Multi-agent based emergency communication resource collaborative scheduling method

Abstract

This invention relates to the field of resource scheduling technology, specifically a multi-agent-based collaborative scheduling method for emergency communication resources. This invention dynamically calculates the link capacity limit based on the current rainfall intensity by querying a rain attenuation zone carrying capacity table; driven by the water diffusion rate, and combined with base station geographical distribution parameters and population density distribution parameters, it predicts the service supply and demand gap and performs pre-allocation of links to be migrated; it calculates the expected signal-to-noise ratio (SNR) of candidate frequency bands based on the rain attenuation coefficient of the operating frequency band, and performs rain attenuation availability screening and interference conflict detection; simultaneously, it establishes an error record based on the deviation between the measured SNR and the theoretical value, enabling adaptive correction in subsequent scheduling processes. This method can improve the accuracy of emergency communication resource scheduling, link stability, and critical service assurance capabilities in disaster environments.

Description

Technical Field

[0001] This invention relates to the field of resource scheduling technology, specifically to a multi-agent-based emergency communication resource collaborative scheduling method. Background Technology

[0002] Emergency communication resource coordination and scheduling is a key support for ensuring multi-departmental collaborative operations in the handling of sudden disasters. Existing emergency communication scheduling technologies mainly adopt the following methods: First, a static scheduling method based on a plan database and rule engine, which allocates fixed communication resources according to a preset emergency level. This method cannot adapt to the dynamic decay of communication resource availability caused by the continuous dynamic changes in the disaster boundary as water spreads. Second, a centralized scheduling method based on single-agent reinforcement learning, which is limited by the computational delay and single-point failure risk caused by the convergence of global states, making it difficult to maintain scheduling stability under the condition that the infrastructure at the disaster site is damaged. Third, a distributed scheduling method based on multi-agent reinforcement learning. Existing solutions generally assume that there is a complete cooperative relationship between the agents. However, in actual scenarios, there is a structural conflict between the communication resource needs of meteorological departments and rescue departments, presenting a hybrid game relationship of partial confrontation and partial cooperation. Existing solutions cannot effectively model this.

[0003] Furthermore, the update frequencies of situational information held by the aforementioned departments differ significantly, and there are also limited information authorization constraints among the departments. This results in each autonomous scheduling agent being able to make independent decisions based only on local information, leading to problems such as channel conflicts, priority inversion, and scheduling gaps. Consequently, it becomes impossible to achieve globally optimal collaborative allocation of communication resources that meets real-time constraints within the critical time window of disaster evolution. Therefore, there is an urgent need for a method that can achieve dynamic priority arbitration and collaborative scheduling of cross-departmental communication resources under multi-source non-stationary disturbances and limited information authorization constraints.

[0004] To address this, a multi-agent-based collaborative scheduling method for emergency communication resources is proposed. Summary of the Invention

[0005] The purpose of this invention is to provide a multi-agent-based emergency communication resource collaborative scheduling method. By combining rainfall intensity, water accumulation diffusion, service supply and demand prediction, and rain attenuation signal-to-noise ratio analysis, it can realize dynamic pre-allocation, frequency screening, and adaptive error correction of emergency communication links, thereby improving the efficiency of communication resource scheduling and service guarantee capabilities in disaster scenarios.

[0006] To achieve the above objectives, the present invention provides the following technical solution: A multi-agent-based emergency communication resource collaborative scheduling method includes: Query the pre-stored rain attenuation partition carrying capacity table of newly arrived resources, and calculate the link carrying capacity limit using the current rainfall intensity as the index; Using the water diffusion rate as a common driving force, the supply-side attenuation and demand-side increment are calculated based on the geographical distribution parameters of base stations and the population density distribution parameters, respectively. The supply and demand gap of each service type at the arrival time is predicted, the links to be migrated are pre-allocated according to the service type with the largest gap, and the feasibility is verified based on the link carrying capacity limit. Based on the service type of the link to be migrated and the current rainfall intensity as input, the expected signal-to-noise ratio of each unoccupied frequency band is calculated by combining the frequency band rain attenuation coefficient; rain attenuation availability screening and interference conflict detection are performed on the candidate frequency bands, and the frequency bands that pass the dual detection are written into the frequency occupancy map in a soft reserved state; Collect the deviation between the measured signal-to-noise ratio and the expected signal-to-noise ratio of each link in the current batch, and write it into the error record table using resource type, operating frequency band, and rainfall intensity range as indexes; when subsequent batches are connected, use the difference between the current and the most recently recorded rainfall intensity as the criterion: if it does not exceed the threshold, the error representative value is corrected to the expected signal-to-noise ratio calculation; if it exceeds the threshold, it is directly calculated using theoretical parameters and a new interval error benchmark is established.

[0007] Preferably, the link carrying capacity limit is calculated using the current rainfall intensity as an index, including: the rain attenuation partition carrying capacity table is indexed by discrete rainfall intensity partitions; each row is pre-recorded based on the rain attenuation characteristics and signal transmission power parameters of the newly arrived resource operating frequency band, and the maximum total service bandwidth that can be carried when the link signal-to-noise ratio in the corresponding rainfall intensity partition meets the minimum service guarantee requirements. Upon receiving an access request from a newly arrived resource, the measured rainfall intensity in the disaster area is extracted from the observation data of the current scheduling cycle. The measured value is compared with the rainfall intensity partition boundaries of each row in the carrying capacity table one by one to locate the partition row that is closest to and does not exceed the current measured rainfall intensity. When the measured rainfall intensity falls between the boundaries of two adjacent partitions, the maximum carrying capacity of the service bandwidth recorded in the row where the lower partition is located is used for each service type link within the scheduling cycle. The maximum total bandwidth that can be carried by the link is read as the upper limit of the link capacity and output to the feasibility verification process.

[0008] Preferably, the pre-allocation process for the links to be migrated includes: along the supply-side path: using the geographical area advanced by the current water accumulation boundary in each scheduling cycle as a quantitative expression of the water accumulation diffusion rate, querying the geographical distribution parameters of base stations in each geographical grid within the area, and obtaining the critical water accumulation depth corresponding to the base station ground elevation and equipment protection level; comparing the predicted water accumulation depth of each grid with the critical water accumulation depth of each base station, including base stations exceeding the critical value in the expected failure base station set, extracting the bearer link list of the base stations in the set from the link management table and grouping them by service type, and accumulating the minimum bearer bandwidth of the links by service type; based on the current cycle's supply-side bandwidth reduction and the newly arrived... The remaining scheduling cycles for the expected arrival time of the resources are used to derive the supply-side attenuation. Along the demand-side path, the geographical area advancing along the flood boundary in each cycle and the population density distribution parameters of each geographical grid within that area are analyzed to derive the scale of the new population due to flood blockade. Based on the historical proportion of the number of links for each business type within the scheduling cycle, the scale of the new population is decomposed into the demand-side increment for each business type. Based on the supply-side attenuation and the demand-side increment, the predicted supply-demand gap for each business type is derived. The business type with the largest predicted supply-demand gap is selected, and the link identifiers and minimum bandwidth requirements of each link in the candidate set of links to be migrated for that type are written into the scheduling pre-occupancy table.

[0009] Preferably, the feasibility verification based on the link capacity limit includes: reading all link entries pre-allocated according to the service type with the largest gap from the scheduling pre-occupancy table, accumulating the minimum capacity bandwidth requirement recorded in each entry, and obtaining the total bandwidth requirement of the pre-allocation scheme; comparing the total bandwidth requirement with the link capacity limit: if the total bandwidth requirement does not exceed the link capacity limit, the pre-allocation scheme is determined to be feasible, the status of the corresponding entry in the scheduling pre-occupancy table is updated to executable, and a feasible pre-allocation scheme is output; if the total bandwidth requirement exceeds the link capacity limit, the link entries are arranged in descending order of their supply-demand gap contribution, and the bandwidth requirement is accumulated sequentially until the accumulated value first exceeds the link capacity limit, at which point the scheme is truncated; the link entries before the truncated position are retained as feasible schemes.

[0010] Preferably, the expected signal-to-noise ratio for each unoccupied frequency band is calculated, including: Taking the service type of the link to be migrated as input, the minimum signal-to-noise ratio threshold value of the service type is read from the correspondence table between service type and minimum signal-to-noise ratio threshold, and used for rain attenuation availability filtering; For each currently unoccupied frequency segment in the frequency occupancy map, perform the following calculations segment by segment: Using the rain attenuation coefficient of the frequency band to which the frequency segment belongs and the current measured rainfall intensity as input, calculate the additional rain attenuation per unit link propagation distance of the frequency segment; read the expected deployment coordinates of the newly arrived resource and the coordinates of the nearest receiving node from the system resource view, and calculate the link propagation distance between the two points; subtract the free space path attenuation determined by the link propagation distance and frequency from the signal transmission power parameter, and then subtract the total rain attenuation obtained by the product of the additional rain attenuation per unit distance and the link propagation distance to obtain the expected receiving signal strength; subtract the expected receiving signal strength from the receiving node noise floor parameter to obtain the expected signal-to-noise ratio of the candidate frequency segment, and pass it along with the candidate frequency segment identifier to the dual detection process.

[0011] Preferably, the dual detection includes: Rain attenuation availability screening stage: The expected signal-to-noise ratio (SNR) of candidate frequency bands is compared segment by segment with the minimum SNR threshold corresponding to the service type of the link to be migrated; candidate frequency bands with expected SNR not lower than the minimum SNR threshold are marked as rain attenuation available and enter the interference collision detection stage; candidate frequency bands with expected SNR lower than the minimum SNR threshold are marked as rain attenuation unavailable and removed from the candidate set; Interference collision detection stage: The center frequency and occupied bandwidth of all currently occupied frequency bands are read from the frequency occupancy map; for each candidate frequency band that passes the rain attenuation availability screening, the absolute value of the difference between its center frequency and the center frequency of each occupied frequency band is calculated and compared one by one with the adjacent channel protection interval threshold specified in the wireless technical specification; candidate frequency bands whose frequency interval with all occupied frequency bands is not less than the adjacent channel protection interval threshold pass the interference collision detection; if any interval is less than the threshold, the candidate frequency band is marked as having interference collision and removed.

[0012] Preferably, the criterion is the difference between the current and the most recently recorded rainfall intensity, including: querying the error record table using a combination index of resource type and operating frequency band, extracting the rainfall intensity value of the most recently written record in the matching entry; calculating the absolute value of the difference between the measured rainfall intensity of the current scheduling cycle and the rainfall intensity value of the most recently recorded record, and comparing it with a preset rainfall intensity change threshold; if the absolute value of the difference does not exceed the preset threshold, then reading the error representative value from the latest record of the matching entry, superimposing the error representative value onto the expected signal-to-noise ratio calculation result of each candidate frequency band in the current batch, and using the corrected expected signal-to-noise ratio value as the basis for judging the availability of rain attenuation; If the absolute value of the difference exceeds the preset threshold, the historical error representative value is not read. Instead, the expected signal-to-noise ratio of each candidate frequency band in the current batch is directly calculated using the theoretical parameters of the frequency band rain attenuation coefficient, and dual detection is performed. After the current batch has completed access and run for a full scheduling cycle, the measured signal-to-noise ratio is collected from each link receiving node. The deviation between the measured signal-to-noise ratio of each link and the theoretically calculated expected signal-to-noise ratio is calculated, and the median of all link deviation values ​​is taken as the error representative value of this batch.

[0013] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. This invention achieves real-time quantitative assessment of communication link carrying capacity under severe weather conditions by introducing a "rain attenuation zoning capacity table" and a dynamic rainfall intensity indexing mechanism. Based on the current measured rainfall intensity, combined with the rain attenuation characteristics of the operating frequency band, signal transmission power parameters, and minimum service guarantee requirements, this invention dynamically calculates the maximum total bandwidth that the link can carry. It also employs a conservative zoning matching strategy of "taking the lower limit rather than the higher limit" to avoid overestimating communication capacity under severe weather conditions. This allows for early identification of link capacity boundaries under complex meteorological conditions, improving the service stability and scheduling reliability of emergency communication networks in rainstorm scenarios, and reducing the risk of communication interruptions due to link mismatch.

[0014] 2. This invention constructs a service gap prediction mechanism based on a dual-path coupling analysis of "supply-side attenuation - demand-side increment," enabling proactive and coordinated scheduling of emergency communication resources. Using the water spread rate as a common driving force, this invention predicts potentially failed base stations based on base station geographical distribution parameters and critical water depth, quantifying supply-side bandwidth attenuation. Simultaneously, it combines population density distribution parameters and historical service proportions to predict new service demands in disaster areas, forming predicted supply-demand gap values ​​for different service types, and pre-allocating links to be migrated accordingly. This improves the resource utilization efficiency and critical service continuity assurance capabilities of emergency communication systems in dynamic disaster environments.

[0015] 3. This invention improves the environmental adaptability and prediction accuracy of frequency resource selection results by constructing an adaptive signal-to-noise ratio (SNR) evaluation mechanism that combines "theoretical parameter calculation and historical error correction." After each batch of scheduling is completed, this invention collects the deviation between the measured SNR and the theoretical expected value, and establishes an error record table according to resource type, operating frequency band, and rainfall intensity range. During subsequent scheduling, it determines whether to introduce a representative error value to correct the expected SNR based on the difference between the current rainfall intensity and historical records, thus achieving dynamic calibration of the propagation model. This significantly improves the accuracy of frequency band selection, reduces misjudgments and frequency conflicts, and increases the success rate of emergency communication link access. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the process of the multi-agent-based emergency communication resource collaborative scheduling method provided by the present invention; Figure 2 This is a schematic diagram of the supply and demand gap prediction process provided by the present invention; Figure 3 A schematic diagram of the error adaptive correction process provided by the present invention. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the invention.

[0018] Example 1: In this embodiment, the multi-agent collaborative scheduling system comprises three types of functional agents: a meteorological agent, a coordination agent, and a resource execution agent. The meteorological agent is responsible for maintaining real-time observation data on rainfall intensity and water spread rate in disaster areas, and pushing observation update messages to the coordination agent according to the scheduling cycle. The coordination agent is responsible for executing the core scheduling process of this method, including querying the rain attenuation zone carrying capacity table, calculating the link carrying capacity limit, predicting the supply-demand gap, performing dual detection, and maintaining the frequency occupancy map and error record table. The resource execution agent is deployed at each emergency communication resource terminal, responsible for collecting measured signal-to-noise ratios, reporting access requests, and executing frequency configuration and link migration instructions issued by the coordination agent. The coordination agent adopts a periodic polling mode to centrally complete scheduling decisions, transforming the multi-department mixed game scenario into a centralized optimization problem under limited information constraints, avoiding channel conflicts and priority inversions caused by distributed decision-making.

[0019] Please see Figure 1 This invention provides a multi-agent-based emergency communication resource collaborative scheduling method, the technical solution of which is as follows: Query the pre-stored rain attenuation partition carrying capacity table of newly arriving resources, and calculate the link carrying capacity limit using the current rainfall intensity as an index; using the water diffusion rate as a common driving force, calculate the supply-side attenuation and demand-side increment based on the geographical distribution parameters of base stations and the population density distribution parameters, predict the supply-demand gap for each service type at the arrival time, pre-allocate links to be migrated according to the service type with the largest gap, and verify feasibility based on the link carrying capacity limit; based on the service type of the link to be migrated and the current rainfall intensity as input, and... The expected signal-to-noise ratio (SNR) of each unoccupied frequency band is calculated based on the combined frequency band rain attenuation coefficient. Rain attenuation availability screening and interference conflict detection are performed on candidate frequency bands, and frequency bands that pass the dual detection are written into the frequency occupancy map in a soft-reserved state. The deviation between the measured SNR and the expected SNR of each link in the current batch is collected and written into the error record table using resource type, operating frequency band, and rainfall intensity range as indexes. When subsequent batches are connected, the difference between the current and the most recently recorded rainfall intensity is used as the criterion: if it does not exceed the threshold, the error representative value is corrected to the expected SNR calculation; if it exceeds the threshold, it is directly calculated using theoretical parameters and a new interval error benchmark is established.

[0020] Furthermore, the link carrying capacity limit is calculated using the current rainfall intensity as an index, including: the rain attenuation partition carrying capacity table is indexed by discrete rainfall intensity partitions; each row is pre-recorded based on the rain attenuation characteristics and signal transmission power parameters of the newly arrived resource operating frequency band, and the maximum total service bandwidth that can be carried when the link signal-to-noise ratio in the corresponding rainfall intensity partition meets the minimum service guarantee requirements. Upon receiving an access request from a newly arrived resource, the measured rainfall intensity in the disaster area is extracted from the observation data of the current scheduling cycle. The measured value is compared with the rainfall intensity partition boundaries of each row in the carrying capacity table one by one to locate the partition row that is closest to and does not exceed the current measured rainfall intensity. When the measured rainfall intensity falls between the boundaries of two adjacent partitions, the maximum total service bandwidth that can be carried by the record in the row where the lower partition is located is taken as the carrying limit of each service type link in this scheduling cycle. The maximum total bandwidth that can be carried by the link is read as the upper limit of the link capacity and output to the feasibility verification process.

[0021] Specifically, the rain attenuation zone carrying capacity table is pre-built during the initialization phase, using several discrete rainfall intensity zones as row indexes. During table construction, the maximum total service bandwidth that can be carried within each rainfall intensity zone is calculated offline using an offline script, combining the specific attenuation coefficients from ITU-R Recommendation P.838 for each frequency band with signal transmission power parameters, free-space path attenuation, and the minimum signal-to-noise ratio (SNR) threshold for each service type. This calculation is loaded into the coordinating agent's memory as a configuration file during the initialization phase, and only table lookup operations are performed during scheduling. For each row corresponding to a rainfall intensity zone, the maximum total service bandwidth that can be carried within that rainfall intensity zone is pre-recorded based on the rain attenuation characteristics of the newly arrived resource's operating frequency band under the corresponding rainfall intensity, and the resource's signal transmission power parameters, ensuring that the link SNR meets the minimum service guarantee requirements. The aforementioned rain attenuation characteristics are taken from the specific attenuation coefficients for each frequency band in ITU-R Recommendation P.838, and the minimum SNR for service guarantee is taken from the quality of service specifications corresponding to the service type carried by the resource. The principle for dividing the boundaries of rainfall intensity zones in the table is as follows: based on historical rainfall intensity data of the disaster area and the forecast grading standards of the meteorological department, the rainfall intensity is divided into several adjacent non-overlapping intervals from zero to the upper limit of extreme rainfall. The boundary values ​​of each zone are written into the configuration file by the meteorological intelligent agent administrator during the system deployment phase and will not be dynamically modified in subsequent scheduling cycles.

[0022] Upon receiving an access request from a newly arrived resource, the coordinating agent extracts the measured rainfall intensity in the disaster area, measured in millimeters per hour, from the observation data of the meteorological agent's current scheduling cycle. Subsequently, the coordinating agent compares this measured value with the rainfall intensity partition boundaries of each row in the carrying capacity table, locating the partition row that is closest to and does not exceed the current measured rainfall intensity. "Closest to and does not exceed" means selecting the row with the largest lower boundary value among all partitions whose lower boundary values ​​do not exceed the current measured rainfall intensity; this is the row corresponding to the lowest rainfall intensity within the interval into which the current measured value falls. When the measured rainfall intensity is exactly equal to a partition boundary value, that partition is the currently located row. When the measured rainfall intensity falls between the boundaries of two adjacent partitions, the maximum total service bandwidth recorded in the row containing the lower partition is taken to ensure that the actual carrying capacity under the current rainfall attenuation conditions is not overestimated, preventing overestimation from causing bandwidth allocation in subsequent scheduling schemes to exceed the actual carrying capacity. The maximum total service bandwidth read is used as the link carrying capacity limit and output to the feasibility verification process.

[0023] For example, if the measured rainfall intensity in the disaster area is 37 mm per hour, and the capacity table uses 20 mm, 30 mm, and 40 mm per hour as continuous partition boundaries, then the current measured value falls within the partition range of 30 mm to 40 mm per hour. Locate the row corresponding to 30 mm per hour and read the maximum total service bandwidth that can be carried in that row as the upper limit of the link capacity.

[0024] In this embodiment, the calculation process for the link carrying capacity limit constitutes the feasibility boundary input for all subsequent scheduling steps. The supply-demand gap prediction results driven by the water diffusion rate, and the pre-allocation scheme for the links to be migrated based on the service type with the largest gap, must all undergo feasibility verification under this upper limit constraint to ensure that the pre-allocation scheme does not exceed the actual carrying capacity under the current rain attenuation conditions.

[0025] Furthermore, referring to Figure 2The pre-allocation process for the migration links includes: along the supply-side path: using the geographical area advanced by the current water accumulation boundary in each scheduling cycle as a quantitative expression of the water accumulation diffusion rate, querying the geographical distribution parameters of base stations in each geographical grid within the area, and obtaining the critical water accumulation depth corresponding to the base station ground elevation and equipment protection level; comparing the predicted water accumulation depth of each grid with the critical water accumulation depth of each base station, including base stations exceeding the critical value in the expected failure base station set, extracting the bearer link list of the base stations in the set from the link management table and grouping them by service type, and accumulating the minimum bearer bandwidth of the links by service type; based on the current cycle's supply-side bandwidth reduction and the newly arrived resources... The remaining scheduling cycles for the expected arrival time of the source are used to derive the supply-side attenuation. Along the demand-side path, the geographical area advancing along the flood boundary in each cycle and the population density distribution parameters of each geographical grid within that area are analyzed to derive the scale of new population added due to flooding. Based on the historical proportion of the number of links for each service type within the scheduling cycle, the scale of new population is decomposed into the demand-side increment for each service type. Based on the supply-side attenuation and the demand-side increment, the predicted supply-demand gap for each service type is derived. The service type with the largest predicted supply-demand gap is selected, and the link identifiers and minimum bandwidth requirements of each link in the candidate set of links to be migrated for that type are written into the scheduling pre-occupancy table.

[0026] Specifically, the quantitative expression of the water spread rate in this invention is: the newly added geographical area covered by the outward advancement of the water boundary within each scheduling cycle, measured in square kilometers per scheduling cycle. This value is calculated by the meteorological agent at the beginning of the current scheduling cycle based on topographic elevation data and measured rainfall intensity, and then reported to the coordinating agent. The water spread rate also serves as a common input for both the supply-side attenuation calculation and the demand-side increment calculation, ensuring that the predictions in both directions are conducted under a unified assumption of water spread advancement.

[0027] Along the supply-side path, based on the geographical area advanced by the current water accumulation boundary in each scheduling cycle, the geographical distribution parameters of base stations stored in each geographical grid within this area are queried. The geographical distribution parameters of each base station include its geographical coordinates, ground elevation, and equipment protection level. The equipment protection level is given according to the national standard for electrical equipment protection levels, and its corresponding critical water accumulation depth is the water depth at which the power supply system fails due to water ingress when the water depth at the bottom of the equipment compartment reaches this value. This parameter is imported from the equipment file during system initialization and is not modified during subsequent scheduling operations. The predicted water accumulation depth of each grid is compared with the critical water accumulation depth of each base station, and base stations whose predicted water accumulation depth exceeds the critical water accumulation depth are included in the set of base stations expected to fail. Subsequently, the list of links currently carried by each base station within the expected set of failed base stations is extracted from the link management table and grouped according to predefined service types such as voice scheduling, video backhaul, and data acquisition. In this embodiment, the typical QoS parameters for each service type are as follows: minimum carrying bandwidth for voice scheduling is 64kbps, and the minimum signal-to-noise ratio threshold is 5dB; minimum carrying bandwidth for video backhaul is 2Mbps, and the minimum signal-to-noise ratio threshold is 15dB; minimum carrying bandwidth for data acquisition is 128kbps, and the minimum signal-to-noise ratio threshold is 8dB. These parameters are written into the service type and QoS parameter correspondence table during system initialization, and are queried during scheduling operation using the service type identifier as the key. The statistical window for the historical proportion of the number of links for each service type is a sliding window of the most recent 10 completed scheduling cycles. The minimum carrying bandwidth of each link under the same service type is accumulated to obtain the reduction in supply-side bandwidth for the current scheduling cycle. The reduction in supply-side bandwidth for the current cycle is multiplied by the remaining scheduling cycle number calculated by the difference between the expected arrival time of the new resource and the current time, and the cumulative supply-side attenuation of each service type at the arrival time is extrapolated.

[0028] Along the demand-side path, the geographical area advancing along the flood boundary in each cycle is multiplied grid-by-grid by the population density distribution parameters of each geographical grid within that area, and the results are accumulated to obtain the scale of the newly added population due to flooding, in people. The population density distribution parameters are imported into the geographic information database by the disaster area administrative department during system deployment, stored with grids as the basic unit, and subsequently dynamically updated periodically, not exceeding the data of the current day. Based on the historical proportion of the number of links of each business type within the scheduling cycle, the scale of the newly added population is decomposed into the demand-side increment of each business type. The so-called historical proportion of the number of links of each business type within the scheduling cycle refers to the ratio of the sum of the number of links of each business type in several historical scheduling cycles that have been completed before the current forecast calculation to the total number of all active links in the current scheduling cycle. This ratio is updated in real time by the coordinating agent after statistically analyzing the historical link record table, and is written once at the end of each scheduling cycle for use in the demand-side forecast of the next cycle.

[0029] The supply-side attenuation and demand-side increment of the same business type are added to obtain the predicted supply-demand gap for each business type. The business type with the largest predicted supply-demand gap is selected. When the predicted supply-demand gap for all business types is zero or negative, the coordination and scheduling agent skips the pre-allocation process for expected migration links this week, reserving the link capacity limit for subsequent cycles, and writes the "no migration demand" status into the scheduling pre-allocation table. When multiple business types have equal predicted gap values ​​and are all the largest positive values, they are selected first according to their business urgency coefficient from highest to lowest. The link identifiers and minimum capacity bandwidth requirements of each link in the candidate set of links to be migrated for that type are written into the scheduling pre-allocation table, associated with the newly arriving resource identifier, completing the link pre-allocation based on the business type with the largest gap.

[0030] Furthermore, the feasibility verification based on the link capacity limit includes: reading all link entries pre-allocated according to the service type with the largest gap from the scheduling pre-occupancy table, accumulating the minimum capacity bandwidth requirement recorded in each entry, and obtaining the total bandwidth requirement of the pre-allocation scheme; comparing the total bandwidth requirement with the link capacity limit: if the total bandwidth requirement does not exceed the link capacity limit, the pre-allocation scheme is determined to be feasible, the status of the corresponding entry in the scheduling pre-occupancy table is updated to executable, and a feasible pre-allocation scheme is output; if the total bandwidth requirement exceeds the link capacity limit, the link entries are arranged in descending order of their supply-demand gap contribution, and the bandwidth requirement is accumulated sequentially until the accumulated value first exceeds the link capacity limit, at which point it is truncated; the link entries before the truncation position are retained as feasible schemes.

[0031] Specifically, the coordinating agent reads all link entries pre-allocated according to the service type with the largest gap from the scheduling pre-allocation table, adds up the minimum bandwidth requirement recorded in each entry, and obtains the total bandwidth requirement of the pre-allocation scheme.

[0032] The total bandwidth requirement is compared with the calculated link capacity limit. If the total bandwidth requirement does not exceed the link capacity limit, the pre-allocation scheme is deemed feasible. The status field of the corresponding entry in the scheduling pre-occupancy table is updated from pending verification to executable. The feasible pre-allocation scheme is output for use in subsequent frequency allocation and dual detection processes.

[0033] If the total bandwidth demand exceeds the link capacity limit, the truncation process begins. Link entries in the scheduling pre-allocation table are rearranged from largest to smallest based on their supply-demand gap contribution. The supply-demand gap contribution refers to the normalized contribution of the ratio of the minimum bandwidth demand of each link to the predicted supply-demand gap for the service type to which the link belongs, multiplied by the total number of links. When multiple links exist under the same service type, each link allocates the supply-demand gap proportionally to its minimum bandwidth demand. After rearranging, the bandwidth demand is accumulated sequentially from the top of the list until the accumulated value first exceeds the link capacity limit, at which point the link is truncated. Link entries before the truncation point are retained as feasible solutions, and their status is updated to executable. Link entries at and after the truncation point are marked as pending processing in subsequent batches. The link identifier, service type, and unmet bandwidth amount of the marked entries are packaged and passed to the next processing flow, awaiting priority inclusion in the pre-allocation candidate when accessing resources in the next batch.

[0034] Further, the expected signal-to-noise ratio for each unoccupied frequency band is calculated, including: Taking the service type of the link to be migrated as input, the minimum signal-to-noise ratio threshold value of the service type is read from the correspondence table between service type and minimum signal-to-noise ratio threshold, and used for rain attenuation availability filtering; For each currently unoccupied frequency segment in the frequency occupancy map, the following calculations are performed segment by segment: Using the rain attenuation coefficient of the frequency band to which the segment belongs and the current measured rainfall intensity as input, calculate the additional rain attenuation per unit link propagation distance for that frequency segment; read the expected deployment coordinates of newly arriving resources and the coordinates of the nearest receiving node from the system resource view, and calculate the link propagation distance between the two points; subtract the free space path attenuation determined by the link propagation distance and frequency from the signal transmission power parameter, and then subtract the total rain attenuation obtained by the product of the additional rain attenuation per unit distance and the link propagation distance, to obtain the expected receiver signal strength; in the above link budget, emergency communication resources are pre-deployed at high locations (such as drones or emergency communication vehicle lifting masts) to maintain line-of-sight propagation conditions; the link budget also retains a system margin of no less than 5dB to cover additional losses caused by environmental uncertainties such as multipath fading. Subtract the expected receiver signal strength from the receiver node noise floor parameter to obtain the expected signal-to-noise ratio of the candidate frequency segment, and pass it along with the candidate frequency segment identifier to the dual detection process.

[0035] Specifically, firstly, using the service type of the link to be migrated as an index, the minimum signal-to-noise ratio (SNR) threshold value corresponding to that service type is read from the correspondence table between service types and minimum SNR thresholds. The minimum SNR threshold value reflects the minimum signal quality requirements of the receiver needed to maintain the normal operation of that service type, and is determined by the corresponding wireless communication technology standard and service system quality of service specifications. For example, for a voice dispatch link, its minimum SNR threshold is usually determined based on the frame error rate requirements of narrowband voice coding; for a video backhaul link, its minimum SNR threshold is determined based on the video coding compression rate and the reliability requirements of reception decoding. These threshold values ​​serve as the basis for judgment during the rain attenuation availability screening stage.

[0036] For each currently unoccupied frequency segment in the frequency occupancy map, the expected signal-to-noise ratio is calculated segment by segment. Using the rain attenuation coefficient of the frequency band to which the segment belongs and the current measured rainfall intensity as input, the additional attenuation per unit link propagation distance for that frequency band is calculated. The rain attenuation coefficient is taken from the specific attenuation coefficient for each frequency band in ITU-R Recommendation P.838. This coefficient reflects the additional signal loss per kilometer propagation distance caused by rain scattering and absorption. Its value varies with the operating frequency and rainfall intensity; the higher the frequency and the greater the rainfall intensity, the larger the specific attenuation coefficient.

[0037] The projected deployment coordinates of the newly arrived resource and the coordinates of the nearest receiving node are read from the resource view. The propagation distance between the two points is calculated in kilometers. The resource view is maintained by the coordinating agent and records the geographical coordinates, access status, and service configuration information of all resources that have been accessed and are expected to be accessed within the current scheduling cycle. The coordinates of the receiving nodes are reported by the receiving nodes during registration and stored in the resource view.

[0038] After obtaining the additional attenuation per unit distance from rain and the link propagation distance, the expected receiver signal strength is obtained by subtracting the free-space path attenuation (determined by the link propagation distance and operating frequency) from the signal transmission power parameter, and then subtracting the total rain attenuation (the product of the additional attenuation per unit distance and the link propagation distance). The result is expressed in decibels per milliwatt (dW). The free-space path attenuation is calculated based on the principle of spherical diffusion loss during electromagnetic wave propagation in free space. The attenuation increases with increasing operating frequency and propagation distance, and its specific value is uniquely determined by the center frequency and link propagation distance of the frequency band. The expected receiver signal strength is then subtracted from the receiver node's noise floor parameter to obtain the expected signal-to-noise ratio (SNR) for the candidate frequency band, expressed in decibels. The noise floor parameter reflects the thermal noise power baseline of the receiver node under the current operating bandwidth and ambient temperature conditions. It is determined by the noise figure parameter given in the receiver node's equipment specifications and the current channel bandwidth, and is entered into the receiver node equipment file during system deployment. The calculated expected SNR, along with the candidate frequency band identifier, is then passed to the dual detection process.

[0039] Furthermore, the dual detection includes: Rain attenuation availability screening stage: The expected signal-to-noise ratio (SNR) of candidate frequency bands is compared segment by segment with the minimum SNR threshold corresponding to the service type of the link to be migrated; candidate frequency bands with expected SNR not lower than the minimum SNR threshold are marked as rain attenuation available and enter the interference collision detection stage; candidate frequency bands with expected SNR lower than the minimum SNR threshold are marked as rain attenuation unavailable and removed from the candidate set; Interference collision detection stage: The center frequency and occupied bandwidth of all currently occupied frequency bands are read from the frequency occupancy map; for each candidate frequency band that passes the rain attenuation availability screening, the absolute value of the difference between its center frequency and the center frequency of each occupied frequency band is calculated and compared one by one with the adjacent channel protection interval threshold specified in the wireless technical specification; candidate frequency bands whose frequency interval with all occupied frequency bands is not less than the adjacent channel protection interval threshold pass the interference collision detection; if any interval is less than the threshold, the candidate frequency band is marked as interfering and removed.

[0040] Specifically, in the rain attenuation availability screening stage: the calculated expected signal-to-noise ratio (SNR) of each candidate frequency band is compared segment by segment with the minimum SNR threshold value corresponding to the service type of the link to be migrated. Candidate frequency bands with expected SNR not lower than the minimum SNR threshold value are marked as rain attenuation available and enter the interference collision detection stage; candidate frequency bands with expected SNR lower than the minimum SNR threshold value are marked as rain attenuation unavailable, removed from the candidate set, and no longer enter the subsequent processing flow.

[0041] Interference Conflict Detection Phase: The center frequencies and occupied bandwidths of all currently occupied frequency bands are read from the frequency occupancy map to form a list of occupied frequency bands. For each candidate frequency band that passes the rain attenuation availability screening, the absolute value of the difference between its center frequency and the center frequencies of each occupied frequency band in the list is calculated, and then compared one by one with the adjacent channel guard interval threshold value specified in the wireless technical specification. The adjacent channel guard interval threshold value refers to the minimum frequency interval that must be maintained between two adjacent frequency bands in the same wireless system to avoid adjacent channel interference. This value is determined by the wireless transmission technical specification adopted. For example, for a system using orthogonal frequency division multiplexing modulation, the adjacent channel guard interval must not be less than the sum of the channel bandwidth and the guard bandwidth. The specific value is written into the configuration file according to the wireless technical standard adopted during system deployment. Candidate frequency bands whose frequency intervals with all occupied frequency bands are not less than the adjacent channel guard interval threshold value pass the interference conflict detection; if any occupied frequency band has a frequency interval less than the threshold value with the candidate frequency band, the candidate frequency band is marked as having an interference conflict and removed from the candidate set.

[0042] Candidate frequency bands detected through the above two stages are written into the frequency occupancy map in a soft-reservation state. The soft-reservation record includes the following fields: frequency band identifier, associated resource identifier, reservation activation time, and reservation expiration time. The reservation expiration time is set to the expected arrival time of the newly arrived resource plus one scheduling cycle, serving as the timeout criterion. At the beginning of each scheduling cycle, the coordinating agent traverses all records in the frequency occupancy map that are in a soft-reservation state, extracts the reservation expiration time field, and compares it with the current time: for records whose reservation expiration time is earlier than the current time, the system restores the corresponding frequency band status from soft-reservation to unoccupied, removes the soft-reservation entry from the frequency occupancy map, and releases the frequency band for reuse in subsequent scheduling cycles; for records whose reservation expiration time is still later than the current time, the soft-reservation state remains unchanged, waiting for the corresponding resource to actually arrive and the link activation to be completed.

[0043] Furthermore, referring to Figure 3 The method uses the difference between the current and most recently recorded rainfall intensity as the criterion, including: querying the error record table using a combination index of resource type and operating frequency band, extracting the rainfall intensity value of the most recently written record in the matching entry; calculating the absolute value of the difference between the measured rainfall intensity of the current scheduling cycle and the rainfall intensity value of the most recently recorded record, and comparing it with a preset rainfall intensity change threshold; if the absolute value of the difference does not exceed the preset threshold, then reading the error representative value from the latest record of the matching entry, superimposing the error representative value on the expected signal-to-noise ratio calculation result of each candidate frequency band in the current batch, and using the corrected expected signal-to-noise ratio value as the basis for judging the availability of rain attenuation; If the absolute value of the difference exceeds the preset threshold, the historical error representative value is not read. Instead, the expected signal-to-noise ratio of each candidate frequency band in the current batch is directly calculated using the theoretical parameters of the frequency band rain attenuation coefficient, and dual detection is performed. After the current batch has completed access and run for a full scheduling cycle, the measured signal-to-noise ratio is collected from each link receiving node. The deviation between the measured signal-to-noise ratio of each link and the theoretically calculated expected signal-to-noise ratio is calculated, and the median of all link deviation values ​​is taken as the error representative value of this batch.

[0044] Specifically, after the current batch of access is completed and a full scheduling cycle is completed, the measured signal-to-noise ratio (SNR) values ​​of all links in the current batch are collected from each link receiving node. These values ​​are then compared with the theoretically calculated expected SNR under the same frequency band and link propagation conditions. A link-by-link difference calculation is performed to obtain the SNR deviation value for each link. The sign of the error representative value is defined as the difference between the measured SNR and the theoretically expected SNR; a positive value indicates that the actual channel is better than the theoretical prediction, and a negative value indicates that the actual channel is worse than the theoretical prediction. The median of all link deviation values ​​is taken as the error representative value for this batch. Using the median instead of the mean is to reduce the impact of abnormal deviations caused by local factors such as obstruction and multipath propagation on the representative value from a few links. This error representative value is written into the error record table using a combined index of resource type, operating frequency band, and the partition to which the current measured rainfall intensity belongs. The record entries also store the specific value of the measured rainfall intensity for the current scheduling cycle.

[0045] When subsequent batches of resources are added, the error record table is queried using a combination index of resource type and operating frequency band. The rainfall intensity value of the most recently written record in the matching entries is extracted as the most recent recorded rainfall intensity. The absolute value of the difference between the measured rainfall intensity of the current scheduling period and the most recent recorded rainfall intensity is calculated and compared with a preset rainfall intensity change threshold.

[0046] The method for determining the preset threshold for rainfall intensity variation is as follows: Based on the nonlinear relationship between the specific attenuation coefficient and rainfall intensity in each frequency band as described in ITU-R Recommendation P.838, when the change in rainfall intensity exceeds a certain range, the corresponding specific attenuation coefficient changes significantly, resulting in a non-negligible jump in the systematic offset between the theoretically calculated expected signal-to-noise ratio and the measured value. During system deployment, technicians select the maximum rainfall intensity variation that ensures the change in specific attenuation coefficient does not exceed a predetermined proportion, based on the specific attenuation coefficient curve corresponding to the frequency band used, as the threshold value, and write this value into the system configuration file. Threshold values ​​can be set separately for different operating frequency bands; higher frequency bands are more sensitive to changes in rainfall intensity, and their corresponding threshold values ​​should be smaller.

[0047] If the absolute value of the difference does not exceed a preset threshold, the representative error value is read from the latest record of the matching entry in the error record table. This representative error value is then superimposed on the expected signal-to-noise ratio (SNR) calculation results of each candidate frequency band in the current batch. The corrected expected SNR value is used as the basis for judging rain attenuation availability. The representative error value reflects the systematic offset of the theoretical calculation model relative to the actual measurement for resources of the same type and frequency band under similar rainfall intensity ranges. Its applicability is limited to access scenarios located in the same geographical region as the recorded batch, and where the difference between the current rainfall intensity and the rainfall intensity at the time of recording does not exceed a preset threshold. When the same batch involves multiple working frequency bands, each frequency band queries its own representative error value using a combined index of "resource type + working frequency band". Each frequency band is corrected independently and not mixed across frequency bands. When the superimposed representative error value is negative, the corrected SNR is lower than the theoretical expectation, tending to exclude frequency bands with worse actual performance; when the representative error value is positive, the corrected SNR is higher than the theoretical expectation, and the corresponding frequency band performs better than the theoretical prediction in the actual environment.

[0048] If the absolute value of the difference exceeds a preset threshold, the historical error representative value is not read. Instead, the expected signal-to-noise ratio (SNR) of each candidate frequency band in the current batch is directly calculated using the theoretical parameters of the frequency band rain attenuation coefficient, and a dual detection process is executed. After the current batch completes access and runs for a full scheduling cycle, the measured SNR is collected, the deviation value of each link is calculated, and the median is taken as the error representative value. The error record table is written with the resource type, operating frequency band, and the interval to which the current measured rainfall intensity belongs as a combined index, establishing the first error benchmark record for that rainfall intensity interval for use by subsequent batches with rainfall intensities in the same interval.

[0049] Example 2: Based on Example 1, this embodiment further illustrates the specific implementation of the dynamic priority queue management mechanism for the links to be processed, and the soft reservation validity re-verification mechanism when rainfall intensity changes significantly.

[0050] The scenario applicable to this embodiment is as follows: a continuous heavy rainfall disaster enters the middle stage of water accumulation and spread, ground base stations continuously lose connection as the water level rises, new batches of emergency communication resources are successively connected, but the link carrying capacity of a single connection is insufficient to cover all links to be migrated, resulting in some links being backed up and not processed in multiple scheduling cycles; at the same time, the rainfall system moves, and the measured rainfall intensity in the disaster area has changed significantly compared to when the previous batch of resources were connected.

[0051] The preliminary steps of the coordinated intelligent agent's method are as follows: Query the pre-stored rain attenuation zone carrying capacity table of the newly arrived resource and calculate the link carrying capacity limit using the current rainfall intensity as an index; drive dual-path calculations on the supply and demand sides using the water diffusion rate to obtain the predicted supply and demand gap values ​​for each service type and write the links to be migrated into the scheduling pre-occupancy table according to the service type with the largest gap; accumulate the minimum carrying bandwidth requirements of all link entries in the scheduling pre-occupancy table, compare them with the link carrying capacity limit, and perform feasibility verification; pre-allocate the working frequency for the newly arrived resource using a dual detection method consisting of rain attenuation availability screening and interference conflict detection, and write it into the frequency occupancy map with a soft reservation status; after the resource has run for a full scheduling cycle, collect the deviation between the measured signal-to-noise ratio and the expected signal-to-noise ratio, write it into the error record table using a combined index of resource type, working frequency band, and rainfall intensity range, and determine whether to use the error representative value to correct the expected signal-to-noise ratio calculation when accessing subsequent batches, based on the difference between the current and most recently recorded rainfall intensities. The above steps are consistent with the description in Example 1 and will not be repeated here.

[0052] Furthermore, dynamic priority queue management is performed on link entries marked as "pending subsequent batch processing," including: For each link entry marked as "to be processed in subsequent batches", record the scheduling cycle number when each entry enters the pending state; after the feasibility verification process of each subsequent scheduling cycle is completed, use the product of the unmet bandwidth of each entry and the number of scheduling cycles that the entry has been waiting for as the dynamic priority score of the entry, and reorder the pending queues from high to low according to the dynamic priority score. Once the newly arrived resources have completed the calculation of the link capacity limit, if there is remaining capacity after deducting the total bandwidth requirement of the first pre-allocated scheme in this batch from the calculated link capacity limit, the coordinating agent will sequentially retrieve link entries from the queue to be processed in descending order of dynamic priority score, and verify whether the minimum bandwidth requirement of each entry does not exceed the current remaining capacity: entries that meet the condition will have their status updated from "to be processed in subsequent batches" to "executable", and the minimum bandwidth requirement of the entry will be deducted from the remaining capacity; entries that do not meet the condition will remain in the pending state and continue to participate in the next cycle of sorting; until the remaining capacity is exhausted or the queue to be processed is traversed. For link entries in the processing queue that have waited for more than the preset maximum waiting period threshold, their service type is marked as "over-limit alarm" and reported to the coordinating agent for manual decision-making on whether to activate the degradation protection scheme for that service type.

[0053] Specifically, after the feasibility verification process is completed, because the link capacity limit of the newly arrived resources in this batch is insufficient to support all the links to be migrated, some link entries are marked as "pending processing in subsequent batches" and transferred to the dynamic priority queue. When an entry enters the queue, the system synchronously records the scheduling cycle number when the entry enters the pending state. This number is stored in the status field of the queue entry and serves as the starting reference for subsequent calculation of the number of scheduling cycles already waited.

[0054] After the feasibility verification of each subsequent scheduling cycle, the coordinating agent updates the priority score of each entry in the queue: the unmet bandwidth demand recorded for each entry is multiplied by the number of scheduling cycles elapsed from when it entered the queue to the end of the current scheduling cycle, and the product of these two values ​​is used as the current dynamic priority score for that entry. The longer the waiting time and the greater the amount of unmet bandwidth, the higher the score, reflecting the scheduling implication that the urgency of link recovery continuously increases with the accumulation of waiting time. After the update, the coordinating agent reorders the queue according to the score from highest to lowest.

[0055] Once the link capacity limit calculation for a new batch of resources is completed and the initial pre-allocation scheme for this batch is executed, the coordinating agent subtracts the total bandwidth requirement of the initial pre-allocation scheme from the link capacity limit for this batch. If the difference is greater than zero, the remaining capacity available for allocation is determined by this difference. The agent then retrieves the remaining capacity in the queue, starting from the first entry, according to the current queue order. It verifies whether the minimum bandwidth requirement of the entry does not exceed the remaining capacity. If the condition is met, the status is updated to executable and the corresponding bandwidth is deducted from the remaining capacity. If the condition is not met, the entry remains in the pending state and continues to be stored in the queue. The process continues until the remaining capacity is exhausted or the queue ends.

[0056] Regarding the acquisition of the maximum waiting period threshold: During system initialization, the equipment administrator calculates this threshold by dividing the maximum communication interruption tolerance time specified for each service type in the disaster response plan by the currently configured scheduling cycle time, taking the integer part of the quotient, and writing it into the system parameter table as a fixed configuration value. For entries that have exceeded this threshold and have not yet been allocated, the coordinating agent marks their service type as an "over-limit alarm" and reports it, allowing manual judgment on whether to activate the degradation protection scheme for that service type.

[0057] Furthermore, the calculation of dynamic priority scores also incorporates a business type urgency coefficient, including: During system initialization, the emergency command center configures corresponding urgency coefficient values ​​for each business type based on the severity of the disaster, and stores the correspondence between the identifier of each business type and its urgency coefficient value in the business urgency coefficient table. The urgency coefficient value is an administrative configuration parameter input from outside the system and remains fixed during operation. The configuration principle is that the urgency coefficient value configured for business types with higher relevance to life safety is higher than that for business types with lower relevance to life safety. For each pending link entry, the product of its unmet bandwidth and the number of scheduling cycles already waited is used as the basic priority value. Then, the urgency coefficient of the service type to which the entry belongs is used as input, and the corresponding value is read from the service urgency coefficient table. The basic priority value is multiplied by this value to obtain the comprehensive priority score of the entry. The comprehensive priority score replaces the dynamic priority score, and the pending queue is reordered to ensure that when the remaining capacity is limited, the available capacity is allocated to the service links with higher comprehensive priority scores first.

[0058] Specifically, during system initialization, the emergency command center configures urgency coefficients for each business type based on the severity of the disaster, writing these coefficients into a business urgency coefficient table. Urgency coefficients are represented by positive integers, and during configuration, it must be ensured that no two business types have equal urgency coefficient values ​​to prevent ties in overall priority scores and guarantee the uniqueness of the ranking results. Business types with stronger support for life safety are configured with higher urgency coefficient values, while those with weaker support are configured with lower values. These hierarchical relationships are explicitly defined and fixed during configuration and remain unchanged during operation.

[0059] When sorting the queue, the coordinating agent first calculates the basic priority value of each item, then reads the urgency coefficient corresponding to the business type of the item from the business urgency coefficient table, and uses the product of the basic priority value and the urgency coefficient as the comprehensive priority score. The comprehensive priority score replaces the basic priority score to reorder the queue. When two items have similar basic priority scores, the item with the higher urgency coefficient of its business type has a higher comprehensive priority score and is placed earlier in the sorting, thus ensuring that when the remaining capacity is limited, business links with higher relevance to life safety are allocated first.

[0060] Furthermore, when it is determined that the rainfall intensity has changed significantly, the validity of the soft reservation records in the frequency occupancy map is revalidated, including: When the absolute value of the difference exceeds the preset threshold and it is determined that the rainfall intensity has changed significantly, the coordinating agent traverses all records in the frequency occupancy graph that are currently in a soft reservation state before performing dual detection of each candidate frequency band in the current batch. For each soft reservation record, using the measured rainfall intensity of the current scheduling cycle as input, the expected signal-to-noise ratio (SNR) of the soft reservation frequency band from the predicted deployment coordinates of the associated resource to the nearest receiving node is recalculated. The recalculated expected SNR is compared with the minimum SNR threshold of the preset service type of the associated resource. For soft reservation records whose recalculated expected SNR does not meet the minimum SNR threshold, their status is updated to "rain attenuation failure," the soft reservation mark of the corresponding frequency band is cleared, and the record is re-added to the candidate frequency band pool of the current batch. At the same time, the corresponding associated resource identifier is written to the queue to be re-reserved. For soft reservation records whose recalculated expected SNR still meets the minimum SNR threshold, the soft reservation status remains unchanged. After the current batch of dual detection processes is completed serially and the results are written to the frequency occupancy map, the coordinating agent restarts the processing of the queue to be re-reserved: based on the current measured rainfall intensity, each associated resource identifier in the queue to be re-reserved is re-executed in the current candidate frequency segment pool for rain attenuation availability screening and interference conflict detection. New candidate frequency segments that pass the dual detection are written to the frequency occupancy map in a soft reservation state, and the corresponding resource identifier and the original reservation failure time are associated to complete the soft reservation update. The processing of the queue to be re-reserved is executed sequentially with the current batch of dual detection, without concurrency, to ensure the consistency of the frequency occupancy map state writing.

[0061] Specifically, when the coordinating agent determines through the process that the absolute value of the difference between the current measured rainfall intensity and the most recently recorded rainfall intensity in the error record table exceeds a preset threshold, that is, when it is determined that the rainfall intensity has changed significantly, it first enters the soft reservation re-verification process before starting the dual detection of the candidate frequency band of the current batch.

[0062] The coordinating agent traverses all records in the frequency occupancy graph that are in a soft-reservation state. For each record, using the measured rainfall intensity of the current scheduling cycle as input, it recalculates the expected signal-to-noise ratio (SNR) of the soft-reserved frequency band from the predicted deployment coordinates of the associated resource to the nearest receiving node according to the link budget calculation logic. The calculation process is completely consistent with that when the soft reservation was first established, except that the old rainfall intensity used when establishing the soft reservation is replaced with the current measured rainfall intensity. The recalculated expected SNR is compared with the minimum SNR threshold of the preset service type of the associated resource: for soft-reserved records that do not meet the threshold, their status is updated to "rain attenuation failure", the soft-reservation mark of the corresponding frequency band is cleared, and the frequency band is re-added to the current batch of candidate frequency bands; at the same time, the corresponding associated resource identifier is written to the queue to be re-reserved; soft-reserved records that still meet the threshold remain unchanged.

[0063] After the soft reservation re-verification traversal is completed, the coordinating agent performs rain attenuation availability screening and interference conflict detection on the candidate frequency bands of the current batch according to the dual detection process. At this time, the candidate frequency band pool contains frequency bands released by re-verification and original unoccupied frequency bands. Frequency bands from both sources participate in the dual detection uniformly without differentiation. After the dual detection of the current batch is completed and all results are written to the frequency occupancy map, the coordinating agent then performs the processing of the queue to be re-reserved: based on the current measured rainfall intensity, each associated resource identifier in the queue is re-verified for rain attenuation availability screening and interference conflict detection in the current candidate frequency band pool. Those that pass are written to the frequency occupancy map in a soft reservation state, associated with the corresponding resource identifier and the original reservation failure time, completing the soft reservation record update. The two rounds of writing operations are executed serially. The processing of the queue to be re-reserved can only be executed after the dual detection writing of the current batch is completed, ensuring that the frequency occupancy map does not show a state of duplicate allocation of the same frequency band.

[0064] Regarding the acquisition of the rainfall intensity change threshold: This threshold is determined by the equipment administrator during system initialization based on the partition boundary spacing width used in the rainfall attenuation partition carrying capacity table of each working frequency band in the system frequency band parameter table. When the rainfall intensity change is sufficient to allow the current rainfall intensity to cross from its current partition to an adjacent partition, the carrying capacity estimation when the original soft reservation was established is no longer applicable to the current partition. Therefore, the minimum rainfall intensity change required to cross to an adjacent partition is used as the basis for threshold determination. This value is calculated from the partition boundary parameters of each frequency band carrying capacity table and written into the system parameter table during initialization for querying during operation.

[0065] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A multi-agent-based emergency communication resource collaborative scheduling method, characterized in that, include: Query the pre-stored rain attenuation partition carrying capacity table of newly arrived resources, and calculate the link carrying capacity limit using the current rainfall intensity as the index; Using the water diffusion rate as a common driving force, the supply-side attenuation and demand-side increment are calculated based on the geographical distribution parameters of base stations and the population density distribution parameters, respectively. The supply and demand gap of each service type at the arrival time is predicted, the links to be migrated are pre-allocated according to the service type with the largest gap, and the feasibility is verified based on the link carrying capacity limit. Based on the service type of the link to be migrated and the current rainfall intensity as input, the expected signal-to-noise ratio of each unoccupied frequency band is calculated in combination with the frequency band rain attenuation coefficient; Rain attenuation availability screening and interference conflict detection are performed on candidate frequency bands. Frequency bands that pass the dual detection are written into the frequency occupancy map in a soft reserved state. Collect the deviation between the measured signal-to-noise ratio and the expected signal-to-noise ratio of each link in the current batch, and write it into the error record table using resource type, operating frequency band, and rainfall intensity range as indexes; when subsequent batches are connected, use the difference between the current and the most recently recorded rainfall intensity as the criterion: if it does not exceed the threshold, the error representative value is corrected to the expected signal-to-noise ratio calculation; if it exceeds the threshold, it is directly calculated using theoretical parameters and a new interval error benchmark is established.

2. The emergency communication resource collaborative scheduling method based on multi-agent systems according to claim 1, characterized in that: The link carrying capacity limit is calculated using the current rainfall intensity as an index, including: the rain attenuation partition carrying capacity table is indexed by discrete rainfall intensity partitions; each row is pre-recorded based on the rain attenuation characteristics and signal transmission power parameters of the newly arrived resource operating frequency band, and the maximum total service bandwidth that can be carried when the link signal-to-noise ratio in the corresponding rainfall intensity partition meets the minimum service guarantee requirements. Upon receiving an access request from a newly arrived resource, the system extracts the measured rainfall intensity in the disaster area from the observation data of the current scheduling cycle. It then compares the measured value with the rainfall intensity partition boundaries of each row in the carrying capacity table to locate the partition row that is closest to and does not exceed the current measured rainfall intensity. When the measured rainfall intensity falls between the boundaries of two adjacent partitions, the system takes the maximum carrying capacity of the service bandwidth recorded in the row where the lower partition is located and selects the links of each service type within the scheduling cycle. The maximum carrying capacity of the service bandwidth is read as the upper limit of the link carrying capacity and output to the feasibility verification process.

3. The emergency communication resource collaborative scheduling method based on multi-agent systems according to claim 1, characterized in that: The pre-allocation process for the links to be migrated includes: along the supply-side path: using the geographical area advanced by the current water accumulation boundary in each scheduling cycle as a quantitative expression of the water accumulation diffusion rate, querying the geographical distribution parameters of base stations in each geographical grid within the area, and obtaining the critical water accumulation depth corresponding to the base station ground elevation and equipment protection level; comparing the predicted water accumulation depth of each grid with the critical water accumulation depth of each base station, including base stations exceeding the critical value in the expected failure base station set, extracting the bearer link list of the base stations in the set from the link management table and grouping them by service type, and accumulating the minimum bearer bandwidth of the links by service type; based on the current cycle's supply-side bandwidth reduction and newly arrived resources... The remaining scheduling cycles for the expected arrival time of the source are used to derive the supply-side attenuation. Along the demand-side path, the geographical area advancing along the flood boundary in each cycle and the population density distribution parameters of each geographical grid within that area are analyzed to derive the scale of new population added due to flooding. Based on the historical proportion of the number of links for each service type within the scheduling cycle, the scale of new population is decomposed into the demand-side increment for each service type. Based on the supply-side attenuation and the demand-side increment, the predicted supply-demand gap for each service type is derived. The service type with the largest predicted supply-demand gap is selected, and the link identifiers and minimum bandwidth requirements of each link in the candidate set of links to be migrated for that type are written into the scheduling pre-occupancy table.

4. The emergency communication resource collaborative scheduling method based on multi-agents according to claim 3, characterized in that: The feasibility verification based on the link capacity limit includes: reading all link entries pre-allocated according to the service type with the largest gap from the scheduling pre-occupancy table, accumulating the minimum capacity bandwidth requirement recorded in each entry, and obtaining the total bandwidth requirement of the pre-allocation scheme; comparing the total bandwidth requirement with the link capacity limit: if the total bandwidth requirement does not exceed the link capacity limit, the pre-allocation scheme is determined to be feasible, the status of the corresponding entry in the scheduling pre-occupancy table is updated to executable, and a feasible pre-allocation scheme is output; if the total bandwidth requirement exceeds the link capacity limit, the link entries are arranged in descending order of their contribution to the supply-demand gap, and the bandwidth requirement is accumulated sequentially until the accumulated value first exceeds the link capacity limit, at which point the scheme is truncated; the link entries before the truncated position are retained as feasible schemes.

5. The emergency communication resource collaborative scheduling method based on multi-agent systems according to claim 1, characterized in that: Calculate the expected signal-to-noise ratio for each unoccupied frequency band, including: Taking the service type of the link to be migrated as input, the minimum signal-to-noise ratio threshold value of the service type is read from the correspondence table between service type and minimum signal-to-noise ratio threshold, and used for rain attenuation availability filtering; For each currently unoccupied frequency segment in the frequency occupancy map, perform the following calculations segment by segment: Using the rain attenuation coefficient of the frequency band to which the frequency segment belongs and the current measured rainfall intensity as input, calculate the additional rain attenuation per unit link propagation distance of the frequency segment; read the expected deployment coordinates of the newly arrived resource and the coordinates of the nearest receiving node from the system resource view, and calculate the link propagation distance between the two points; subtract the free space path attenuation determined by the link propagation distance and frequency from the signal transmission power parameter, and then subtract the total rain attenuation obtained by the product of the additional rain attenuation per unit distance and the link propagation distance to obtain the expected receiving signal strength; subtract the expected receiving signal strength from the receiving node noise floor parameter to obtain the expected signal-to-noise ratio of the candidate frequency segment, and pass it along with the candidate frequency segment identifier to the dual detection process.

6. The emergency communication resource collaborative scheduling method based on multi-agents according to claim 5, characterized in that: The dual detection process includes: Rain attenuation availability screening phase: The expected signal-to-noise ratio (SNR) of candidate frequency bands is compared segment by segment with the minimum SNR threshold corresponding to the service type of the link to be migrated; candidate frequency bands with an expected SNR not lower than the minimum SNR threshold are marked as rain attenuation available and proceed to the interference collision detection phase; candidate frequency bands with an expected SNR lower than the minimum SNR threshold are marked as rain attenuation unavailable and removed from the candidate set; Interference collision detection phase: The center frequency and occupied bandwidth of all currently occupied frequency bands are read from the frequency occupancy map; for each candidate frequency band that passes the rain attenuation availability screening, the absolute value of the difference between its center frequency and the center frequencies of each occupied frequency band is calculated and compared one by one with the adjacent channel protection interval threshold specified in the wireless technical specification; candidate frequency bands whose frequency intervals with all occupied frequency bands are not less than the adjacent channel protection interval threshold pass the interference collision detection; if any interval is less than the threshold, the candidate frequency band is marked as interfering and removed.

7. The emergency communication resource collaborative scheduling method based on multi-agent systems according to claim 1, characterized in that: The criteria for determining rainfall intensity are based on the difference between the current and most recently recorded rainfall intensity. This includes: querying the error record table using a combination of resource type and operating frequency band as an index, extracting the rainfall intensity value of the most recently written record from the matching entries; calculating the absolute value of the difference between the measured rainfall intensity of the current scheduling cycle and the most recently recorded rainfall intensity value, and comparing it with a preset rainfall intensity change threshold; if the absolute value of the difference does not exceed the preset threshold, then reading the error representative value from the latest record of the matching entries, superimposing the error representative value onto the expected signal-to-noise ratio calculation results of each candidate frequency band in the current batch, and using the corrected expected signal-to-noise ratio value as the basis for judging rain attenuation availability screening. If the absolute value of the difference exceeds the preset threshold, the historical error representative value is not read. Instead, the expected signal-to-noise ratio of each candidate frequency band in the current batch is directly calculated using the theoretical parameters of the frequency band rain attenuation coefficient, and dual detection is performed. After the current batch has completed access and run for a full scheduling cycle, the measured signal-to-noise ratio is collected from each link receiving node. The deviation between the measured signal-to-noise ratio of each link and the theoretically calculated expected signal-to-noise ratio is calculated, and the median of all link deviation values ​​is taken as the error representative value of this batch.

Images