one of claims 1 to 3, characterized in that the graph Laplace semi-supervised interpolation method recovers information on each virtual node in graph topology by: establishing a weighted adjacency matrix reflecting K neighbors of all nodes in the graph topology, wherein weights in the weighted adjacency matrix are based on the position relation among the nodes in the graph topology; establishing a Laplace matrix by weighting the adjacency matrix; and establishing a smoothness equation based on the Laplace matrix, converting the smoothness equation into a problem function for solving missing values in the topology of the graph, and after the problem function is solved, completing information recovery on the virtual nodes.

5. The method of claim 4, wherein the indoor positioning radio map reconstruction method based on the map signal,

the weighted adjacency matrix is W _i,j In the graph signal, the closer the relationship between the node i and the node j is, the more W _i,j The greater the value of (A);

wherein m is _i,j A binary (0,1) adjacency matrix being a K neighbor, if node j is a K neighbor node of node i, then m _i,j 1, otherwise 0; g _i,j Is the Euclidean distance between node pair (i, j).

6. The method of claim 4, wherein the problem function of the missing value is:

where y is the signal to be recovered and x _i For the remaining reference nodes, S _K L is the laplacian matrix for the remaining set of reference nodes.

7. The method for indoor positioning radiomap reconstruction based on map signals as claimed in claim 4, characterized in that the problem function of missing values is solved by the CVX tool box of matlab.

8. The method of claim 4, wherein the smoothness equation is:

TV(x):＝x ^T Lx＝∑ _i≠j W _ij (x _i -x _j ) ² (2)。

9. an indoor positioning radio map reconstruction system based on map signals, comprising:

the frame construction unit is used for constructing an indoor positioning frame and obtaining an indoor initial radio map;

the map topology establishing unit is used for filling virtual nodes in the initial radio map to obtain a radio map to be reconstructed, regarding reference nodes in the radio map to be reconstructed as nodes of a map topology, and regarding fingerprint information on each reference node as information on the nodes of the map topology;

the recovery unit is used for recovering information on each virtual node in the graph topology by a graph Laplace semi-supervised interpolation method; a reconstructed radiomap is obtained.

10. The system of claim 9, further comprising:

an RSS recovery accuracy evaluation unit for evaluating the reconstructed radio map by RSS recovery accuracy analysis;

and a DNN neural network evaluation unit for evaluating the newly reconstructed radiomap through the DNN neural network.

Images