Collaborative driving scene-oriented adaptive RAN slice resource management method

Abstract

The invention discloses a cooperative driving scene-oriented adaptive RAN slice resource management method, relates to the technical field of vehicle cooperation, and has the core that differential service quality requirements of two types of services of cooperative sensing and cooperative control are simultaneously met through a dynamic closed-loop mechanism. The method comprises the following steps: firstly, establishing a double-slice architecture scheduled by different time granularities and a corresponding exclusive and shared resource pool; dynamically selecting a resource allocation strategy parameter based on a real-time network state; then, on two time scales of a time slot and a micro time slot, resource allocation is executed according to the selected strategy, and non-orthogonal multiple access modes such as punching or power domain superposition are introduced into a shared resource pool so as to realize flexible multiplexing; finally, key performance indexes are calculated and fed back to the strategy selection step, a continuously optimized self-adaptive closed loop is formed, and therefore intelligent and self-adaptive wireless resource management is achieved in the complex car networking environment. According to the invention, the overall utilization rate of wireless spectrum resources is effectively improved.

Description

Technical Field

[0001] This invention relates to the field of vehicle cooperative technology, and more specifically to an adaptive RAN slice resource management method for cooperative driving scenarios. Background Technology

[0002] The evolution of 5G V2X communication technology has driven the development of vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) communication, providing the necessary communication conditions for collaborative perception and control in collaborative driving scenarios. However, although existing Radio Access Network (RAN) slicing technology can allocate physical radio resources on demand and provide logically isolated transmission channels for different services, it still has significant limitations when facing highly dynamic, multi-service-coexisting vehicle-to-everything (V2X) scenarios such as collaborative driving. First, collaborative perception services and collaborative control services have fundamentally different communication requirements: the former pursues high throughput to support the transmission of large-scale environmental perception data, while the latter has extremely strict requirements for reliability and end-to-end latency. Existing slicing mechanisms often struggle to simultaneously accommodate the differentiated Quality of Service (QoS) requirements of these two types of services in the same network environment, leading to a situation of mutual constraints in resource allocation.

[0003] Furthermore, current resource management is largely limited to optimization within individual slices, lacking dynamic coordination capabilities across slices and making it difficult to adjust in real time based on network load and service bursts. The independent architecture between slices also leads to insufficient global joint optimization, hindering efficient resource sharing and overall system performance improvement in multi-service environments. Finally, in complex environments with high-speed vehicle movement and rapidly changing channel conditions, existing mechanisms have limited overall adaptive capabilities, making it difficult to maintain stable service quality and ensure efficient resource utilization, thus restricting further development of cooperative driving systems in terms of safety and efficiency. Summary of the Invention

[0004] To better balance the differentiated service quality requirements of various collaborative tasks while ensuring resource utilization efficiency, this invention proposes an adaptive RAN slice resource management method for collaborative driving scenarios, including the following steps: S1: Establish two types of slices, including collaborative sensing slices for V2I communication allocated in time slots and collaborative control slices for V2V communication allocated in micro time slots. S2: Divide the wireless resources into mutually orthogonal sensing-dedicated resource pools, control-dedicated resource pools, and shared resource pools that support dynamic sharing of the two types of slices; S3: Based on the state of the vehicle-to-everything (V2X) network, dynamically select resource allocation strategy parameters including cooperative control slice priority, cooperative perception slice priority, fairness weight, and network complexity limit; S4: Based on the selected resource allocation strategy parameters, at the beginning of the current time slot, allocate radio resources to the cooperative sensing vehicles occupying the dedicated sensing resource pool, and allocate sub-channels in the radio resources to the cooperative sensing vehicles occupying the shared resource pool. S5: Based on the selected resource allocation strategy parameters, allocate radio resources to the cooperative control vehicles occupying the dedicated resource pool in each micro-time slot within the current time slot, and use a non-orthogonal multiple access method to allocate the corresponding required radio resources to the cooperative control vehicles and cooperative sensing vehicles occupying sub-channels in the same shared resource pool. S6: Calculate the data transmission rate loss and packet loss rate of the collaborative sensing task caused by the non-orthogonal multiple access method and feed it back to step S3 to update the resource allocation strategy parameters.

[0005] This invention establishes dedicated slices for collaborative sensing and collaborative control services, allocated using time slots and micro-time slots respectively, and divides them into dedicated and shared resource pools. This fundamentally achieves resource isolation and flexible sharing between the two types of services, effectively improving the overall utilization of wireless spectrum resources while ensuring their respective minimum performance requirements. Furthermore, by introducing a dynamic resource sharing mechanism based on non-orthogonal multiple access, the system can intelligently reuse services within the resource sharing pool, alleviating resource contention issues in high-load scenarios and ensuring high reliability and low latency transmission for collaborative control services.

[0006] Furthermore, in step S5, the non-orthogonal multiple access method includes the following two methods: Punching method: Interrupt the data transmission of the cooperative sensing vehicle on the shared sub-channel, and allocate the sub-channel to the cooperative control vehicle for exclusive data transmission; Superposition method: By introducing power domain non-orthogonal multiple access technology, sub-channels in the shared resource pool are reused.

[0007] Furthermore, in step S5, the specific steps for allocating the necessary radio resources to the cooperative control vehicle and cooperative sensing vehicle occupying the same sub-channel in the shared resource pool include: Enumerate all possible combinations of non-orthogonal multiple access methods for each sub-channel in the shared resource pool; For each combination, under the conditions of satisfying the transmit power constraint and the receive signal-to-interference-plus-noise ratio constraint, calculate the optimal transmit power for the cooperative control vehicle and the cooperative sensing vehicle. Based on the calculated optimal transmission power, the resource allocation utility function value for each combination is calculated using an optimization function. Based on the bipartite graph matching algorithm, the combination that maximizes the resource allocation utility function value is selected as the final solution.

[0008] Furthermore, the optimization function aims to meet the real-time QoS requirements in the cooperative driving scenario, and the formula is expressed as follows: In the formula, For time slot numbering, For micro-timeslot labeling, In time slot micro-time slots Based on resource allocation strategy parameters The optimization function, To coordinate the control of slice priority, To coordinate the perception of slice priority, For fairness weighting, Due to network complexity constraints, To control the dedicated resource pool, To perceive the exclusive resource pool, To share the resource pool, In time slot micro-time slots A set of V2V communication links consisting of mutually coordinated control vehicles. for A set of V2V communication links in In time slot micro-time slots The location represents the sub-channel. Assigned to link The binary indicator of the state. In time slot micro-time slots Link A binary indicator of data transmission packet loss status. For a collaborative sensing vehicle ensemble, collaborative sensing vehicles Together with the base station, they form a V2I communication link. In time slot The location represents the sub-channel. Assigned to collaborative sensing vehicles The binary indicator of the state. In time slot micro-time slots Collaborative sensing vehicles Data transmission rate, In time slot micro-time slots Collaborative sensing vehicles Maximum data transfer rate To collaboratively perceive vehicles The average data transmission rate during the current sensing period. In time slot micro-time slots Sub-channels located within the shared resource pool The nonorthogonality degree introduced by nonorthogonal reuse.

[0009] Furthermore, the degree of non-orthogonality is calculated using the following formula: In the formula, These are the runtime complexity coefficients for the punching method and the stacking method, respectively. A binary indicator to represent the status of the punching method. A binary indicator of the state is used to represent the superposition method. In time slot Collaborative sensing vehicles Occupy sub-channel In case of a link Interference channel gain, In time slot Link Occupy sub-channel In this situation, the vehicle is collaboratively perceived from the transmitting end. Interference channel gain, In time slot Collaborative sensing vehicles The V2I communication link is located in the sub-channel Channel gain at that location, In time slot V2V communication link In sub-channel Channel gain at that location.

[0010] Furthermore, in step S3, the vehicle network state is represented by the following state matrix: In the formula, For time slots The state matrix at that point, In time slot Collaborative sensing vehicles Data transmission rate, In time slot Collaborative sensing vehicles Maximum data transfer rate The proportion of data transmission for collaborative sensing services relative to independent transmission. In time slots The activation probability, number of activations, and number of packet losses for internal collaborative control services. In time slot The network complexity at that location, For wireless resources, Total duration.

[0011] Furthermore, in step S3, the resource allocation strategy parameters are dynamically selected using the following formula: In the formula, For the selected resource allocation strategy parameters, For wireless resource management strategies, This represents the total number of time slots. For time slot numbering, As a discount incentive, These are the weighting coefficients for the collaborative sensing slice and the collaborative control slice, respectively. In time slot The specific wireless resource allocation method. The utility function for collaborative control slices. The utility function for co-perceived slices.

[0012] Furthermore, the utility function of the collaborative sensing slice aims to increase the instantaneous data transmission rate, reduce network complexity, and improve the completion rate of collaborative sensing tasks within the total time slots. The utility function of the collaborative control slice aims to reduce the instantaneous packet loss rate and the cumulative packet loss rate.

[0013] Compared with the prior art, the present invention has at least the following beneficial effects: (1) The present invention proposes an adaptive RAN slice resource management method for cooperative driving scenarios. By establishing dedicated slices with time slots and micro time slots as allocation units for cooperative sensing services and cooperative control services respectively, and dividing dedicated and shared resource pools, the isolation and flexible sharing of resources between the two types of services are fundamentally realized, thereby effectively improving the overall utilization rate of wireless spectrum resources while ensuring their respective minimum performance requirements. (2) Introducing a dynamic resource sharing mechanism based on non-orthogonal multiple access enables the system to intelligently reuse services in the resource sharing pool, alleviates resource contention in high-load scenarios, and ensures high reliability and low latency transmission of collaborative control services; (3) Through the designed two-layer optimization framework, the strategy parameters such as slice priority and fairness weight can be intelligently adjusted according to the real-time network status and performance feedback, so as to always maintain the efficient satisfaction of the QoS requirements of the two types of services in the complex and ever-changing Internet of Vehicles environment. Attached Figure Description

[0014] Figure 1 A flowchart illustrating the steps of an adaptive RAN slice resource management method for cooperative driving scenarios; Figure 2 This is a schematic diagram of a collaborative driving scenario; Figure 3 This is a schematic diagram of a wireless resource pool; Figure 4 Performance curves for different collaborative sensing data packets; Figure 5 Performance curves for different collaborative control data arrival rates. Detailed Implementation

[0015] The following are specific embodiments of the present invention, which are described in conjunction with the accompanying drawings. However, the present invention is not limited to these embodiments.

[0016] In practical applications of cooperative driving, the wireless access network needs to support two distinct types of services simultaneously: one is cooperative perception services that rely on vehicle-to-infrastructure communication, characterized by large data volumes, relatively continuous transmission, and high throughput requirements; the other is cooperative control services based on direct vehicle-to-vehicle communication, which are extremely sensitive to transmission latency and reliability, with any jitter or packet loss potentially impacting driving safety. Existing static-strategy-based wireless resource slicing methods struggle to cope with the dynamic fluctuations in service demands within the vehicle-to-everything (V2X) environment, failing to ensure both efficient transmission of cooperative perception data and extremely low latency and ultra-high reliability of cooperative control commands. Therefore, this invention proposes an adaptive RAN slicing resource management method for cooperative driving scenarios. By constructing a dynamic resource partitioning framework comprising dedicated and shared resource pools, introducing a non-orthogonal multiple access mechanism supporting puncturing and power domain superposition, and designing a multi-time-scale hierarchical optimization algorithm, it achieves adaptive quality-of-service assurance for both types of services in the complex and ever-changing V2X environment. Figure 1 As shown, this resource management method mainly includes the following steps: S1: Establish two types of slices, including collaborative sensing slices for V2I communication allocated in time slots and collaborative control slices for V2V communication allocated in micro time slots. S2: Divide the wireless resources into mutually orthogonal sensing-dedicated resource pools, control-dedicated resource pools, and shared resource pools that support dynamic sharing of the two types of slices; S3: Based on the state of the vehicle-to-everything (V2X) network, dynamically select resource allocation strategy parameters including cooperative control slice priority, cooperative perception slice priority, fairness weight, and network complexity limit; S4: Based on the selected resource allocation strategy parameters, at the beginning of the current time slot, allocate radio resources to the cooperative sensing vehicles occupying the dedicated sensing resource pool, and allocate sub-channels in the radio resources to the cooperative sensing vehicles occupying the shared resource pool. S5: Based on the selected resource allocation strategy parameters, allocate radio resources to the cooperative control vehicles occupying the dedicated resource pool in each micro-time slot within the current time slot, and use a non-orthogonal multiple access method to allocate the corresponding required radio resources to the cooperative control vehicles and cooperative sensing vehicles occupying sub-channels in the same shared resource pool. S6: Calculate the data transmission rate loss and packet loss rate of the collaborative sensing task caused by the non-orthogonal multiple access method and feed it back to step S3 to update the resource allocation strategy parameters.

[0017] Taking a vehicle-to-vehicle V2X communication system as an example, we will explain several preliminary arrangements of the present invention.

[0018] In this invention, the total available bandwidth of the V2X communication system is divided into... An orthogonal sub-channel. Denoted as Using indexes This indicates that each sub-channel has the same bandwidth, denoted as . This invention will consider two different granularities in the time dimension: time slots and mini-slots. A single time slot is composed of... It consists of mini-slots. The former is indexed using... Indicates; the latter's index is used This invention studies a cooperative driving scenario in a highway environment. This scenario mainly involves two types of core vehicle-to-everything (V2X) services and establishes corresponding RAN slices for these two types of services: Collaborative Perception Slice (CP Slice): Figure 2 In the image, a group of Connected and Automated Vehicles (CAVs) located in the area marked by the red dashed line communicates via their own V2I (Vehicle-to-Infrastructure) links. The vehicle transmits its own sensory data (such as location, speed, and obstacle information) to a roadside base station. The base station then processes the data from each cooperative sensing vehicle. The data is fused and processed to generate the target area ( Figure 2 The global environment awareness results (black dashed area) are shown. This service has relatively low sensitivity to latency (typically in the range of 50 to 100 ms), and its V2I communication and allocation use slots as the smallest time granularity.

[0019] Collaborative Control Slice (CC Slice): Figure 2 In the diagram, the area indicated by the yellow dashed line represents a set of CAVs, denoted as . The CAVs within this set need to send their own driving decision information to other collaboratively controlled vehicles through their own V2V communication links (e.g., the platoon's vehicle-to-the-front communication topology) according to the collaborative driving requirements. Because collaborative control services require high-throughput transmission of environmental perception data and have strict requirements on communication latency, their V2V communication and allocation use mini-slots as the smallest time granularity.

[0020] At the same time, such as Figure 3 As shown, in the V2X communication system designed in this invention, the base station configures a dedicated radio resource pool for each type of RAN slice, and also establishes a shared resource pool to support flexible and adaptive resource allocation. These resource pools are all composed of a set of sub-channels for V2X communication, and the configuration is as follows: Slice-specific resource pools: Each slice is allocated a set of mutually orthogonal sub-channels to achieve resource isolation between slices and ensure QoS of V2X communication within a slice. The collaborative sensing and control-specific resource pools are denoted as follows: and ,satisfy .

[0021] Slice Shared Resource Pool: A set of sub-channels is reserved as a resource pool shared by all slices, denoted as... ,satisfy Base stations can dynamically utilize sub-channels in the shared resource pool in an orthogonal or non-orthogonal manner based on real-time service load (such as bursty data traffic), thereby improving the utilization efficiency of spectrum resources.

[0022] Among these considerations, since collaborative sensing slices primarily carry vehicle environment sensing services, which are characterized by large data volumes, continuous transmission, and strong periodicity, high throughput and spectrum utilization efficiency are of paramount importance. By adopting a slot-based allocation method, continuous blocks of wireless resources can be allocated for sensing data, ensuring coding gain and link stability, thereby meeting the needs of large-scale sensing information transmission.

[0023] Therefore, this invention considers a total number of time slots of Execute within the time window The corresponding sensing time interval is [number] times for each collaborative sensing event. In the first During the cycle of this collaborative perception task, the collaborative perception vehicles participating in the collaboration The current sensing data needs to be transmitted to the base station via the V2I communication link, where the definition is... To collaboratively perceive vehicles Transmission delay within the current sensing period.

[0024] The base station needs to perform fusion processing on the sensing data transmitted by CAVs within the current sensing period. If the data is aggregated at this time... Collaborative sensing vehicles Unable to The internal transmission of sensing data to the base station results in spatiotemporal errors in the sensing data, which may significantly affect the accuracy and efficiency of the base station in performing sensing data fusion.

[0025] Therefore, this invention obtains information about the collaborative sensing vehicle for each sensing cycle. Maximum transmission delay Does it meet the maximum fusion latency? This represents the accuracy of the base station data fusion result. The proportion of all sensing cycles that complete the collaborative sensing task is defined as (in bold). It means that when The value is 1 when the condition is met and 0 when the condition is not met. (1).

[0026] Collaborative control slicing primarily serves control-related services such as vehicle platooning and trajectory coordination. These services are characterized by small, bursty packets, requiring extremely high latency and reliability. By employing a mini-slot-level allocation method, transmission can be triggered instantly without waiting for a complete slot, and the ability to preempt low-priority services is also available. This effectively reduces end-to-end latency and jitter, ensuring the real-time and deterministic transmission of control information.

[0027] This invention will be located in the time slot Medium and micro time slots The set of vehicles requiring coordinated control is abstracted as a group of V2V communication links. In V2V communication links In the middle, it is divided into a collaborative control sending end. and collaborative control receiver It should be noted that a single vehicle under collaborative control cannot simultaneously act as both a transmitter and a receiver.

[0028] To meet the low latency requirements of collaborative control services, data packets for these services must be transmitted within a single micro-slot; otherwise, the packets will be discarded. The definition is as follows: (2), in, In time slot micro-time slots Link A binary indicator of data transmission packet loss status. In time slot micro-time slots Link Data transmission rate, To coordinate and control the data packet size of the service. Therefore, with a total number of time slots of Within the time window, the reliability of the collaborative control slice is measured by the packet drop ratio (PDR), which is defined as: (3).

[0029] Building upon the above, the design objective of this invention is to provide intelligent adaptive wireless network slicing for both collaborative sensing and collaborative control services when they coexist. To this end, this invention introduces a wireless resource pool partitioning strategy that includes dedicated resource pools for collaborative sensing and collaborative control, and a shared resource pool, to achieve adaptive resource allocation across slices. The specific mechanism is as follows: (1) When the data volume of both types of slices is low, the collaborative sensing slice and the collaborative control slice preferentially occupy the sub-channels in their respective exclusive resource pools, that is, the traditional orthogonal resource allocation method is adopted. (2) When the data volume of any slice exceeds the carrying capacity of its dedicated resource pool, the slice may additionally occupy the sub-channel in the shared resource pool.

[0030] Since the two types of slices are allocated at the slot and mini-slot levels respectively, that is, at the beginning of a time slot, a sub-channel is first allocated to the cooperative sensing slice, and then a sub-channel is allocated to the cooperative control slice in each micro-time slot of the current time slot. In the orthogonal mode, the two types of slices occupy independent radio resources, and there is no interference or resource contention between them. Therefore, its operation mechanism is relatively simple, and this embodiment will not elaborate on it in detail. This invention focuses on the situation where the two types of slices need to coexist on shared sub-channels. In this case, resource allocation and interference management are more complex, and therefore have greater research value. In the time slot... At that time, it is assumed that the system has already provided cooperative perception for vehicles. Sub-channels were allocated from the shared resource pool. ,Right now In the time slot micro-time slots In this invention, binary variables are used. Indicates whether the sub-channel is simultaneously allocated to a V2V communication link. ,like =1, then there are two non-orthogonal multiple access methods to achieve resource sharing between slices: (1) Punching method: in the sub-channels already allocated to the collaborative sensing service Above, a sudden service demand arises in the collaborative control service. Because this service has a higher priority, the transmission of the collaborative sensing service is replaced by the transmission of the control service; that is, the collaborative sensing vehicle needs to be interrupted at this time. Data transmission, enabling V2V communication links Exclusive access to this sub-channel Binary pointer Indicates whether or not a drilling method is used.

[0031] (2) Overlay method: When the traffic of both collaborative sensing and collaborative control slices is under high load, power domain non-orthogonal multiple access (PD-NOMA) technology can be introduced to reuse sub-channels in the shared resource pool. PD-NOMA allows collaborative sensing vehicles in V2I communication links to reuse sub-channels. and V2V communication link exist Parallel transmission is achieved through differentiated transmission power allocation. Among these, This is a binary indicator for PD-NOMA technology. In this technology, to ensure high reliability of cooperative control service transmission, cooperative control V2V communication links are used. High transmit power must be allocated to ensure that the signal can be preferentially decoded at the receiving end.

[0032] Prior to this, this embodiment has detailed the resource pool partitioning and allocation principles, as well as the "strategy" for non-orthogonal access proposed in this invention. However, these details remain at a functional and conceptual level. Therefore, to transform them into quantifiable mathematical expressions, it is necessary to establish a precise signal transmission and interference model for collaborative sensing and collaborative control services.

[0033] For coordinated control services, considering that data packets in the coordinated control slice are treated as short packets, a more accurate approximation of short data packets is needed using finite block length theory. (In time slots) micro-time slots V2V communication link The data transmission rate is: (4), in, The length of a time slot The bandwidth of each sub-channel, The block length of the data packet. In communication theory, the Q-function is used to describe the probability of the tail of the standard normal distribution. For effective decoding error probability; The formula for calculating channel dispersion is: ; For V2V communication links The signal-to-interference-plus-noise ratio (SIR) within the current micro-timeslot is defined as: (5), Here, In time slot micro-time slots V2V communication link In sub-channel The transmission power at that location; Noise power on each subchannel; In the time slot V2V communication link In sub-channel The channel gain at that location remains stable within a time slot. Defined as in time slot micro-time slots V2V communication link In sub-channel The non-orthogonal interference received at a point is represented as: (6), in, In time slot micro-time slots Collaborative sensing vehicles In sub-channel The transmission power at that location, In time slot Collaborative sensing vehicles Occupy sub-channel In cases from the communication link The interference channel gain.

[0034] For collaborative sensing services, the aim is to ensure the transmission rate of sensing users, thereby improving the overall completion rate of collaborative sensing tasks. (Collaborative sensing vehicles) In the time slot The data transmission rate can be considered as The sum of the transmission rates of each micro-slot: (7), in, Collaborative sensing vehicles In the time slot micro-time slots The data rate and signal-to-interference-plus-noise ratio at the location, among which The definition is as follows: (8), Here, In time slot micro-time slots Collaborative sensing vehicles In sub-channel The transmission power; In the time slot Collaborative sensing vehicles The V2I communication link is located in the sub-channel Channel gain at that location, This is in the time slot micro-time slots The V2I communication link is affected by the link. Non-orthogonal interference: (9), in, In time slot Link Occupy sub-channel In this situation, the vehicle is collaboratively perceived from the transmitting end. The interference channel gain.

[0035] While non-orthogonal multiple access (NOR) can improve spectrum utilization, it also introduces cross-slice interference, SIC decoding coupling, and implementation overhead for scheduling and power allocation, increasing overall operational and computational complexity. Therefore, this invention defines a non-orthogonality degree (ND) in RAN slices to quantify this cost. Based on ND, this invention achieves a trade-off between "spectral efficiency and reliability / complexity" in resource scheduling and power allocation. ND is defined as follows: (10) This item indicates the time slot. micro-time slots The sub-channel located within the shared resource pool. The non-orthogonality degree introduced in [the text]. These are the operational complexity coefficients for punching and stacking methods, respectively. However, it should be noted that to ensure the overall system has a low degree of non-orthogonality, it is necessary to ensure... .

[0036] So, in the time slot The total nonorthogonality at the location is as follows: (11).

[0037] Based on the established signal transmission model, the differentiated quality of service requirements of collaborative sensing and collaborative control services are refined into a multi-objective optimization problem, thereby integrating the aforementioned technical elements into a clear mathematical solution objective. In this invention, this mathematical solution objective aims to provide corresponding wireless resources for collaborative sensing and collaborative control slices with low resource consumption. This enables the proposed framework to flexibly select optimal resource allocation strategy parameters based on service load and network conditions, effectively guaranteeing the differentiated requirements of the two types of services. To achieve this objective, this invention constructs an objective function that simultaneously encompasses the performance of collaborative sensing slices and collaborative control slices: (12) in, For wireless resource management strategies, As a discount factor, The weighting coefficients of the sensing slice and the control slice are coordinated respectively. The utility function for collaborative control slices. For the utility function of collaboratively perceived slices, For time slots The specific wireless resource allocation method is as follows: .

[0038] Since the two types of RAN slices have different time granularities, and as shown in formula (12), the optimization objective function includes both the performance of the collaborative control service and the utility function. Therefore, to ensure the performance of both types of slices simultaneously, the objective problem needs to be decomposed into two relatively independent multi-timescale decision problems: The utility function of collaborative sensing slices: (12a) The main tasks of collaborative sensing slicing are as follows: increasing instantaneous throughput, avoiding high-complexity operation, and improving the overall time slot performance. The completion rate of the perception task, in formula (12a) , , These are the weighting coefficients for these three items, respectively.

[0039] The utility function of collaborative control slices: (12b) Within the collaborative control slice, a penalty term is introduced to reduce the instantaneous packet loss rate. and cumulative packet loss rate To ensure the high reliability requirements of collaborative control operations, pay attention to... .

[0040] To solve this multi-objective optimization problem, this invention proposes to transform the objective function (12) into a two-level optimization problem, and innovates a two-level optimization algorithm based on the prior arrangement, thereby ensuring that the robustness of the system is improved while improving resource utilization.

[0041] In real-world vehicular network (V2V) environments, network conditions and service requirements are often complex and constantly changing. This invention proposes a resource management strategy that can automatically adjust RAN slice priorities based on network conditions and service characteristics. To this end, this invention introduces the Option-Critic framework and utilizes reinforcement learning algorithms to autonomously learn strategy switching schemes on a macroscopic time scale, thereby achieving dynamic optimization of slice priorities and adaptive selection of resource allocation strategies. Specifically, the agent can automatically select two distinct resource allocation strategies based on the actual scenario: one that prioritizes cooperative perception services, and the other that caters to low-latency, high-reliability cooperative control services.

[0042] The state space is a holistic description of the RAN slice, network state, and vehicular network environment, and is crucial for ensuring algorithm effectiveness. This invention establishes the current time-space... The state matrix is: in, For time slots The state matrix at that point, In time slot Collaborative sensing vehicles Maximum data transfer rate The proportion of data transmission for collaborative sensing services relative to independent transmission. In time slots The activation probability, number of activations, and number of packet losses for internal collaborative control services. In time slot The network complexity at that location.

[0043] Since the ultimate goal of the upper-level algorithm is to simultaneously ensure the performance of the collaborative sensing slice and the control slice, according to formulas (12a) and (12b), we can obtain the time slot... Reward function: .

[0044] Specifically, dynamic resource allocation strategy By decomposing the strategy into selection sub-strategies With termination sub-policy This can effectively enable the termination and switching of RAN slice resource allocation modes, where: Select sub-strategy , Used to guide the allocation of low-level real-time wireless resources, specifically defined as a set of parameters: , To coordinate the control of slice priority, To coordinate the perception of slice priority, For fairness weighting, Due to network complexity constraints.

[0045] Termination Sub-strategy , The aim is to dynamically evaluate the effectiveness of the current strategy, and when the network state changes significantly, determine whether the optimal resource allocation strategy has changed and adjust it in a timely manner. The probability of switching resource allocation strategies.

[0046] At the lower-level algorithm level, it's in the time slot. and micro-time slots Based on the wireless resource allocation strategy determined by the upper layer, real-time and fine-grained wireless resource allocation is performed at the time slot level or even the micro-time slot level. Its goal is to meet the real-time QoS requirements of V2X communication services within the slice. Lower-layer optimization function: (13) in, In time slot micro-time slots Based on resource allocation strategy parameters The optimization function, In time slot micro-time slots Collaborative sensing vehicles Maximum data transfer rate To collaboratively perceive vehicles In the current sensing cycle The average data transmission rate within the network. And the network complexity constraint introduced in this invention... The specific constraints are as follows: C1: ; C2: ; C3: ; C4: ; C5: ; C6: ; C7: ; Wherein: C1 guarantees that each subchannel is allocated to at most one V2V communication link in each microtimeslot. To avoid resource conflicts, C2 restricts each subchannel to be assigned to only one cooperative sensing vehicle. C3 imposes a maximum transmit power constraint; C4 sets the maximum network complexity introduced by non-orthogonal multiple access methods; C5 restricts the use of only puncturing or PD-NOMA methods on a shared subchannel for cooperative sensing and cooperative control users; C6 guarantees the minimum transmission rate for cooperative control services; C7 constrains the data transmission rate range for cooperative sensing users. .

[0047] This invention requires allocating collaborative sensing slices, and then allocating collaborative control slices based on these slices to obtain the optimal resource allocation scheme. Considering that collaborative sensing slices are allocated at the time slot level and collaborative control slices are allocated at the micro-time slot level, this invention proposes a two-step resource allocation method: first allocating collaborative sensing slices, and then allocating collaborative control slices, thereby achieving the optimal solution for overall resource allocation.

[0048] First, in time slots Initially, we used the PF allocation method to assign each cooperative sensing vehicle within the cooperative sensing slice. Perform sub-channel and power allocation, that is: (14) (15) in, To set the maximum transmission power, This achieves the highest signal-to-interference-plus-noise ratio for collaborative sensing. It is worth noting that collaborative sensing vehicles transmitting on shared sub-channels... Because the access method with the cooperative control vehicle is non-orthogonal, it is necessary to coordinate with the cooperative control vehicle to determine the timing at the micro-time slot.

[0049] Then, collaborative control slice allocation needs to be performed based on collaborative sensing slice allocation. For the lower-level optimization function (13), firstly, constraints C1 and C2 are ignored, and let =1, meaning it's assumed the resource allocation variable is a known constant. Under this condition, the original problem... It can be decomposed into There are several independent subproblems, each with its own closed-form solution. Based on this, and by reconsidering constraint C1, a globally optimal resource allocation scheme can be constructed using the aforementioned independent solutions. At this point, the problem... It is reformulated as a joint optimization problem.

[0050] (16) Therefore, each subproblem is: (17).

[0051] The solution approach for subproblem (17) is as follows, since this problem contains discrete variables. With continuous variables This is a mixed integer programming problem, subject to constraints C3–C7. Solve it using the following steps: Iterate through all discrete decision variable sets that satisfy constraint C4. The feasible combinations of values ​​are used to optimize power allocation while satisfying C3–C7. This yields the optimal power allocation. Based on the type and occupancy status of the resource pool to which the sub-channel belongs, three feasible usage modes can be determined: (1) Orthogonal access method (this method is used by default for dedicated resource pools): V2V communication link Orthogonal transmission is performed using a dedicated sub-channel, with the following transmit power: (18) in, , indicating V2V communication link In the time slot The lowest acceptable signal-to-interference-plus-noise ratio The signal-to-interference-plus-noise ratio for coordinated control.

[0052] (2) Drilling method: In this mode, the V2V communication link Preempting the transmission of collaborative sensing vehicles, at this time the V2V communication link Transmit power reference formula (18), and at the same time .

[0053] (3) Overlay method: The collaborative sensing and collaborative control tasks are transmitted in the same shared sub-channel using a non-orthogonal power domain method, as shown in the following formula: , .

[0054] By calculating the resource allocation methods under all discrete combinations, the corresponding values ​​of formula (13) are obtained, and then the optimal method is selected using the maximum weight matching method on the bipartite graph. .

[0055] In summary, in the two-layer optimization algorithm constructed in this invention, the upper-layer agent can perform macro-level planning and strategy formulation for wireless resources based on environmental information and the overall network status, and use this strategy to guide the real-time resource allocation of the lower layer. The lower layer then performs resource scheduling at the specific time slot and micro-time slot levels, and promptly feeds back the resulting QoS performance indicators and communication channel status to the upper layer. The upper layer uses these feedback results to evaluate the current wireless resource management strategy, thereby determining its effectiveness, and uses this as an important basis for further optimization decisions. In the context of vehicle-to-everything (V2X) scenarios, this effectively ensures the communication needs of collaborative perception and collaborative control.

[0056] In a preferred embodiment, the adaptive RAN slice resource management method proposed in this invention is evaluated through a set of simulation experiments. For simulation parameters and deployment, this invention considers a highway scenario and uses SUMO to simulate the environment of a real road. To further evaluate the algorithm's performance, this invention is compared with a scheme that optimizes the number of sub-channels in a dedicated resource pool at a fixed time granularity. The simulation experiment variables are shown in Table 1.

[0057] Table 1: Simulation Experiment Variables Table To evaluate the performance of the proposed solution under different collaborative sensing service load conditions, this invention constructs a vehicle-to-everything (V2X) simulation scenario from low load to high load by adjusting the data packet size of the collaborative sensing service. Figure 4 The paper presents a comparison of the proposed solution with a benchmark solution in this scenario, focusing on the collaborative sensing task completion rate and collaborative control packet loss rate. It can be observed that as the size of the collaborative sensing data packets gradually increases, the benchmark solution's collaborative sensing task completion rate significantly decreases, while the packet loss rate of the collaborative control link rapidly increases. This indicates that under high load conditions, resource contention and interference accumulation can severely impact system performance. In contrast, the proposed solution effectively mitigates resource conflicts under high load conditions. The collaborative perception task completion rate has always remained at a high level, and it is still significantly better than the benchmark solution even in the case of large data packets; The collaborative control system consistently maintains a low packet loss rate, demonstrating its advantage in ensuring highly reliable business operations.

[0058] To evaluate the adaptability of the proposed scheme to different levels of burstiness of control services, this invention constructs vehicle-to-everything (V2X) simulation scenarios under various load conditions by adjusting the average arrival rate of collaborative control services. Figure 5The performance comparison results between the proposed solution and the benchmark solution are presented. As can be seen from the figure, under low traffic conditions, the benchmark solution can maintain a certain performance level; however, as the average arrival rate of the collaborative control service increases, its collaborative sensing task completion rate drops rapidly, and the packet loss rate of the collaborative control link also increases significantly, demonstrating its sensitivity to sudden service disruptions. In contrast, the proposed solution maintains stable performance under the same conditions. In low-volume scenarios, the completion rate and packet loss rate are significantly better than the benchmark solution, indicating that this solution can still make efficient use of resources under light load conditions. In high-volume scenarios, the completion rate and packet loss rate can still be maintained within a relatively stable range, indicating that this solution can effectively mitigate the performance degradation caused by sudden surges in collaborative control business.

[0059] In summary, the adaptive RAN slice resource management method proposed in this invention for cooperative driving scenarios establishes dedicated slices for cooperative sensing services and cooperative control services, with time slots and micro-time slots as allocation units, and divides them into dedicated and shared resource pools. This fundamentally achieves resource isolation and flexible sharing between the two types of services, thereby effectively improving the overall utilization rate of wireless spectrum resources while ensuring their respective minimum performance requirements.

[0060] By introducing a dynamic resource sharing mechanism based on non-orthogonal multiple access, the system can intelligently reuse services in the resource sharing pool, alleviate resource contention in high-load scenarios, and ensure high reliability and low latency transmission of collaborative control services.

[0061] Through the designed two-layer optimization framework, the system can intelligently adjust strategy parameters such as slice priority and fairness weight based on real-time network status and performance feedback, thereby ensuring efficient QoS satisfaction of both types of services in the complex and ever-changing vehicle network environment.

[0062] It should be noted that all directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of the present invention are only used to explain the relative positional relationship and movement of each component in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indication will also change accordingly.

[0063] Furthermore, in this invention, descriptions involving terms such as "first," "second," and "a" are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0064] In this invention, unless otherwise explicitly specified and limited, the terms "connection," "fixed," etc., should be interpreted broadly. For example, "fixed" can mean a fixed connection, a detachable connection, or an integral part; it can mean a mechanical connection or an electrical connection; it can mean a direct connection or an indirect connection through an intermediate medium; it can mean the internal communication of two components or the interaction between two components, unless otherwise explicitly limited. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0065] Furthermore, the technical solutions of the various embodiments of the present invention can be combined with each other, but only if they are feasible for those skilled in the art. If the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such combination of technical solutions does not exist and is not within the scope of protection claimed by the present invention.

Claims

1. An adaptive RAN slice resource management method for cooperative driving scenarios, characterized in that, Including the following steps: S1: Establish two types of slices, including collaborative sensing slices for V2I communication allocated in time slots and collaborative control slices for V2V communication allocated in micro time slots. S2: Divide the wireless resources into mutually orthogonal sensing-dedicated resource pools, control-dedicated resource pools, and shared resource pools that support dynamic sharing of the two types of slices; S3: Based on the state of the vehicle-to-everything (V2X) network, dynamically select resource allocation strategy parameters including cooperative control slice priority, cooperative perception slice priority, fairness weight, and network complexity limit; S4: Based on the selected resource allocation strategy parameters, at the beginning of the current time slot, allocate radio resources to the cooperative sensing vehicles occupying the dedicated sensing resource pool, and allocate sub-channels in the radio resources to the cooperative sensing vehicles occupying the shared resource pool. S5: Based on the selected resource allocation strategy parameters, allocate radio resources to the cooperative control vehicles occupying the dedicated resource pool in each micro-time slot within the current time slot, and use a non-orthogonal multiple access method to allocate the corresponding required radio resources to the cooperative control vehicles and cooperative sensing vehicles occupying sub-channels in the same shared resource pool. S6: Calculate the data transmission rate loss and packet loss rate of the collaborative sensing task caused by the non-orthogonal multiple access method and feed it back to step S3 to update the resource allocation strategy parameters.

2. The adaptive RAN slice resource management method for cooperative driving scenarios as described in claim 1, characterized in that, In step S5, the non-orthogonal multiple access method includes the following two methods: Punching method: Interrupt the data transmission of the cooperative sensing vehicle on the shared sub-channel, and allocate the sub-channel to the cooperative control vehicle for exclusive data transmission; Superposition method: By introducing power domain non-orthogonal multiple access technology, sub-channels in the shared resource pool are reused.

3. The adaptive RAN slice resource management method for cooperative driving scenarios as described in claim 1, characterized in that, In step S5, the specific steps for allocating the necessary radio resources to the cooperative control vehicle and cooperative sensing vehicle that occupy sub-channels in the same shared resource pool include: Enumerate all possible combinations of non-orthogonal multiple access methods for each sub-channel in the shared resource pool; For each combination, under the conditions of satisfying the transmit power constraint and the receive signal-to-interference-plus-noise ratio constraint, calculate the optimal transmit power for the cooperative control vehicle and the cooperative sensing vehicle. Based on the calculated optimal transmission power, the resource allocation utility function value for each combination is calculated using an optimization function. Based on the bipartite graph matching algorithm, the combination that maximizes the resource allocation utility function value is selected as the final solution.

4. The adaptive RAN slice resource management method for cooperative driving scenarios as described in claim 3, characterized in that, The optimization function aims to meet the real-time QoS requirements in the cooperative driving scenario, and its formula is as follows: In the formula, For time slot numbering, For micro-timeslot labeling, In time slot micro-time slots Based on resource allocation strategy parameters The optimization function, To coordinate the control of slice priority, To coordinate the perception of slice priority, For fairness weighting, Due to network complexity constraints, To control the dedicated resource pool, To perceive the exclusive resource pool, To share the resource pool, In time slot micro-time slots A set of V2V communication links consisting of mutually coordinated control vehicles. for A set of V2V communication links in In time slot micro-time slots The location represents the sub-channel. Assigned to link The binary indicator of the state. In time slot micro-time slots Link A binary indicator of data transmission packet loss status. For a collaborative sensing vehicle ensemble, collaborative sensing vehicles Together with the base station, they form a V2I communication link. In time slot The location represents the sub-channel. Assigned to collaborative sensing vehicles The binary indicator of the state. In time slot micro-time slots Collaborative sensing vehicles Data transmission rate, In time slot micro-time slots Collaborative sensing vehicles Maximum data transfer rate To collaboratively perceive vehicles The average data transmission rate during the current sensing period. In time slot micro-time slots Sub-channels located within the shared resource pool The nonorthogonality degree introduced by nonorthogonal reuse.

5. The adaptive RAN slice resource management method for cooperative driving scenarios as described in claim 4, characterized in that, The degree of nonorthogonality is calculated using the following formula: In the formula, These are the runtime complexity coefficients for the punching method and the stacking method, respectively. A binary indicator to represent the status of the punching method. A binary indicator of the state is used to represent the superposition method. In time slot Collaborative sensing vehicles Occupy sub-channel In case of a link Interference channel gain, In time slot Link Occupy sub-channel In this situation, the vehicle is collaboratively perceived from the transmitting end. Interference channel gain, In time slot Collaborative sensing vehicles The V2I communication link is located in the sub-channel Channel gain at that location, In time slot V2V communication link In sub-channel Channel gain at that location.

6. The adaptive RAN slice resource management method for cooperative driving scenarios as described in claim 5, characterized in that, In step S3, the vehicle network state is represented by the following state matrix: In the formula, For time slots The state matrix at that point, In time slot Collaborative sensing vehicles Data transmission rate, In time slot Collaborative sensing vehicles Maximum data transfer rate The proportion of data transmission for collaborative sensing services relative to independent transmission. In time slots The activation probability, number of activations, and number of packet losses for internal collaborative control services. In time slot The network complexity at that location, For wireless resources, Total duration.

7. The adaptive RAN slice resource management method for cooperative driving scenarios as described in claim 1, characterized in that, In step S3, the resource allocation strategy parameters are dynamically selected using the following formula: In the formula, For the selected resource allocation strategy parameters, For wireless resource management strategies, This represents the total number of time slots. For time slot numbering, As a discount incentive, These are the weighting coefficients for the collaborative sensing slice and the collaborative control slice, respectively. In time slot The specific wireless resource allocation method. The utility function for collaborative control slices. The utility function for co-perceived slices.

8. The adaptive RAN slice resource management method for cooperative driving scenarios as described in claim 7, characterized in that, The utility function of the collaborative sensing slice aims to increase the instantaneous data transmission rate, reduce network complexity, and improve the completion rate of collaborative sensing tasks within the total time slots. The utility function of the collaborative control slice aims to reduce the instantaneous packet loss rate and the cumulative packet loss rate.

Images