Urban internet of things communication distributed access precision regulation system based on edge collaboration

Abstract

The application discloses an urban Internet of Things communication distributed access precision regulation system based on edge cooperation and belongs to the technical field of communication regulation. Theoretical deviation fields are corrected by confidence weighting through the fusion of node mutual listening data and mobile terminal reporting data, the success / failure records of terminal access are taken as feedback true values, refined deviation fields with spatial continuity and physical authenticity are output, and through weighted Voronoi partition, overlapping grids are uniquely attributed to optimal nodes, eliminating the interference accumulation caused by multiple node repeated coverage. For hollow grids, directional enhancement is realized through antenna electric downtilt angle adjustment and sector power compensation mask, and donor-relay two-hop forwarding links are constructed for stubborn hollows. Based on local responsibility attribution, the rapid judgment and contract forwarding of access requests are realized, and the bottleneck of centralized processing is eliminated. Through real-time monitoring of the access failure rate of hollow grids, incremental detection and local contract regeneration of invalid areas are automatically triggered.

Description

Technical Field

[0001] This invention relates to the field of communication control technology, specifically to a precise control system for distributed access to urban IoT communication based on edge collaboration. Background Technology

[0002] Urban IoT edge nodes are often installed using existing facilities such as utility poles, traffic lights, and building facades. Due to limitations in space, power supply, and municipal permits, node locations often cannot achieve an ideal cellular layout. This can lead to overlapping coverage and mutual interference among multiple edge nodes in some areas, as well as coverage gaps in other areas. Distributed access control algorithms are typically optimized based on the node's own perception of terminal distribution and channel occupancy. However, if the node coverage does not match the design assumptions, the optimal resource allocation calculated by the algorithm may not be consistent with the actual terminal reachability. For example, if a terminal is in an area where the coverage of two nodes overlaps, both nodes may allocate resources to it simultaneously, resulting in duplicate scheduling. Terminals in areas with coverage gaps are left unmanaged, leading to a high access failure rate. This results in defects such as unreasonable edge node deployment leading to coverage overlap and gaps, and a mismatch between access control results and the actual wireless environment. Summary of the Invention

[0003] The purpose of this invention is to provide a precise control system for distributed access of urban Internet of Things communication based on edge collaboration, in order to solve the technical problems mentioned in the background art.

[0004] The objective of this invention can be achieved through the following technical solutions: A city-wide IoT communication distributed access precision control system based on edge collaboration includes: Coverage deviation field generation and compensation output module: Each edge node periodically broadcasts a detection reference frame carrying the node identifier and transmission power level, and at the same time collects the received power of the detection reference frames of adjacent edge nodes, constructs an initial measured coverage reachability matrix in the node × grid dimension, and compares it with the theoretical coverage matrix grid by grid to generate the original coverage deviation field. The original coverage deviation field is subjected to deviation field compensation calculation, and the compensated refined coverage deviation field is output. Coverage responsibility area dynamic reshaping module: Based on the generated refined coverage deviation field and measured coverage reachability matrix, the coverage responsibility area is reshaped: For the set of overlapping grids, a weighted generator is constructed based on the measured received power of each candidate edge node in the grid and the margin relative to the lowest available power of the terminal. Through weighted Voronoi partitioning, each overlapping grid is uniquely assigned to the edge node with the largest margin, forming a responsibility assignment table; For the set of hole grids, the donor node with the strongest measured signal to the hole grid is selected, and an antenna electrical downtilt adjustment amount and sector power compensation mask are generated for it to expand its coverage boundary. A relay node is assigned to the hole grid that is still unreachable after expansion, and a two-hop forwarding link is constructed from the donor node through the relay node to the hole grid terminal. Distributed access control and execution module: When a terminal access request arrives, the edge node verifies whether the grid where the requesting terminal is located belongs to the scope of its responsibility attribution table. If it does, access resources are allocated; otherwise, the request is forwarded to the responsible node according to the contract. When a terminal in an overlapping area simultaneously initiates access to two edge nodes, the responsibility token arbitration mechanism is triggered. Only the node holding the responsibility token for that grid performs resource allocation, while the other node reclaims the reserved resources. The access failure rate of each hole grid is statistically analyzed in real time, and the coverage deviation field is dynamically updated based on the analysis results.

[0005] Furthermore, while the detection reference frame is being broadcast, a parallel acquisition process for two types of sensing data is initiated: the first type of sensing data is data from mutual listening between nodes, and the second type of sensing data is data reported by mobile IoT terminals; the two types of discrete raw data are then mapped to a unified spatial reference frame.

[0006] Furthermore, when performing deviation field compensation calculation on the original coverage deviation field, the actual terminal access success / failure records of each grid in the previous control cycle are extracted as feedback truth values. The prediction-actual state inconsistency rate of each grid is calculated, and the confidence decay factor is used to perform weighted correction on the deviation values ​​of the corresponding grids in the original coverage deviation field. Grids that have been disproven by feedback truth values ​​for multiple consecutive cycles are marked as environmental change anchor points. Spatial interpolation compensation is performed on the neighboring low-confidence grids using the environmental change anchor points, and the compensated refined coverage deviation field is output.

[0007] Furthermore, the inconsistency rate between the predicted and actual states at grid g for edge node n. The calculation expression is: ;in, The total number of terminal attempts to access edge node n in grid g during the previous control cycle; This represents the number of times the connection failed. `g` is a preset regularization variable; `g` is a raster index variable, `g=1,2,…,G`, where `G` is the total number of raster cells; `n` is an edge node index variable, `n=1,2,…`. , This represents the total number of edge nodes within the deployment area.

[0008] Furthermore, for each grid g in the overlapping grid set, a weighted Voronoi partition is performed, and each overlapping grid is uniquely assigned to the edge node with the largest power margin. The weighted generator is defined as the average power margin of the edge node n across all overlapping grids. For any overlapping grid, define its weighted distance to edge node n, and select the edge node with the smallest weighted distance as the sole responsible node of that grid.

[0009] Furthermore, when performing consistency checks on the generated responsibility attribution table, a load balancing factor is defined. Let be the number of overlapping grid cells assigned to edge node n in the responsibility attribution table; if there exists any node n that satisfies This triggers the conflict resolution process; This is the preset maximum number of responsible grid cells per node.

[0010] Furthermore, for each grid cell in the identified set of void grid cells, donor nodes that can expand coverage to fill the void are selected; for the selected donor nodes, an antenna electrical downtilt adjustment is generated to expand their coverage boundaries.

[0011] Furthermore, a sector power compensation mask is generated for the donor node to directionally enhance the transmission power toward the hole grid. For the hole grid that still cannot be directly covered by the donor node after the previous steps, a two-hop forwarding link is constructed from the donor node through the relay node to the hole grid terminal.

[0012] Furthermore, when located at geographical coordinates When a target terminal initiates an access request to edge node n, node n first calculates the grid index where the terminal is located based on the Global Navigation Satellite System location information reported by the terminal. ; Obtain the raster index Then, edge node n performs attribution verification: querying the local responsible raster list. ,judge If the condition is met, it indicates that the grid where the terminal is located falls within the responsibility scope of this node, and node n has the right to allocate access resources to the terminal; if the condition is not met, the subsequent forwarding process will proceed.

[0013] Furthermore, when the access failure rate is greater than the access failure rate threshold, the corresponding grid is determined to be a failed grid. For the grids determined to be failed, the failed areas are further marked, and the set of failed areas is defined as the union of all failed grids and their spatial neighboring grids. Upon receiving a failed area marker, incremental detection sampling is triggered, updating only the failed area and not the entire city area.

[0014] Compared to existing solutions, the beneficial effects achieved by this invention are: This invention achieves automated and large-scale sensing with city-level coverage by integrating node mutual listening data and mobile terminal reported data, eliminating the need for manual drive testing. Using terminal access success / failure records as feedback truth values, the theoretical deviation field is corrected with confidence weighting, enabling the model to adaptively track dynamic environmental changes and output a refined deviation field with spatial continuity and physical authenticity. This solves the misjudgment problem caused by the mismatch between the theoretical model and the actual environment from the source.

[0015] By using weighted Voronoi subdivision, overlapping grids are uniquely assigned to the optimal node, eliminating the interference accumulation caused by multiple nodes overlapping coverage. For hole grids, directional enhancement is achieved through antenna electrical downtilt adjustment and sector power compensation masking. Furthermore, a donor-relay two-hop forwarding link is constructed for persistent holes to achieve coverage filling with minimal hardware modifications.

[0016] The system achieves rapid decision-making and contract forwarding of access requests based on local responsibility attribution, eliminating the bottleneck of centralized processing; the responsibility token arbitration mechanism fundamentally solves the resource conflict of concurrent access in overlapping areas; by monitoring the access failure rate of void grids in real time, it automatically triggers incremental detection and local contract regeneration of failed areas, forming a closed-loop self-healing capability of perception, reshaping, regulation, and re-perception, which can quickly respond to environmental changes without human intervention, significantly improving the system's robustness and intelligent operation and maintenance level in dynamic urban environments. Attached Figure Description

[0017] The invention will now be further described with reference to the accompanying drawings.

[0018] Figure 1 This is a flowchart illustrating the operation of the edge-collaboration-based urban IoT communication distributed access precision control system of the present invention.

[0019] Figure 2 This is a flowchart illustrating the process of generating the refined coverage deviation field in this invention.

[0020] Figure 3 This is a flowchart of the donor node selection and adjustment process in this invention.

[0021] Figure 4 This is a flowchart illustrating the process of updating the faulty region marker in this invention. Detailed Implementation

[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] like Figure 1 As shown, this invention is a city IoT communication distributed access precision control system based on edge collaboration, comprising: Coverage deviation field generation and compensation output module: Each edge node periodically broadcasts a detection reference frame carrying the node identifier and transmission power level, and at the same time collects the received power of the detection reference frames of adjacent edge nodes, constructs an initial measured coverage reachability matrix in the node × grid dimension, and compares it with the theoretical coverage matrix grid by grid to generate the original coverage deviation field. The original coverage deviation field is subjected to deviation field compensation calculation, and the compensated refined coverage deviation field is output. When implementing the periodic broadcasting of the detection reference frame and the acquisition of multi-source sensing data, each edge node follows the system's preset detection period. The detection period is measured in seconds, and in this embodiment, it is preferably set to 0.1 seconds, corresponding to a broadcast frequency of 10Hz. Detection reference frames are periodically broadcast on the specified air interface resources. The physical layer payload of each detection reference frame carries the following mandatory fields: a globally unique node identifier. and the current transmit power level indication value of the node. The unit is dBm; the node identifier is used to distinguish the signal source, and the transmit power level is used for subsequent path loss calculation; While the reference frame is being broadcast, a parallel acquisition process for two types of sensing data is initiated: The first type of sensing data is the data from mutual listening between nodes: each edge node n, n=1, 2, ... , To determine the total number of edge nodes within the deployment area, the receiver is used to scan the air interface channel in the transmission time slot of a node other than its own to listen for signals from other neighboring edge nodes. The broadcast probe reference frame was used to measure the received power of its reference signal. ;definition For node n to receive node The received power of the reference signal in the probe reference frame, in dBm; this measurement directly reflects the attenuation of the wireless propagation path between the two nodes. The second type of sensing data is data reported by mobile IoT terminals: The system continuously receives location marker signal strength samples reported by mobile IoT terminals installed on urban mobile carriers such as city buses, taxis, and sanitation vehicles through the core network interface. Each sample, reported at time t, contains the following triplet information: the terminal's current geographical location coordinates. The received power is provided by the terminal's built-in Global Navigation Satellite System receiver and the downlink reference signal measured by the terminal from edge node n. The unit is dBm; all reported samples are initially cached according to timestamp and spatial coordinates to form the original sample pool; It should be noted that this step is used to establish a unified benchmark for signal broadcasting and acquisition. Without relying on manual drive testing, it actively stimulates and captures the response of the wireless environment, providing raw data for subsequent coverage analysis. The beneficial effect is that by using the existing network infrastructure and existing mobile terminals as sensing nodes, it realizes automated, low-cost, and large-scale acquisition of coverage data, avoiding the drawbacks of traditional drive testing schemes that are time-consuming, labor-intensive, and have sparse data.

[0024] When mapping the two types of discrete raw data to a unified spatial reference frame, the target city area is divided into G uniform geographic grids according to a preset spatial resolution, where G is the total number of grids, each grid g covers a square area, and g is the grid index variable, g=1, 2, ..., G, with a side length of... The unit is meters, and in this embodiment, it is preferably set to 50 meters; the geometric center coordinates of each grid are used. , which serves as the reference position for this grid; For the second type of sensing data, since the location reported by the mobile terminal is discrete and random, with the geometric center of grid g as the center, Define a spatially related neighborhood for the radius; where, The signal strength measurements reported by all terminals within this neighborhood from edge node n. Let q be the index variable of valid samples that fall within the neighborhood of grid g and report the measurement value of node n, where q = 1, 2, ... , Let n be the total number of valid samples that fall within the neighborhood of grid g and report the measurement value of node n; perform robust statistics on this set based on the median of signal strength, and define the benchmark value of the measured signal strength from edge node n at grid g. Its expression is: ; where median{} represents the operation of taking the median of the set; the logic of using the median instead of the arithmetic mean is that the median has a natural robustness to outliers and can effectively suppress abnormally low values ​​caused by the terminal being in a state of deep fading, such as when a vehicle passes through a tunnel entrance or is blocked by a large billboard, so that the signal strength benchmark value mapped to the grid is more stable and more representative of the typical coverage level of the grid.

[0025] For the first type of sensing data, since inter-node listening is a discrete measurement occurring at fixed node coordinates, it needs to be mapped to the raster level through spatial interpolation; the spatial coordinates of node n are defined as follows: Based on all Location of each node and mutual listening measurement value Using the inverse distance weighted interpolation algorithm, the equivalent received signal strength indication value at each grid point g relative to the edge node n is calculated. Among them, the nodes participating in interpolation weights It is inversely proportional to spatial distance, and the specific expression is: ;in, Represents a node The Euclidean distance from the spatial coordinates to the geometric center coordinates of the grid g; The index variable is the reciprocal of the node distance; the denominator is the sum of the reciprocals of the distances to all nodes, ensuring that the sum of all weights is 1; the design logic of weights being inversely proportional to distance is based on the attenuation characteristic of spatial correlation in wireless propagation: the closer the nodes are, the better their mutual listening results reflect the propagation characteristics of that location, thus giving them higher interpolation weights. The calculation expression is: ; This value represents the expected received signal strength of the virtual receiver at grid g for edge node n, which is derived from the inter-node listening data.

[0026] It should be noted that this step is used to unify spatially discrete and temporally asynchronous heterogeneous raw data into a standardized raster coordinate framework, solving the fundamental problem of spatial alignment of multi-source data. The beneficial effects achieved are: by using median statistics and inverse distance weighted interpolation, the anti-interference capability of terminal-reported data and the spatial continuity of node-to-node data are enhanced, respectively. This provides spatially aligned and physically meaningful input tensors for subsequent matrix construction, avoiding the spatial mismatch and computational complexity explosion problems caused by directly using discrete points.

[0027] Based on the two types of raster-level measurements obtained, when constructing a unified measured coverage reachability matrix, an initial measured coverage reachability matrix is ​​constructed. The dimension of this matrix is Row index corresponds to edge node n, column index corresponds to raster g; matrix elements This is a Boolean variable, representing whether the signal of the edge node n can effectively cover the grid g at the level of pure measured data. Constructing a fusion judgment threshold The unit is dBm. This threshold is preset by the system based on the receiver sensitivity and demodulation signal-to-noise ratio requirements. In this embodiment, based on the sensitivity specifications of typical IoT communication modules, it is preferably set to [value missing]. Define the weighted fusion value of the two types of measured evidence. Its expression is: ;in, and The preset fusion weight coefficients satisfy... , and 0 < , <1; This embodiment preferably sets =0.7, =0.3, meaning a higher confidence level is given to the measured signal strength directly reported by the terminal. The logic is as follows: the data reported by the terminal comes from actual user experience and is a direct reflection of the air interface quality; while the data from mutual listening between nodes is an indirect inference, although it can cover sparse areas of the terminal, its confidence level is slightly lower; based on the fusion decision threshold, the initial measured coverage can reach the matrix elements. The assignment logic is as follows: ; when When, it indicates that the measured data supports edge node n to form effective coverage of grid g; when When the time is specified, it indicates that the measured data shows that the node has failed to provide effective coverage of the grid.

[0028] It should be noted that this step is used to merge two types of grid-level measurements with different physical meanings and confidence levels into a unified Boolean coverage determination, forming an objective and comprehensive measured description of the current wireless coverage status. The beneficial effects are as follows: through the weighted fusion strategy, the complementary advantages of dual-source data are effectively utilized, terminal data is guaranteed to be true and reliable, and mutual listening data fills in the sparse gaps, avoiding the problem of inaccurate evaluation in areas with insufficient samples due to a single data source; the Boolean output directly meets the subsequent comparison requirements with the theoretical matrix, converting continuous measurements into a decisionable coverage status and unifying the comparative metric.

[0029] like Figure 2 As shown, the theoretical coverage matrix is ​​derived based on the installation coordinates of the edge nodes and the designed antenna pattern, and compared grid-by-grid with the measured coverage matrix. When generating the deviation field, the theoretical coverage matrix is ​​derived. Their dimensions are the same For edge node n and grid g, the theoretical received power is calculated based on the free space path loss model and the antenna gain pattern. Its expression is: ;in, The transmit power of node n, in dBm, is obtained by mapping the broadcast power level indication value; Let be the gain of the antenna at node n in the direction of grid g, expressed in dBi. This gain value is obtained by two-dimensional linear interpolation of the designed antenna pattern function at node n. and These are the azimuth and elevation angles of the grid g relative to the node n, respectively, calculated from spatial geometric relationships; is the horizontal distance from node n to the center of grid g, in meters; The path loss is expressed in dB, using the logarithmic distance path loss model commonly used in urban macrocell environments: ;in, For reference distance Average path loss at that location, The value in meters is given by the system calibration under a standard test environment; in this embodiment, it is taken as... dB; The path loss index is taken as 3.5 to 4.5 for urban environments; this embodiment uses 4.5. ; To conform to a mean of 0 and a standard deviation of In the theoretical derivation stage, the log-normal shadowed fading random variable is taken as... That is, only the deterministic path loss component is calculated. The logic is that the theoretical matrix should reflect stable and reproducible physical laws, rather than random fluctuations, so that the subsequent deviation field can purely reflect the systematic deviation between the measured and the theoretical.

[0030] based on Theoretically covering matrix elements The assignment logic is consistent with the previous steps, that is: ; Subsequently, the initial measured coverage reachability matrix was... With theoretical coverage matrix Element-by-element comparison within the same grid space generates the original coverage deviation field. Original Coverage Deviation Field The dimension is Its elements The deviation value is a real number and is defined as follows: ;in, For theoretical power derivation, This is the measured fusion power; when A value greater than 0 indicates that the theoretically predicted coverage strength is higher than the measured value, meaning the theoretical model is overly optimistic and there are unmodeled additional losses; when... When the value is less than 0, it indicates that the measured signal strength exceeds the theoretical prediction, meaning that there are unmodeled favorable propagation conditions, such as reflection or waveguide effects. The calculation logic of this deviation value is that it directly quantifies the systematic difference between the theoretical model and the actual environment, rather than random measurement error.

[0031] It should be noted that this step is based on the deterministic physical model to extrapolate the theoretical coverage of the expected results and to make a quantitative comparison with the measured perception to generate a spatially distributed deviation field, which transforms environmental uncertainty into a quantifiable deviation value. The beneficial effect achieved is that the original deviation field describes the systematic deviation pattern of the theoretical model in the complex urban environment. This deviation is not a simple error, but contains implicit information of comprehensive environmental characteristics such as building shading, multipath reflection, and vegetation attenuation. It provides a spatially continuous data base for subsequent refinement and compensation, and fundamentally solves the core contradiction that pure theoretical models are difficult to adapt to complex dynamic environments.

[0032] When introducing actual terminal access behavior as feedback truth value to objectively quantify the confidence level of the current coverage prediction, the system extracts the previous control cycle. Records of successful and failed terminal accesses in each grid cell; control cycle. The unit is seconds, determined by system configuration, with a typical value of 60 seconds; for each grid g, the behavior of all terminals attempting to access edge node n within this period is counted, and the following definition is used: Defined as the total number of terminal attempts to access edge node n in grid g during the previous control cycle. Let g be the number of access failures, then the inconsistency rate between the predicted and actual states for edge node n at grid g is... The calculation expression is: ;in, As a preset small amount of regularization, this embodiment takes Used to prevent when At that time, a division by zero exception occurred, which also caused the grid where no access attempts were made to... Approaching 0, meaning no confidence decay is introduced; the logic of this expression is based on a fundamental fact: terminal access failure is the most direct and reliable evidence of insufficient coverage or deteriorated link quality; numerator The denominator reflects the absolute number of actual failures. Given the total number of attempts, the ratio of the total number of attempts to the failure rate is the inconsistency rate. This directly reflects the accuracy of the prediction of the link quality between the terminal and node n in grid g based on the current coverage matrix: The higher the value, the less reliable the predictions made by the theoretical or experimental model at that grid point.

[0033] It should be noted that this step utilizes the inherent access behavior of existing network terminals as a high-confidence feedback signal to quantitatively evaluate the credibility of the prediction model at each grid-node dimension. The beneficial effects are: a large amount of real user experience data can be obtained without additional measurement overhead, and the statistical granularity is fine down to the node × grid dimension, which gives the confidence assessment a high degree of spatial resolution. It can accurately identify which regions and nodes have prediction models with systematic biases, providing a quantitative basis based on real data for subsequent weighted correction, and transforming the optimization decision from open-loop model-driven to closed-loop data-driven.

[0034] Using the statistical inconsistency rate between prediction and actual state as a confidence decay factor, the original bias field is weighted and corrected. Simultaneously, when environmental abrupt changes that are consistently falsified by feedback truth values ​​are identified as anchor points, the corrected bias value is defined. Its expression is: The logic of this modified expression is: when When the value is 1, it indicates that the prediction of the previous adjustment cycle was completely accurate, and the model is highly reliable at this grid point. In this case, the original deviation value multiplied by the coefficient 1 is completely preserved. When the value increases, it indicates a larger deviation between the prediction and the actual result, and a lower model confidence level. In this case, the contribution of the original deviation value is proportionally suppressed. When the value approaches 1, the original deviation value approaches zero, meaning the original deviation information of that grid is completely discarded. This multiplicative correction method, rather than a simple threshold truncation, ensures the spatial smoothness of the confidence transition and avoids discontinuous jumps in deviation values ​​between adjacent grids caused by hard decisions. At the same time, define continuous observation windows. In this embodiment, the preset is 5 consecutive control cycles. For grid g and edge node n, if there are in each control cycle within the observation window... , The preset falsification threshold is preferably set to [value] in this embodiment. If the prediction-to-actual inconsistency rate remains consistently high and exceeds 80%, then the raster g is marked as an environmental mutation anchor point along the dimension of node n; define the environmental mutation anchor point indicator variable. Its assignment logic is as follows: ; when When the time dimension is defined, it indicates that the grid has persistent environmental change factors that are not captured by the existing model in the coverage prediction of the corresponding node, such as the occlusion of newly built high-rise buildings or temporary construction barriers. These grids are regarded as reliable feedback anchors with high confidence. The logic of this time dimension cumulative decision is to prevent misjudgment caused by instantaneous high inconsistency rate due to occasional events (such as a single vehicle occlusion). Only persistent high inconsistency rate verified by multiple cycles is identified as a real environmental change.

[0035] It should be noted that this step adaptively weights and corrects the original bias field using a confidence decay factor, while simultaneously identifying anchor grids that have experienced real environmental mutations over time. The beneficial effects are twofold: firstly, it prevents unreliable bias information from contaminating subsequent optimization decisions, thus suppressing the spatial spread of erroneous biases; secondly, the marking of environmental mutation anchors provides high-confidence spatial reference points for subsequent spatial interpolation, ensuring that the refinement compensation has reliable data source support. Furthermore, the continuous multi-cycle falsification conditions ensure the robustness of the anchor markings, avoiding system oscillations caused by frequent false triggers.

[0036] Using labeled environmental mutation anchors Spatial interpolation compensation is performed on neighboring low-confidence gratings to generate the final refined coverage bias field. At that time, the interpolation operation is performed independently for each edge node n. Assume that for node n, there are a total of Each raster is marked as an environmental mutation anchor point, and its raster index set is denoted as . The corrected deviation values ​​at these anchor grid points It is considered as a precisely known interpolation reference point; for any non-anchor raster Its refinement deviation value It is calculated using the following weighted interpolation expression based on spatial correlation: ;in, Let be the interpolation weight coefficient for the k-th anchor point relative to the target grid g. The principle for determining this weight is that anchor points closer to each other receive higher weights, and the sum of all weights is 1. The specific expression is: ;in, For the k-th anchor point grid The geometrical distance to the target grid g, in meters; The preset spatial length parameters are in meters; in this embodiment, they are preferably set to... The meter is used to control the smoothness of interpolation. The logic behind the weight expression using the Gaussian radial basis function is that the correlation of the spatial propagation environment decreases exponentially with increasing distance. The closer the anchor point is, the more similar the propagation environment (occlusion, reflection, diffraction, etc.) it experiences. Therefore, its deviation value has higher reference value for the target grid and should be given greater weight. This determined a smoothing radius equivalent to the building scale in an urban environment, corresponding to a 200-meter influence range. This effectively fills the information gaps in low-confidence gratings without over-smoothing and erasing local propagation heterogeneity. In the special case where there are no anchor points, simply let That is, the degenerates to the corrected deviation value without interpolation adjustment. This degenerate logic ensures that the system can still work in scenarios where high confidence feedback is lacking. For the grid cells marked as anchor points, i.e. The corrected deviation value is directly retained. After traversing all edge nodes n and all grids g, the complete refined coverage deviation field matrix is ​​obtained. Its dimensions are .

[0037] It should be noted that this step utilizes anchor points as a high-confidence spatial benchmark. Through weighted interpolation based on spatial correlation, the discrete anchor point information is expanded into a continuous bias field covering the entire raster. The beneficial effects are: it enables rasteres that do not have enough terminal access samples to obtain bias compensation knowledge from neighboring anchor points, significantly improving the spatial coverage and continuity of the refined bias field and avoiding optimization blind spots caused by sample sparsity. At the same time, the form of the Gaussian kernel function ensures the spatial smoothness and physical rationality of the interpolation field, making the refined bias field both rooted in the anchor point truth and conforming to the basic physical laws of continuous change in the propagation environment.

[0038] Based on the refined coverage deviation field of the output When correcting the equivalent coverage determination result of edge node n to grid g, the coverage indicator value after refinement and compensation is defined. Its expression is: ;in, For the defined measured fusion power, To refine the deviation value, the sum of the two is equivalent to the final equivalent received power estimate after deviation correction. The judgment logic is as follows: the fused measured power reflects the actual perceived signal strength, and the refined deviation value reflects the systematic bias of the theoretical model relative to the measured value. The sum of the two is equivalent to generating a comprehensive coverage estimate that is both rooted in real-time sensing data and incorporates historical feedback knowledge, based on the measured value and compensated by the deviation. This estimate is closer to the real long-term propagation conditions than pure measured or pure theoretical values. Based on the revised coverage indicator value Perform the following set identification: Overlapping grid sets A grid is defined as the set of grids where at least two distinct edge nodes simultaneously satisfy the coverage condition, and its mathematical description is: This set represents the overlapping coverage area where the measured signals of multiple edge nodes simultaneously meet the standards, and is a key candidate area for subsequent collaborative merging or interference coordination optimization. Hollow grid set , defined as the set of graticles where all edge nodes have a coverage indicator value of 0, and its mathematical description is: This set represents coverage gaps where the measured signals from all nodes are substandard. These are target areas where coverage enhancement is needed by adjusting node transmit power, beam direction, or adding supplementary nodes.

[0039] It should be noted that this step is used to quantize the continuous refinement deviation field into a discrete set of coverage states with clear physical meaning, providing direct input for spatial partitioning in subsequent optimization modules. The beneficial effects are as follows: the clear distinction between overlapping and holed grids enables the system to accurately locate areas that "need coordination" and "need blind patching," providing a spatial partitioning basis for the formulation of differentiated optimization strategies; at the same time, converting the complex continuous deviation field into Boolean indicator values ​​significantly reduces the data complexity of cross-module interfaces, allowing subsequent coherent spectrum clustering discriminators, logical port dynamic mapping modules, etc., to make decisions directly based on clear coverage states, thus avoiding interference deterioration caused by forced merging in areas that should not be merged.

[0040] In this embodiment, by constructing a city-level coverage perception capability independent of traditional drive testing, an initial coverage deviation field combining measured and theoretical data is established using massive mobile terminal reported data and inter-node listening data. Furthermore, by using terminal access success / failure records as feedback truth values ​​and the prediction-actual inconsistency rate as a confidence decay factor, the deviation field is adaptively weighted and corrected, enabling the model to track dynamic changes in the urban environment. Long-term falsified graticules are marked as environmental mutation anchor points, and spatial interpolation is used to transmit high-confidence information from these anchor points to neighboring low-confidence graticules, ultimately outputting a refined coverage deviation field and a set of overlapping / hole graticules. This step provides a high-fidelity spatial input basis for subsequent coherent spectrum clustering discrimination, logical port dynamic mapping, and delay slope tracking compensation, fundamentally avoiding the systemic defect where coverage enhancement caused by the mismatch between the theoretical model and the real environment is misjudged as interference enhancement. This significantly improves the system's coverage optimization efficiency and engineering deployment robustness in complex and dynamic urban environments.

[0041] Coverage responsibility area dynamic reshaping module: Based on the generated refined coverage deviation field and measured coverage reachability matrix, the coverage responsibility area is reshaped: For the set of overlapping grids, a weighted generator is constructed based on the measured received power of each candidate edge node in the grid and the margin relative to the lowest available power of the terminal. Through weighted Voronoi partitioning, each overlapping grid is uniquely assigned to the edge node with the largest margin, forming a responsibility assignment table; For the set of hole grids, the donor node with the strongest measured signal to the hole grid is selected, and an antenna electrical downtilt adjustment amount and sector power compensation mask are generated for it to expand its coverage boundary. A relay node is assigned to the hole grid that is still unreachable after expansion, and a two-hop forwarding link is constructed from the donor node through the relay node to the hole grid terminal. To ensure data consistency, the following preprocessing operations are performed: For any edge node n and any raster g, if If the actual measurement shows that the node does not effectively cover the grid, then the corresponding refinement deviation value will be forcibly adjusted. Setting it to 0 is logical because the correction effect of the deviation field should only be effective within areas where the actual coverage has been confirmed, avoiding the incorrect application of deviation compensation to links that do not meet the coverage conditions, and preventing false coverage estimates; after preprocessing, the effective coverage estimate value is defined. Its expression is: ; The purpose of this preprocessing step is to align the refined bias field with the measured coverage reachability matrix in terms of physical meaning, ensuring that the bias correction only applies to verified links. The beneficial effect is that it avoids the bias field correction from misclassifying non-existent coverage links as reachable, thus ensuring the physical authenticity of subsequent responsibility area division.

[0042] like Figure 3 As shown, for each grid in the identified set of overlapping grids, all candidate edge nodes that effectively cover that grid (i.e., satisfying...) are calculated. When the power margin of a node is sufficient, define the minimum available power of the terminal. The unit is dBm. This parameter is preset by the system based on the terminal demodulation threshold and service quality requirements. In this embodiment, based on typical IoT uplink demodulation sensitivity, it is preferably set to -115dBm for overlapping grids. And candidate edge nodes, define power margin Its expression is: ;in, Effective coverage estimate defined for the preceding steps, in dBm; power margin The unit is dB, representing the received power provided by the node at that grid. The greater the margin relative to the minimum power requirement of the terminal, the better the coverage quality of the node at that grid, and the more qualified it is to be the sole responsible node for that grid. The logic of this expression is that in overlapping coverage areas, the decision of attribution should not be based solely on the maximum absolute value of the received power, but rather on the margin relative to the terminal's requirements. This is because different terminal types may have different minimum power requirements, and a uniform approach should be adopted. As a benchmark, it can eliminate decision bias caused by terminal heterogeneity.

[0043] It should be noted that this step uniformly maps the coverage intensity of each candidate node in the overlapping grid to a power margin relative to the minimum requirements of the terminal, providing a comparable metric for subsequent weighted Voronoi subdivision. The beneficial effect of this is that by using the margin rather than absolute power as the decision criterion, the responsibility attribution decision is aligned with the actual needs of the terminal, avoiding suboptimal attribution decisions in areas where the absolute signal values ​​are all high but the margins differ significantly.

[0044] For overlapping grid sets For each grid g in the dataset, perform a weighted Voronoi subdivision, uniquely assigning each overlapping grid to the edge node with the largest power margin, and defining a weighted generator. Let n be the average power margin of the edge node n across all overlapping grids, and its expression is: ;in, This represents a subset of all rasters whose edge node n is effectively covered in the overlapping raster set. The number of elements in this subset, i.e., the number of grid cells, is expressed in dB. It represents the average coverage superiority of the node in the overlapping coverage area and is used as a weighting factor for subsequent Voronoi subdivisions. For any overlapping grid Define its weighted distance to edge node n. Its expression is: ;in, is the Euclidean distance from node n to the grid center g, in meters; As a preset small amount of regularization, this embodiment takes dB, used to prevent when A division by zero error occurs when the value is too small; The value is positive; if a node's After calculation, it is less than Then let it equal to To ensure numerical stability; The unit is meters / dB, and its physical meaning is: the effective extension capability of a node's coverage radius under unit power margin. The larger the margin, the smaller the weighted distance, and the more competitive the node is in obtaining the responsibility of the grid. The design logic of this weighted distance is that the essence of Voronoi partitioning is to allocate spatial points to the nearest generator based on a certain metric. However, in this step, generators should not only compete based on spatial distance, but should compete based on the "equivalent distance after power margin weighting", so that nodes with superior coverage quality can obtain a larger responsibility area. Based on weighted distance, the unique owner node of overlapping grids Determined by the following formula: ; That is, select the edge node with the smallest weighted distance as the sole responsible node of the grid; If multiple nodes have the same weighted distance and are all at the minimum value, then select one of them. The larger node is chosen as the home node to break a tie; all nodes are traversed. Then, a responsibility assignment table for overlapping rasters is generated. .

[0045] It should be noted that this step uses a weighted Voronoi partitioning algorithm to uniquely assign overlapping grids, which were originally covered by multiple nodes simultaneously, to the optimal responsible node, eliminating resource waste and interference risks caused by coverage redundancy. The beneficial effects are as follows: the weighted distance design combines spatial distance with coverage quality, so that the responsibility assignment considers both the rationality of the node's physical location and the quality of the coverage margin, avoiding the defect of pure spatial Voronoi partitioning, which may assign grids to neighboring nodes with insufficient coverage margin; unique assignment avoids the interference accumulation caused by multiple nodes repeatedly transmitting in the same grid, and refines the division of coverage responsibilities from a spatial dimension.

[0046] For the generated responsibility attribution table Define the load balancing factor when performing consistency checks. , where n is the number of overlapping graticles assigned to the edge node in the responsibility attribution table, is expressed as: If there exists any node n such that... , The maximum number of responsible grid cells per node is preset; in this embodiment, it is preferably set to [value]. That is, if the average number of responsibilities is rounded up to 1.2, the conflict resolution process is triggered. The specific method for conflict resolution is as follows: For each overlapping grid g that an overloaded node is responsible for, calculate the second-best belonging node for that grid. and reassign the grid However, a redistribution is required. The load does not exceed ; If the overload state is not eliminated after one traversal, then gradually increase the speed. Repeat the above process until the load on all nodes does not exceed 1.5 times the average value. After the conflict is resolved, update the responsibility attribution table.

[0047] It should be noted that this step ensures that the responsibility allocation of overlapping grids is feasible in engineering, avoiding the excessive coverage responsibility borne by a single node, which could lead to backhaul link congestion or power resource exhaustion. The beneficial effect of this is that, through conflict resolution under load balancing constraints, the responsibility allocation result is not only optimal in terms of signal quality, but also fair and feasible in terms of network resource allocation. This avoids theoretically optimal but engineering-infeasible allocation schemes, thereby improving the actual deployment reliability of the system.

[0048] For the identified set of hole grids For each grid g in the dataset, when selecting donor nodes that can expand the coverage to fill the hole, the measured signal strength from the edge node n at the hole grid g is defined. ;for Since the definition of a hole grid is that the measured signals of all nodes are substandard, for any node n, we have and For finite values ​​(non- )or The system uses the following rules to filter donor nodes: For each hole grid g, in all areas with available signals (i.e. Not for Among the edge nodes, the node with the highest measured signal strength is selected as the donor candidate node. Its expression is: If no node has a usable signal, that is, all All If the grid is completely isolated, it will not be screened for donor nodes, but will instead be marked as a candidate location for adding nodes later. The selected donor nodes represent the candidates among the currently deployed nodes that are closest to meeting the coverage requirements of the hole grid. Although their signal strength at the hole grid is below the threshold, they have the highest potential for improvement among all available links.

[0049] It should be noted that this step identifies a donor node for each hole grid that is most likely to achieve coverage expansion through parameter adjustments, providing a clear target for subsequent antenna adjustments and power compensation. The beneficial effect is that by using the maximum measured signal strength as the donor selection criterion, the node with the best initial conditions is selected for enhancement, reducing the cost required for antenna adjustments and power increases, and avoiding the ineffective adjustment or excessive power consumption caused by blindly selecting weak signal nodes.

[0050] For the selected donor nodes When generating antenna electrical downtilt adjustment to extend its coverage boundary, the donor node is defined. The current antenna electrical downtilt angle is The unit is degrees, and a positive value indicates downward tilt. The nodal antenna mounting height is... The unit is meters, and the horizontal distance between the hole grid g and the node is... The unit is meters, defining the target coverage extension distance. ; Based on antenna propagation geometry, the theoretical electrical downtilt angle required to make the antenna main lobe point towards the void grid is... It is given by the following formula: ;in, The height is the typical height of the terminal, in meters. In this embodiment, based on the deployment scenario of urban IoT terminals, it is preferably set to 1.5 meters. The expression is based on the arctangent function, with the result in degrees. It directs the antenna's main lobe towards the ground position of the hole grid. However, actual wireless propagation is limited by the antenna's vertical beamwidth and sidelobe characteristics, so the electrical downtilt angle cannot be directly adjusted. Otherwise, it may cause near-field cover collapse; Define the electric downtilt adjustment amount The unit is degrees, and its expression is: ;in, For the limiting function, Limited to the range Inside; This represents the maximum allowable electrical downtilt angle adjustment within a single adjustment cycle, expressed in degrees. In this embodiment, based on the hardware response capability of the electrically adjustable antenna, it is preferably set to [value missing]. This is to prevent drastic changes in coverage caused by a single, large adjustment, which could lead to dropped calls and access failures; the final antenna electrical downtilt adjustment command for the donor node is... .

[0051] It should be noted that this step calculates the precise antenna electrical downtilt adjustment for the donor node, so that its main lobe points towards the direction of the hole grid, thereby improving the signal coverage strength in that direction. The beneficial effect is that the calculation of the adjustment is based on geometric relationships and hardware constraints, pointing towards the target direction while limiting the magnitude of a single change, ensuring the smoothness and feasibility of the coverage expansion process, and avoiding the appearance of new holes in the original coverage area due to excessive adjustment.

[0052] donor node When generating sector-specific power compensation masks to directionally enhance transmit power towards the void grid, donor nodes are defined. The total downlink transmit power is The unit is dBm, and this node supports Each sector has independently configurable power; in this embodiment, this is preset to 3, corresponding to three sectors. Sector; Define a holed grid g with an azimuth angle relative to the donor node. The unit is degrees, and this angle determines which sector the hole belongs to in terms of coverage; the original transmit power of sector s is defined as... , The unit is dBm, and it meets the following requirements. In the linear power domain, where power is measured in watts, the summation relationship is satisfied; in the dBm domain, the operation needs to be converted to the linear domain and then converted back to dBm. The system identifies the target sector to which the hole grid g belongs. The judgment logic is as follows: Falling within the azimuth coverage area of ​​the sector; Define power compensation mask. The transmit power adjustment for sector s is generated according to the following rule: For the target sector , , This is the preset maximum power compensation amount for a single sector, expressed in dB. In this embodiment, it is preferably set to 3 dB. This represents the maximum permissible transmit power per sector of a node, expressed in dBm; for non-target sectors... , Furthermore, if multiple void grid cells belong to the same target sector from the same donor node, the maximum compensation amount required for all voids is taken to prevent overcompensation due to repeated accumulation; the new transmit power of the target sector after compensation is... The system needs to keep the total power below the node's maximum total transmit power limit, so a total power normalization check is performed before actually issuing commands.

[0053] It should be noted that this step generates a directional power compensation mask for the donor node, concentrating limited power resources in the direction facing the hole, avoiding waste and interference spread caused by increasing power proportionally in all directions; the beneficial effect is that the sector compensation strategy makes the power enhancement spatially directional, increasing the transmission power only in the required direction, which effectively fills the hole and minimizes the interference increment to existing coverage in other directions, thus achieving efficient use of power resources.

[0054] For hollow grids that still cannot be directly covered by the donor node after the previous steps, a two-hop forwarding link is constructed from the donor node through the relay node to the hollow grid terminal. For a hollow grid g, the link is defined as follows: at the donor node... After antenna and power adjustments, calculate the adjusted equivalent coverage estimate. Its expression is: ;in, The change in antenna gain caused by the electrical downtilt adjustment is expressed in dB and is determined by the gain difference of the antenna's vertical radiation pattern function before and after the adjustment. like If the hole is considered to have been directly covered by the donor node, then no relay link needs to be built; if If so, it is determined that the hole needs to be solved through a relay method; For a holed grid g that needs to be relayed, the system filters for relay nodes from all deployed edge nodes. The filtering criteria include: (1) Relay node At the donor node Within the effective coverage area, that is ,in, The grid to which the relay node belongs; (2) Relay node Effective coverage estimate at hole grid g The largest among all candidate nodes is defined as follows: ; If multiple candidates meet the conditions, select one. The largest one becomes the relay node; if all All If the grid cannot be resolved in two hops using the existing network, it is marked as a location where a new node needs to be deployed. The specific configuration for constructing a two-hop forwarding link is as follows: donor node In the downlink direction, the destination address of the data packet destined for the terminal in the hole grid g is redirected to the relay node. The access address is specified, and a forwarding identifier is added; relay node Upon receiving the data packet, based on the forwarding identifier, it is retransmitted on the local node's air interface resources, using the relay node's normal downlink power. This forwarding link configuration is encapsulated as a routing table entry. .

[0055] It should be noted that this step establishes a two-hop forwarding path for stubborn gaps that cannot be covered by a single hop from the donor node, and achieves coverage coverage through multi-hop network collaboration. The beneficial effects are: without adding new physical nodes, the problem of extreme coverage blind spots can be solved through the collaborative forwarding of existing nodes, which saves hardware deployment costs and makes full use of the relay capabilities of edge nodes; the hierarchical design of the donor and relay ensures that the signal quality of the forwarding link is guaranteed in both the backhaul and access directions.

[0056] The generated responsibility attribution table Antenna electrical downtilt adjustment amount Aggregate and sector power compensation mask Set, two-hop forwarding link routing table entries The set is encapsulated sequentially as a consistency control contract; During the encapsulation process, an internal consistency check of the contract is performed: for any g, if g appears in both the responsibility attribution table and the two-hop forwarding link routing table entry set, it is determined to be a conflicting configuration. In this case, the system prioritizes retaining the attribution assignment in the responsibility attribution table and deletes the corresponding entry in the two-hop forwarding link routing table entry set, because direct overwriting takes precedence over relay forwarding; after the check passes, the contract is serialized into a binary instruction stream.

[0057] It should be noted that this step integrates multiple control instructions into a structured contract data package, ensuring that all configurations are logically consistent and conflict-free. The beneficial effect of this is that contractual encapsulation allows for a global consistency review of the configuration instructions output by different modules before release, avoiding control chaos caused by conflicts in responsibility attribution and forwarding routes.

[0058] In this embodiment, based on the refined coverage deviation field and the measured coverage reachability matrix, differentiated spatial reshaping strategies are implemented for two special regions: overlapping grids and hole grids. For overlapping grids, weighted Voronoi partitioning is used to uniquely assign the area covered by multiple nodes to the optimal responsible node, and load balancing conflict resolution ensures the engineering feasibility of the allocation. For hole grids, the coverage boundary is expanded step by step through three measures: donor node screening, antenna electrical downtilt adjustment, and sector power compensation mask. For stubborn holes that are still unreachable, a donor-relay two-hop forwarding link is further constructed. All control instructions are finally encapsulated into a consistent control contract and sent to each node for execution. This achieves a leap from perception to analysis to decision-making to execution, transforming the coverage quality assessment results into specific and executable network parameter adjustment instructions. This fundamentally solves the problem of interference accumulation in overlapping areas and insufficient coverage in hole areas in distributed micro-site deployment, laying a stable spatial coverage pattern foundation for subsequent logical port dynamic mapping and latency compensation.

[0059] Distributed access control and execution module: When a terminal access request arrives, the edge node verifies whether the grid where the requesting terminal is located belongs to the scope of its responsibility attribution table. If it does, access resources are allocated; otherwise, the request is forwarded to the responsible node according to the contract. When a terminal in an overlapping area simultaneously initiates access to two edge nodes, the responsibility token arbitration mechanism is triggered. Only the node holding the responsibility token for that grid performs resource allocation, while the other node reclaims the reserved resources. The access failure rate of each hole grid is statistically analyzed in real time, and the coverage deviation field is dynamically updated based on the analysis results.

[0060] After receiving the contract subset, each edge node loads the contract content into its local control rule engine. At the next system frame boundary, node n synchronously applies antenna and power configuration parameters and sends back an acknowledgment identifier via the physical layer. Once all involved nodes have completed their confirmation, the responsibility grid list and forwarding mapping table take effect immediately and are used as the basis for judging subsequent access requests. It should be noted that this step transforms the generated centralized control contract into a set of executable rules local to each edge node, ensuring that each node has a consistent understanding of grid ownership and forwarding logic. The beneficial effects are: local storage avoids the network overhead and latency of querying the central database for each access request, the distributed decision mechanism significantly reduces access response latency, and synchronous activation ensures the consistency of configuration switching across the entire network.

[0061] When at geographical coordinates When the target terminal initiates an access request to edge node n, node n first calculates the grid index where the terminal is located based on the Global Navigation Satellite System location information reported by the terminal. The expression for calculating the raster index is: ;in, The coordinates of the reference origin for the defined urban area are in meters. This refers to the grid side length, in meters. This represents the number of grid columns in the X direction for the urban area. To ensure the result is an integer raster index, the calculation is rounded down. If the terminal cannot obtain a GNSS positioning due to indoor environment, the uplink sounding reference signal time difference positioning method is used. At least three edge nodes jointly measure the signal time difference to calculate the estimated position of the terminal, which is then mapped to the raster index. ; Obtain raster index Then, edge node n performs attribution verification: querying the local responsible raster list. ,judge If the condition is met, it indicates that the grid where the terminal is located falls within the responsibility scope of this node, and node n has the right to allocate access resources to the terminal; if the condition is not met, the subsequent forwarding process will proceed. It should be noted that this step maps the terminal location of the access request to a grid space and quickly matches it with the local responsibility attribution to determine whether the current node is a legitimate service node for the request. The beneficial effects are: the gridded location decision discretizes continuous geographic coordinates into a finite number of grid indices, making the attribution query degenerate into a set lookup operation with constant time complexity, which greatly improves the processing speed of access requests, and the location backup scheme ensures availability in indoor scenarios without global navigation satellite system signal coverage.

[0062] The processing judgment is as follows When an edge node n receives an access request, it queries its local forwarding mapping table. Determine if there is a target for the grid. A valid forwarding entry; if it exists, node n will encapsulate the access request packet into a forwarding frame, with the following fields in the frame header: original target raster. Original request terminal identifier, donor node identifier n, relay node identifier The forwarding frame is sent to the relay node via an inter-node backhaul link, such as fiber optic or microwave. After receiving the forwarded frame, the relay node identifies the request as a non-directly attributable request forwarded by the donor node based on the forwarding identifier in the frame header. The relay node then queries its own responsibility grid list. Then the relay node directly allocates access resources to the terminal; if The relay node continues to query its own forwarding mapping table to forward the next hop, repeating the above process until a responsible node is found, or the number of hops exceeds the preset maximum number of forwarding hops. In this embodiment The default value is 2; If the forwarding mapping table does not exist for... The entry, and node n's responsibility grid list does not contain it. If the node n replies to the terminal with an access rejection response, it will carry the identifier of the node to which the grid belongs in the response. The node will then query the central controller to guide the terminal to the correct responsible node and re-initiate the access request.

[0063] It should be noted that this step, following the forwarding link defined in the contract, guides non-homed access requests from the incorrectly arrived node to the correct responsible node, ensuring that each terminal's access request is ultimately processed by the legitimate service node. The beneficial effects are: the two-hop forwarding mechanism allows the terminal to be unaware of complex network topology and affiliation relationships, and only needs to initiate access to any node with the strongest signal, with the network side handling the request routing, reducing the complexity of terminal implementation; the maximum hop limit prevents forwarding dead loops, ensuring the robustness of the system.

[0064] When handling potential concurrent access conflicts in overlapping grid areas, since each overlapping grid has been uniquely assigned to a single responsible node in the preceding steps, theoretically, requests from the same terminal in the same grid should only be responded to by the responsible node. However, in actual networks, signal fluctuations or terminal measurement errors may cause a terminal to simultaneously request access from two edge nodes during the access process. and Send the access preamble sequence, and both nodes are determined. and Both being true means that the overlapping raster is included in the responsibility raster lists of both nodes. This should not happen in theory, but it may coexist briefly during the instantaneous window of contract switching. To resolve the aforementioned concurrency conflicts, the system introduces a responsibility token mechanism: each overlapping grid... In the contract, a unique responsibility token variable is generated by the central controller. The initial holder of the token is the one who will be determined. The token is encoded as a random number seed, generated by the central controller and synchronously distributed to all nodes along with the contract; when edge nodes... and Simultaneously received from the same grid During the access preamble sequence, both nodes enter the token arbitration process: before allocating access resources, node n first sends a token request message to the central controller or its designated token management agent, the message containing the grid. And the node identifier n; when the token management agent receives multiple requests, it only returns token authorization for the request from the current token holder. Requests to other nodes return The token will be displayed along with the identifier of the current holder; the current token holder is denoted as... ; The expression for the authorization determination logic is: ; get The node performs access resource allocation and, after the resource allocation is completed, reports the allocation confirmation to the token management agent; obtains... The node immediately reclaims the physical resource blocks reserved for this access request, including time-frequency resources and preamble sequences, and returns a waiting instruction to the terminal, or guides it to re-initiate access to the holder node; the token management agent does not transfer token holding rights after each arbitration, that is, tokens in the same grid will only be transferred according to the new responsibility when a new contract is generated.

[0065] It should be noted that in the critical scenario of overlapping grids, this step ensures that the access request of the same terminal is processed by only one legitimate node at any given time through a centralized token arbitration mechanism, thus eliminating resource conflicts and interference caused by dual access. The beneficial effects are: the token mechanism is triggered only when concurrent access occurs, avoiding the additional signaling overhead in normal scenarios; the arbitration result is clear and unique, and reserved resources are promptly reclaimed, avoiding resource loss and air interface efficiency degradation due to failed competition.

[0066] After successful arbitration, when the responsible node allocates access resources to the terminal, and edge node n obtains the legitimate right to process the terminal's access request through attribution verification or token arbitration, node n performs the following resource allocation operations: (1) Based on the buffer status report and service type reported by the terminal, allocate uplink and downlink time-frequency resources from the available physical resource block pool of this node. The resource block allocation result is recorded as follows: Includes the starting physical resource block index and duration ; (2) Based on the generated sector power compensation mask Determine the transmit power configuration of the sector where the target terminal is located, and carry a power adjustment indication in the downlink control information; (3) Generate a temporary wireless network identifier and establish a wireless resource control connection between the terminal and the node; After allocation, node n sends a resource authorization signaling message to the terminal and reports a connection establishment record to the central controller. The record contains a four-tuple. ; For connection timestamps; If, during the resource allocation process, the node load exceeds the capacity limit (i.e., the current number of connections reaches the node's maximum concurrent users), resulting in the inability to allocate resources, then node n returns a resource shortage response to the terminal and simultaneously reports an overload alarm to the central controller, triggering a load balancing adjustment request.

[0067] It should be noted that after the attribution of responsibility and arbitration are completed, this step involves the actual allocation of resources and connection establishment, completing the final stage of terminal access. The beneficial effect achieved is that the power compensation mask in the control contract is implemented in the actual transmit power configuration, so that the accessed terminal can directly benefit from the generated sector enhancement configuration. The connection establishment record is reported, providing a data source for subsequent failure rate statistics.

[0068] like Figure 4 As shown, for each grid in the identified set of hole grids, when calculating the access failure rate of terminals within that grid in real time, an observation window is defined. The unit is seconds, which is preset to 30 seconds in this embodiment. At the end of each observation window, the system performs failure rate statistics on the holed grid. The statistical data comes from the connection establishment records reported by all nodes and the records of access rejection responses generated. The total number of access requests initiated by terminals in grid g within the observation window is defined as follows: The number of successfully established connections was: The access failure rate of grid g is then... The calculation expression is: ;in, The preset regularization value is set to a value in this embodiment. Used to prevent when A division-by-zero exception occurs, simultaneously reducing the failure rate of grids with no access attempts to 0; when hour, It does not trigger any action; The value range is [0, 1]. The larger the value, the more difficult it is for the terminal in the hole grid to successfully access the network. The logic of this statistical expression is that it uses the number of access attempts actually initiated by the terminal as the denominator and the number of failures as the numerator, directly representing the degree of service quality degradation that the terminal can perceive, rather than the coverage estimate self-tested by the network side.

[0069] It should be noted that this step continuously monitors the actual service quality of the hole grid based on the actual access behavior of the terminal, forming a data foundation for closed-loop feedback. The beneficial effect is that the failure rate statistics come directly from the objective access experience of the terminal, avoiding secondary errors in coverage estimation by the network side, and providing a real and reliable quantitative basis for the accurate determination of failure areas.

[0070] When determining whether to trigger a closed-loop update process based on the statistical access failure rate, a preset access failure rate threshold is defined. In this embodiment, based on the typical service quality requirements of network operators, the preferred setting is as follows: That is, a 25% failure rate, if and only if At that time, grid g is determined to be a failed grid; For a grid g that is determined to be faulty, the system further marks the faulty region and defines a set of faulty regions. Let be the union of all failed rasters and their spatial neighboring rasters. A neighborhood is defined as a distance from the geometric center of a failed raster that does not exceed a neighborhood radius threshold. The set of all rasters, neighborhood radius threshold The unit is meters, and in this embodiment, it is preferably set to meters. The expression for the failure area marking logic is as follows: ;in, Candidate neighborhood grid identifier; Union symbol; subscript To traverse all graticles, only graticles g that satisfy the failure condition are selected for neighborhood expansion. This expression includes the failed grid and its surrounding potentially affected areas into the failed area. The logic is that an increase in the access failure rate is often not an isolated event, but reflects the overall deterioration of coverage or increased interference in the area. It is necessary to conduct incremental detection and replanning on a regional basis. After marking, the system transmits the information of the failed area set to the trigger interface.

[0071] It should be noted that this step converts the quantified access failure rate into a spatial set of failed areas, which serves as a geographical indication of the area to trigger closed-loop updates. The beneficial effects are: the neighborhood expansion mechanism ensures that the scope of the triggered update covers potentially affected areas, avoiding new problems at the boundary caused by repairing only a single failed grid; the threshold is set at 25%, which is neither too sensitive to avoid frequent triggering of closed loops, nor too sluggish to ensure timely response to service quality deterioration.

[0072] Upon receiving a failed area marker, incremental probing and sampling is triggered, updating only the failed areas rather than the entire city area. The specific operation of incremental probing and sampling is as follows: The system is directed to the set of failure regions. Mobile IoT terminals within the system (i.e., the defined second type of sensing data source) send detection scheduling instructions, requesting them to perform detection for the next period of time. Within this embodiment, it is assumed that... The system reports location marker signal strength samples at an encrypted reporting frequency (twice the normal reporting frequency) every second; simultaneously, it instructs edge nodes within the failure area to increase the number of received measurement samples in the direction of the failure grid during mutual listening measurements, thereby improving the estimation accuracy of spatial angle of arrival and Doppler frequency shift. After the incremental sampling is completed, the system re-executes the previous steps of generating the original coverage deviation field, but the processing scope is limited to the set of grates and edge nodes within the set of failed regions. Specifically, the new detection reference frame broadcasting, data acquisition, rasterization mapping, construction of the measured coverage reachability matrix, and generation of the original coverage deviation field are only calculated for the set of failed regions. For grates outside the region, the already output refined coverage deviation field remains unchanged. After the incremental update is completed, the system uses the updated refined coverage deviation field to replace the old values ​​of the corresponding regions in the original deviation field, forming a locally updated coverage deviation field.

[0073] It should be noted that this step limits the closed-loop trigger to incremental sensing and deviation field updates within the failure area, avoiding the computational overhead of global recalculation. The beneficial effects are: the incremental update strategy significantly shortens the closed-loop response time from minutes to ten seconds for full updates. At the same time, encrypted reporting and encrypted mutual listening measurements increase the data sampling density in the failure area, making deviation field correction more accurate and realizing an efficient operation and maintenance mode of focusing on problem areas.

[0074] In this embodiment, the generated static control contract is transformed into dynamically running access processing rules: through responsibility attribution verification and contract forwarding, it is ensured that the access request of each terminal is guided to the correct service node, eliminating access chaos caused by coverage overlap; through the responsibility token arbitration mechanism, the resource conflict during concurrent access in overlapping areas is fundamentally resolved, eliminating interference accumulation and resource waste; through real-time monitoring of access failure rate and threshold triggering of hole grids, terminal experience degradation is automatically transformed into closed-loop update requests, and incremental detection sampling and local contract reshaping strategies are adopted to realize a self-healing cycle of perception-reshaping-control, enabling the system to have the adaptive ability to resist dynamic environmental changes. When the urban environment changes (e.g., the construction of new obstructions) and coverage deteriorates, the system can automatically perceive, make decisions, and adjust without manual intervention, significantly improving the intelligence level and response speed of network operation and maintenance. At the same time, the incremental update strategy ensures the efficiency of the closed-loop process and global stability.

[0075] In the several embodiments provided by this invention, it should be understood that the disclosed methods can be implemented in other ways. For example, the embodiments of the invention described above are merely illustrative; for example, the division of modules is only a logical functional division, and there may be other division methods in actual implementation.

[0076] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; they may be located in one place or distributed across several network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0077] Furthermore, the functional modules in the various embodiments of this invention can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The integrated module can be implemented in hardware or in a combination of hardware and software functional modules.

[0078] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the essential characteristics of the present invention.

[0079] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A city IoT communication distributed access precision control system based on edge collaboration, characterized in that, include: Coverage deviation field generation and compensation output module: Each edge node periodically broadcasts a detection reference frame carrying the node identifier and transmission power level, and at the same time collects the received power of the detection reference frames of adjacent edge nodes, constructs an initial measured coverage reachability matrix in the node × grid dimension, and compares it with the theoretical coverage matrix grid by grid to generate the original coverage deviation field. The original coverage deviation field is subjected to deviation field compensation calculation, and the compensated refined coverage deviation field is output. Coverage responsibility area dynamic reshaping module: Based on the generated refined coverage deviation field and measured coverage reachability matrix, the coverage responsibility area is reshaped: For the set of overlapping grids, a weighted generator is constructed based on the measured received power of each candidate edge node in the grid and the margin relative to the lowest available power of the terminal. Through weighted Voronoi partitioning, each overlapping grid is uniquely assigned to the edge node with the largest margin, forming a responsibility assignment table; For the set of hole grids, the donor node with the strongest measured signal to the hole grid is selected, and an antenna electrical downtilt adjustment amount and sector power compensation mask are generated for it to expand its coverage boundary. A relay node is assigned to the hole grid that is still unreachable after expansion, and a two-hop forwarding link is constructed from the donor node through the relay node to the hole grid terminal. Distributed access control and execution module: When a terminal access request arrives, the edge node verifies whether the grid where the requesting terminal is located is within the scope of its responsibility attribution table. If it is, access resources are allocated; otherwise, the request is forwarded to the responsible node according to the contract. When a terminal in an overlapping area simultaneously initiates access to two edge nodes, the responsibility token arbitration mechanism is triggered. Only the node holding the responsibility token for that grid performs resource allocation, while the other node reclaims the reserved resources. The system performs real-time statistics and analysis on the access failure rate of each hole grid, and dynamically updates the coverage deviation field based on the analysis results.

2. The urban IoT communication distributed access precision control system based on edge collaboration according to claim 1, characterized in that, While the detection reference frame is being broadcast, a parallel acquisition process for two types of sensing data is initiated: the first type of sensing data is data from mutual listening between nodes, and the second type of sensing data is data reported by mobile IoT terminals; the two types of discrete raw data are then mapped to a unified spatial reference frame.

3. The urban IoT communication distributed access precision control system based on edge collaboration according to claim 1, characterized in that, When performing deviation field compensation calculation on the original coverage deviation field, the actual terminal access success / failure records of each grid in the previous control cycle are extracted as feedback truth values. The prediction-actual state inconsistency rate of each grid is calculated. The confidence decay factor is used to perform weighted correction on the deviation values ​​of the corresponding grids in the original coverage deviation field. Grids that have been disproven by feedback truth values ​​for multiple consecutive cycles are marked as environmental change anchor points. Spatial interpolation compensation is performed on the neighboring low-confidence grids using the environmental change anchor points, and the compensated refined coverage deviation field is output.

4. The urban IoT communication distributed access precision control system based on edge collaboration according to claim 3, characterized in that, Inconsistency rate between predicted and actual states for edge node n at grid g The calculation expression is: ;in, The total number of terminal attempts to access edge node n in grid g during the previous control cycle; This represents the number of times the connection failed. `g` is a preset regularization variable; `g` is a raster index variable, `g=1,2,…,G`, where `G` is the total number of raster cells; `n` is an edge node index variable, `n=1,2,…`. , This represents the total number of edge nodes within the deployment area.

5. The urban IoT communication distributed access precision control system based on edge collaboration according to claim 1, characterized in that, For each grid g in the set of overlapping grids, perform a weighted Voronoi partitioning, and uniquely assign each overlapping grid to the edge node with the largest power margin. Define a weighted generator as the average power margin of the edge node n across all overlapping grids. For any overlapping grid, define its weighted distance to edge node n, and select the edge node with the smallest weighted distance as the sole responsible node of that grid.

6. The urban IoT communication distributed access precision control system based on edge collaboration according to claim 5, characterized in that, When performing consistency checks on the generated responsibility attribution table, define the load balancing factor. Let be the number of overlapping grid cells assigned to edge node n in the responsibility attribution table; if there exists any node n that satisfies This triggers the conflict resolution process; This is the preset maximum number of responsible grid cells per node.

7. The urban IoT communication distributed access precision control system based on edge collaboration according to claim 6, characterized in that, For each grid cell in the identified set of void grid cells, a donor node that can expand coverage to fill the void is selected; for the selected donor node, an antenna electrical downtilt adjustment is generated to expand its coverage boundary.

8. The urban IoT communication distributed access precision control system based on edge collaboration according to claim 7, characterized in that, A sector power compensation mask is generated for the donor node to directionally enhance the transmit power toward the hole grid. For hole grids that still cannot be directly covered by the donor node after the previous steps, a two-hop forwarding link is constructed from the donor node through the relay node to the hole grid terminal.

9. The urban IoT communication distributed access precision control system based on edge collaboration according to claim 8, characterized in that, When at geographical coordinates When a target terminal initiates an access request to edge node n, node n first calculates the grid index where the terminal is located based on the Global Navigation Satellite System location information reported by the terminal. ; Obtain the raster index Then, edge node n performs attribution verification: querying the local responsible raster list. ,judge If the condition is met, it indicates that the grid where the terminal is located falls within the responsibility scope of this node, and node n has the right to allocate access resources to the terminal; if the condition is not met, the subsequent forwarding process will proceed.

10. The urban IoT communication distributed access precision control system based on edge collaboration according to claim 9, characterized in that, When the access failure rate is greater than the access failure rate threshold, the corresponding grid is determined to be a failed grid. For the grid determined to be failed, the failed area is further marked, and the set of failed areas is defined as the union of all failed grids and their spatial neighboring grids. Upon receiving a failed area marker, incremental detection sampling is triggered, updating only the failed area and not the entire city area.

Images