LoRa relative position positioning algorithm and system

Abstract

The invention discloses a LoRa relative position positioning algorithm and system, and the method comprises the steps: S1, deploying LoRa nodes: deploying LoRa child nodes in a target region, so as to guarantee that a network covers the whole region; s2, beacon collection: sending, receiving and recording of beacon signals are completed, and an RSSI signal matrix is formed; s3, an RSSI conversion distance model: through actual measurement and fitting, establishing a signal attenuation model between an RSSI signal and a distance, and converting an RSSI value into a distance estimation value to form a distance matrix; s4, matrix filling, wherein the problem of matrix data missing is solved through matrix filling; s5, carrying out multi-dimensional scale dimension reduction amp on the MDS; and obtaining position information: carrying out dimension reduction processing on the distance matrix by using a multi-dimensional scaling method MDS, and finally obtaining the position information of each node. The method does not depend on a base station, is suitable for an environment with irregular node distribution, and reduces the deployment and maintenance cost; nodes can be easily added on a large scale, and the situation of non-dense distribution or non-uniform distribution does not need to be worried about; the robustness is high, and the influence on the positioning effect is small due to the missing condition of some nodes; moreover, the method does not depend on a central server, each node can work independently, and the robustness and fault tolerance of the system are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of Internet of Things, and in particular to a LoRa relative position positioning algorithm and system. Background Art

[0002] The Internet of Things (IoT) technology is developing rapidly. The LoRaWAN network has been widely used in the field of IoT due to its low power consumption and long-distance communication characteristics.

[0003] Most of the LoRa positioning methods on the market are based on base stations. This is because the positioning capability of LoRa technology depends largely on the gateway (base station) in the LoRaWAN network architecture. The LoRaWAN network consists of nodes, gateways, and network servers. The gateway is responsible for receiving the node's signal and uploading the data to the network server, while the server executes the positioning algorithm to determine the node's location.

[0004] Currently commonly used LoRa positioning methods include:

[0005] (1) RSSI (Received Signal Strength Indication)-based positioning: This method uses signal strength to estimate distance. Although simple, the positioning accuracy is greatly affected by environmental factors and is suitable for scenarios where accuracy requirements are not high.

[0006] (2) Positioning based on TDOA (Time Difference of Arrival): The position is calculated by measuring the time difference between the signal reaching different gateways. Usually at least three gateways are required to achieve two-dimensional positioning. It is one of the more accurate LoRa positioning technologies on the market.

[0007] (3) Positioning based on TOA (Time of Arrival): Positioning is based on the absolute time when the signal arrives at each gateway. This requires strict time synchronization between all gateways and is more complex to implement.

[0008] (4) Positioning based on AOA (Angle of Arrival): Positioning is performed by measuring the angle of arrival of the signal. This method requires a specially designed antenna array and can be used in combination with technologies such as TDOA to improve positioning accuracy.

[0009] These positioning methods are all in the LoRaWAN network, using base stations (gateways) to receive signals and the network server to calculate to determine the location of the node. In addition, in current practical applications, base station-based positioning is the mainstream, especially in scenarios that require coverage of large areas, long distances, and centralized management.

[0010] However, while using base station positioning has brought significant advantages in the fields of wireless communications and the Internet of Things, it also has some disadvantages. Especially for the positioning needs of terminal devices in large-scale plant lights or street lights, using base stations for positioning may encounter the following shortcomings:

[0011] (1) Network capacity and reliability: In large-scale deployment scenarios, such as plant light and street light networks, LoRa networks may face capacity bottlenecks. A large number of nodes sending data at the same time may cause network congestion, affecting the reliability of data transmission and the accuracy of positioning. Network reliability is also affected by the density of base stations. If the base stations are unevenly distributed, some areas may not have signal coverage, affecting the positioning quality. In addition, a sufficient density of base stations is required to ensure positioning accuracy and coverage. If the number of base stations is insufficient, the positioning effect will be greatly reduced.

[0012] (2) Energy consumption: Although LoRa technology is known for its low power consumption, high-frequency positioning needs, such as the need to frequently update location information, may accelerate battery consumption and affect the maintenance cycle and operating costs of lamps.

[0013] (3) Cost and infrastructure: More base stations and complex signal processing equipment are required. Large-scale deployment of LoRa base stations and corresponding positioning infrastructure may require high deployment and maintenance costs. In addition, in order to improve positioning accuracy and network coverage, the density of base stations may need to be increased, further increasing costs.

[0014] (4) Limited spectrum resources: LoRa uses free unlicensed frequency bands, which means that in crowded frequency bands, other wireless devices may interfere with the LoRa signal, affecting data transmission and positioning accuracy. In addition, due to the distribution of base stations and signal propagation characteristics, the positioning effect will be affected in some areas with complex terrain or severe obstruction.

[0015] In addition, in closed environments such as indoor factories and greenhouses, the existing LoRa positioning system may face some challenges, especially in complex and changing indoor environments. The following challenges still exist and areas need to be improved:

[0016] (1) Signal reflection and attenuation: In indoor environments, objects such as walls, equipment, and metal structures may cause signal reflection and attenuation, affecting positioning accuracy.

[0017] (2) Multipath effect: The signal may reach the receiver through multiple paths, making it difficult for the positioning algorithm to accurately determine the true location.

[0018] (3) Environmental dynamics: Changes in the internal layout of factories and greenhouses and the movement of equipment may affect the stability of the positioning system.

[0019] (4) Device density: High-density device deployment may increase positioning difficulty, especially in scenarios with small device monitoring. Summary of the invention

[0020] In response to at least one of the above problems, the present invention provides an efficient LoRa relative position positioning algorithm and system that does not rely on base stations and is based on RSSI combined with multidimensional scaling (MDS), aiming to provide more accurate, reliable, convenient and efficient positioning services, which can be suitable for the positioning needs of nodes in scenarios such as large-scale plant lights or street lights.

[0021] To achieve the above object, the present invention adopts the following technical solutions:

[0022] A LoRa relative position positioning algorithm includes the following steps:

[0023] S1. Deploy LoRa nodes: Deploy LoRa subnodes in the target area to ensure that the network covers the entire area;

[0024] S2, beacon collection: complete the sending, receiving and recording of beacon signals to form an RSSI signal matrix;

[0025] S3, RSSI conversion distance model: through actual measurement and fitting, establish the signal attenuation model between RSSI signal and distance, which is used to convert RSSI value into distance estimation value and form distance matrix;

[0026] S4, Matrix filling: Use matrix filling to solve the problem of missing matrix data;

[0027] S5. MDS multidimensional scaling dimensionality reduction & obtaining location information: Use the multidimensional scaling method MDS to reduce the dimension of the distance matrix. The MDS algorithm can retain the information of the relative distance of the nodes, map the high-dimensional distance information to the low-dimensional space, and finally obtain the location information of each node.

[0028] As a preferred embodiment of the above solution, step S2 specifically includes:

[0029] S21, beacon signal transmission: the gateway master node controls the sub-nodes to send beacon signals in sequence to ensure order and avoid signal conflicts;

[0030] S22, beacon signal reception and recording: after receiving the beacon signal, the neighboring node records the unique identification number SN and RSSI value of the sending node to provide raw data for subsequent analysis;

[0031] S23, multi-node signal recording: Each node records multiple SNs and their RSSI values, eventually forming an N*N RSSI signal matrix, which indicates the signal strength between two nodes.

[0032] As a preferred embodiment of the above solution, step S23 specifically includes:

[0033] S231, beacon collection is initiated by the master node, and requests are sent one by one according to the SN list. A certain time interval is left between each request to ensure that each node can send out its own beacon;

[0034] S232, when other collecting nodes receive the beacon of the current node, they record the SN of the node and the corresponding RSSI signal to form a signal set;

[0035] S233, the master node controls polling to allow all nodes to complete beacon transmission, and ensures the stability of RSSI by calculating the average value through multiple rounds of polling.

[0036] S234. The master node collects RSSI signal sets of all nodes one by one.

[0037] As a preferred embodiment of the above solution, in step S3, the conversion model between RSSI and distance is generated according to the following formula (1):

[0038] d=N*Math.Log10(RSSI)-RSSI0 (1)

[0039] Where RSSI0 is the RSSI value measured at a specific distance; N is the path loss index, which depends on the environment. RSSI0 and N can be obtained by fitting multiple sets of data through actual testing; d is the distance, usually in meters.

[0040] As a preferred embodiment of the above solution, step S4 specifically includes:

[0041] S41, symmetrical filling: using the theoretical symmetry of the RSSI signal matrix, if there are null values ​​in the matrix, check whether there is valid data at the symmetrical position, and fill the control with the data at the symmetrical position;

[0042] S42, K nearest neighbor algorithm: For missing values ​​that cannot be solved by symmetric filling, the K nearest neighbor algorithm KNN is used to estimate the missing distance value based on the distance information of the nearest K nodes, thereby completing the matrix filling.

[0043] As a preferred embodiment of the above solution, step S42 specifically includes:

[0044] S421. Construct distance metric: Define a method to measure the distance between nodes;

[0045] S422, select K value: select an appropriate K value, that is, the number of neighbors used to fill the missing values;

[0046] S423, K nearest neighbor filling: For each missing value, find the known values ​​of its K nearest neighbors and calculate the weighted average of these neighbors as the filling value.

[0047] As a preferred embodiment of the above solution, step S5 specifically includes:

[0048] S51. According to actual needs, a two-dimensional dimension reduction method is used to perform calculations based on MDS;

[0049] S52. Visualize the relative positions of the nodes and compare and verify them with known positions.

[0050] A LoRa relative position positioning system includes multiple LoRa sub-nodes, a LoRa gateway master node and a host computer connected to each node. The sub-nodes are responsible for signal measurement and mutual transmission, the gateway master node is responsible for controlling each sub-node to send beacon signals in sequence and collect signal data, and the host computer is responsible for implementing a positioning algorithm.

[0051] As a preferred embodiment of the above scheme, the host computer mainly includes a signal measurement module, a data processing module and a position calculation module.

[0052] Due to the above structure, the beneficial effects of the present invention are:

[0053] 1. Independent of base stations: Independent of base stations, and suitable for environments with irregular node distribution;

[0054] 2. Low cost: No base station deployment is required, thus reducing deployment and maintenance costs;

[0055] 3. Scalability: You can easily add nodes on a large scale without worrying about the distribution being sparse or uneven.

[0056] 4. Strong robustness: The missing of some nodes has little impact on the positioning effect;

[0057] 5. Decentralization: Without relying on a central server, each node can work independently, improving the robustness and fault tolerance of the system;

[0058] 6. Strong adaptability: It can adapt to a variety of different scenarios. For example, in scenario one, in scenarios such as large-scale plant lights or street lights, when a node fails, it can respond quickly based on direct communication between nodes without relying on base stations, which means that the fault detection and positioning process can be faster because the data does not need to be uploaded to the central server for processing, reducing delays. This is crucial for timely recovery or replacement of faulty equipment. Scenario two: In large-scale indoor factories or greenhouses, light capture is required, and there is no positioning system, making on-site maintenance difficult. The advantage of using this patent is that the existing conditions (wireless communication) can be used to solve the positioning problem of the equipment without adding other hardware costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for describing the embodiments are briefly introduced below.

[0060] Figure 1 is a flow chart of the method of the present invention;

[0061] Figure 2 A schematic diagram of the deployment of LoRa nodes of the present invention;

[0062] Figure 3 It is a distance matrix schematic diagram of the present invention;

[0063] Figure 4 Flowchart collected for the beacon of the present invention;

[0064] Figure 5 Flow chart of converting RSSI matrix into distance matrix of the present invention;

[0065] Figure 6 A flowchart of matrix filling of the present invention;

[0066] Figure 7 It is a complete graph restored by calculation after losing different data proportions in the present invention;

[0067] Figure 8 It is an actual node physical location diagram of the present invention;

[0068] Fig. 9 For the present invention and Figure 8 Corresponding operation node location diagram. DETAILED DESCRIPTION

[0069] The technical solution of the present invention will be described clearly and completely below in conjunction with the accompanying drawings of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of protection of the present invention.

[0070] like Figure 1 As shown, this embodiment provides a LoRa relative position positioning algorithm, including the following steps:

[0071] S1. Deploy LoRa nodes:

[0072] Deploy LoRa subnodes in the target area to ensure network coverage of the entire area, such as smart street light networks, which lays the foundation for the subsequent sending and receiving of label washing signals. Figure 2 As shown in the figure, a 4-point LoRa network is deployed. Assuming that each node can receive the signal sent by other nodes, a 4*4 matrix can be constructed (such as Figure 3As shown), theoretically, the matrix is ​​a symmetric matrix and the elements on the diagonal are 0 (from itself to itself).

[0073] S2. Beacon collection:

[0074] Complete the sending, receiving and recording of beacon signals to form an RSSI signal matrix (such as Figure 4 ), including:

[0075] S21, beacon signal transmission: the gateway master node controls the sub-nodes to send beacon signals in sequence to ensure order and avoid signal conflicts;

[0076] S22, beacon signal reception and recording: after receiving the beacon signal, the neighboring node records the unique identification number SN and RSSI value of the sending node to provide raw data for subsequent analysis;

[0077] S23, multi-node signal recording: Each node records multiple SNs and their RSSI values, and finally forms an N*N RSSI signal matrix, which indicates the signal strength between two nodes; specifically, it includes:

[0078] S231, beacon collection is initiated by the master node, and requests are sent one by one according to the SN list. A certain time interval is left between each request to ensure that each node can send out its own beacon;

[0079] S232, when other collecting nodes receive the beacon of the current node, they record the SN of the node and the corresponding RSSI signal to form a signal set;

[0080] S233, the master node controls polling to allow all nodes to complete beacon transmission, and ensures the stability of RSSI by calculating the average value through multiple rounds of polling.

[0081] S234. The master node collects RSSI signal sets of all nodes one by one.

[0082] S3, RSSI conversion distance model:

[0083] Through actual measurement and fitting, a signal relationship model between RSSI signal and distance is established to convert RSSI value into distance estimation value and form a distance matrix (such as Figure 5 The algorithm for converting RSSI to actual distance (D) can be implemented by using a signal attenuation model. Here is a simplified algorithm example based on the Free Space Path Loss Model for estimating distance in an unobstructed open space.

[0084] The conversion model between RSSI and distance is generated according to the following formula (1):

[0085] d=N*Math.Log10(RSSI)-RSSI0 (1)

[0086] Where RSSI0 is the RSSI value measured at a specific distance; N is the path loss index, which depends on the environment. RSSI0 and N can be obtained by fitting multiple sets of data through actual testing; d is the distance, usually in meters; a typical one is as follows: RSSI0 = 31.06, N = -14.47.

[0087] S4. Matrix filling:

[0088] Due to distance problems or missing data, matrix filling is required. Because not every node can obtain the RSSI signals of all other nodes and convert them into distance matrices, there will be a problem of missing matrix data. When processing RSSI data of multiple nodes, it is found that the signals between some nodes cannot be received, resulting in an incomplete RSSI matrix. If the data is incomplete, subsequent algorithm calculations cannot be performed. Therefore, two methods can be used to fill the matrix, such as Figure 6 As shown, specifically including:

[0089] S41, Symmetric filling: Symmetric matrix filling is a reasonable choice, because in many practical applications, the RSSI values ​​between nodes should be symmetrical (that is, the distance value from A to B is the same as the distance value from B to A). In this case, the symmetric filling method can be used to fill the missing distance value, that is, the RSSI signal matrix is ​​theoretically symmetrical. If there is an empty value in the matrix, check whether there is valid data at the symmetric position, and fill the control with the data at the symmetric position;

[0090] S42, K-nearest neighbor algorithm: If there are missing values ​​at symmetrical positions, it is an effective method to use the K-nearest neighbors (KNN) method to fill the missing values ​​in the distance matrix. That is, for missing values ​​that cannot be solved by symmetrical filling, the K-nearest neighbor algorithm KNN can be used. The KNN method predicts the missing values ​​by finding the K closest known values. This method can make better use of existing data, especially when the distribution of distance values ​​has a certain regularity, the effect will be better, including:

[0091] S421. Construct distance metric: Define a method to measure the distance between nodes;

[0092] S422, select K value: select an appropriate K value, that is, the number of neighbors used to fill the missing values;

[0093] S423, K nearest neighbor filling: For each missing value, find the known values ​​of its K nearest neighbors, and calculate the weighted average of these neighbors as the filling value. The advantages of using the K nearest neighbor algorithm are: (1) Simple and easy to use: The K nearest neighbor algorithm is simple in concept, easy to understand and implement, and does not require complex parameter adjustments, which helps to quickly deploy the LoRa positioning algorithm. (2) No training process required: K nearest neighbor is a lazy learning algorithm that does not require an explicit training phase and directly uses all samples for prediction, which improves the real-time performance of the algorithm. (3) Strong adaptability: K nearest neighbor can process various types of data, including numerical and categorical data, and has a wide range of applications and can process RSSI data in different forms. (4) High flexibility: The performance of the K nearest neighbor algorithm can be optimized by adjusting the K value. It has high flexibility and can select the optimal parameters according to actual conditions. (5) Large impact of local data: Since K nearest neighbor relies on neighboring data points, it can handle data with significant local features well, which helps to fill missing values ​​and process noisy data, and improve the integrity and accuracy of the data. (6) Filling missing data with high precision: During the matrix filling process, the K nearest neighbor algorithm can effectively estimate missing values ​​using the information of similar nodes, thus ensuring the integrity and accuracy of the RSSI matrix.

[0094] The problem of missing matrix data in actual scenarios can be solved by using the symmetric filling and K-nearest neighbor algorithm filling methods. Figure 7 Shown is 0% ( Figure 7 a, i.e. no missing data) and the data loss rate is 10% ( Figure 7 b) 20% ( Figure 7 c) 30% ( Figure 7 d) 40%( Figure 7 e) is a comparison chart showing the integrity of the data after being processed. It has been proven that even with a data loss rate of 10%, 20%, 30%, or 40%, the integrity of the data can be well guaranteed.

[0095] S5, MDS multidimensional scaling dimensionality reduction & obtaining location information: Use the multidimensional scaling method MDS to reduce the dimension of the distance matrix. The MDS algorithm can retain the information of the relative distance of the nodes, map the high-dimensional distance information to the low-dimensional space, and finally obtain the location information of each node. Specifically include:

[0096] S51. According to actual needs, a two-dimensional dimension reduction method is used to perform calculations based on MDS;

[0097] S52, visualize the relative positions of the nodes and compare and verify them with known positions. Figure 8 is a 10*10 actual node physical location, Fig. 9 It is the position of the operation node. It can be seen from the figure that the node after positioning by MDS can well reflect the relative position of the original coordinates.

[0098] The advantages of using multidimensional scaling (MDS) are: (1) Strong dimensionality reduction capability: MDS can reduce high-dimensional data to low-dimensional space while maintaining the distance relationship between data points as much as possible, which helps to accurately restore the node position in the LoRa positioning algorithm. (2) Wide adaptability: MDS is applicable to a variety of data types, including distance matrices and similarity matrices, and can process various forms of RSSI data, providing a flexible dimensionality reduction solution. (3) Preserving distance information: The data points reduced by MDS can retain the distance relationship between the original data points as much as possible, which is particularly important in the positioning algorithm and helps to improve positioning accuracy. (4) Simplifying complex problems: MDS can simplify complex high-dimensional space problems into low-dimensional problems that are easy to understand and handle, which is helpful for data analysis and interpretation. (5) Unsupervised learning method: MDS does not require labels or supervisory information, so it is suitable for unsupervised learning tasks and helps to discover potential structures in RSSI data.

[0099] This embodiment also provides a LoRa relative position positioning system, including multiple LoRa subnodes, a LoRa gateway master node and a host computer connected to each node, the subnodes are responsible for signal measurement and mutual transmission, the gateway master node is responsible for controlling each subnode to send beacon signals in sequence and collect signal data, and the host computer is responsible for implementing the positioning algorithm. Among them, the host computer mainly includes a signal measurement module, a data processing module and a position calculation module.

[0100] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A LoRa relative position positioning algorithm, characterized in that: The steps include: S1. Deploy LoRa nodes: Deploy LoRa subnodes in the target area to ensure that the network covers the entire area; S2, beacon collection: complete the sending, receiving and recording of beacon signals to form an RSSI signal matrix; S3, RSSI conversion distance model: through actual measurement and fitting, establish the signal attenuation model between RSSI signal and distance, which is used to convert RSSI value into distance estimation value and form distance matrix; S4, Matrix filling: Use matrix filling to solve the problem of missing matrix data; S5. MDS multidimensional scaling dimensionality reduction & obtaining location information: Use the multidimensional scaling method MDS to reduce the dimension of the distance matrix. The MDS algorithm can retain the information of the relative distance of the nodes, map the high-dimensional distance information to the low-dimensional space, and finally obtain the location information of each node.

2. A LoRa relative position positioning algorithm according to claim 1, characterized in that: Step S2 specifically includes: S21, beacon signal transmission: the gateway master node controls the sub-nodes to send beacon signals in sequence to ensure order and avoid signal conflicts; S22, beacon signal reception and recording: after receiving the beacon signal, the neighboring node records the unique identification number SN and RSSI value of the sending node to provide raw data for subsequent analysis; S23, multi-node signal recording: Each node records multiple SNs and their RSSI values, eventually forming an N*N RSSI signal matrix, which indicates the signal strength between two nodes.

3. A LoRa relative position positioning algorithm according to claim 2, characterized in that: Step S23 specifically includes: S231, beacon collection is initiated by the master node, and requests are sent one by one according to the SN list. A certain time interval is left between each request to ensure that each node can send out its own beacon; S232, when other collecting nodes receive the beacon of the current node, they record the SN of the node and the corresponding RSSI signal to form a signal set; S233, the master node controls polling to allow all nodes to complete beacon transmission, and ensures the stability of RSSI by calculating the average value through multiple rounds of polling; S234. The master node collects RSSI signal sets of all nodes one by one.

4. A LoRa relative position positioning algorithm according to claim 1, characterized in that: In step S3, the conversion model between RSSI and distance is generated according to the following formula (1): d=N*Math.Log10(RSSI)-RSSI0 (1) Where RSSI0 is the RSSI value measured at a specific distance; N is the path loss index, which depends on the environment. RSSI0 and N can be obtained by fitting multiple sets of data through actual testing; d is the distance in meters.

5. A LoRa relative position positioning algorithm according to claim 1, characterized in that: Step S4 specifically includes: S41, symmetrical filling: using the theoretical symmetry of the RSSI signal matrix, if there are null values ​​in the matrix, check whether there is valid data at the symmetrical position, and fill the control with the data at the symmetrical position; S42, K nearest neighbor algorithm: For missing values ​​that cannot be solved by symmetric filling, the K nearest neighbor algorithm KNN is used to estimate the missing distance value based on the distance information of the nearest K nodes, thereby completing the matrix filling.

6. A LoRa relative position positioning algorithm according to claim 5, characterized in that: Step S42 specifically includes: S421. Construct distance metric: Define a method to measure the distance between nodes; S422, select K value: select an appropriate K value, that is, the number of neighbors used to fill the missing values; S423, K nearest neighbor filling: For each missing value, find the known values ​​of its K nearest neighbors and calculate the weighted average of these neighbors as the filling value.

7. A LoRa relative position positioning algorithm according to claim 1, characterized in that: Step S5 specifically includes: S51. According to actual needs, a two-dimensional dimension reduction method is used to perform calculations based on MDS; S52. Visualize the relative positions of the nodes and compare and verify them with known positions.

8. A LoRa relative position positioning system, applied to a LoRa relative position positioning algorithm as described in any one of claims 1-7, characterized in that: It includes multiple LoRa sub-nodes, a LoRa gateway master node and a host computer connected to each node. The sub-nodes are responsible for signal measurement and mutual transmission. The gateway master node is responsible for controlling each sub-node to send beacon signals in sequence and collect signal data. The host computer is responsible for implementing the positioning algorithm.

9. A LoRa relative position positioning system according to claim 8, characterized in that: The host computer mainly includes a signal measurement module, a data processing module and a position calculation module.

Images