A seamless handover method for a hybrid communication network oriented to automatic driving

Abstract

The application relates to the technical field of wireless communication networks, and discloses a seamless switching method for a hybrid communication network for automatic driving, which comprises the following steps: a network function of a core network receives planning track information of an automatic driving terminal and continuously gathers a plurality of terminal conventional wireless measurement reports in a network; in response to the planning track, instant statistics are performed on the gathered reports to generate dynamic wireless live data representing a current optimal service cell mapping relationship; the planning track is projected to the dynamic wireless live data to calculate a certain switching point and a target cell; wireless resources are reserved in the target cell in advance according to an instruction; when the automatic driving terminal reaches the certain switching point, the automatic driving terminal is triggered to perform instantaneous switching without resource negotiation by using the reserved wireless resources; the application changes the switching management from passive response to active pre-service, generates a dynamic decision basis by using group real-time measurement data, eliminates resource negotiation delay, and guarantees the continuity of automatic driving communication.

Description

Technical Field

[0001] This invention relates to a seamless switching method for hybrid communication networks for autonomous driving, belonging to the field of wireless communication network technology. Background Technology

[0002] Currently, mobility management is the core mechanism for ensuring terminal connection continuity in wireless communication networks. The industry consensus and technical assumption is that the network passively determines whether a terminal has moved to the edge of needing a handover by listening to the terminal's reports on past and current wireless signal measurements, and only initiates the handover process at that moment. This passive response method, which relies on historical and current measurement data, has been proven stable and effective in traditional communication services. However, as the next generation of mobile communication core networks evolves to support ultra-low latency and ultra-high reliability services such as autonomous driving, the above-mentioned traditional mechanism based on passive measurement shows its limitations. When an autonomous vehicle is driving at high speed towards an area where the signal is known to be rapidly attenuating, although its upper-layer navigation system has actively planned the future path, the wireless network carrying the communication is unaware of this active planning intention. The network side still adopts its passive response process, that is, it must wait for the vehicle's measurement report to reach the handover threshold before it begins to search for and allocate resources for the terminal in the target cell. At this time, the vehicle's physical location is often already deep in the signal attenuation area. This inherent time window caused by the network mechanism's perception lag and resource negotiation process directly causes the interruption of critical services.

[0003] To alleviate this problem, while optimization methods such as increasing the frequency of measurement reporting or lowering the handover decision threshold have been attempted in the field, these improvements are merely optimizations within the existing passive framework and do not change the fundamental timing of the perception, decision-making, and action process, thus failing to eliminate decision lag in principle. Regarding the above issues, some solutions in the field attempt to optimize resource allocation strategies, but their focus is often on adjusting data transmission priorities based on the vehicle's current operating state, rather than proactively managing the handover process based on future mobility intentions. For example, Chinese invention patent CN118901273A discloses a communication network for autonomous vehicles. The core of this scheme for dynamic allocation of network resources lies in determining a data transmission priority strategy based on one or more operating states of the autonomous vehicle (AV), such as current environmental operating conditions, driving events, or AV operating conditions, and dynamically allocating communication resources for auxiliary data accordingly. However, this method is essentially still a passive response to the current state. It optimizes the priority of data content but does not solve the core problem that this invention addresses: that is, because the network side cannot predict the precise future trajectory of the vehicle, the handover decision is still lagging when crossing cell boundaries at high speed, and the risk of connection interruption due to resource negotiation at inevitable handover points cannot be eliminated.

[0004] Therefore, the technical problem to be solved by this invention is how to provide a method that enables wireless communication networks to obtain the future driving intentions of terminals and allocate resources in advance to achieve seamless switching. Summary of the Invention

[0005] This invention provides a seamless switching method for hybrid communication networks for autonomous driving. Its main purpose is to solve the problem that existing wireless communication networks rely on passive measurement, which leads to decision lag and makes it impossible to obtain the future intentions of the terminal for proactive resource reservation, thus making it difficult to meet the seamless switching requirements of high-reliability services such as autonomous driving.

[0006] To achieve the above objectives, this invention provides a seamless handover method for hybrid communication networks for autonomous driving, applied to the core network side of a wireless communication network. The network functions in the core network pre-store a basic wireless environment map, which represents the static mapping relationship between geographical road topology and serving cells and candidate serving cells. The method includes:

[0007] In the receiving step, the network function receives the planned trajectory information reported by the autonomous driving terminal, which represents its future driving intention;

[0008] In the aggregation step, the network function continuously aggregates routine wireless measurement reports from multiple terminals in the network, which include their current geographical location and wireless measurement values.

[0009] In the dynamic live feed generation step, the network function responds to the planned trajectory information and performs an instant statistical analysis on the reports in the regular wireless measurement reports whose geographical locations are located within the area along the planned trajectory, so as to generate a dynamic live feed based on the consensus of the group, representing the statistical results of the geographical locations along the planned trajectory and the current wireless measurement values.

[0010] In the pre-image step, the network function projects the planned trajectory information onto the basic wireless environment map and combines it with the statistical results of the dynamic wireless situation to identify the geographical location where the optimal serving cell on the planned trajectory changes as the inevitable handover point, and determines the serving cell corresponding to the inevitable handover point as the target cell.

[0011] The resource reservation step involves the network function instruction target cell pre-allocating and maintaining the radio resources required for handover before the autonomous driving terminal reaches the necessary handover point.

[0012] In the instantaneous handover process, when the autonomous driving terminal reaches the necessary handover point, the core network triggers the autonomous driving terminal to perform the handover. The handover does not go through resource negotiation and directly utilizes the radio resources reserved in the target cell.

[0013] Preferably, before performing the resource reservation step, the network function further includes: a congestion prediction step: querying a future resource reservation ledger maintained for the target cell to determine whether performing the resource reservation step at the target time slice corresponding to the inevitable handover point will lead to resource conflict; a spatiotemporal decoupling step: when it is predicted that a resource conflict will occur, the network function first finds a candidate serving cell with available resources in the target time slice in the basic radio environment map, and resets the resource reservation step to the candidate serving cell; or when no candidate serving cell with available resources is found, it finds an adjacent time slice with available resources in the future resource reservation ledger of the target cell and performs the resource reservation step at an offset.

[0014] Preferably, after the resource reservation step and before the instantaneous handover step, the method further includes: an active detection step: the network function instructs the target cell to perform an internal radio quality probe on the radio resources reserved for the autonomous driving terminal to obtain the instantaneous interference level of the reserved resources; and an arbitration step: the network function, based on... The rules confirm or reject instantaneous switching steps, among which, The measurement value represents the instantaneous disturbance level. This represents a preset handover execution threshold. When the rule is met, the instantaneous handover step is confirmed. Error correction step: When the instantaneous handover step is rejected, the network function immediately selects a candidate serving cell from the basic radio environment map that is geographically available at the inevitable handover point, and repeats the resource reservation step, active detection step, and arbitration step for the candidate serving cell.

[0015] Preferably, in the receiving step, when the network function does not receive the planned trajectory information, the receiving step is replaced by a trajectory inference step, which includes: a historical data reuse step: the network function acquires and analyzes the historical sequence of the regular mobility measurement reports reported by the autonomous driving terminal; a vector and path matching step: based on the historical sequence, the current movement vector of the autonomous driving terminal is calculated, and the current movement vector is matched to the geographic road topology represented by the basic wireless environment map; a trajectory generation step: along the geographic road topology, the current movement vector is prospectively projected to generate an inferred future trajectory; and the dynamic real-time generation step and the pre-mapping step are performed based on the inferred future trajectory.

[0016] Preferably, the pre-mapping step utilizes the statistical results of dynamic wireless conditions to cover any pre-stored static wireless measurement data based on historical drive tests or simulations in the core network to perform identification and determination.

[0017] Preferably, the planned trajectory information is simplified into a sequence of geographic coordinates of a set of key path inflection points.

[0018] Preferably, the receiving step further includes: receiving the current speed vector reported by the autonomous driving terminal; and the pre-mapping step further includes: calculating the expected arrival timestamp of the inevitable switching point based on the planned trajectory information and the speed vector; the resource reservation step is triggered within a preset time window before the arrival timestamp.

[0019] Preferably, the vector and path matching step calculates the current movement vector as the velocity vector; and the pre-mapping step further includes: calculating the expected arrival timestamp of the inevitable switching point based on the inferred future trajectory and velocity vector; the resource reservation step is triggered within a preset time window before the arrival timestamp.

[0020] Preferably, the dynamic real-time generation step further includes: a deviation identification step: comparing the statistical results of the dynamic wireless real-time with the static wireless measurement data representing the geographical area pre-stored in the core network to identify whether there is a statistical deviation between the two; and a dynamic entry generation step: when the statistical deviation exceeds a preset threshold, generating a time-sensitive dynamic temporary coverage area entry, which is used to cover the static wireless measurement data in the pre-mapping step.

[0021] Preferably, it also includes: a trajectory cancellation step: when the network function receives a trajectory cancellation signal from the autonomous driving terminal, it immediately releases the wireless resources reserved for the planned trajectory; and repeats the dynamic real-time generation step, the pre-mapping step, and the resource reservation step according to the new planned trajectory information reported by the autonomous driving terminal.

[0022] Compared with the prior art, the beneficial effects of the present invention are:

[0023] 1. By receiving and processing the planned trajectory information representing the future driving intention of the terminal on the core network side of the wireless communication network, a new information input path is established. This path enables network-side functional entities (such as MIPAP) to obtain deterministic prediction of the future mobility state of the terminal, thereby enabling them to proactively reserve radio resources before the terminal reaches the inevitable handover point. This operation mode transforms the passive response process that relies on lag measurement reports in traditional mobility management into a proactive pre-service process, avoiding the time delay inherent in the temporary negotiation of resources during handover.

[0024] 2. The wireless environment digital twin database used in this method passively aggregates routine real-time reports from multiple terminals in the network, containing their current geographical location and wireless measurement values. Based on group consensus, it performs real-time verification of static mapping relationships, constructing a dynamic correction mechanism. This mechanism utilizes existing standard signaling data streams in the network, enabling it to generate timely dynamic temporary coverage area entries in the face of temporary and dynamic wireless environmental changes, such as crowd gatherings or temporary obstructions. This ensures that the environmental data used in the pre-mapping step closely approximates the actual physical reality, avoiding reliance on outdated static data. This leads to the failure of handover decisions. Meanwhile, by maintaining a future resource reservation ledger for the target cell, an active congestion prediction step for the target time slice is added before executing the resource reservation step. When congestion is predicted to occur, the system initiates a spatiotemporal decoupling strategy, or selects a suboptimal alternative cell with available resources, or executes at a slight offset from the adjacent available time slice of the target cell. This mechanism solves the resource conflict caused by high-density terminals, such as convoys, requesting reservations in close proximity along the same trajectory. It ensures that each terminal in the group can obtain effective resource reservations, so that the active handover method remains available under complex traffic flow.

[0025] 3. After the resource reservation step and before the instantaneous handover step, an active probing step for specific reserved radio resources is introduced, executed by the target cell under network function instructions. By acquiring and verifying the instantaneous interference level of the reserved resource, the network side completes the final confirmation of the true availability of the target channel before the terminal performs the handover. If the arbitration fails, the error correction step is immediately initiated, a suboptimal alternative cell is selected, and the probing process is repeated. This closed-loop verification process ensures that the channel migrated to during the final zero-negotiation handover is clean and available, avoiding the terminal handover to a resource that has been contaminated by instantaneous interference. When the terminal does not actively report the planned trajectory information, a trajectory inference step is also provided. This step acquires and analyzes the historical sequence of the regular mobility measurement reports reported by the terminal, calculates its current movement vector, matches it to the geographical road topology in the wireless environment digital twin database, and then projects it to generate the inferred future trajectory. This alternative path utilizes the existing standard signaling in the network, enabling the active reservation mechanism on the core network side to be applied to existing terminals that do not support the new protocol, thus expanding the applicability and backward compatibility of this technical solution. Attached Figure Description

[0026] Figure 1 This is a schematic diagram of the main process of the seamless switching method of the present invention;

[0027] Figure 2 This is a comparison chart of the RSRP of dynamic wireless real-time data and static map data in this invention;

[0028] Figure 3This is a schematic diagram of the core network architecture and interaction of the present invention, which includes MIPAP functionality. Detailed Implementation

[0029] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the scope of protection of the present invention.

[0030] This invention provides a seamless handover method for hybrid communication networks for autonomous driving. Its core logic is executed on the core network side of the wireless communication network, specifically by a newly added or embedded network function (NF) entity within an existing AMF, SMF, or PCF, such as a Mobility Intent Awareness and Proactive Pre-provisioning (MIPAP) function. The method's architecture primarily includes a high-priority intent receiving module, a dynamic real-time generation module based on group consensus, a pre-mapping module fusing static topology and dynamic real-time data, an active resource reservation module, and a zero-negotiation instantaneous handover execution module. These modules work together to transform network mobility management from a passive response mode to a proactive pre-service mode. One initial state of this method is that the network function in the core network pre-stores a basic radio environment map. This map is not a real-time radio quality map, but rather a static topology data structure representing the network's physical deployment. Specifically, it stores the geographical road topology managed by the operator (e.g., G2 highway K100 to K1). The static logical correspondence between the 10-segment road and all serving cells along that segment, such as cell gNodeB-ID-A and its alternative serving cells gNodeB-ID-B and gNodeB-ID-C, is generated by network planning tools or historical road test data and serves as the geographical topology benchmark for subsequent pre-mapping steps. To obtain inputs that are not typically available in existing networks, the network function is configured in the receiving step to establish a specific session with the autonomous driving terminal when it is detected to receive the planned trajectory information reported by the terminal, which represents its future driving intention. To reduce the signaling overhead of the air interface, this information is preferably simplified into a set of geographical coordinate sequences of key path inflection points, such as [(lat1, lon1), (lat2, lon2)]. These inflection points, such as highway ramps and urban road curves, are sufficient to represent the driving path for the next 30 seconds. At the same time, to achieve accurate time prediction, the receiving step also includes receiving the current speed vector reported by the terminal, such as a vector containing [heading angle, speed]. The vector.

[0031] Due to the static limitations of basic wireless environment maps, which cannot reflect instantaneous changes in the wireless environment such as temporary obstructions and dynamic interference caused by crowd gatherings, the network function acquires dynamic data through an aggregation step. It is configured to continuously and passively aggregate routine wireless measurement reports from all terminals within the network coverage area, including non-autonomous driving terminals. These reports are part of a standard process and contain [terminal GPS coordinates, timestamp, serving cell ID, RSRP / RSRQ measurement values]. Then, in the dynamic real-time generation step, in response to the received planned trajectory information of a specific terminal, the network function immediately performs an instant statistical query on the aggregated reports, retrieving only... Reports whose geographical locations are within a specific area along the planned trajectory (e.g., within 50 meters on either side of the trajectory route) and whose timestamps are within the last 5 minutes are generated. By statistically analyzing retrieved consensus data, such as 100 reports for a specific road segment, for example, calculating the average RSRP or the vote rate of the best cell, a dynamic wireless situation is generated, representing the geographical location along the planned trajectory and the statistical results of current wireless measurements. In subsequent processing, this dynamic wireless situation is prioritized higher than any pre-stored static wireless measurement data based on historical drive tests or simulations in the core network, thus achieving coverage of outdated data. It should be noted that in some network deployment scenarios, the core network, in addition to... In addition to storing the static topology information contained in the basic wireless environment map, it may also pre-store static wireless measurement data representing specific geographical areas under typical conditions, generated based on historical drive tests or offline simulations, such as the average RSRP / RSRQ baseline value for a specific location. In this case, the dynamic live-line generation step may further include a deviation identification step, which compares the statistical results of the measurement values ​​generated in the dynamic live-line data through real-time statistics with the aforementioned pre-stored static wireless measurement data to identify whether there is a statistical deviation exceeding a preset threshold. This deviation may indicate a temporary change in network status or environmental interference. Accordingly, when a statistical deviation exceeding a preset threshold is identified, the system will take appropriate action. When the threshold is reached, a dynamic entry generation step can also be performed. This involves generating a dynamic temporary coverage area entry with a preset Time To Live (TTL) for the geographic area in the basic wireless environment map or associated database. This entry records the current best cell and its measurement statistics based on the dynamic wireless situation and is given a higher query priority. In the subsequent pre-mapping steps, when the system queries the wireless environment information of a specific geographic location, it will prioritize using the data in the dynamic temporary coverage area entry. If there is no valid temporary entry or the entry has expired, it will fall back to using the dynamic wireless situation results or basic map information generated through real-time statistics.

[0032] In the pre-image step, the network function performs a data fusion calculation. First, it projects the terminal's planned trajectory information onto a basic radio environment map to identify all potential cell boundary crossing points. Then, using the statistical results of dynamic radio surveillance, these crossing points are verified and confirmed to identify the geographical locations where the optimal serving cell on the planned trajectory has indeed changed. These locations are determined as inevitable handover points, and the corresponding optimal serving cell, confirmed by dynamic radio surveillance, is identified as the target cell. Simultaneously, this step combines the received velocity vector (… ) and the path length of the planned trajectory, calculate the expected arrival timestamp of the inevitable switching point. For example, if PHP is in In addition, the arrival timestamp is calculated to be 30 seconds later; to eliminate negotiation delay during the handover process, this method performs a resource reservation step, and the network function is triggered within a preset time window based on the calculated arrival timestamp, for example, before the arrival timestamp. The core network actively sends instructions to the target cell, requesting it to pre-allocate and maintain, for example, all radio resources required for switching to Warm-Standby mode, such as PDCCH and PUSCH / PUCCH channel resources, for the specific terminal. When the physical location of the autonomous driving terminal actually reaches the necessary handover point, the core network, for example, through the AMF, no longer waits for the terminal's measurement report, but actively triggers an instantaneous handover step, sending a handover instruction containing reserved resource information to the terminal. After receiving the instruction, the terminal instantaneously synchronizes on the reserved resources of the target cell without any measurement, random access, or resource negotiation process, thereby compressing the handover interruption time window to near zero.

[0033] To ensure backward compatibility of the method, in the receiving step, when the network function does not receive the planned trajectory information (e.g., the terminal does not support it or a timeout occurs), the receiving step is replaced by a trajectory inference step. In this alternative path, the network function performs a historical data reuse step, acquiring and analyzing the historical sequence (e.g., data from the past 15 seconds) that the autonomous driving terminal has reported for routine mobility measurement reports. Through a vector and path matching step, based on the [GPS, timestamp] data in the historical sequence, the current movement vector of the terminal is calculated, for example, heading due north, speed. The system matches this vector to the geographic road topology represented by the basic wireless environment map, for example, matching it to the northbound lane of the G2 expressway. Then, in the trajectory generation step, a forward projection is performed on the current movement vector along the extension direction of the G2 expressway, for example, projecting the path for the next 30 seconds to generate an inferred future trajectory. This inferred future trajectory and its calculated velocity vector are used as inputs to subsequent steps, enabling the system to trigger subsequent active reservation processes without requiring active reporting from the terminal. In the trajectory inference step, to ensure the engineering reproducibility of the vector-path matching, a particle filter algorithm can be used. This algorithm uses the calculated current movement vector (including position, speed, and heading) as the observation input and utilizes the geographic road topology in the basic wireless environment map as a constraint for state transitions to output a matching path with the highest confidence on the map. The subsequent trajectory generation step dynamically determines the range of the forward projection based on this matching path. Specifically, the system retrieves the distance from the current location to the next critical decision point (e.g., a highway ramp) from the geographic information database of the basic wireless environment map. And set a maximum inference time window. (For example, calibrated to 60 seconds), the final inferred projected distance of the future trajectory identified as The driving distance corresponding to this time window, i.e. ,in This procedure replaces the method of projecting at fixed intervals (e.g., 30 seconds) with the smaller value of the current movement vector rate; to address the proactive congestion problem that may be caused by high-density traffic such as convoys, the network function is configured to perform a congestion prediction step before performing the resource reservation step. This step queries a time-slice-based database maintained for each target cell. A future resource reservation ledger, indexed by unit, is used to determine whether performing a new reservation in the target time slice corresponding to a certain handover point will lead to a resource conflict, for example, if the reservation count exceeds 80% of the available resource threshold. When a conflict is predicted, a spatiotemporal decoupling step is immediately initiated. This step first attempts spatial decoupling, i.e., in the basic radio environment map, finding a geographically similar but resource-available alternative serving cell, and resetting the resource reservation step to that alternative serving cell. If spatial decoupling fails, for example, if no available alternative cell exists, then temporal decoupling is performed, searching for a resource-available adjacent time slice in the target cell's future resource reservation ledger, for example... or The resource reservation step will be slightly offset to be executed in the adjacent time slice.

[0034] To avoid misjudgments caused by the slight time difference between the dynamic radio situation statistics period and the instantaneous handover execution time, such as when reserved resources are contaminated by sudden interference at the last moment, this method introduces a closed-loop verification after the resource reservation step and before the instantaneous handover step. This active probing step involves the network function commanding the target cell to perform an internal, real-time radio quality probe on the specific radio resources (e.g., a specific time-frequency resource block) reserved for the terminal, such as a silent noise floor scan, to obtain the instantaneous interference level of the reserved resources. During the arbitration process, the network function operates based on a preset switching threshold. For example, a threshold representing the minimum signal-to-noise ratio required for URLLC services. The judgment of rules, among which, The measurement value represents the instantaneous disturbance level. The method represents a preset handover execution threshold. When this rule is met, the handover is confirmed. If it is rejected, an error correction step is immediately initiated. A candidate serving cell is selected from the basic radio environment map, and the resource reservation, active detection, and arbitration process is repeated for the candidate serving cell until a resource with an interference level that meets the threshold is found. Finally, to cope with real-time changes in the terminal's driving intention, the method also includes a trajectory cancellation step. When the network function receives a trajectory cancellation signal triggered by a path change from the autonomous driving terminal, such as a temporary lane change by the user or navigation replanning, the network function is configured to immediately release the radio resources reserved for the old planned trajectory to avoid invalid resource occupation. The network function then restarts and repeats the dynamic real-time generation step, the pre-mapping step, and the resource reservation step according to the new planned trajectory information reported by the terminal to ensure that the pre-service always matches the latest intention of the terminal. To make the dynamic real-time generation step deterministic in engineering implementation, this step includes a sample size verification and data cleaning procedure when performing real-time statistics. Specifically, the system presets a minimum sample size threshold. This threshold The value can be set according to the business type and road grade. For example, for highway URLLC business, the value is set to 10 reports. When the number of valid reports in a specific geographic area and time window is less than 10, the value can be set to 10 reports. If this step is not performed, the process reverts to performing pre-mapping using static data from the basic wireless environment map, while if the sample size meets the requirements... Under the premise of ensuring data cleanliness, statistical procedures prioritize data cleaning. For example, the Z-Score method is used to remove outlier reports whose RSRP measurements deviate from the group mean by more than three standard deviations. Finally, on the cleaned sample set, the median is used for statistical analysis of continuous values ​​such as RSRP to avoid the influence of extreme values ​​on the mean. For the determination of the optimal cell, the method with the highest vote rate is adopted.

[0035] Example 1: In a high-density highway scenario with autonomous vehicles, a typical challenge faced by the 5G28 core network is ensuring the continuity of URLLC services when traversing signal-shadowed areas such as large overpasses or short tunnels. Traditional measurement-based passive handover methods, with their inherent perception lag and resource negotiation delays, are prone to service interruptions in such scenarios of rapid signal attenuation. This problem is further exacerbated when the wireless environment undergoes dynamic changes (e.g., temporary construction obstructions causing instantaneous signal degradation in the optimal cell in static planning). An autonomous vehicle is... The speed approached the shadow area of ​​an overpass, and its in-car navigation system had already planned for the future. The vehicle receives the driving path and, following the receiving steps, simplifies it into planned trajectory information for key path inflection points, reporting this information, along with its speed vector, to the MIPAP function in the 5G28 core network. The MIPAP function immediately responds to this planned trajectory, projecting it onto a pre-stored basic radio environment map to identify the topological boundary between cell A (the current serving cell) and cell B (the statically optimal target cell) at the interchange entrance K105+300m. Simultaneously, the MIPAP function executes a dynamic real-time generation step, using the area along the planned trajectory as an index to perform an instantaneous analysis of its continuously aggregated conventional radio measurement report database. A query revealed that over the past 5 minutes, all terminals passing through the interchange entrance, including other vehicles and ordinary mobile phones, consistently reported abnormally low RSRP / RSRQ statistics for cell B, while the RSRP / RSRQ statistics for the alternative serving cell C were optimal. This dynamic radio situation result overlaid the static data in the basic radio environment map. Therefore, in the pre-mapping step, the MIPAP function integrated the above information, identifying the interchange entrance K105+300m as a necessary handover point, and determining cell C, verified by the dynamic radio situation, as the target cell. Subsequently, the MIPAP function... Using the velocity vector and distance, the estimated arrival timestamp of the vehicle is calculated to be 32.7 seconds later; the 5G28 core network does not wait, but instead... At 32.2 seconds, the resource reservation step is executed, and the target cell C is actively instructed to pre-allocate radio resources for the terminal. When the vehicle arrives at the inevitable handover point at 32.7 seconds, the core network immediately triggers the instantaneous handover step and sends a handover instruction containing the reserved resource information of cell C to the terminal. The terminal directly uses the reserved resources to access cell C without negotiation.

[0036] Example 2: To objectively evaluate the radio resource management effect of the method of the present invention on the handover performance of autonomous driving services on the 5G28 core network side, an experimental platform based on network simulation and road test data playback was built. This platform includes a server cluster simulating the 5G28 core network (including AMF, SMF, and MIPAP functions) and multiple radio frequency channel simulators simulating the radio access network (gNodeB). The test object is an autonomous driving terminal (OBU) simulator carrying URLLC (Ultra-Reliable Low-Latency Communication) services. The test scenario is set as the signal shadow area of ​​the highway overpass in Example 1. The OBU simulator uses... The vehicle was traveling at a speed towards the area, which is characterized by a sharp drop in RSRP (Reference Signal Received Power) within a 50-meter range. To verify the synergistic effect of the dynamic real-time generation step and the resource reservation step in the method of this invention, especially its effectiveness in the face of dynamic changes in the wireless environment, a temporary signal blockage was introduced at the coverage edge of the static optimal target cell B marked on the basic wireless environment map 10 seconds before the OBU simulator reached the inevitable handover point. The dynamic interference source (simulating the parking of large engineering vehicles) caused the statically suboptimal cell C to become the actual optimal cell at that moment. The experiment involved four operating modes undergoing 1000 repeated tests. The key performance indicators were Handover Interruption Time (HCI) and Handover Success Rate. Control group 1 was configured to run the standard passive handover method, with the OBU simulator only reporting routine wireless measurement reports. Control group 2 was configured to run an incomplete reserved method, where the MIPAP function received the planned trajectory information but the dynamic real-time generation step was not enabled, and its pre-mapping step... The first group is configured to run the basic wireless environment map (static data) and perform the resource reservation step. The second group is configured to run another incomplete reservation method, namely, the MIPAP function performs the receiving step, the aggregation step, the dynamic situation generation step, and the pre-mapping step, but the resource reservation step is not enabled. When the handover is triggered at the inevitable handover point, it must perform real-time resource negotiation with the target cell. The third group is configured to run the complete method in the specific implementation method, including the receiving step, the dynamic situation generation step, the pre-mapping step, the resource reservation step, and the instantaneous handover step. Under the above experimental settings, the statistical mean results of each group test are shown in Table 1.

[0037] Table 1: Performance comparison of different switching methods in dynamic interference scenarios.

[0038]

[0039] Referring to the experimental data in Table 1, control group 1 experienced a high failure rate in its immediate negotiation process because of perception lag, triggering the handover only after the OBU had penetrated deep into the signal shadow area. Although control group 2 utilized planned trajectory information for pre-mapping, its reliance on static data from the basic radio environment map led to the incorrect identification of the interfered cell B as the target cell, resulting in a high handover failure rate due to resource reservation being performed on the wrong cell. Control group 3 correctly identified cell C as the actual optimal cell through a dynamic real-time generation step, improving the handover success rate; however, its lack of resource reservation resulted in a 28.3ms immediate negotiation delay. However, it still cannot meet the specific requirements of URLLC services; the prototype of this invention obtains the predicted time through trajectory planning, avoids dynamic interference through the dynamic real-time generation step (correctly selects cell C), and locks resources in advance through the resource reservation step. Finally, its instantaneous handover step achieves an interruption time of 1.8ms and a handover success rate of 99.99%; the test data shows that the combination of the three steps of trajectory planning information, dynamic real-time generation and resource reservation included in the method of this invention is necessary for achieving the high reliability and low latency connection continuity required for autonomous driving on the 5G28 core network side.

[0040] To further clarify in principle the indispensability of the dynamic wireless real-time generation step in avoiding dynamic interference and ensuring the correctness of pre-handover decisions, the following comparative example is set up.

[0041] Comparative Example 1: This Comparative Example 1 aims to verify the technical effectiveness of an active handover method that relies solely on planned trajectory information and a basic wireless environment map but does not configure the dynamic wireless situation generation step when facing dynamic wireless environment changes. The test platform, network configuration, OBU simulator parameters, and dynamic interference scenario (i.e., introducing a dynamic interference source with temporary signal blockage of 20dB at the coverage edge of the static optimal cell B) used in this Comparative Example are completely consistent with the settings described in Embodiment 2 of this invention. The handover method used in this Comparative Example has the same operating mechanism as Control Group 2 described in Embodiment 2. The specific execution steps are as follows: Receiving step: The MIPAP function in the core network receives the planned trajectory information representing its future driving intention and its current speed vector (120km / h) reported from the OBU simulator; Pre-mapping step: The MIPAP function does not execute the convergence step and the dynamic wireless situation generation step. The network function directly projects the received planned trajectory information onto the pre-stored map. On the basic wireless environment map, since the static map data shows that cell B is the optimal serving cell at location K105+300m, the MIPAP function, based solely on this static data, determines this location as a necessary handover point and incorrectly identifies cell B as the target cell. In the resource reservation step, the MIPAP function, based on the estimated arrival timestamp calculated from the velocity vector and the planned trajectory, actively instructs cell B, which has been contaminated by dynamic interference sources, to pre-allocate and maintain wireless resources for the OBU simulator within a preset time window. In the instantaneous handover step, when the OBU simulator reaches the necessary handover point, the core network triggers the OBU simulator to perform an instantaneous handover to cell B without resource negotiation. Under the above settings, the method of Comparative Example 1 was repeatedly tested 1000 times on the experimental platform and compared with the control group 1 using the standard passive handover method in Example 2, and the sample group of this invention using the complete method of this invention. The measured statistical mean results are shown in Table 2.

[0042] Table 2: Performance comparison of Comparative Example 1 and related methods.

[0043]

[0044] Experimental data shows that although the method in Comparative Example 1 utilizes planned trajectory information to achieve proactive pre-service, it completely lacks a dynamic wireless situation generation step, and its decision-making is based solely on outdated static data in the basic wireless environment map. In scenarios where the wireless environment changes dynamically, this method cannot perceive the physical reality that the static optimal cell B has been severely interfered with, and instead continues to mistakenly select it as the target cell and perform resource reservation. When the OBU simulator is instructed to instantly switch to this incorrect cell B, its access is likely to fail due to the severe degradation of the reserved resource (channel) quality. This failure causes the OBU simulator to spend additional time (e.g., waiting for relevant timers to time out) to abort the failed handover after entering the signal attenuation area, and to fall back to the standard measurement-based passive network search process. At this time, the wireless conditions are worse than when passive handover is triggered in Control Group 1. Therefore, the statistical results of the handover interruption time (122.1ms) and handover success rate (75.2%) in Comparative Example 1 are even worse than those of the traditional passive handover method (Control Group 1).

[0045] Example 3: This example combines Figures 1 to 3 This section describes a seamless switching method for hybrid communication networks used in autonomous driving, such as... Figure 1 As shown, during the receiving step, the terminal acquires the planned trajectory information. The system then determines whether the planned trajectory has been received. If not, the trajectory inference step is initiated to generate the inferred future trajectory. If the trajectory has been received, or after trajectory inference is completed, the system enters the dynamic real-time generation step. This step generates dynamic wireless real-time data based on group consensus. The required data source comes from the aggregation step, which is responsible for aggregating the regular wireless measurement reports from multiple terminals. The generated dynamic real-time data, along with the static road topology and cell mapping information represented by the pre-stored basic wireless environment map, are used together in the pre-mapping step to fuse the trajectory, dynamic real-time data, and basic map, thereby identifying inevitable handover points and target cells. After identification, the system enters the congestion prediction step by querying future data... The source reservation ledger is used to determine whether there is a predicted conflict. If a conflict exists, a spatiotemporal decoupling step is executed, and the request is redirected to the alternative serving cell or adjacent time slice. If there is no conflict, or after the spatiotemporal decoupling is completed, a resource reservation step is executed, instructing the target cell or alternative cell to reserve radio resources. After the reservation is completed, an active detection and arbitration step is initiated to verify the instantaneous interference level of the reserved resources. If the arbitration is rejected, for example due to excessive interference, an error correction step is triggered, an alternative serving cell is selected, and the reservation and detection process is repeated. If the arbitration confirms that the quality meets the standard, the process waits for the terminal to reach the handover point, and finally triggers the instantaneous handover step when the autonomous driving terminal reaches the necessary handover point. This handover does not involve resource negotiation, enabling the autonomous driving terminal to complete a seamless handover using the reserved resources.

[0046] like Figure 2As shown in the figure, this diagram illustrates the comparison of signal strength RSRP (dBm) along the planned trajectory. The dashed line represents the static map data RSRP at location K105+300m, where the signal strength is approximately -92dBm. The solid line represents the dynamic real-time RSRP, where the signal strength drops sharply to -105dBm at this location (within a temporary interference zone). This dynamic real-time data reveals temporary interference that the static map fails to reflect, thus providing a more accurate decision-making basis for the pre-mapping step. Figure 3 As shown, this architecture introduces a new network function entity, MIPAP (Mobility Intent Awareness and Proactive Provisioning), in addition to existing core network functions such as AMF / SMF. The autonomous driving terminal reports standard signaling, such as planned trajectory information, to the existing core network functions via a wireless connection through the current serving cell. Other terminals, acting as sources of regular measurement reports, report regular measurement reports through other cells. These reports are stored in a dynamic real-time data source database of regular measurement reports. The MIPAP function is responsible for querying / reading this database and the static topology of the basic radio environment map, and querying / updating the future resource reservation ledger congestion prediction. The MIPAP function sends its decision results, such as proactive reservation instructions, to the existing core network functions, which then issue resource reservation / probe instructions to the target cell or alternative serving cell. Finally, during handover, the existing core network functions issue an instantaneous handover instruction to the current serving cell, which then forwards it to the autonomous driving terminal.

[0047] Example 4: When deploying the MIPAP function in the 5G28 core network to support autonomous driving URLLC services, relevant control parameters need to be calibrated to ensure handover reliability and avoid invalid occupation of radio resources; among them, the handover execution threshold used for arbitration in the active detection step... and the preset time window triggered by the resource reservation step. The setting procedure for these two parameters is described below; to determine the switching execution threshold To determine the value, perform the following calibration procedure: First, clarify that the target URLLC service is autonomous driving control signaling, and its key performance indicator requirement is that the maximum tolerable packet error rate (PER) is set to [value missing]. The minimum data throughput is Secondly, a network simulation platform is used, which needs to simulate air interface channel conditions and inject controllable interference to establish a test link including an OBU simulator, a target cell gNodeB simulator, and an interference source. On this link, the target URLLC service flow is run, and the interference source injects additive white Gaussian noise (AWGN) interference of different power levels into the time-frequency resource blocks allocated to this service flow. Interference power from Beginning, with The step size increases incrementally; at each interference power level, the operating service flow transmission is no less than Group the data into groups and record the PER and throughput corresponding to each interference level; analyze the test data to identify groups that can meet all preset KPIs (PER). And throughput Maximum interference power threshold If the test shows The KPIs were still met at that time, but... PER exceeds Then determine Finally, based on statistical analysis of interference fluctuations in the network or pre-set system robustness requirements, a... The safety margin value will switch the execution threshold. Set as Subtracting this safety margin value, i.e. This calibration value is then configured in the arbitration step of the MIPAP function.

[0048] To determine the preset time window triggered by the resource reservation step. The following calculation procedure is executed: First, obtain the various system latency parameters required for resource reservation, including the signaling latency from the core network to the target cell. and the processing time within the target community The statistical average or preset upper limit value is obtained by monitoring the performance of the deployed network; here, the value is taken as [value]. and Minimum time for terminal to process handover command The value is determined based on the terminal's reported capabilities or technical specifications; here it is taken as [value]. The basic delay is obtained. Secondly, set a safe buffer distance. This value is determined based on the known maximum error range of the positioning technology used, and is set here to [value missing]. Based on the legally mandated maximum speed limit of the geographical section of road where the necessary switching point is located. (Can be obtained from the geographic information database associated with the basic wireless environment map), this is for (about ), computation time buffer Finally, an engineering scheduling margin is added, set by the network operator based on network load and operation and maintenance strategies. , here is The preset time window is calculated. This calculation result represents the lead time of the MIPAP function triggering resource reservation step relative to the expected arrival time stamp. This procedure correlates the time window setting with measurable network parameters, geographically relevant traffic rules, and operator policies. The result obtained through the above calibration and calculation procedure... and The parameters are applied to the deployment of the 5G28 core network MIPAP function, enabling the method to set the timing of its proactive probing and resource reservation behaviors.

[0049] Example 5: In an urban canyon environment, an autonomous vehicle is approaching a necessary handover point according to its planned trajectory. The MIPAP function has performed a pre-image step to determine cell X as the target cell, and is expected to arrive before the timestamp. The resource reservation step was executed; immediately afterwards, the MIPAP function triggered the active probe step, instructing cell X to perform internal radio quality probe on the specific radio resources reserved for this terminal; after performing the probe, cell X reported the probe results, including the instantaneous interference level it calculated. for The MIPAP function, during the arbitration step, compares this value with the switching execution threshold specified in Example 3. Comparison, because The arbitration result was a rejection; the MIPAP function immediately initiated error correction steps, immediately canceling the reservation on cell X and querying the basic radio environment map to determine that there was a replacement serving cell Y at that geographical location; the MIPAP function repeated the resource reservation step and active detection step for cell Y; the instantaneous interference level reported by cell Y after detection was... ,satisfy Under the conditions, arbitration was passed; finally, when the terminal reached the necessary handover point, the core network triggered an instantaneous handover to cell Y, and the handover process was successfully completed; this process demonstrates that the active detection and arbitration mechanism identified and avoided the pollution of reserved resources on the target cell caused by sudden co-channel interference before the handover was executed.

[0050] Example 6: In a highway platooning scenario, a convoy of 5 autonomous vehicles... Speed, Maintain They travel at intervals, sharing the same planned trajectory, and will arrive at approximately the same time (due to the interval, approximately...). The time difference between the arrival at the same inevitable handover point PHP, the target cell corresponding to PHP is cell P; the resource capacity of cell P is evaluated, and its active congestion threshold is set at a certain threshold in a single cell. A maximum of three handover reservation requests can be processed simultaneously within a time slice; as the convoy approaches PHP, the OBUs of each vehicle sequentially report trajectory information to the MIPAP function; the MIPAP function calculates the estimated arrival timestamps of PHP for the first three vehicles in the convoy. All fell into the target time slice Inside, MIPAP performs a congestion prediction step, queries the future resource reservation ledger of cell P, and discovers... The current time slice reservation count is 0, therefore the resource reservation step has been successfully executed for these 3 vehicles, and... The reserved count is updated to 3; when processing the request for the 4th vehicle, its expected arrival timestamp is... Still fall into MIPAP performs congestion prediction again and finds that the current count of 3 plus the current request of 1 will equal 4, exceeding the threshold of 3, and determines that a resource conflict will occur. MIPAP initiates the spatiotemporal decoupling procedure, first attempting spatial decoupling. A query of the basic radio environment map reveals a backup serving cell Q at the PHP location, and cell Q is... The future resource reservation ledger count for the time slice is 0, so MIPAP resets the resource reservation step for the 4th vehicle to cell Q and updates the ledger in cell Q; when processing the request for the 5th vehicle, its expected arrival timestamp is... Still fall into MIPAP anticipates a conflict in cell P and attempts spatial decoupling, only to find that cell Q is also full (count 1, assuming its threshold is 1). Therefore, it performs temporal decoupling, searching the future resource reservation ledger of cell P. Adjacent time slices were found With available capacity in the time slice (count 0), MIPAP slightly offsets the resource reservation step for the 5th vehicle to be executed in the adjacent time slice and updates the corresponding ledger. Ultimately, the instantaneous switchover command issued to the fleet is adjusted accordingly, ensuring that all vehicles in the fleet utilize the reserved resources (distributed across time slices in cell P). Time slice of community Q and the time slice of community P The seamless transition was completed.

[0051] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0052] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A seamless handover method for hybrid communication networks for autonomous driving, applied to the core network side of a wireless communication network, characterized in that, The network functions in the core network pre-store a basic radio environment map, which represents the static mapping relationship between geographical road topology and serving cells and candidate serving cells; the method includes: In the receiving step, the network function receives the planned trajectory information reported by the autonomous driving terminal, which represents its future driving intention; In the aggregation step, the network function continuously aggregates routine wireless measurement reports from multiple terminals in the network, which include their current geographical location and wireless measurement values. The dynamic live feed generation step involves the network function responding to the planned trajectory information by performing an instant statistical analysis on reports from regular wireless measurement reports whose geographical locations are within the planned trajectory area. This analysis generates a dynamic live feed based on group consensus, representing the statistical results of the geographical locations along the planned trajectory and the current wireless measurement values. In the pre-image step, the network function projects the planned trajectory information onto the basic wireless environment map and combines it with the statistical results of the dynamic wireless situation to identify the geographical location where the optimal serving cell on the planned trajectory changes as the inevitable handover point, and determines the serving cell corresponding to the inevitable handover point as the target cell. The resource reservation step involves the network function instruction target cell pre-allocating and maintaining the radio resources required for handover before the autonomous driving terminal reaches the necessary handover point. In the instantaneous handover process, when the autonomous driving terminal reaches the necessary handover point, the core network triggers the autonomous driving terminal to perform the handover. The handover does not go through resource negotiation and directly utilizes the radio resources reserved in the target cell.

2. The seamless switching method for hybrid communication networks for autonomous driving according to claim 1, characterized in that, Before executing the resource reservation step, the network function also includes: a congestion prediction step: querying a future resource reservation ledger maintained for the target cell to determine whether executing the resource reservation step at the target time slice corresponding to the inevitable handover point will lead to resource conflict; a spatiotemporal decoupling step: when it is predicted that a resource conflict will occur, the network function first finds a candidate serving cell with available resources in the target time slice in the basic radio environment map and resets the resource reservation step to the candidate serving cell; or when no candidate serving cell with available resources is found, it finds an adjacent time slice with available resources in the future resource reservation ledger of the target cell and performs the resource reservation step at an offset.

3. The seamless switching method for hybrid communication networks for autonomous driving according to claim 1, characterized in that, After the resource reservation step and before the instantaneous handover step, the process also includes: an active probing step: the network function instructs the target cell to perform an internal radio quality probe on the radio resources reserved for the autonomous driving terminal to obtain the instantaneous interference level of the reserved resources; and an arbitration step: the network function... The rules confirm or reject instantaneous switching steps, among which, The measurement value represents the instantaneous level of disturbance. This represents a preset handover execution threshold. When the rule is met, the instantaneous handover step is confirmed. Error correction step: When the instantaneous handover step is rejected, the network function immediately selects a candidate serving cell from the basic radio environment map that is geographically available at the inevitable handover point, and repeats the resource reservation step, active detection step, and arbitration step for the candidate serving cell.

4. The seamless switching method for hybrid communication networks for autonomous driving according to claim 1, characterized in that, In the receiving step, when the network function does not receive the planned trajectory information, the receiving step is replaced by a trajectory inference step, which includes: a historical data reuse step: the network function acquires and analyzes the historical sequence of regular mobility measurement reports reported by the autonomous driving terminal; a vector and path matching step: based on the historical sequence, the current movement vector of the autonomous driving terminal is calculated and matched onto the geographic road topology represented by the basic wireless environment map; a trajectory generation step: along the geographic road topology, the current movement vector is prospectively projected to generate an inferred future trajectory; and the dynamic real-time generation step and the pre-mapping step are performed based on the inferred future trajectory.

5. The seamless switching method for hybrid communication networks for autonomous driving according to claim 1, characterized in that, The pre-mapping step utilizes the statistical results of dynamic wireless conditions to cover any pre-stored static wireless measurement data based on historical drive tests or simulations in the core network to perform identification and determination.

6. The seamless handover method for hybrid communication networks for autonomous driving according to claim 1, characterized in that, The planned trajectory information is simplified into a sequence of geographic coordinates of a set of key path inflection points.

7. The seamless switching method for hybrid communication networks for autonomous driving according to claim 1, characterized in that, The receiving step also includes: receiving the current speed vector reported by the autonomous driving terminal; and the pre-mapping step also includes: calculating the expected arrival timestamp of the inevitable switching point based on the planned trajectory information and the speed vector; the resource reservation step is triggered within a preset time window before the arrival timestamp.

8. A seamless switching method for hybrid communication networks for autonomous driving according to claim 4, characterized in that, The vector and path matching step calculates the current movement vector as the velocity vector; Furthermore, the pre-mapping step also includes: calculating the expected arrival timestamp of the inevitable switching point based on the inferred future trajectory and velocity vector; the resource reservation step is triggered within a preset time window before the arrival timestamp.

9. A seamless switching method for hybrid communication networks for autonomous driving according to claim 1, characterized in that, The dynamic live data generation step also includes: a deviation identification step: comparing the statistical results of the dynamic live wireless data with the static wireless measurement data representing the geographic area pre-stored in the core network to identify whether there is a statistical deviation between the two; and a dynamic entry generation step: when the statistical deviation exceeds a preset threshold, a time-sensitive dynamic temporary coverage area entry is generated, which is used to cover the static wireless measurement data in the pre-mapping step.

10. A seamless switching method for hybrid communication networks for autonomous driving according to claim 1, characterized in that, Also includes: Track cancellation procedure: When the network function receives a track cancellation signal from the autonomous driving terminal, it immediately releases the wireless resources reserved for the planned track; Based on the new planned trajectory information reported by the autonomous driving terminal, the dynamic real-time generation step, the pre-mapping step, and the resource reservation step are repeated.

Images