A method for dynamic resource allocation in vehicle-to-everything (V2X) networks based on a confidence-based service time window

By introducing a reliable service time window into the Internet of Vehicles (IoV) and combining multi-factor evaluation of vehicle access and resource allocation, the problem of service timeouts caused by link failures and improper resource allocation in the IoV is solved, thereby improving service completion rate and resource utilization efficiency.

CN123124110APending Publication Date: 2026-11-17SOUTHWEAT UNIV OF SCI & TECH
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611317945.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-28
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

During the allocation of vehicle-to-everything (V2X) resources, high-speed vehicle movement and link quality fluctuations can lead to communication link failures. Improper allocation of wireless resources when multiple services arrive simultaneously can cause data retransmission, queue backlog, and timeouts for critical services.

Method used

A method based on a confidence-based service time window is adopted, which combines geometric coverage conditions, service deadline, link quality stability and control time overhead. The probability of continuous service availability of the link is predicted by perturbation samples, the service completion value and resource allocation priority are evaluated, and the vehicle access method is selected and resources are allocated.

Benefits of technology

It improved the service completion rate before link loss, optimized the allocation of wireless resources, ensured that critical services were completed before the deadline, and reduced average latency and data retransmission rate.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123124110A_ABST
    Figure CN123124110A_ABST
Patent Text Reader

Abstract

This invention discloses a dynamic resource allocation method for vehicle-to-everything (V2X) networks based on a confidence-based serviceable time window. During each scheduling cycle, vehicle motion status and link quality information are collected to construct a candidate link set. Combining short-term motion prediction, the probability of continued service availability for candidate links is calculated using disturbance samples to obtain a confidence-based serviceable time window. The effective completion value of services is calculated, link resource allocation weights are determined, and a conflict graph weighted greedy algorithm considering the resource block usage benefits of links is used to complete resource allocation. After the cycle ends, the service queue and link risk trigger values ​​are updated based on transmission results and feedback. This invention addresses the problem of link failure and critical service transmission failure caused by the lack of effective assessment of the time a link can be used for communication transmission in existing methods. It can improve the service completion rate and the amount of data actually successfully completed before connection loss, thereby improving service transmission continuity and resource allocation effectiveness.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of vehicle-to-everything (V2X) wireless communication and resource management technology, specifically to a dynamic resource allocation method for V2X based on a confidence-based serviceable time window. Background Technology

[0002] Vehicle-to-everything (V2X) resource allocation involves allocating limited wireless resources to communication links between vehicles and roadside units or other vehicles, based on the vehicles' communication needs and link status, to support service data transmission. Because vehicle location and link quality are constantly changing, and different services have different data volumes, priorities, and deadlines, the system needs to dynamically determine the vehicle's access method and the link resource allocation results.

[0003] However, the following problems still exist in the process of allocating vehicle network resources: 1) High-speed vehicle movement and link quality fluctuations may cause communication links to fail before the remaining service transmission is completed. Furthermore, access establishment, handover, and resource adjustment will further compress the actual transmission time, resulting in only partial transmission of sensing data or control information, causing data retransmission, queue backlog, and access reselection; 2) When multiple types of services arrive at the same time, limited wireless resources need to be allocated among different links. If the deadline, priority, and feasibility of the service before link failure cannot be determined simultaneously, resources may be occupied by services that are difficult to complete in time, resulting in critical services exceeding the deadline or failing to complete before the link fails.

[0004] Therefore, there is an urgent need for a communication resource allocation method in the Internet of Vehicles environment that can accurately assess the time that a link can be used for communication transmission, and thereby evaluate the value of service completion and determine the priority of resource allocation. Summary of the Invention

[0005] This invention provides a dynamic resource allocation method for vehicle-to-everything (V2X) networks based on a confidence-based serviceable time window. This method does not rely solely on instantaneous link quality or a single deterministic connection time as the sole decision-making criterion. Instead, it maps geometric coverage conditions, service cutoff times, link quality stability, control time overhead, and prediction uncertainty together as a confidence-based serviceable time window for candidate links. This window is then used as a unified variable to achieve service availability assessment, vehicle access selection, link resource allocation, and risk-triggered updates.

[0006] S1: In each cycle, collect data such as the location, speed, acceleration, direction of movement, and link quality of vehicles with communication needs, and construct a candidate link set;

[0007] S2: Based on the predicted distance, signal-to-noise ratio, and control time overhead of the candidate links, the probability of the candidate links remaining serviceable is predicted by perturbation samples, and the confidence serviceable time window of the candidate links is obtained.

[0008] S3: Based on the confidence service time window, calculate the maximum amount of business that can be completed within the window, the degree of business completion, and the effective value of business completion, which are used to determine whether the link can reliably complete the business before the connection is lost;

[0009] S4: Based on the candidate link's credible service time window, effective service completion value, control time overhead, access object load, and future failure risk, select the vehicle to use direct connection or cluster head relay access method, and perform resource allocation;

[0010] S5: After each cycle ends, update the service queue, link risk trigger value, and confidence level based on the actual transmission results and feedback.

[0011] As a further optimization of the present invention, in step S1, a candidate link set is constructed based on vehicle-related information, and the specific process is as follows:

[0012] At the start of the scheduling cycle, roadside units or centralized control nodes collect the motion status of vehicles with communication needs, including position, speed, acceleration, and heading; collect service status, including remaining queue length, service priority, waiting time, and deadline; collect link status, including current link quality, resource reuse interference, transmission power, and ACK / NACK feedback; for cluster head relay scenarios, they also collect the number of members in candidate cluster heads, access load, backhaul link status, and remaining forwarding capacity; the collected information is used for subsequent prediction, and when information is missing, the most recent valid measurement value or short-time filtering results can be used to supplement it;

[0013] For vehicles The set of roadside units that meet the roadside unit coverage conditions and minimum link quality requirements is denoted as . , will be located in the vehicle The set of vehicles within the workshop communication range that have relay capabilities and can establish a backhaul link is denoted as […]. The set of candidate access objects is represented as:

[0014]

[0015] The candidate access object set includes direct connection access objects and cluster head relay access objects, and the two access methods can be compared in a unified subsequent evaluation process; objects that do not meet the current coverage conditions, minimum link quality requirements, or forwarding capability requirements are not included in the candidate set to avoid making invalid predictions for obviously unusable objects;

[0016] Construct a candidate link set based on the candidate access object set: For , build vehicles To roadside unit direct link ;for , build vehicles To candidate cluster head Access side link And construct candidate cluster heads To its reach roadside unit backhaul link The candidate link set is represented as:

[0017]

[0018] When the candidate access object is a candidate cluster head, the corresponding relay path is... and The structure involves evaluating the two links separately in subsequent calculations, and constraining the relay path with the reachable rate of the weaker link from end to end. This preserves the accurate evaluation at the link level while allowing the relay path to be compared as a candidate access method during access decision-making.

[0019] As a further optimization of the present invention, in step S2, the confidence service time window of the candidate link is calculated, and the specific process is as follows:

[0020] For candidate links First calculate the deterministic serviceable time window. This window takes the minimum of the geometrically available service time, the service cutoff allowable service time, and the stable service time, and deducts the control time overhead, representing the net service time that can be used to support the service under the current forecast results:

[0021]

[0022] when A result of zero indicates that the candidate link does not have a valid duration available for service transmission in the current period;

[0023] command link The sending and receiving ends are respectively and ;vehicle At any moment Over time The subsequent method for calculating the predicted location is as follows:

[0024]

[0025] in, , and vehicles At any moment Position, velocity, and acceleration;

[0026] link At any moment Predicted elapsed time The distance after that is expressed as:

[0027]

[0028] in, For link The corresponding coverage radius, when When it is a direct link from the vehicle to the roadside unit or a cluster head backhaul link, Take the effective coverage radius of the roadside unit; when When it is an inter-vehicle link, Determine the communication radius between vehicles;

[0029] Geometric service time The longest time the predicted distance does not exceed the coverage radius:

[0030]

[0031] in, For the time domain; this definition requires the time from the current moment to... The coverage condition is met throughout the entire interval, rather than only at the end of the interval, thus avoiding misjudging cases that have left the coverage and then re-entered the coverage as continuously available.

[0032] Business deadline and permitted service time By link Deadline of the services carried With the current moment The difference is determined:

[0033]

[0034] If the current time has reached or exceeded the business deadline Setting it to zero indicates that subsequent transmissions will no longer generate valid value for completion before the deadline; incorporating this time into the deterministic serviceable time window can prevent the link from still being connected but the service has exceeded the deadline from continuing to occupy priority resources;

[0035] To predict link quality, the link gain is first calculated based on the predicted distance; link At any moment Predicted link gain The calculation method is as follows:

[0036]

[0037] in, The path gain constant. This is the path loss index. This refers to the shadow fading coefficient;

[0038] Calculate the link quality value based on the link gain. :

[0039]

[0040] in, For link The transmission power, For noise power, For link A set of links where there are resource reuse conflicts or interference conflicts;

[0041] Stable service time This represents the longest period of time during which the link quality value remains at or above this threshold. The calculation method is as follows:

[0042]

[0043] in This is the minimum serviceable link quality threshold;

[0044] Controlling time expenditure It consists of access establishment time, control signaling and handover time, and local rescheduling time.

[0045]

[0046] in, Indicates the time required to establish a new access relationship. This indicates the time required for control signaling interaction and handover. This indicates the calculated time for the affected link to perform a local rescheduling.

[0047] Due to uncertainties in vehicle motion measurement, short-term prediction, shadow fading, resource reuse interference, and service arrival, only using The actual service duration may be overestimated or underestimated; therefore, M disturbance samples are generated for the uncertainty: the disturbance range or sampling distribution is determined based on the positioning error, historical motion fluctuations, link quality measurement errors and service arrival statistics.

[0048] In the Under one disturbance sample, the link In the future Serviceable events within are represented as follows:

[0049]

[0050] in, This indicates the downlink of the perturbation sample in the future duration. The system continuously meets constraints related to coverage, link quality, service deadline, and remaining queues. This indicates that at least one condition is not met; To allow for the upper limit of the remaining queue, This is an indicator function; by setting the above conditions, it distinguishes between "the physical link is still connected" and "the link can still effectively carry the current service";

[0051] link In the future The probability of continued serviceability within the given range is calculated by the proportion of serviceable events occurring in the M perturbation samples:

[0052]

[0053] link At any moment Confidence level The confidence-to-service time window is represented as follows:

[0054]

[0055] The confidence-based service time window represents the period when the probability of continued service availability is not less than [a certain value]. Under certain conditions, this is the longest time that can be used for service transmission after deducting control time overhead; compared with the deterministic serviceable time window, this window explicitly retains the predictive reliability: when the disturbance samples are relatively consistent in their judgment of the link failure time, the window is more stable; when the motion or channel uncertainty is large, the window that meets the confidence level is shortened accordingly, thereby reducing the possibility that a short-lived good link is preferentially selected.

[0056] As a further optimization of the present invention, in step S3, the maximum amount of business that a candidate link can complete within the window, the degree of business completion, and the effective value of business completion are calculated. The specific process is as follows:

[0057] After obtaining the confidence-available service time window, calculate the link. effective rate :

[0058]

[0059] in, For effective load factor, For the bandwidth of a single resource block, For link At any moment The link quality value; this effective rate is used to estimate the amount of data the link can handle at a given actual payload ratio;

[0060] link Maximum number of transactions that can be completed within the trusted service time window Represented as:

[0061]

[0062] The maximum achievable traffic volume combines link rate with future sustainable service duration; for links with high instantaneous rates but short confidence service windows, the maximum achievable traffic volume may still be small; therefore, this metric is more suitable than using instantaneous rate alone to determine whether a service can be completed before the link is lost.

[0063] set up For link At any moment Remaining queue length, business completion rate Represented as:

[0064]

[0065] in, To prevent extremely small positive numbers with a denominator of zero; This indicates that, given the current effective rate and the confidence available service time window, the link has the capability to complete all remaining queues; This indicates that only a portion of the business is expected to be completed; the business completion rate will take into account the business scale in the evaluation, avoiding judging the feasibility of completion solely based on the link window length or rate.

[0066] To reflect the changing characteristics of business priority and time-sensitive business value with waiting time and deadline, the current value of the business is calculated. :

[0067]

[0068] in, This is the business priority weighting coefficient. This is the value decay coefficient. For business in the link The waiting time on This is the deadline for business operations; once the deadline is reached, the indicator function is set to zero, so that business operations that have exceeded the deadline will no longer occupy priority resources with their original time value.

[0069] Further calculate the effective completion value of the business. :

[0070]

[0071] The effective value of a business transaction is determined by its current value, the projected completion rate, and the probability of continued service availability within the specified window. When a business has a high priority but a low completion rate, or a high completion rate but insufficient probability of continued service availability, All of these will decrease accordingly; only when the business value, feasibility, and link reliability are all high simultaneously will the effective completion value of the business be significant; this indicator serves as a positive benefit for subsequent access selection.

[0072] As a further optimization of the present invention, in step S4, the vehicle access method is selected and resource allocation is performed. The specific process is as follows:

[0073] For vehicles and candidate access objects Calculate the overall access utility :

[0074]

[0075] in, , and vehicles Through candidate access objects The confidence service window of the formed link or relay path, the value of effective service completion, and the time overhead of access and handover control; Candidate access objects The load value; For vehicles Through candidate access objects The risk of future failure of the formed link; , and As a normalized benchmark; , , , and These are non-negative weighting coefficients;

[0076] The future failure risk value is obtained from the probability of continued serviceability within the confident serviceable time window:

[0077]

[0078] The comprehensive access utility makes the continuous service capability and business completion benefits of the candidate link a positive term, and makes the access control overhead, access object load and future failure risk a penalty term.

[0079] When the risk trigger value of the vehicle's current link is 0, the current access object is maintained to avoid frequent switching when the link still meets service requirements; when the risk trigger value is 1 or the vehicle has not yet established an access relationship, the object with the highest overall access utility is selected from the candidate access object set.

[0080]

[0081] when Belongs to the roadside unit set At that time, the vehicle Use direct connection; when When a vehicle belongs to the candidate cluster head set, It is added to the corresponding vehicle cluster and forwarded to the roadside unit via the candidate cluster head;

[0082] After the vehicle access selection is completed, the service links formed for the selected access objects are... Calculate resource allocation weights :

[0083]

[0084] in, This reflects the queue pressure that needs to be processed within the unit's remaining service time. Reflects the percentage of business that has not yet been completed as expected. Reflects the current value of the business. This reflects the reliability of continued service availability within the window; the weight is not driven solely by the total queue, but by the urgency of completing the target service within the remaining valid window.

[0085] A set of pending service links is constructed based on the selection results of vehicle access objects. and resource block set Pending service links Use resource blocks The revenue is defined as:

[0086]

[0087] in, For link Use resource blocks The achievable rate at that time; this benefit reflects both the urgency of link scheduling and the transmission capacity that a specific resource block can provide.

[0088] set up Assign variables to binary resources. Indicates the resource block Assigned to link Otherwise, take 0; the resource allocation objective is to maximize the weighted revenue of all allocated links:

[0089]

[0090] Resource allocation must at least satisfy the following constraint: For resource blocks allocated exclusively, each resource block may be allocated to at most one non-multiplexed link:

[0091]

[0092] For links with conflicting edges in the conflict graph, the same resource block cannot be used simultaneously:

[0093]

[0094] in For the set of conflicting edges; for vehicles via candidate cluster head Forwarded to roadside unit The relay path's end-to-end rate is limited by the weaker segment between the access and backhaul sides:

[0095]

[0096] For candidate cluster heads The total access traffic of its member vehicles cannot exceed the remaining forwarding capacity of the cluster head. :

[0097]

[0098] In the solution process, first follow The link-resource block combinations are arranged from high to low. Candidate combinations are checked in turn to see if they meet the conflict constraints, relay end-to-end rate constraints, and cluster head remaining forwarding capacity constraints. If they meet the constraints, the allocation is completed and the available resources and related capacity are updated. If they do not meet the constraints, the combination is skipped until the resource blocks are exhausted or there are no feasible combinations. A weighted greedy algorithm is used to prioritize the allocation of resources to links with high urgency and transmission benefits within the confidence service time window while maintaining low computational complexity.

[0099] As a further optimization of the present invention, in step S5, the service queue, link risk trigger value, and confidence level are updated based on the actual transmission results and feedback. The specific process is as follows:

[0100] At the end of the scheduling period, the link service queue is updated based on the actual amount of data successfully completed as confirmed by ACK; let the scheduling period length be... Then the link The remaining queue length at the start of the next cycle is represented as follows:

[0101]

[0102] in, This represents the actual amount of successfully completed data confirmed by ACK during the previous scheduling cycle. For new traffic arriving at the start of the next scheduling cycle; only the amount of data confirmed with ACK is deducted from the queue, which can prevent data that was not successfully received or needs to be retransmitted from being mistakenly counted as completed traffic;

[0103] Link risk trigger value The initial value is 0, and in subsequent periods it is updated based on the confidence serviceable time window, business completion rate, probability of continued service availability, effective business completion value, and cumulative NACK count:

[0104]

[0105] in, To establish a confidence threshold for the available service time window, This represents the threshold for business completion. This is the minimum threshold for the probability of continued service. The threshold for effective business completion. The cumulative number of NACKs within the most recent N scheduling cycles. This is the threshold for the number of NACKs; when any condition is met, This indicates that the link's continuous service capability, service availability, or actual transmission reliability is insufficient, and access reselection or partial resource rescheduling needs to be performed in the next cycle.

[0106] Risk trigger values ​​do not require a global reconfiguration every cycle: when a link triggers a risk, only the candidate access objects for the vehicle corresponding to that link, the links that conflict with it, and the affected resource blocks are recalculated; links that have not triggered risks and are not affected by conflict changes continue to use the original access relationships and resource allocation results; through local updates, control time overhead can be reduced, and the control time overhead can be fed back in the next confidence service time window calculation;

[0107] Update confidence level First, calculate the expected amount of data to be transmitted within the period. The calculation method is as follows:

[0108]

[0109] in For binary variables, For link Use resource blocks Time rate;

[0110] Next, calculate the deviation in resource allocation. The calculation method is as follows:

[0111]

[0112] in, To prevent extremely small positive numbers with a denominator of zero;

[0113] Confidence level Update according to the following calculation method:

[0114]

[0115] in, The updated confidence level; Update the step size to reflect the confidence level; This is a range operator, and its calculation method is as follows: If the value is less than the lower limit, the lower limit is used; if it is greater than the upper limit, the upper limit is used; if it falls within the range, the original value remains unchanged. and They are respectively The minimum and maximum thresholds, The allowable error for the target.

[0116] This invention employs the above-mentioned technical methods and introduces the concept of a confidence-to-service time window. Compared with existing methods, the advantages of this invention are:

[0117] 1) Unify the geometric coverage duration, service cutoff constraints, link quality stabilization duration, and control time overhead into an effective time range that can be used for scheduling, and use disturbance samples to give the probability of continuous serviceability, thereby reducing the overestimation of link serviceability by a single deterministic prediction.

[0118] 2) Calculate the maximum achievable service volume, service completion rate, and effective service completion value using a confidence-based service time window, so that access selection and resource allocation can distinguish between "services that can be completed before the link is lost" and "services that can only be partially transmitted", and prioritize the allocation of limited radio resources to links with a higher probability of completion before the link is lost. Attached Figure Description

[0119] Figure 1 This is a schematic diagram of the overall process of the vehicle-to-everything (V2X) dynamic resource allocation method based on a credible serviceable time window according to the present invention.

[0120] Figure 2 A schematic diagram of the physical modeling of the communication connection between the vehicle, relay vehicle, and roadside unit; Example

[0121] The present invention will be further described below with reference to the accompanying drawings and embodiments. For example... Figure 1 As shown in the system flowchart, the technical solution in the invention content provides a complete workflow for this invention, from candidate link construction, calculation of the confidence-based service time window, calculation of the value of effective service completion, access selection, resource allocation to feedback updates. The following embodiments will not repeat the above derivation step-by-step, but will instead illustrate how this invention can be deployed and implemented in conjunction with specific road scenarios, vehicle trajectories, service configurations, and operational processes. The parameters described below are example values ​​for ease of explanation and can be adjusted according to road type, communication standard, vehicle density, and service requirements.

[0122] The example selected is a highway vehicle communication scenario. Figure 2 Taking a vehicle i with communication needs, a candidate relay vehicle h, and a roadside unit r as an example, the physical relationship of the communication connection is explained: Vehicle i can directly access the roadside unit when the coverage and link quality requirements are met; or when the communication ranges of vehicle i and relay vehicle h overlap and relay vehicle h is within the service range of the roadside unit, a relay access path is formed through "vehicle i - relay vehicle h - roadside unit r". Figure 2 This is used to illustrate the physical connection relationship between a single candidate relay path and a direct path. In reality, there are multiple vehicles, multiple roadside units, and multiple candidate access objects at the same time.

[0123] Three roadside units are deployed along the road, each connected to the roadside edge controller via wired or wireless backhaul. The roadside edge controller is responsible for receiving vehicle status, maintaining candidate links, performing access selection and radio resource allocation, and processing ACK / NACK feedback at the end of the scheduling cycle. Roadside units are evenly distributed along the road section covered by the vehicle trajectories, with their longitudinal position close to the center of the road. The effective coverage radius of a single roadside unit is set to 275 meters. The effective communication radius between vehicles is set to 95 meters. To prevent a single candidate cluster head from carrying too many members, the maximum size of a vehicle cluster is set to 12 vehicles. At the beginning of each scheduling cycle, the controller filters direct connection access objects and candidate cluster heads based on the distance between the vehicle and the roadside unit, the distance between vehicles, and the current link quality.

[0124] Candidate cluster heads are generated from vehicles currently able to maintain backhaul connections with roadside units. During implementation, priority is given to vehicles that are close to the roadside unit, have minimal speed changes, move in the same direction as surrounding vehicles, and have high remaining forwarding capacity. For the same vehicle seeking access, if both a directly connected object and multiple candidate cluster heads exist, the controller creates a candidate record for each, but does not immediately switch access relationships. Instead, it determines the final access object after evaluating future service time and business feasibility.

[0125] The vehicle motion state was obtained using the highD highway vehicle trajectory public dataset. The highD dataset is a dataset of autonomous driving trajectories on German highways collected by drone aerial photography. It records a large number of vehicles' continuous motion processes in real traffic environments, including vehicle number, position coordinates, speed, acceleration, lane number, and vehicle size. It can be used to construct realistic vehicle movement scenarios and supports simulation verification of vehicle trajectory prediction, communication link changes, and vehicle-to-everything (V2X) resource allocation algorithms. The example reads vehicle number, frame number, lateral position, longitudinal position, lateral speed, longitudinal speed, lateral acceleration, longitudinal acceleration, and lane number from this dataset, and reads the sampling frame rate from the recorded information file.

[0126] In this embodiment, the maximum number of stable vehicles within the selected time interval is set to 80, and the scheduling cycle is run continuously for 260 cycles. The sampling frame rate of highD data is 25 frames per second. In this embodiment, the vehicle status is read every 5 frames, and the time interval between adjacent scheduling cycles is approximately 0.2 seconds.

[0127] For trajectory records lacking speed, acceleration, or lane number information, adjacent valid samples are used to supplement them. In engineering deployment, vehicle position and speed can be provided by satellite positioning units, inertial measurement units, or road cooperative sensing systems, while link quality can be obtained by vehicle and roadside units based on reference signal received power, signal-to-noise ratio, or signal-to-interference-plus-noise ratio. If a measurement fluctuates significantly, the range of variation for that variable is correspondingly expanded in the disturbed samples.

[0128] This embodiment configures 12 available resource blocks, with each pending link receiving a maximum of 2 resource blocks within a scheduling cycle. The bandwidth of a single resource block is set to 0.42MHz, and the payload factor is set to 0.64 to deduct non-service data such as pilot signals, protocol headers, and coding redundancy. The transmit power for the vehicle-to-roadside unit link is set to 20dBm, the transmit power for the inter-vehicle link is set to 18.5dBm, and the receiver noise power is set to -96dBm.

[0129] Link propagation employs a simplified model incorporating distance loss and shadowing fading. The path loss exponent is set to 2.68, the shadowing fading standard deviation to 4.5 dB, and the minimum serviceable link quality threshold to 8.2 dB. The implementation incorporates resource congestion and co-channel interference based on the number of links scheduled within the same period, ensuring that links within the coverage area may still experience unstable transmission due to increased interference.

[0130] The future state prediction time domain is set to 2.2 seconds, with a prediction step size of 0.2 seconds. For each candidate link, 20 sets of disturbance samples are generated at each prediction time to reflect location prediction errors, vehicle speed fluctuations, shadowing fading, and interference changes. The initial confidence level is set to 0.86, which serves only as the starting threshold for the first calculation window of a new link. After the link is actually scheduled, the confidence level for the next period is updated based on its allocated resources and the number of successful ACK confirmations.

[0131] The access establishment time is set to 0.03 seconds. When the access object changes, a 0.08-second handover time is added. When accessing via a candidate cluster head relay, a 0.05-second cluster joining and relay relationship establishment time is added. In risk assessment, the confidence-to-service time window threshold is set to 0.30 seconds, the service completion threshold is set to 0.36, the minimum continuous service probability threshold is set to 0.66, and the NACK count threshold is set to 2. When the link window is too short, the expected service completion rate is too low, the continuous service probability is insufficient, or there have been consecutive recent transmission failures, the controller marks the link as needing re-examination. These thresholds are used to construct a reproducible experimental environment and do not limit the invention to using only these values.

[0132] This embodiment sets up three types of data: security services, collaborative perception services, and general services. The data volume of security services is between 18kbit and 48kbit, with a deadline of 0.40 seconds and a priority weight of 3.2; the data volume of collaborative perception services is between 480kbit and 1450kbit, with a deadline of 1.15 seconds and a priority weight of 2.1; and the data volume of general services is between 700kbit and 2200kbit, with a deadline of 2.20 seconds and a priority weight of 1.0.

[0133] Within each scheduling cycle, services are randomly generated for vehicles appearing in the road scene. The single-cycle arrival probabilities for safety services, cooperative perception services, and general services are set to 0.52, 0.20, and 0.10, respectively. To ensure fairness in comparisons between different resource allocation strategies, the program pre-generates a fixed service arrival sequence before each strategy is executed, and all strategies use the exact same vehicle trajectories and service arrival results.

[0134] Each service record includes at least the service number, the vehicle to which it belongs, the service type, the initial data volume, the remaining data volume, the arrival time, the deadline, the priority, and the value decay coefficient. A service is considered validly completed if it is fully completed before the deadline and the link it relies on remains valid; if it is not completed by the deadline, it is considered expired. At the end of the cycle, only the amount of data confirmed with ACK is used to update the remaining queue; data that failed to be transmitted or needs to be retransmitted remains in the queue.

[0135] At the start of the scheduling cycle, the controller first reads the motion status of all vehicles in the current frame and adds newly arriving services to the corresponding vehicle queue. For vehicles without pending services, only their motion and backhaul status required to serve as candidate cluster heads is retained, and no access-side resources are allocated to them; for vehicles with pending services, the controller establishes their direct connection candidate objects and relay candidate objects.

[0136] In this embodiment, when the number of resource blocks is insufficient, links that are ranked lower may not receive resources in this cycle; their services will remain in the queue and will participate in the evaluation again in the next cycle.

[0137] After resource allocation is completed, vehicles transmit according to the selected access relationship. ACK results are generated based on actual link quality, probability of continued service availability, and resource congestion level. Upon receiving an ACK, the amount of data actually successfully transmitted is deducted from the corresponding service queue; if no ACK is received, a NACK record is added to the vehicle's record.

[0138] At the end of the cycle, the controller saves the throughput, effective throughput, ACK rate, queue length, number of access handovers, number of vehicles checked by the control, and window mismatch status for this cycle, and then enters the next scheduling cycle. Since the new confidence level is only used in the next cycle and does not modify the decisions already completed in the current cycle, there is no problem of repeated calculations within the same cycle during the entire operation.

[0139] To verify the effectiveness of different technical aspects, multiple comparison strategies were implemented under the same vehicle trajectory and service arrival sequence: The method of this invention is a complete method that simultaneously uses the confidence-based service time window, the effective completion value of the service, and access risk assessment; the no_confidence method removes the confidence mechanism strategy and uses only the deterministic service time; the no_value method removes the service value strategy, still predicts the service window, but does not distinguish between service priority and timeliness value. The comparison methods include: the instant_rate method uses an instantaneous rate strategy, prioritizing links with higher current rates; the geom_lifetime method considers link lifespan, primarily based on the time before the vehicle leaves the coverage area; and the direct_only method is a direct-connection strategy, disallowing the use of candidate cluster head relays.

[0140] The implementation output includes cycle-by-cycle recording and overall summary. Cycle-by-cycle recording is used to observe the number of vehicles, the number of vehicles waiting to be served, throughput, effective throughput, ACK rate, average queue length, number of handovers, control check range, and window mismatch. Evaluation metrics include pre-loss service completion rate, effective completion throughput, pre-deadline completion rate, transmission success rate, and average latency. The pre-loss service completion rate represents the proportion of services transmitted before a vehicle leaves the coverage area or the communication link fails, used to measure whether resource allocation prioritizes resources for links with sufficient service time and the ability to complete services in a timely manner. Effective completion throughput represents the amount of data successfully transmitted per unit time after ACK confirmation, within the link's validity period and the service deadline; this metric not only reflects transmission speed but also excludes timeouts, connection interruptions, and incomplete data, better reflecting the actual ability of resources to be converted into effective service transmission. The deadline completion rate represents the proportion of all arriving services that are fully transmitted before their respective deadlines. It is used to evaluate the scheduling method's ability to guarantee time constraints. A higher rate indicates that the system prioritizes urgent services and reduces the number of services that lose their usability due to late completion. The transmission success rate represents the proportion of transmitted data that has received ACK confirmation. It primarily measures the reliability of wireless transmission and whether the selected links and resource blocks are suitable for the current communication state. A lower rate typically indicates unstable link quality, significant interference, or a mismatch between resource allocation and actual link capacity. Average latency represents the average time elapsed from when a service enters the queue to when it successfully completes transmission. It is used to evaluate the system's response speed and queue scheduling efficiency. A lower average latency indicates shorter service waiting and transmission times. However, it should be analyzed in conjunction with the deadline completion rate and effective throughput to avoid achieving lower latency by only serving a small number of easily completed services.

[0141] The relevant experimental data are as follows:

[0142] Table 1 Comparison of resource allocation performance under different ablation conditions

[0143] Business completion rate before losing contact Effective throughput / Mbps Completion rate before deadline Transmission success rate Average latency / ms Method of the present invention 0.8016 10.03 0.8749 0.8274 300.67 no_confidence method 0.4676 6.19 0.7932 0.6294 364.48 no_value method 0.3847 8.02 0.4546 0.8039 1056.29

[0144] Table 1 presents the ablation experiment results. As shown in Table 1, the method of this invention outperforms both the no_confidence method (removing the confidence window mechanism) and the no_value method (removing the service value mechanism) across all five indicators. This indicates that both mechanisms jointly influence resource allocation results from two aspects: "judging the link's sustainable service capability" and "distinguishing between service priority and timeliness." Compared to the no_confidence method, the method of this invention improves the pre-disconnection service completion rate, effective throughput, pre-deadline completion rate, and transmission success rate; simultaneously, the average latency decreases from 364.48ms to 300.67ms. This demonstrates that by introducing a confidence-based service time window, the system can more effectively identify links that still possess stable transmission conditions for a future period, avoiding excessive resource allocation to links that are about to fail or have insufficient actual transmission capacity. Therefore, the improvement in pre-disconnection completion rate, effective throughput, and transmission success rate is particularly significant. Compared to the no_value method, the method of this invention also improves the service completion rate before disconnection and the completion rate before the deadline; the average latency decreases from 1056.29ms to 300.67ms. It is worth noting that the transmission success rate of the no_value method still reaches 0.8039, which is only 2.35 percentage points different from the 0.8274 of the method of this invention, but its completion rate before the deadline and average latency have deteriorated significantly. This shows that simply ensuring that "data can be successfully transmitted" cannot guarantee that "valuable and urgent services can be completed in a timely manner". The main function of the effective service completion value mechanism is to prioritize resources for services with higher priority, longer waiting time, tighter deadlines, and the ability to be completed before disconnection. Therefore, its main improvement is reflected in service timeliness, the ability to complete the service before disconnection, and queue waiting time, rather than just the physical layer transmission success rate.

[0145] Table 2 Performance Comparison of the Invention Method and Typical Resource Allocation Strategies

[0146] Business completion rate before losing contact Effective throughput / Mbps Completion rate before deadline Transmission success rate Average latency / ms Method of the present invention 0.8016 10.03 0.8749 0.8274 300.67 instant_rate method 0.2777 5.22 0.3808 0.7909 905.10 geom_lifetime method 0.3284 5.05 0.4197 0.7991 656.91 direct_only method 0.1628 4.55 0.2563 0.7143 1360.61

[0147] As shown in Table 2, the method of this invention exhibits significant advantages in terms of service completion rate before disconnection, effective completion throughput, completion rate before the deadline, and average latency. This indicates that by considering the confidence-based service window and the value of effective service completion, it can make fuller use of limited communication time and wireless resources. Compared with the instant_rate method, the method of this invention significantly improves the service completion rate before disconnection, effective completion throughput, and completion rate before the deadline; the average latency is reduced from 905.10ms to 300.67ms. This suggests that selecting a link solely based on the instantaneous rate tends to favor links with higher current rates but shorter remaining service time, while the method of this invention focuses more on whether the link can continue until the service is completed. Compared to the geom_lifetime method, the method of this invention significantly improves the service completion rate before disconnection, effective completion throughput, and completion rate before the deadline, while reducing the average latency by approximately 54.23%. This indicates that although the geom_lifetime method considers the geometric connection duration, it does not fully reflect factors such as link quality fluctuations, control time overhead, and service time limits. Therefore, relying solely on the geometric lifetime cannot accurately represent the actual effective time available for service transmission. Compared to the direct_only method, the advantages of the method of this invention are most obvious: the service completion rate before disconnection increases from 0.1628 to 0.8016, the effective completion throughput increases from 4.55Mbps to 10.03Mbps, and the completion rate before the deadline increases from 0.2563 to 0.8749. This shows that when only direct connection between vehicles and roadside units is used, the available service paths are significantly limited when vehicles are at the coverage edge or the direct connection link quality is poor. Introducing candidate cluster head relays can expand the range of available access choices, providing some vehicles with transmission paths with longer durations or more stable quality. In summary, Table 2 shows that the advantages of the method of the present invention are not only reflected in the instantaneous transmission rate, but also in the fact that a higher proportion of services can be completed before the link fails and before the deadline, and more allocated wireless resources are converted into actual effective transmission data, while significantly reducing service waiting and completion latency.

[0148] In summary, this embodiment completes status reading, candidate link establishment, future service capability evaluation, service value judgment, access selection, resource block allocation, actual transmission feedback, and next cycle parameter update in a real vehicle trajectory and semi-synthetic communication environment. Specific road lengths, vehicle numbers, roadside unit numbers, service types, and thresholds can all be adjusted according to the actual deployment environment. As long as the ability of a link to continuously carry services is evaluated using a confidence-based service time window, and access selection and resource allocation are based on the value of effective service completion, the method described in this invention can be implemented.

Claims

1. A method for dynamic resource allocation based on confidence serviceable time window for vehicle-to-everything (V2X), characterized in that, Includes the following steps: S1: In each cycle, collect data such as the location, speed, acceleration, direction of movement, and link quality of vehicles with communication needs, and construct a candidate link set; S2: Based on the predicted distance, signal-to-noise ratio, and control time overhead of the candidate links, the probability of the candidate links remaining serviceable is predicted by perturbation samples, and the confidence serviceable time window of the candidate links is obtained. S3: Based on the confidence service time window, calculate the maximum amount of business that can be completed within the window, the degree of business completion, and the effective value of business completion, which are used to determine whether the link can reliably complete the business before the connection is lost; S4: Based on the candidate link's credible service time window, effective service completion value, control time overhead, access object load, and future failure risk, select the vehicle to use direct connection or cluster head relay access method, and perform resource allocation; S5: After each cycle ends, update the service queue, link risk trigger value, and confidence level based on the actual transmission results and feedback. 2.The method of claim 1, wherein, In step S1, the candidate link set is constructed as follows: First, a candidate set of access objects is constructed , , for a set of vehicles; the construction method is: wherein, representing a vehicle at time a set of selectable candidate access objects, representing a set of roadside units that can satisfy the direct link coverage and minimum link quality requirements, representing a set of vehicles that are within a communication range of the vehicle and have relay capability; Secondly, construct a candidate link set. : in Indicates vehicle At any moment Select access object The links that form a chain.

3. The method for dynamic resource allocation in vehicle-to-everything (V2X) networks based on a confidence-based serviceable time window as described in claim 1, characterized in that... In step S2, the confidence-to-serve time window is calculated as follows: First, calculate candidate links. At any moment Deterministic service time window , Represented as: in, For link At any moment Over time Predicted geometric serviceability time; For link At any moment The permitted service period ends when the business is closed. For link At any moment The link quality is consistently above the minimum threshold for a stable service time. To control time expenditure; , , and The calculation method is as follows: 1) Command Link The sending and receiving ends are respectively and ,vehicle At any moment Predicted elapsed time The position is The calculation method is as follows: in, , , vehicles At any moment Position, velocity, and acceleration; link At any moment Predicted elapsed time distance The calculation method is as follows: The calculation method is as follows: in, For link The corresponding coverage radius, when When the direct link is used, the coverage radius of the roadside unit is taken. When considering the workshop link, the communication radius between vehicles should be taken. For the time domain; 2) The calculation method is as follows: in, For link The deadline for undertaking the service; 3) Computation link At any moment Over time Predicted link gain The calculation method is as follows: in, The path gain constant. To represent the distance correction term, it is a small positive number with the same dimensions as the distance. This is the path loss index. This refers to the shadow fading coefficient; Computational Link At any moment Predicted elapsed time signal-to-noise ratio The calculation method is as follows: in, For link The transmission power, For noise power, For link A set of links where resource reuse conflicts or interference conflicts exist; The calculation method is as follows: in, This is the minimum serviceable link quality threshold; 4) The calculation method is as follows: in, , , These represent the time for access establishment, control signaling and handover, and local rescheduling, respectively. Secondly, in the Determine the link under each disturbance sample. In the future Serviceable events within , Represented as: in, A value of 1 indicates that the value is in the first position. Downlink perturbation sample In the future The service is continuously available; a value of 0 indicates that the service is not available. In the first Downlink perturbation sample At any moment Predicted elapsed time distance, In the first Downlink perturbation sample At any moment Predicted elapsed time The signal-to-noise ratio, In the first Downlink perturbation sample At any moment Predicted elapsed time The remaining queue length, To allow for the upper limit of the remaining queue, This is an indicator function that takes the value 1 when the condition is true, and 0 otherwise; Calculation in Downlink perturbation sample In the future Probability of continuous service within , The calculation method is as follows: in, The total number of perturbation samples; Computational Link At that moment Confidence level The following confidence service time window , The calculation method is as follows: in The initial value is It is updated at the end of each cycle.

4. The method for dynamic resource allocation in vehicle-to-everything (V2X) networks based on a confidence-based serviceable time window as described in claim 1, characterized in that, In step S3, the calculation methods for the maximum achievable business volume, business completion rate, and effective business completion value are as follows: Computational Link At any moment effective rate The calculation method is as follows: in For effective load factor, For the bandwidth of a single resource block, For link At any moment Signal-to-noise ratio; Computational Link Within the confidence-available service time window Maximum number of transactions that can be completed within the specified timeframe Calculation method: Computational Link At any moment Business completion rate The calculation method is as follows: in, For link At any moment The remaining queue length, To prevent extremely small positive numbers with a denominator of zero; Computational Link At any moment Current business value and effectively achieve value in business The calculation method is as follows: in, This is the business priority weighting coefficient. This is the value decay coefficient. For business in the link The waiting time; when A value of 0 indicates that the service could not be completed before the loss of contact; when... The timeframe indicates the effective value that the business could have achieved before the loss of contact.

5. The method for dynamic resource allocation in vehicle-to-everything (V2X) networks based on a confidence-based serviceable time window as described in claim 1, characterized in that, In step S4, vehicle access selection and resource allocation are performed as follows: First, calculate the vehicle Select candidate access object At any moment Comprehensive access efficiency The calculation method is as follows: in, , and vehicles Through candidate access objects The formed link at time The confidence available service window, the value of effective business completion, and the time overhead of access and switching control. Candidate access objects The load value, For vehicles Through candidate access objects The formed link at time The future failure risk value; , and As a normalized benchmark, , , , and These are non-negative weighting coefficients; The calculation method is as follows: in, For vehicles Through candidate access objects The probability that the established link will remain serviceable within the confidence service time window; Secondly, select the vehicle access method. The selection method is as follows: in The link formed between the vehicle and the selected access object. Indicates link The risk trigger value, when Time indicates vehicle Established links If the situation deteriorates and communication requirements cannot be met, a new vehicle access method must be selected. when When belonging to a roadside unit set, vehicles use direct connection to enter; when When a vehicle belongs to a candidate cluster head set, it joins the corresponding cluster and accesses the network via a cluster head relay. The link formed for the vehicle's selected access object Calculate resource allocation weights The calculation method is as follows: in, For link At any moment The remaining queue length, For link At any moment The confidence available service time window For link At any moment Business completion rate For link At any moment The current value of the business For link At any moment The probability of continued availability within the lower confidence service time window; Finally, a weighted greedy algorithm based on conflict graphs is used to implement resource allocation. The steps are as follows: Step 1: Construct a set of service links based on the selection results of the vehicle access objects. and resource block set Calculate the revenue from resource blocks used by each link. The calculation method is as follows: in For link Use resource blocks Time rate; Step 2: Construct a set of conflict edges based on resource reuse interference and relay dependency relationships. ; Step 3: Sort the revenue of the set of links to be served from high to low; Step 4: Establish the objective function for resource allocation ,in It is a binary variable; a value of 1 indicates a resource block. Assigned to link A value of 0 indicates that the resource allocation is unallocated; the constraints to be satisfied by resource allocation are: ; ; ; ; Among them, condition 1 means that each resource block can be allocated to at most one non-multiplexed link; condition 2 means that conflicting links cannot share the same resource block; condition 3 means that the end-to-end rate of the relay path is limited by the weaker segment of the access side and the backhaul side; and condition 4 means that the total access traffic of cluster head members cannot exceed the remaining forwarding capacity of the cluster head. ; A greedy algorithm is used to solve the objective function to obtain the resource allocation result.

6. The method for dynamic resource allocation in vehicle-to-everything (V2X) networks based on a confidence-based serviceable time window as described in claim 1, characterized in that, In step S5, the update methods for the business queue, link risk trigger value, and confidence level are as follows: The business queue is updated at the end of each cycle as follows: in, , These are the remaining queue lengths at the start of the next cycle and the end of the previous cycle for a single link, respectively. This refers to the actual amount of successfully completed data confirmed by ACK within one cycle on a single link. This refers to the new traffic volume arriving at the start of the next cycle for a single link; Risk trigger value It is 0 in the initial period, and is updated in subsequent periods according to the following calculation method: in, To establish a confidence threshold for the available service time window, This represents the threshold for business completion. This is the minimum threshold for the probability of continued service. The threshold for effective business completion. This represents the cumulative number of NACKs over the most recent N periods. The threshold for the number of NACKs; Update confidence level : Expected data transfer volume within the calculation period The calculation method is as follows: in For binary variables, For link Use resource blocks Time rate; Refers to a period of time; Computational resource allocation implementation deviation The calculation method is as follows: in This represents the actual amount of successfully completed ACK confirmations for all links in the previous cycle. To prevent extremely small positive numbers with a denominator of zero; Confidence level Update according to the following calculation method: in The updated confidence level; Update the step size to reflect the confidence level; This is a range operator, and its calculation method is as follows: If the value is less than the lower limit, the lower limit is used; if it is greater than the upper limit, the upper limit is used; if it falls within the range, the original value remains unchanged. and They are respectively The minimum and maximum thresholds; The allowable error for the target.