Self-organizing routing method and system based on geometric constraint and direction finding cooperation

Abstract

The invention discloses a self-organizing routing method and system based on geometric constraint and direction finding cooperation, and mainly solves the problems of routing failure and serious interference caused by complex physical boundaries, dense nodes and non-line-of-sight propagation in a limited space under the condition of no GPS (Global Positioning System) in the prior art. According to the implementation scheme, a virtual spline guide rail representing the physical topology trend is constructed through collaborative direction finding to serve as a global navigation reference; carrying out vector fusion on the reference and the node aggregation direction sensed in real time to generate a guide vector; according to the vector, a judgment basis based on a geometric deviation angle is introduced to realize distributed judgment of a node role; and forming a controlled interception area with directional beam constraint by combining direction finding information and a space angle punishment mechanism according to a judgment result. According to the invention, through dynamic cooperation of geometric guidance and real-time perception, a directional controlled interception area is established in a limited space, so that a data transmission path can be adaptive to a curved shape of a roadway, multipath and sidelobe interference is effectively inhibited, the reliability and throughput performance of a network in a dynamic complex environment are remarkably improved, and the reliability of the network is improved. The method can be used for ad hoc network scenes without absolute coordinate reference, such as roadways and tunnels.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of communication technology, and particularly relates to a self-organizing routing method and system, which can be applied to directional data transmission of distributed ad hoc network in a restricted space without GPS signal support. BACKGROUND

[0002] With the rapid development of mines and underground construction, higher requirements are put forward for wireless ad hoc network transmission with high reliability and anti-interference in a restricted space. The rejection of GPS signals makes the traditional geographic routing relying on absolute coordinates completely invalid, and the narrow and winding physical boundary further causes serious non-line-of-sight transmission difficulties and multipath effects. In particular, the existing routing lacks the ability to perceive the physical topology, which easily causes the interception of directional beam boundaries, and the correlation failure caused by the spatial aggregation of nodes. In addition, if the sidelobe interference in a high-density network cannot be effectively physically isolated, the system throughput will be severely restricted. Therefore, it is very necessary to design a self-organizing routing method based on geometric constraints and direction finding cooperation, to build a blocking relay network structure through spatial diversity screening, so that the network can be free from coordinate dependence and suitable for high-resilience directional communication in a GPS strong interference environment.

[0003] Halford et al. in their published paper "Halford T R, Chugg K M. Barrage relay networks [C] / / 2010 Information Theory and Applications Workshop (ITA). IEEE, 2010: 1-8." proposed a method for establishing an interception zone isolation based on TDMA time slots, controlled flooding and RTS / CTS control. The implementation steps of the method are as follows: (1) M time slot TDMA synchronization, source node broadcasts in the first time slot, subsequent time slots are forwarded according to the hop number, physical layer control single forwarding, and multiple nodes are parallel to send to obtain diversity gain by means of phase jitter; (2) the source and the destination obtain the shortest path S through RTS / CTS, the nodes determine the role according to the hop number, the buffer nodes form an interception ring, the spatial reuse of CBR is realized, and the forwarding rule is clear; (3) three-layer logical channel time division multiplexing, the source is specified to send in the DLC time slot, the nodes loop the BRN relay, and the buffer nodes isolate the interference. However, this method has poor adaptability, low network throughput, and great difficulty in system deployment and maintenance due to the need for time slot synchronization and control overhead of buffer nodes, and the need for fine design in interference coupling and parameter selection.

[0004] Yu Zhu et al. in their published paper "Zhu Y, Guo K, Dong C, et al. Barrage Relay Network Assisted Multicast Routing Protocol for Spectrum Dissemination in UAV Networks [C] / / ICC 2025-IEEE International Conference on Communications. IEEE, 2025: 3069-3074." proposes a multicast controlled interception zone establishment method based on all relay surrounding buffer ring and distributed role algorithm. The implementation steps are: (1) the source node broadcasts an extended RTS containing a multi-destination list, the destination node returns a CTS containing the topology position, and the source node determines the internal relay node set of the multicast spanning tree accordingly; (2) all nodes on the source-destination path are selected as relays, buffer nodes are deployed around the relay nodes to form the boundary of the multicast interception zone, and the multicast flow is isolated by suppressing the forwarding of the buffer nodes; (3) candidate nodes independently determine whether they are key path nodes of the multi-beam based on local information, and optimize the control message format to reduce channel overhead by reusing the RTS / CTS framework; (4) internal nodes forward data at designated time slots to support parallel transmission of multiple destinations, buffer nodes suppress forwarding to maintain the MCBR boundary, and reduce redundant relays to improve transmission throughput. The disadvantages of this method are: the buffer / relay role determination and control message design are more complex, the buffer ring is sensitive to topology changes, and there is additional control load.

[0005] The patent document with application number CN201610344697.3 discloses a power transmission line information routing system and method based on directional antennas. The implementation steps are: (1) obtaining and packaging the monitoring information of the monitoring node; (2) determining the next hop node by querying the routing table, and adjusting the position and transmission direction of the directional antenna of the node according to the position of the next hop node; (3) adjusting the transmission power according to the distance between the node and the next hop node to transmit the packaged monitoring information to the next hop node; (4) the next hop node receives the monitoring information, and fuses and packages the received monitoring information with the monitoring information of the node itself; (5) return to step (2) until the monitoring information is sent to the base station or the substation. The disadvantages of this method are: high implementation complexity, weak adaptability to dynamic topology, and high dependence on GPS. SUMMARY

[0006] The present application aims to overcome the shortcomings of the prior art by proposing a self-organizing routing method and system based on geometric constraints and direction finding cooperation in a restricted space, to avoid dependence on GPS, reduce deployment costs, improve network throughput, and enhance environmental adaptability.

[0007] The technical idea for achieving the object of the present application is: a virtual spline guide rail fitting the physical tunnel direction is constructed as a navigation reference through cooperative direction finding; a weighted guidance model is generated by fusing environmental confidence factors and real-time sensing information, and an elastic routing is implemented relying on the spline tangent vector; a distributed mechanism based on geometric deviation angle is introduced to realize autonomous determination of the node role; a directional controlled interception zone and a spatial diversity forwarding architecture are constructed by combining the Top-K screening strategy of direction finding information and angle penalty.

[0008] According to the above idea, the technical scheme of the present application comprises the following:

[0009] 1. A self-organizing routing method based on geometric constraint and cooperative direction finding, characterized in that it comprises:

[0010] (1) The relative position information of the current node and the neighbor nodes is obtained by using the cooperative direction finding and the neighbor discovery mechanism between nodes, and a virtual spline guide rail fitting the physical topology direction is constructed by the Catmull-Rom interpolation algorithm to serve as a global navigation reference in the environment without absolute coordinates;

[0011] (2) The tangent vector of the virtual spline guide rail at the current position is extracted as a geometric reference, which is vector fused with the real-time aggregation direction based on neighbor signal sensing, and an environmental confidence factor is introduced to dynamically adjust the two to generate a weighted guidance vector;

[0012] (3) The source node initiates directional detection according to the weighted guidance vector, the receiving node calculates the geometric deviation angle of the probe signal to obtain the geometric deviation angle of the probe signal relative to the guidance vector, and based on the deviation angle, a distributed threshold decision is performed to establish the function role of the node as a potential relay node or a blocking node;

[0013] (4) A directional controlled interception zone is formed to isolate sidelobe interference based on the establishment of the node role, and for the established potential relay node, a screening strategy combined with angle penalty is used to lock the best candidate set to construct a forwarding architecture with spatial diversity capability.

[0014] Further, in the (1), the relative position information of the current node and the neighbor nodes is obtained by using the cooperative direction finding and the neighbor discovery mechanism between nodes, which comprises:

[0015] (1a) After the system is started, the node is first in omnidirectional reception mode, and periodically broadcasts a Hello beacon on the control channel, which contains its own state information and the one-hop neighbor table and observation distance currently maintained;

[0016] (1b) The node measures the signal wavefront arrival direction of the node relative to the node by switching directional sectors where the accuracy of the measurement is limited by the antenna beam width 2 ;

[0017] (1c) Calculate the received signal strength indication :

[0018] (1d) Estimate the relative distance of the node from the node according to the received signal strength indication :

[0019] (1e) Each node takes itself as the origin (0,0), represents the neighbors in local polar coordinates , and converts them into local Cartesian coordinates: .

[0020] Further, the (1) constructing a virtual spline guide rail fitting the physical topology trend by Catmull-Rom interpolation algorithm, comprising:

[0021] (1f) Set the angle threshold according to the topological characteristics of the current area, and perform screening of the hierarchical anchor point sequence according to the angle change between the predecessor node and the successor node of the node :

[0022] (1g) Cascade linearization of local coordinates:

[0023] (1h) Interpolate the limited Catmull-Rom spline to obtain the tangent vector of any point on the path.

[0024] Further, the (4) for the established potential relay node, the screening strategy combined with angle penalty locks the best candidate set, and constructs a forwarding architecture with spatial diversity capability, comprising:

[0025] (4a) Calculate the comprehensive fitness of each potential relay node ,

[0026] (4b) Perform iterative screening on the remaining potential relay nodes, and calculate the diversity correction weight :

[0027] (4c) Select the node with the largest diversity correction weight to join the candidate set until the number of candidate sets reaches the preset value, and obtain the spatial diversity forwarding architecture.

[0028] 2. A self-organizing routing system based on geometric constraints and direction finding cooperation, comprising:

[0029] a topology awareness and relative positioning module for, in a limited space without GPS, establishing and dynamically updating the local relative coordinates of nodes through cooperative direction finding and distance estimation, and geometric error calibration, and providing neighbor topology awareness information for subsequent routing;

[0030] a geometric trajectory generation and guidance module for identifying hard anchors based on corner features and soft anchors based on link quality, generating a smooth virtual guide rail that fits the actual physical topology, and extracting continuous tangent vectors as geometric reference for directional antenna beam transmission using the Catmull-Rom spline interpolation algorithm;

[0031] a directional routing control and tunnel construction module for dynamically determining the optimal beam pointing by weighted fusion of geometric spline tangent vectors and real-time direction finding awareness vectors, and realizing the determination of the receiving node's autonomous deflection angle through the directional RTS carrying geometric parameters, and quickly constructing a controlled interception zone;

[0032] a candidate set screening and hierarchical optimization module for calculating the fitness function of comprehensive link quality, energy, and angle matching degree for potential relay nodes, and introducing a spatial angle penalty mechanism to suppress nodes with similar spatial positions, and greedily selecting a candidate forwarding node set with the maximum spatial diversity gain.

[0033] Compared with the prior art, the present application has the following advantages:

[0034] Firstly, the present application uses the cooperative direction finding and neighbor discovery mechanism between nodes to obtain the relative position information of the current node and the neighbor nodes, and uses the Catmull-Rom interpolation algorithm to construct a virtual spline guide rail that closely fits the actual physical topology, which is used as a global navigation reference in the absolute coordinate-free environment, breaking the dependence of traditional routing on GPS absolute coordinates, and effectively solving the problem of directional addressing failure caused by non-line-of-sight winding paths in a limited space.

[0035] Secondly, the present application forms a directional controlled interception zone to effectively isolate sidelobe interference through a node deflection angle-based role establishment mechanism, and locks the best candidate set for the determined potential relay nodes by combining the angle penalty-based greedy selection strategy, constructs a spatial diversity forwarding architecture with anti-blocking characteristics, significantly improves the anti-interference ability and transmission robustness in high-density networks, and effectively solves the problems of correlation failure and path oscillation caused by node aggregation. BRIEF DESCRIPTION OF DRAWINGS

[0036] Figure 1is the implementation flowchart of the self-organizing routing method based on geometric constraint and direction finding cooperation of the present application;

[0037] Figure 2 is the schematic diagram of the controlled interception area with spatial diversity capability constructed by the method of the present application;

[0038] Figure 3 is the system block diagram of the self-organizing routing system based on geometric constraint and direction finding cooperation of the present application;

[0039] Figure 4 is the comparison diagram of the end-to-end delay simulation results of the present application and the existing method under different loads;

[0040] Figure 5 is the comparison diagram of the network throughput simulation results of the present application and the existing method under different interference environments;

[0041] Figure 6 is the comparison diagram of the packet delivery rate simulation results of the present application and the existing method under different node densities;

[0042] Figure 7 is the comparison diagram of the routing overhead simulation results of the present application and the existing method under different moving speeds. DETAILED DESCRIPTION

[0043] In order for those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the accompanying drawings of the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application but not all. Based on the embodiments in the present application, other embodiments obtained by those skilled in the art without creative labor should be within the protection scope of the present application.

[0044] Embodiment 1, self-organizing routing method based on geometric constraint and direction finding cooperation.

[0045] Reference Figure 1 The implementation steps of the present example include the following:

[0046] Step 1: Relative topology awareness of cooperative direction finding.

[0047] This step aims to construct a high-credibility local relative coordinate in a limited space without GPS signal coverage, and to calibrate the positioning error caused by multipath effect through a cooperative mechanism. Traditional ad hoc network routing often only relies on hop count as a measure, but in a limited space, wireless signals often form non-line-of-sight propagation due to rock wall reflection, resulting in that less hop count does not mean short physical distance, and routing lacking direction awareness is prone to choosing the wrong direction at the branch of the tunnel. Therefore, the present embodiment adopts the following awareness mechanism based on cooperative direction finding and geometric verification, the implementation of which includes:

[0048] (1.1) Neighbor discovery and direction finding:

[0049] (1.1.1) After the system starts, the node is first initialized to an omnidirectional receiving mode to ensure that it can capture the signal from any direction, and the node periodically broadcasts a Hello beacon on the public control channel;

[0050] Unlike the traditional beacon containing only the source node ID and the survival state, the message payload of the Hello beacon in this embodiment also encapsulates the one-hop neighbor table and the observed distance currently maintained by the node. The intention of this design is to enable the receiving node to obtain the topology information of the two-hop neighbor, thereby providing necessary data support for subsequent verification of link quality using triangular geometry in the absence of a global coordinate system.

[0051] (1.1.2) In the signal receiving phase, the node is not in the omnidirectional mode all the time, but performs directional sector switching through the control of the antenna array, that is, when the node receives a broadcast signal, the direction of wavefront arrival is calculated through the MUSIC direction finding algorithm, and the azimuth angle of the neighbor node relative to itself is locked , where the measurement accuracy is limited by the antenna beam width 2 ;

[0052] Most existing positioning schemes in limited spaces only use RSSI measurement, which cannot distinguish whether the node is in the front extension of the tunnel or in the rear area that has passed, while the present application introduces the azimuth angle to accurately determine the direction of the source node.

[0053] (1.1.3) In terms of distance estimation, the node automatically measures the received signal strength indication after receiving a signal from node :

[0054] ,

[0055] where, is the power of the node when transmitting a signal, with the unit of dBm; is the reference point of the path loss model, usually 1m; is the loss value at the reference distance ; is the path loss parameter, in a limited space, due to the waveguide effect of the tunnel, may be smaller than the value in free space, i.e. , while in areas where scattering is serious due to rough rock walls, May be large, in this embodiment, by the early channel sounding dynamic fitting the value; For the shadow fading random variable obeying zero-mean Gaussian distribution;

[0056] (1.1.4) According to the received signal strength indication , combined with the lognormal shadow model to estimate the relative distance of the node to the node :

[0057] ,

[0058] In actual calculation, in order to reduce the random error of single measurement, the average value , the shadow fading term is reduced to close to 0 by multiple measurement average;

[0059] (1.1.5) Each node does not depend on external absolute geographical coordinates, but establishes a local reference coordinate system with itself as the origin, and represents the observation data of neighbor nodes as local polar coordinates , in order to facilitate subsequent geometric spline based routing calculation, the local polar coordinates are converted into local Cartesian coordinates by transformation matrix ;

[0060] Adopting this relative measurement based coordinate construction mechanism avoids deploying expensive and vulnerable absolute coordinate anchors in a limited space; At the same time, compared with the traditional inertial navigation scheme, this method can effectively eliminate the drift error accumulated over time, and can provide stable and high-precision geometric judgment basis for next hop routing selection;

[0061] (1.2) Cooperative error calibration based on two-hop information:

[0062] (1.2.1) Node obtains the two-hop neighbor list of neighbor nodes by analyzing the Hello beacon broadcast in step (1.1.1), when the neighbor list of nodes and has a common neighbor node , the three nodes form a triangular topology in physical space, according to the cosine law, the theoretical value of the unknown side of the triangle is calculated, and it is compared with the actual observation value , and the closure error representing the topology credibility is calculated:

[0063] ,

[0064] wherein, , are the actual ranging values from node to node , ; is the angle between node and node measured by the direction finding algorithm;

[0065] (1.2.2) The system sets a closure error threshold value in advance according to the actual deployment environment , compares the closure error with the threshold value, and judges the state of the link corresponding to the edge :

[0066] If , it is determined that the ranging is distorted due to serious multipath effect or signal blockage, the system will directly mark the link as untrusted and eliminate it, without using traditional smoothing methods such as average filtering for compensation;

[0067] If , it is determined that the link has high credibility. On this basis, in order to improve the estimation accuracy of the relative coordinates of the nodes and avoid the deviation introduced by discarding effective observation values or relying on a single measurement result, the weighted least squares method is further used to optimize and correct the position of the target node;

[0068] (1.2.3) The weighted least squares method is used to optimize and correct the position of the target node:

[0069] The current node first obtains two types of observation data: one is the direct ranging data and direction finding data of the target node by the node itself ; the other is the auxiliary observation data from the common neighbor node ,

[0070] According to the two types of data obtained, the node constructs a residual sum of squares objective function about the coordinates of the target node :

[0071] ,

[0072] wherein, is the total number of observation links participating in the calculation, is the known coordinate of the source node corresponding to the th observation link,​​ the actual ranging value of the first link, the weight factor of the first observed link, which is proportional to the RSSI of the link;

[0073] the optimal coordinate estimation value of the target node coordinate is obtained by solving the objective function minimizing , and the corrected high-precision coordinate is updated to the local neighbor topology table. This step effectively suppresses the influence of measurement noise and improves the robustness of position estimation, providing accurate control points for subsequent construction of Catmull-Rom spline curves.

[0074] This step is designed to construct a virtual guide spline in a restricted environment lacking global coordinate reference, based on the local relative coordinates of each node. The virtual guide path is geometrically smooth and must physically pass through the relay node. The tangent direction of the path is used to guide the beam pointing of the directional antenna, thereby achieving continuous alignment of the communication link.

[0075] Step 2: Construct a constrained virtual guide spline.

[0076] This step aims to construct a virtual guide path that is geometrically smooth and must physically pass through the relay node in a restricted environment lacking global coordinate reference, based on the local relative coordinates of each node. The tangent direction of the path is used to guide the beam pointing of the directional antenna, thereby achieving continuous alignment of the communication link.

[0077] Traditional geometric routing is usually based on the Euclidean distance greedy forwarding strategy. In a restricted environment, the straight path often causes communication interruption due to obstacles, and at corners, it is easy to cause the phenomenon of not being able to find the next hop node that satisfies the distance minimization due to sparse distribution of neighboring nodes.

[0078] Therefore, the embodiment proposes a trajectory planning mechanism based on hierarchical anchor point screening and Catmull-Rom spline interpolation, which includes the following steps:

[0079] (2.1) Screening of hierarchical anchor point sequence:

[0080] In order to screen out nodes that can effectively represent the key geometric features of the topology as anchor points, the embodiment divides the anchor points into hard anchor points and soft anchor points according to the angle change characteristics and link quality indicators of the current region, which is implemented as follows:

[0081] (2.1.1) Set the angle threshold according to the topological characteristics of the current region , and calculate the angle change amount between the predecessor and successor nodes of the node , compare it with the angle threshold to determine the identity of the node:

[0082] If​ , then determine the node which is at the geometric turning point of the path and mark it as a hard anchor point, this kind of node plays a key role in the connectivity of the area, which can ensure the necessary turning of the virtual path at this location, thereby avoiding the interruption of the communication link caused by detouring;

[0083] If , and the node is the last node in the path , and the Euclidean distance between the front and rear neighbor nodes is greater than a preset threshold , then determine that the node is in a long straight corridor, and select the backbone node with the highest link quality SNR as a soft anchor point, wherein can be adjusted according to the single-hop communication distance of the node;

[0084] The routing process of the traditional controlled interception area usually selects a relay node only according to the hop count, ignoring the geometric constraints of the physical environment. The present example introduces a hierarchical anchor point mechanism, which ensures the connectivity of the path at the geometric turning point through the hard anchor point, and the soft anchor point is used to maintain the continuity of the path in the long distance section. Through this mechanism, the system only retains the key nodes as the control points of the spline interpolation, realizes the sparsification of the path representation, and significantly reduces the computational complexity of the subsequent Catmull-Rom spline interpolation.

[0085] (2.2) Cascaded linearization of local coordinates:

[0086] (2.2.1) Let the current node be the origin, and the relative coordinates of the next-hop anchor point be , and the local coordinates of the next-next-hop anchor point relative to be , at this time, is not determined in the reference system of , and needs to be mapped to this reference system through coordinate transformation;

[0087] (2.2.2) Define a 3x3 homogeneous transformation matrix containing a translation component and a rotation component :

[0088] ,

[0089] wherein represents the relative displacement of the next-hop anchor point , and represents the deflection angle of the local reference axis of relative to the reference axis of ;

[0090] (2.2.3) Using homogeneous transformation matrix The anchor point In The coordinate in the reference frame is mapped to The reference frame, and the coordinate in the reference frame is calculated In The coordinate in the reference frame :

[0091] ,

[0092] Wherein, The local coordinate of the anchor point In The reference frame;

[0093] Through cascading changes, the current node Can obtain the control point coordinate sequence in the unified representation of its own reference frame, which includes the predecessor anchor point , the own anchor point , the successor anchor point And the next successor anchor point ;

[0094] (2.2.4) Recursion of the transformation process of step (2.2.3) to calculate the coordinates of the subsequent anchor points, and obtain the multi-hop anchor unified reference frame coordinates;

[0095] This step can effectively solve the problem of limited routing view in traditional ad hoc networks. By unifying the multi-hop anchor coordinates to the reference frame of the current node, the system can obtain the geometric trend of the future path in advance, thereby providing accurate geometric input for subsequent Catmull-Rom spline interpolation and dynamic beam alignment of directional antennas;

[0096] (2.3) Interpolation of the limited Catmull-Rom spline:

[0097] In a wireless multi-hop network, the control points correspond to the actual deployed anchor nodes. If the curve generated by path planning does not accurately pass through these control points, it means that the beam of the directional antenna will point to an area where there is no node, resulting in the inability to relay and forward data, and further causing communication link interruption;

[0098] Therefore, the embodiment selects the Catmull-Rom spline, which is an interpolation curve. Its mathematical properties ensure that the curve must strictly pass through all given control points, unlike the approximate spline such as Bézier or B-spline, thereby ensuring that the planned path is physically realizable and supports continuous multi-hop communication, which is implemented as follows:

[0099] (2.3.1) Calculate the basis matrix under the constraint of Catmull-Rom spline :

[0100] ,

[0101] wherein, is a tension parameter, taking value 0.5, which is used to adjust the bending degree of the curve, so that it will not produce excessive overshoot or shock;

[0102] (2.3.2) Control point sequence based on unified coordinates and base matrix , calculate the smooth curve segment between control points to :

[0103] ,

[0104] wherein, is a normalization parameter, whose continuous change can generate a complete smooth path from to ;

[0105] (2.3.3) For any point coordinate on the path , the first derivative of is taken to obtain the tangent vector of the point on the path :

[0106] ,

[0107] When , the tangent vector corresponds to the initial direction of the path at the current node , which can be used as the optimal beam pointing direction of the directional antenna. This direction not only points to the next hop node , but also implicitly contains the trend component of pointing to the next hop , so as to realize the smooth transition of the beam in multi-hop transmission.

[0108] In contrast, the traditional greedy forwarding routing strategy often leads to a path in the form of a broken line in areas such as alley corners, which will cause a sharp switching of the beam direction. By introducing the derivative direction of the Catmull-Rom spline, the present embodiment effectively suppresses the direction mutation and improves the stability and continuity of the communication link.

[0109] Step 3: Direction-finding assisted directional communication tunnel construction.

[0110] ​​After completing the construction of the virtual guidance path based on Catmull-Rom splines, the system has obtained a theoretically optimal transmission path that conforms to the geometric constraints of physical space. However, due to the time-varying characteristics of the wireless channel and the high node density in the confined space, it may not be able to accurately align with the actual optimal receiving direction of the next hop node. It is necessary to control the signal coverage through directional transmission to avoid disordered flooding.

[0111] This step dynamically adjusts the beam direction of the transmitting node by integrating geometric planning and real-time direction-finding sensing data, and autonomously determines the node's role in the network using signaling interaction carrying geometric constraint information. This allows for the construction of a controlled interception zone in physical space, supported by directional beams and capable of continuous multi-hop transmission. The implementation is as follows:

[0112] (3.1) Weighted fusion of guiding vectors and beamforming:

[0113] (3.1.1) Extract the current node Forward anchor point With subsequent anchor points The relative coordinate difference is normalized to obtain the theoretical guiding vector. :

[0114] ,

[0115] in, For the precursor anchor point At the current node Coordinates in a local reference frame For subsequent anchor points At the current node Coordinates in a local reference frame It is the Euclidean distance between the two;

[0116] This vector represents from point to The unit direction can be used as a rough reference for beam pointing, and is consistent with the direction obtained by spline differentiation. Redundant verification is formed;

[0117] (3.1.2) Based on the neighbor discovery mechanism, within the forwarding target area of ​​the current node, filter out... Find the effective neighbor nodes and extract their respective angle of arrival measurements to form a set. ;

[0118] (3.1.3) on Angle of arrival Calculate its average value As the average azimuth of neighboring nodes within the region, a unit-aware guidance vector is constructed. :

[0119] ,

[0120] in, Within the target area The average arrival direction of each neighboring node;

[0121] It represents the physical direction where the signal energy is most concentrated at the current moment. Its magnitude is 1, and it can be used as a real-time reference for beam pointing. This vector can be used to correct possible deviations in the theoretical path, thereby avoiding communication link interruptions caused by uneven node distribution or dynamic changes.

[0122] (3.1.4) Guide the theoretical vector With perceptual guidance vector Combined into a direction vector matrix, a dual-modal vector set containing both geometric prior information and physically measured information is constructed. :

[0123] ;

[0124] (3.1.5) Utilizing environmental confidence factors Constructing weight vectors :

[0125] ,

[0126] in, Environmental confidence factors;

[0127] (3.1.6) Matrix characteristics and weight vector Perform matrix multiplication, then normalize the result using Euclidean norm to obtain the weighted guiding vector. :

[0128] ,

[0129] in, As a theoretical guiding vector, For sensing guide vectors;

[0130] In engineering implementation, regardless of and Regardless of how the angle changes, the final output is always the guiding vector. This avoids the impact of amplitude fluctuations on beamforming gain and ensures the energy focusing efficiency of directional transmission;

[0131] (3.2) The role of the receiving node is self-established based on the deviation angle:

[0132] After the source node sends out the directional probe signal carrying geometric constraints, in order to determine the spatial position relationship of itself relative to the virtual guiding spline, any receiving node located in the signal coverage range The following geometric solving steps will be performed:

[0133] (3.2.1) Receiving node Measuring the angle of arrival of the directional probe signal from the source node , so as to determine the link direction vector of the receiving node relative to the source node :

[0134] ,

[0135] Wherein, represents the spatial direction from the receiving node pointing to the source node, and the length of the module is 1, and the direction is obtained by taking the negative of the angle of arrival;

[0136] (3.2.2) Receiving node Analyzing the received probe signal, performing signal unpacking operation, and extracting the weighted guiding vector embedded by the source node As a reference benchmark, the vector is obtained from the formula for calculating the transmission guiding vector in step (3.1.6), which integrates the spline geometric trend and real-time perception information, and represents the ideal transmission axis direction expected by the source node at the current time, which is used as the main reference benchmark for geometric judgment in this step;

[0137] (3.2.3) Calculate the geometric deviation angle between the link direction vector and the weighted guiding vector :

[0138] ;

[0139] (3.2.4) Take the obtained geometric deviation angle as the quantitative input of the subsequent role judgment algorithm, which is used to distinguish whether the node is a potential relay node or a blocking node;

[0140] Compared with the traditional geographic routing protocol which relies on the absolute coordinates of the receiving node to calculate the distance from the target path, the embodiment converts the spatial position deviation into an angle deviation by calculating the geometric deviation angle between the link direction vector and the weighted guiding vector , without establishing a global coordinate system, and only relying on the relative geometric relationship between nodes, so as to effectively quantify the degree of deviation of the receiving node from the ideal transmission path, which is suitable for limited space environment lacking global positioning information;

[0141] (3.2.5) Receiving Node After completing the geometric deviation angle After calculation, the deviation angle tolerance threshold is set based on the half-power beamwidth property of the directional antenna. At the same time, the signal-to-noise ratio threshold is set based on the receiving sensitivity of the node. Based on the two parameters mentioned above, the following role determination logic is executed:

[0142] like ,and If so, the receiving node itself will be identified as a potential relay node and will be prepared to participate in subsequent forwarding competition.

[0143] like ,and If the receiving node status is set to ignore, the current probe signal is discarded, and the omnidirectional network allocation vector timer is set to remain silent.

[0144] like ,regardless If so, the receiving node itself will be designated as an obstructing node, an omnidirectional network allocation vector timer will be set, and it will remain silent during subsequent transmission phases.

[0145] This step constructs an isolation zone in the physical space through a decentralized adaptive decision mechanism, which activates nodes near the main axis of the virtual guidance path to participate in relay transmission, while actively suppressing nodes in the antenna sidelobe coverage area. This mechanism can effectively eliminate sidelobe interference in directional communication, reduce the energy consumption of unnecessary nodes, and provide an isolation boundary for spatial reuse, significantly improving network throughput in confined spaces.

[0146] Step 4: Cooperative candidate set locking based on spatial diversity.

[0147] After selecting a set of potential relay nodes that meet both geometric and physical constraints, to improve the robustness of data transmission and avoid communication interruptions due to the failure of a single routing path, it is necessary to further select nodes with reasonable spatial distribution and optimal overall performance from the candidate set, thus constructing a forwarding candidate set with spatial diversity gain. Its implementation includes:

[0148] (4.1) Calculate each potential relay node Overall adaptability The node with the highest overall fitness is selected as the primary candidate node and added to the candidate set.

[0149] ,

[0150] in, , , These are the weight coefficients for the energy term, link quality term, and geometric matching term, respectively. ,

[0151] The current remaining energy of the node. Let be the initial energy of the node. The signal-to-noise ratio of the current link. The system's preset ideal link signal-to-noise ratio reference value. For nodes The geometric deviation angle relative to the virtual spline guide;

[0152] (4.2) Overall fitness The highest-ranking node is selected as the primary candidate node and is added to the final candidate set first. middle;

[0153] (4.3) For any candidate node that has not yet been selected Calculate its diversity correction weight after spatial penalty. :

[0154] ,

[0155] in, The penalty coefficient is... and These are the azimuth angles of the candidate node and the nodes in the selected candidate set, respectively. This is the set of currently selected candidate nodes;

[0156] (4.4) Further fill the candidate set until When the number of nodes reaches the preset maximum value K, the spatial forwarding architecture is obtained, and its iterative process includes:

[0157] (4.4.1) Based on the formula in step (4.3), using the current... For all nodes, update and calculate the diversity correction weights for all remaining potential nodes. ;

[0158] (4.4.2) Select the current Add the largest node to the candidate set. ;

[0159] (4.4.3) Repeat steps (4.4.1) and (4.4.2) to gradually build a spatially balanced candidate set. As the candidate set expands, new penalty factors will be added to subsequent calculations until the candidate set is fully expanded. When the number of nodes reaches a preset scale K, a collaborative forwarding architecture with spatial diversity characteristics is obtained, such as... Figure 2 As shown.

[0160] This architecture can ensure the overall performance of nodes while ensuring that the selected nodes are physically separated from each other through spatial distribution constraints, thereby achieving path diversity.

[0161] It should be noted that the sequence numbers of the steps in the above embodiments and the sequence numbers in the claims are only for the purpose of clearly and completely describing the embodiments of the present invention and for ease of understanding, and their order is not limited.

[0162] Example 2: Self-organizing routing system based on geometric constraints and direction finding coordination

[0163] Reference Figure 3 The system includes: a topology sensing and relative positioning module 1, a geometric trajectory generation and guidance module 2, a directional route control and tunnel construction module 3, and a candidate set screening and hierarchical optimization module 4. The topology sensing and relative positioning module includes a neighbor discovery and direction finding submodule 11, a distance estimation and coordinate construction submodule 12, and a collaborative geometric error calibration submodule 13.

[0164] The working principle of the entire system is as follows:

[0165] The topology sensing and relative positioning module 1 forms the foundation for the entire system's topology environment perception. It constructs the relative geometric topology between nodes in an environment without absolute coordinates, providing high-precision underlying data support for the geometric trajectory generation and guidance module 2. The neighbor discovery and direction finding submodule 11 is used to realize the initial state perception and spatial orientation measurement of the network. Its input is the signal received omnidirectionally by the antenna during the initialization phase. Upon startup, all nodes in the network are in omnidirectional reception mode and periodically broadcast Hello beacons carrying their own state and one-hop neighbor information. By interacting with these beacons, nodes can gradually build and maintain their neighbor lists. Simultaneously, this submodule performs sector switching based on the directional antenna, calculating the precise azimuth angle of each neighbor node relative to its own node by measuring the signal characteristics received in different sectors. ; and the azimuth information The neighbor list maintained by this node is output to the distance estimation and coordinate construction submodule 12; the distance estimation and coordinate construction submodule 12 is used to convert signal strength to geometric distance, and establish a local relative coordinate system based on this, combined with the node azimuth information. The system maintains a neighbor list and signal strength, and uses the current node's own position as the origin to calculate the approximate Euclidean distance between nodes using a log-normal shadowing fading propagation model. This sets the coordinates of each neighboring node in a polar coordinate system with the current node as the reference. Convert to local Cartesian coordinates The system outputs a neighbor list containing the relative coordinates of each neighbor node to the collaborative geometric error calibration submodule 13. The collaborative geometric error calibration submodule 13 is used to perform multi-sided collaborative correction of the relative positions of nodes to eliminate measurement errors introduced by environmental factors such as multipath effects. It uses the neighbor list containing the relative coordinates of each node to identify common neighbor nodes with the target node, and uses the cosine theorem to establish a triangular closure relationship. It detects errors in the direct ranging link, and for links whose errors do not exceed a preset tolerance threshold, it uses weighted least squares to smooth the relative position of the target node, generating a high-confidence neighbor topology. Links exceeding the preset tolerance threshold are discarded. The system then outputs the collaboratively calibrated list containing the corrected relative coordinates of the nodes. The highly reliable neighbor list is output to the geometry trajectory generation and guidance module 2;

[0166] The geometric trajectory generation and guidance module 2 is used to plan and generate a smooth virtual transmission path, which utilizes the corrected relative coordinates of the nodes. Calculate the rate of change of the angle between the lines connecting the nodes. By using the included angle threshold By comparison, hard and soft anchors in the network topology are identified, forming a hierarchical anchor sequence, and then transformed using a homogeneous transformation matrix. The coordinates of multi-hop anchor points distributed in different local coordinate systems are transformed to a unified reference coordinate system. Finally, the Catmull-Rom spline algorithm is used to smoothly fit the anchor point sequence, generating a continuous curve passing through the preset physical relay nodes. By differentiating this curve, the tangent vector at any parameter position on the path is extracted. And output it to the directional routing control and tunnel construction module 3;

[0167] The directional routing control and tunnel construction module 3 is used for path planning and real-time sensing, dynamic control of signal transmission, and spatial isolation, combined with theoretical tangent vectors. Based on the local real-time sensed neighbor direction vectors, and according to the environmental confidence factor By weighting and combining the two factors, the optimal launch direction can be calculated. Node along The receiving node transmits a directional probe message carrying path geometry parameters. By measuring the angle of arrival of the signal corresponding to this message, the receiving node calculates its own geometric deviation angle relative to the virtual spline tangent. Nodes according to It makes autonomous judgments based on preset tolerance thresholds and signal-to-noise ratio thresholds to determine its own role. In the process of dynamically constructing a controlled interception zone, it determines the set of potential relay nodes and outputs them to the candidate set screening and hierarchical optimization module 4.

[0168] The candidate set screening and hierarchical optimization module 4 is used to screen and optimize the set of forwarding nodes with spatial diversity characteristics from the set of potential relay nodes. It comprehensively considers the energy state, SNR and geometric projection position of each candidate node on the spline curve, and performs hierarchical evaluation and sorting of nodes according to the spatial diversity principle. It constructs a cooperative forwarding path with reasonable spatial distribution and redundancy fault tolerance capability, and outputs the finally determined cooperative forwarding node sequence with the maximum spatial diversity gain, thereby forming a data transmission architecture with dual optimization characteristics of geometry and link, and finally determining the set of candidate forwarding nodes.

[0169] It should be noted that the above functional modules can be implemented, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, they can be implemented, in whole or in part, as program instruction products. A program instruction product includes one or a set of program instructions. When the program instructions are loaded and executed on a computer, the described process or function is generated, in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The program instructions can be stored in a computer-readable and writable storage medium, or transferred from one computer's readable and writable storage medium to another.

[0170] In this embodiment, the direct coupling or communication connection between the modules can be achieved through indirect coupling or communication connection via interfaces, devices, or modules. The functional modules and sub-modules in this embodiment can dynamically reside within a single processing unit, or each module can exist physically independently, or two or more modules can dynamically reside within a single processing unit. When these dynamic components are implemented as software functional modules and sold or used as independent products, they can also be stored in a computer-readable and writable storage medium. This storage medium can be a memory, disk, or optical disc, etc.

[0171] The effects of this invention are further illustrated by the following simulation experiments:

[0172] 1. Simulation conditions:

[0173] The hardware platform for the simulation experiment is an Intel i5-6400 CPU with a clock speed of 2.90GHz and 16GB of memory.

[0174] The software platform for the simulation experiment was Windows 10 operating system and MATLAB R2019a.

[0175] The simulation experiment scenario is as follows: The rectangular region is defined; the nodes are set as antenna arrays with beamforming capabilities; the fading environment is set as the channel follows a log-normal shadowing model; and the path loss exponent is set. The interference model is set as the cumulative interference model.

[0176] 2. Simulation Content

[0177] Simulation 1: Under the conditions of the above simulation experiment, single-hop link data transmission under different loads was performed using the present invention and three existing routing algorithms (G-CBR, DCC, and MCBR). The end-to-end delays were compared, and the results are as follows: Figure 4 As shown in the figure, the horizontal axis represents network load, and the vertical axis represents average end-to-end latency.

[0178] The G-CBR algorithm is a traditional controlled interception zone algorithm based on a greedy strategy. It determines the number of hops in a triangle based on control messages, automatically identifies relay nodes and buffer nodes within the controlled interception zone, and realizes autonomous cooperative forwarding within TDMA time slots.

[0179] The DCC algorithm is a routing algorithm based on dynamic cell clustering. It dynamically adjusts the cell size and transmission range according to the local node density, and performs routing by combining geographical location and fuzzy link status, so as to realize infinitely expandable MANET adaptive communication.

[0180] The MCBR algorithm is a controllable interception zone established based on multi-channel broadcast. It extends unicast CBR to multicast, forms the interception zone through the union of multicast trees, and each node independently determines whether it is on the shortest path to any destination, thus realizing distributed multicast routing without a central coordination.

[0181] from Figure 4 As can be seen, as network load continues to increase, the end-to-end latency of existing comparison algorithms rises sharply, and its growth curve shows typical packet queuing and backlog characteristics. In contrast, the latency growth of the present invention remains linear and gradual under the same load conditions, indicating that the present invention can effectively avoid network congestion and has higher throughput and load adaptability.

[0182] Simulation 2: Under the conditions of the above simulation test scenario, the present invention and three existing routing algorithms (G-CBR, DCC, and MCBR) were used to transmit service data in networks under different interference environments. The overall throughput was compared, and the results are as follows: Figure 5 As shown, the horizontal axis represents the link signal-to-noise ratio, and the vertical axis represents the average network throughput.

[0183] from Figure 5 As can be seen, with the improvement of the link signal-to-noise ratio, the network throughput of all comparison schemes shows an upward trend. However, the existing comparison algorithms always maintain a low throughput level in low signal-to-noise ratio environments, while the present invention can still maintain link connectivity under low signal-to-noise ratio conditions. During the process of improving the signal-to-noise ratio, its throughput growth is more significant and always maintains a high level.

[0184] Simulation 3: Under the conditions of the above simulation test scenario, controlled interception zones were constructed using the present invention and three existing routing algorithms (G-CBR, DCC, and MCBR) at different node densities. The received packet delivery rates were compared, and the results are as follows: Figure 6 As shown in the figure, the horizontal axis represents node density, and the vertical axis represents group delivery rate.

[0185] from Figure 6 As can be seen, in the scenario described, as the node density increases, the existing comparison algorithm shows a significant and sharp decrease in data packet delivery rate after the network node density exceeds 20 nodes / km, while the delivery rate of the present invention remains basically stable with the change of node density, demonstrating extremely strong robustness.

[0186] Simulation 4: Under the conditions of the above simulation test scenario, the routing topology was established using the present invention and three existing routing algorithms (G-CBR, DCC, and MCBR) for different node movement speeds. The routing costs were compared, and the results are as follows: Figure 7 As shown in the figure, the horizontal axis represents the node density and movement speed, and the vertical axis represents the routing cost.

[0187] from Figure 7 As can be seen, in the scenario described, the routing overhead of the existing comparative schemes increases significantly with the increase in movement speed, while the overhead curve of the present invention grows the most gently, indicating that its route establishment and maintenance process is least affected by node mobility and has stronger topology dynamic adaptability.

[0188] In summary, compared with existing algorithms, this invention can effectively overcome communication bottlenecks caused by strong electromagnetic interference, high node density access, frequent node movement, and heavy network load in complex communication environments with limited space. It significantly improves the system's network throughput, packet delivery reliability, routing maintenance efficiency, and end-to-end real-time performance. In particular, it demonstrates excellent robustness and adaptability in non-line-of-sight transmission scenarios without GPS support.

Claims

1. A self-organizing routing method based on geometric constraints and direction finding coordination, characterized in that, include: (1) By utilizing the collaborative orientation finding and neighbor discovery mechanism between nodes, the relative position information between the current node and its neighboring nodes is obtained, and a virtual spline guide rail that fits the physical topology is constructed using the Catmull-Rom interpolation algorithm, which serves as the global navigation reference in an environment without absolute coordinates. (2) Extract the tangent vector of the virtual spline guide at the current position as the geometric reference, and perform vector fusion with the real-time aggregation direction based on neighbor signal perception. At the same time, introduce the environmental confidence factor to dynamically adjust the two and generate a weighted guide vector. (3) The source node initiates directional probing based on the weighted guiding vector, the receiving node calculates the probing signal to obtain its own geometric deviation angle relative to the guiding vector, and performs distributed threshold decision based on the deviation angle to establish its own functional role as a potential relay node or blocking node. (4) Based on the establishment of the node role, a directional controlled interception zone is formed to isolate sidelobe interference. For the established potential relay nodes, the best candidate set is locked in combination with the angle penalty screening strategy, and a forwarding architecture with spatial diversity capability is constructed.

2. The method according to claim 1, characterized in that, The process in (1) utilizes a collaborative direction finding and neighbor discovery mechanism between nodes to obtain the relative position information between the current node and its neighboring nodes, which includes: (1a) After the system starts, the node first enters the omnidirectional receiving mode and periodically broadcasts the Hello beacon on the control channel, which contains its own status information as well as the currently maintained one-hop neighbor table and observation distance; (1b) The node measures the node by switching the directional sector. Relative to node Signal wavefront arrival direction The accuracy of the measurement is limited by the antenna beamwidth. ; (1c) Calculate the received signal strength indication : ， in, This refers to the power of signals transmitted by neighboring nodes. The baseline for the path loss model. Reference distance The loss value at that location, For path loss parameters, The shadowed fading random variable follows a Gaussian distribution; (1d) Based on the received signal strength indication Node estimation using a log-normal shading model Relative to node relative distance : ； (1e) Each node Using itself as the origin (0,0), select the neighbors. Represented as local polar coordinates And convert to local Cartesian coordinates: .

3. The method according to claim 1, characterized in that, The virtual spline guide rail that fits the physical topology and orientation is constructed by the Catmull-Rom interpolation algorithm in (1), which includes: (1f) Set the included angle threshold based on the topological characteristics of the current region. According to the node Change in the angle between the predecessor and successor nodes Then, perform hierarchical anchor point sequence filtering: If node Observed changes in the included angle Then determine Located at the corner, marked as a hard anchor point; If node Observed changes in the included angle And the distance between the preceding and following neighbor nodes is greater than a specific Euclidean distance. Then, the backbone node with the best link quality SNR is selected and marked as the soft anchor point, where It can be adjusted according to the single-hop communication distance of the node; (1g) Cascaded linearization of local coordinates: (1g1) Let the current node be... Origin, next jump anchor point The relative coordinates are Next anchor point Compared to The coordinates are ; (1g2) through homogeneous transformation matrix get exist Coordinates in a reference frame : ； in For translation coordinates With rotational components The homogeneous transformation matrix under the following conditions Represents a node relative coordinates, Represents a node The local reference axis relative to the node The deflection angle of the reference axis; (1g3) node Through direct measurement and indirect calculation, a unified coordinate sequence of spline control points centered on themselves is obtained: ; (1h) Interpolation of the restricted Catmull-Rom spline: (1h1) Calculate the basis matrix constants under Catmull-Rom spline constraints. : ， in, This is the tension parameter, with a value of 0.5; (1h2) Unify the coordinate sequence based on spline control points and basis matrix constants compute nodes arrive Smooth path segments between : ， in Indicates from arrive The normalization process; (1h3) Coordinates of any point on the path Find the first derivative to obtain the tangent vector at any point on the path. : 。 4. The method according to claim 1, characterized in that, In step (2), the tangent vector of the virtual spline guide at the current position is extracted as a geometric reference, and it is vector-fused with the real-time aggregation direction based on neighbor signal perception. The implementation includes: (2a) Extract the virtual spline guide at the current position The tangent vector at the point is used to calculate the theoretical guiding vector by normalizing the relative coordinate difference between the two anchor points. : , in, for Predecessor node coordinates for Successor node coordinates, It is the Euclidean distance between the two; (2b) Based on the average angle of arrival of neighboring nodes within the target area obtained by the neighbor discovery mechanism, the perception guidance vector is derived. ,in for Set of the arrival angles of the neighbors The average direction; (2c) Guide the theoretical vector With perceptual guidance vector A bimodal vector set containing both geometric prior information and physical measured information is constructed by uniformly mapping the vectors to the local Cartesian coordinate system of the current node. This serves as the standardized input for the subsequent generation of the weighted guiding vector.

5. The method according to claim 1, characterized in that, The generation of the weighted guiding vector in (2) is implemented by: (2d) Treat the bimodal vector set as a feature matrix: ; (2e) Construct a weight vector using environmental confidence factors: ,in, Environmental confidence factors; (2f) Matrix characteristics and weight vector Perform matrix multiplication, then normalize the result using Euclidean norm to obtain the weighted guiding vector. : ， in, As a theoretical guiding vector, For perception guidance vectors.

6. The method according to claim 1, characterized in that, The receiving node in (3) calculates the probe signal to obtain its own geometric deviation angle relative to the guiding vector, which includes: (3a) Receiving node Measure the angle of arrival of the directional probe signal from the source node. This determines the link direction vector of the receiving node relative to the source node. : ； (3b) Receiving node The received probe signal is parsed, the signaling is unpacked, and the weighted bootstrap vector embedded by the source node is extracted. As a reference benchmark; (3c) Calculate the link direction vector With weighted guiding vector Geometric deviation angle between : 。 7. The method according to claim 1, characterized in that, The distributed threshold decision based on the geometric deviation angle in (3) establishes its own functional role as a potential relay node or blocking node, which includes: (3d) Set the deviation angle tolerance threshold according to the half-power beamwidth property of the directional antenna. And set the signal-to-noise ratio threshold based on the receiving sensitivity of the node. The geometric offset angle of the receiving node and Compare and use the measured signal-to-noise ratio of the current link. and Compare these points to determine the node's identity: like ,and If so, the receiving node itself will be identified as a potential relay node and will be prepared to participate in the subsequent forwarding competition. like ,and If the receiving node status is set to ignore, the current probe signal is discarded, and the omnidirectional network allocation vector timer is set to remain silent. like ,regardless If so, the receiving node itself will be designated as an obstructing node, an omnidirectional network allocation vector timer will be set, and it will remain silent during subsequent transmission phases.

8. The method according to claim 1, characterized in that, In step (4), for the established potential relay nodes, the optimal candidate set is locked by combining the angle penalty screening strategy, and a forwarding architecture with spatial diversity capability is constructed, which includes: (4a) Calculate each potential relay node Overall adaptability The node with the highest overall fitness is selected as the primary candidate node and added to the candidate set. ， in, , , These are the weight coefficients for the energy term, link quality term, and geometric matching term, respectively. , The current remaining energy of the node. Let be the initial energy of the node. The signal-to-noise ratio of the current link. The system's preset ideal link signal-to-noise ratio reference value. For nodes The geometric deviation angle relative to the virtual spline guide; (4b) Perform iterative screening on the remaining potential relay nodes and calculate the diversity correction weights. : ， in, The penalty coefficient is... and These are the azimuth angles of the candidate node and the nodes in the selected candidate set, respectively. This is the set of currently selected candidate nodes; (4c) Select the node with the largest diversity correction weight and add it to the candidate set until the number of candidates reaches the preset value to obtain the spatial diversity forwarding architecture.

9. A self-organizing routing system based on geometric constraints and direction finding coordination, characterized in that, include: Topology sensing and relative positioning modules are used in confined spaces without GPS to achieve direction finding and relative positioning through cooperative inter-node positioning. Distance estimation establishes and dynamically updates the local relative coordinates of nodes, and performs geometric error calibration to provide neighbor topology-aware information for subsequent routing; The geometric trajectory generation and guidance module is used to identify hard anchor points based on corner features and screen soft anchor points based on link quality. It uses the Catmull-Rom spline interpolation algorithm to generate a smooth virtual guide rail that fits the actual physical topology and extracts continuous tangent vectors as the geometric reference for directional antenna beam transmission. The directional routing control and tunnel construction module is used to weightedly fuse geometric spline tangent vectors and real-time direction-finding sensing vectors to dynamically determine the optimal beam pointing, and to determine the autonomous deviation angle of the receiving node through a directional RTS carrying geometric parameters, thereby quickly constructing a controlled interception zone. The candidate set selection and hierarchical optimization module is used to calculate the fitness function of comprehensive link quality, energy and angle matching degree for potential relay nodes, and introduces a spatial angle penalty mechanism to suppress nodes with similar spatial locations, and greedily selects the set of candidate forwarding nodes with the maximum spatial diversity gain.

10. The system according to claim 9, characterized in that, The topology sensing and relative positioning module includes: The Neighbor Discovery and Direction Finding submodule is used to control the switching of nodes between omnidirectional receiving mode and directional sector scanning mode, periodically broadcast Hello beacons containing its own status, and measure the arrival direction of the signal wavefront of neighboring nodes to obtain the relative azimuth angle. The distance estimation and coordinate construction submodule is used to inversely calculate the relative distance between nodes based on the log-normal shadowing model using the received signal strength indication, and to establish a local coordinate system with itself as the origin and convert it into local Cartesian coordinates by combining azimuth data. The collaborative geometric error calibration submodule is used to calculate the geometric closure error based on two-hop common neighbor information to identify and eliminate links severely affected by multipath interference, and to use the weighted least squares method to smoothly correct and update the relative position coordinates of neighboring nodes.

Images