An OBU multi-modal communication protocol optimization method

Abstract

This invention provides an OBU multimodal communication protocol optimization method, relating to the field of multimodal communication technology. The method includes: acquiring multimodal network environment parameters and service data to be transmitted from the On-Board Unit (OBU); inputting the network environment parameters into a spatiotemporal correlation channel prediction model to deduce a predicted link quality topology sequence; performing transport layer decoupling and segmentation of the service data based on this topology sequence to obtain a set of service data slices; using a multidimensional utility maximization objective function to perform communication mode optimization matching for each data slice; calculating the dynamic protocol stack encapsulation parameters corresponding to the slices based on the matching mapping relationship; updating the protocol stack control parameters according to the encapsulation parameters, and performing customized message encapsulation to complete adaptive optimization configuration. This invention achieves dynamic reconstruction and cross-layer collaboration of protocol layer parameters, significantly improving the latency determinism and reliability of multimodal concurrent transmission in highly dynamic vehicle-to-everything (V2X) environments.

Description

Technical Field

[0001] This invention relates to the field of multimodal communication technology, and in particular to an optimization method for OBU multimodal communication protocol. Background Technology

[0002] As a key infrastructure of intelligent connected vehicles, the on-board unit is typically equipped with various heterogeneous radio frequency interfaces, such as dedicated short-range communication, cellular vehicle-to-everything (V2X) communication, and wireless local area network (WLAN). Existing on-board multimodal communication systems are mostly based on a layered protocol stack architecture. At the network layer or media access control layer, static physical layer parameters such as received signal strength indication or signal-to-noise ratio are evaluated to perform threshold hard handover or simple data offloading of a single communication link, in order to support different types of data interaction between vehicles and between vehicles and infrastructure.

[0003] With the evolution of collaborative intelligent transportation systems, vehicular communication services are increasingly exhibiting strong heterogeneity and high concurrency characteristics, placing extremely high demands on the latency determinism and reliability of multi-dimensional data transmission. Current communication protocol optimization is shifting from simple passive physical layer link selection to multimodal collaborative resource allocation, dynamic configuration of cross-layer protocol stack parameters, and service-aware proactive network topology adaptive scheduling technologies. The aim is to achieve seamless and transparent collaboration between the underlying heterogeneous transmission resource pool and upper-layer application services.

[0004] Existing passive feedback-based communication protocol processing mechanisms exhibit significant hysteresis when facing highly dynamic vehicular channels. Traditional vehicular multimodal protocol stacks typically employ fixed data segment lengths and static congestion control mechanisms. When performing heterogeneous link handover, the system cannot deeply couple the time-varying attributes of the underlying physical links with the service quality characteristics of upper-layer services. Due to the lack of prior knowledge of future channel state evolution trends and the collaborative reconstruction capability of protocol stack underlying parameters, communication systems are highly susceptible to frequent link reselection, resource gaps, and packet retransmissions in severely fading channels, leading to uncontrollable end-to-end transmission delays and drastic fluctuations in overall system throughput. Summary of the Invention

[0005] To overcome the shortcomings of the prior art, the purpose of this invention is to provide an OBU multimodal communication protocol optimization method. This invention solves the problem that the existing passive link switching and static protocol encapsulation mechanisms are difficult to balance the determinism and reliability of transmission delay when facing highly dynamic vehicle environments.

[0006] To achieve the above objectives, the present invention provides the following solution: An optimization method for OBU multimodal communication protocol includes: Obtain the multimodal network environment parameters and service data to be transmitted from the on-board unit (OBU); The multimodal network environment parameters are input into a preset spatiotemporal correlation channel prediction model to perform time-series state deduction, thereby obtaining a predicted link quality topology sequence. Based on the service data to be transmitted and the predicted link quality topology sequence, the service data to be transmitted is decoupled and segmented at the transport layer through a dynamic protocol stack orchestration mechanism to obtain a service data slice set composed of multiple data slices. Based on the predicted link quality topology sequence, communication mode optimization matching is performed on each data slice in the service data slice set to determine the target communication mode mapping relationship corresponding to each data slice. Based on the target communication mode mapping relationship, calculate the dynamic protocol stack encapsulation parameters corresponding to each data slice; The control parameters of the communication protocol stack of the on-board unit (OBU) are updated according to the dynamic protocol stack encapsulation parameters. Combined with the target communication mode mapping relationship, the communication protocol stack with updated control parameters is used to perform customized message encapsulation on the corresponding data slice in the service data slice set to complete the adaptive optimization configuration of the multimodal communication protocol.

[0007] An OBU multimodal communication protocol optimization system includes: The data acquisition module is used to acquire the multimodal network environment parameters and the service data to be transmitted from the on-board unit (OBU). The timing state deduction module is used to input the multimodal network environment parameters into a preset spatiotemporal correlation channel prediction model to perform timing state deduction and obtain a predicted link quality topology sequence. The transport layer decoupling and segmentation module is used to decouple and segment the service data to be transmitted based on the service data to be transmitted and the predicted link quality topology sequence through a dynamic protocol stack orchestration mechanism, so as to obtain a service data slice set composed of multiple data slices. The communication mode optimization and matching module is used to perform communication mode optimization and matching on each of the data slices in the service data slice set based on the predicted link quality topology sequence, and determine the target communication mode mapping relationship corresponding to each of the data slices; The encapsulation parameter calculation module is used to calculate the dynamic protocol stack encapsulation parameters corresponding to each of the data slices based on the target communication mode mapping relationship. The protocol configuration and customized encapsulation module is used to update the control parameters of the communication protocol stack of the on-board unit (OBU) according to the dynamic protocol stack encapsulation parameters, and combine the target communication mode mapping relationship to perform customized message encapsulation on the corresponding data slice in the service data slice set using the communication protocol stack with updated control parameters, so as to complete the adaptive optimization configuration of the multimodal communication protocol.

[0008] The present invention discloses the following technical effects: This invention provides an optimization method for OBU multimodal communication protocols. By constructing a spatiotemporally correlated channel prediction model to prospectively extrapolate link topology sequences, this invention overcomes the hysteresis of traditional feedback mechanisms and the "ping-pong effect" caused by frequent handovers. Furthermore, it deeply couples the prediction results with service characteristics, adaptively reconstructing the underlying control parameters of the protocol stack (fragment length and congestion backoff coefficient) to perform customized message encapsulation. This method overcomes the rigidity of protocol layer parameters, achieving seamless and transparent collaboration between the underlying heterogeneous physical links and the upper-layer service flows. It effectively reduces the retransmission and packet loss rate under severe channel fading, significantly improving the latency determinism of multimodal concurrent transmission and the overall system throughput. Attached Figure Description

[0009] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0010] Figure 1 A flowchart of an OBU multimodal communication protocol optimization method provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of the multimodal heterogeneous physical link architecture of the on-board unit (OBU) provided in an embodiment of the present invention. Figure 3 This is a schematic diagram illustrating the specific process of communication mode optimization and matching for data slices provided in an embodiment of the present invention; Figure 4 This is a schematic diagram of an OBU multimodal communication protocol optimization system provided in an embodiment of the present invention.

[0011] Figure label: 1-Data acquisition module, 2-Timing state deduction module, 3-Transmission layer decoupling and segmentation module, 4-Communication mode optimization and matching module, 5-Encapsulation parameter calculation module, 6-Protocol configuration and customized encapsulation module. Detailed Implementation

[0012] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0013] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0014] like Figure 1 As shown, this invention provides an OBU multimodal communication protocol optimization method, comprising: Step 100: Obtain the multimodal network environment parameters and service data to be transmitted from the On-Board Unit (OBU); Step 200: Input the multimodal network environment parameters into a preset spatiotemporal correlation channel prediction model to perform time-series state deduction and obtain the predicted link quality topology sequence; Step 300: Based on the service data to be transmitted and the predicted link quality topology sequence, the service data to be transmitted is decoupled and segmented at the transport layer through a dynamic protocol stack orchestration mechanism to obtain a service data slice set composed of multiple data slices; Step 400: Based on the predicted link quality topology sequence, perform communication mode optimization matching on each data slice in the service data slice set to determine the target communication mode mapping relationship corresponding to each data slice; Step 500: Based on the target communication mode mapping relationship, calculate the dynamic protocol stack encapsulation parameters corresponding to each data slice; Step 600: Update the control parameters of the communication protocol stack of the on-board unit (OBU) according to the dynamic protocol stack encapsulation parameters, and combine the target communication mode mapping relationship, use the communication protocol stack with updated control parameters to perform customized message encapsulation on the corresponding data slice in the service data slice set, so as to complete the adaptive optimization configuration of the multimodal communication protocol.

[0015] The On-Board Unit (OBU) multimodal heterogeneous physical link architecture uses the centrally deployed OBU as the core protocol processing hub. Downward, it connects bidirectionally to the on-board application processing system via a data bus, receiving service data to be transmitted in real time and sensing the priority level of upper-layer services. Upward, it concurrently connects multiple heterogeneous physical links. Specifically, such as... Figure 1 As shown, the multiple heterogeneous physical links include: M1 cellular communication link, configured to provide wide-area, high-bandwidth network access; M2 dedicated short-range communication and direct communication link for vehicle-to-everything (V2X) communication, configured to support high-frequency, extremely low-latency vehicle-road cooperative safety interaction; M3 satellite communication link, configured to provide seamless wide-area fallback connectivity in cellular signal blind spots; and M4 wireless LAN link, configured to support low-cost offloading and transmission of large-capacity data in specific parking scenarios. During system operation, each of the heterogeneous physical links is equipped with a channel-aware node (e.g., Figure 1The SN0, SN3, and other RF front-end detection nodes (marked at the top) are used as the execution entities for underlying channel detection. These nodes collect low-level physical parameters such as signal-to-noise ratio (SNR), received signal strength indication (RSSI), and available bandwidth in real time, and continuously transmit and report these parameters as multi-mode network environment parameters to the On-Board Unit (OBU). The OBU integrates the reported environmental parameters from the lower layers with the carrying requirements of upper-layer services, internally performing adaptive reconstruction of protocol control parameters and customized message encapsulation. Finally, the reconstructed messages are distributed downwards to the corresponding heterogeneous physical links for concurrent network transmission according to the optimal modal mapping relationship, thus forming the hardware physical foundation upon which the method of this invention operates.

[0016] Furthermore, the specific implementation process of step 100 is as follows: This embodiment intercepts and parses the service data to be transmitted from the upper-layer application in real time through the application processing chip inside the vehicle unit. This embodiment listens to the raw service data stream transmitted on the vehicle Ethernet or controller area network bus and extracts the service priority and tolerance latency constraint fields in the packet header, thereby providing a service-aware benchmark for subsequent dynamic protocol stack slicing at the physical layer, ensuring that data streams with ultra-reliable and low-latency requirements can be assigned accurate service level labels before the underlying radio frequency transmission.

[0017] To collect parameters for multimodal network environments, this embodiment drives multiple heterogeneous RF transceiver baseband modules at the underlying level to concurrently perform active channel quality detection. This embodiment accurately captures the real-time signal-to-noise ratio, received signal strength indication, and available bandwidth of multiple heterogeneous physical links by frequently polling the underlying feedback level indicators in the physical layer control registers of each communication mode. This constructs a comprehensive physical panorama of the available heterogeneous network resources for the current transmission cycle, effectively eliminating blind spots caused by high-speed vehicle movement and single-link physical fading.

[0018] To achieve predictive extrapolation of channel states, this embodiment collaboratively utilizes the vehicle-mounted satellite navigation and positioning hardware and the inertial measurement unit to capture vehicle spatiotemporal trajectory data. This embodiment reads data frames from the aforementioned sensor hardware bus interface in real time, simultaneously acquiring the vehicle's current geographic coordinates, speed, and heading angle. The vehicle's dynamic physical displacement parameters are converted into digitized spatiotemporal perception feature vectors, providing accurate mobility physical boundary input conditions for subsequent prediction models and significantly improving the scene-adaptive tracking accuracy for cross-modal link state prediction.

[0019] Furthermore, the specific implementation process of step 200 is as follows: This embodiment uses a pre-deployed spatiotemporal correlation channel prediction model within the vehicle's computing unit to perform temporal state extrapolation on the acquired multimodal network environment parameters. The spatiotemporal correlation channel prediction model is a deep neural network architecture that integrates spatial feature perception and temporal memory evolution, aiming to capture the implicit mapping relationship between vehicle geographic displacement and physical channel fading. This embodiment first normalizes and aligns real-time signal-to-noise ratio, received signal strength indication, and vehicle spatiotemporal trajectory data, then concatenates them to construct the current multimodal network environment parameter vector. To ensure the smoothness of the input features, this embodiment extracts parameter vectors from the past 10 consecutive millisecond-level sampling periods as a historical observation window. The spatiotemporal feature extraction network at the model's front end performs nonlinear feature mapping, compressing and projecting the original high-dimensional physical observation data into a low-dimensional abstract feature space, thereby obtaining the channel spatiotemporal correlation feature vector.

[0020] In the data processing of the aforementioned feature mapping, this embodiment utilizes the encoder mapping function within the spatiotemporal feature extraction network to perform spatial dimension feature aggregation operations on the multimodal network environment parameter vector to be processed. This embodiment performs matrix multiplication of the input original physical layer parameters with a preset feature extraction weight matrix, supplemented by bias compensation, and then performs activation processing through a linear rectified function to filter out white noise interference from the underlying physical sensor acquisition. After the above encoding transformation, this embodiment outputs a 128-dimensional channel spatiotemporal correlation feature vector. This feature vector, while filtering out redundant environmental background information, highly condenses the signal-to-noise ratio evolution gradient of the heterogeneous physical links at the current moment and the instantaneous spatial curvature of vehicle movement, providing a clean contextual memory carrier for subsequent time-series extrapolation.

[0021] Subsequently, in this embodiment, the channel spatiotemporal correlation feature vector is input into the long short-term memory network node in the spatiotemporal correlation channel prediction model for time-series evolution deduction. This embodiment utilizes the forget gate, input gate, and output gate mechanisms within the long short-term memory network node to read the historical hidden state of the previous transmission cycle and, combined with the currently input channel spatiotemporal correlation feature vector, iteratively transmits state information through the weight parameters updated in the hidden state. The weight parameters updated in the hidden state originate from the prior channel fading laws accumulated during the offline training phase of the model based on massive amounts of real drive test data, and their value range is typically strictly constrained within a floating-point range of -1 to +1. Through this recursive calculation with a memory attenuation mechanism, this embodiment can accurately simulate the long-term dynamic impact of Doppler shift on channel quality and successfully deduce the time-series predicted hidden state for five consecutive time steps in the future.

[0022] For the temporal prediction hidden state obtained through deduction, this embodiment further utilizes the output layer parameters of link quality prediction for dimensionality reduction decoding. This embodiment performs a fully connected mapping between the temporal prediction hidden state and the output layer parameters, and uses a logistic nonlinear activation function to strictly project the output layer values ​​to a probability threshold range of zero to one. The logistic nonlinear activation function introduced here plays a core data processing role in normalizing the probability distribution. Through this decoding and activation process, this embodiment independently calculates the confidence level of each heterogeneous network link in different quality states such as available, congested, or interrupted for the current time and various future deduction steps, and then combines them to generate a link quality state probability matrix for multiple consecutive future time steps. For example, for three heterogeneous physical links concurrently configured in an on-board unit, this matrix can quantify and provide an accurate probability assessment value for each link to meet low-latency transmission requirements at a specific future evolution time.

[0023] After obtaining the link quality state probability matrix, this embodiment extracts the state node with the highest probability at each time step based on the principle of local optimal decision-making to construct a predicted link quality topology sequence. This embodiment traverses the time dimension vector in the link quality state probability matrix, and within each independent time step slice, horizontally compares the availability confidence of multiple heterogeneous physical links. This embodiment sets a preset probability confidence lower bound, whose default value is usually configured as 0.60 as the threshold interception threshold. When a probability peak exists in multiple links and this peak exceeds the preset probability confidence lower bound, this embodiment selects the link state corresponding to the probability peak as the dominant network quality evolution node at that time step. After extracting and sequentially concatenating continuous state nodes according to the time sequence, this embodiment finally generates a predicted link quality topology sequence reflecting the future short-range multimodal channel alternation evolution map. This sequence provides a dynamic reference for the network carrying capacity with forward-looking physical entities for subsequent service data decoupling and segmentation at the transport layer.

[0024] Specifically, the expression for the spatiotemporal correlation channel prediction model is as follows: ; in, This represents the multimodal network environment parameter vector at time t. For feature mapping function, for eigenvectors, To predict the hidden states of a time-series prediction network, The weight parameters are updated for the hidden state. The output layer parameters for link quality prediction. For activation functions (such as the logical sigmoid). Predicting the number of steps for the future, This represents the quality prediction vector for L links at time t+k, and the predicted spatiotemporal link quality topology sequence is denoted as . .

[0025] Furthermore, the specific implementation process of step 300 is as follows: This embodiment, based on the service data to be transmitted and the predicted link quality topology sequence, decouples and segments the service data to be transmitted at the transport layer through the dynamic protocol stack orchestration mechanism within the vehicle computing platform. Specifically, this embodiment first extracts the service priority field and tolerance latency constraint label attached to the packet header of the service data to be transmitted issued by the application layer. Simultaneously, this embodiment parses the predicted link quality topology sequence (i.e., parses the total available bandwidth represented by the predicted link quality topology sequence) to obtain the network carrying capacity threshold. The process of obtaining this network carrying capacity threshold comprehensively considers the encapsulation overhead of the underlying media access control layer, and its value accurately quantifies the maximum effective payload data volume that the heterogeneous physical interface of the vehicle unit can stably transmit without packet loss within a preset future time window (e.g., five consecutive millisecond-level time steps), thereby providing a strict physical space benchmark for subsequent congestion avoidance and cross-layer traffic shaping.

[0026] After obtaining the network carrying capacity threshold, this embodiment configures and distributes it to the protocol stack scheduler, using the network carrying capacity threshold as the upper limit of resource allocation for the underlying transmission queue of the protocol stack. Within the hard capacity constraint of the resource allocation upper limit, this embodiment performs fine-grained flow control review on the raw data in the buffer queue, intercepting data segments from the data to be transmitted that meet high service priority (e.g., critical safety-related vehicle coordination messages marked with priority levels 1 to 3) and low latency tolerance constraints (e.g., braking warning commands with a latency tolerance threshold of less than 20 milliseconds). These highly time-sensitive and highly sensitive discrete data segments are redirected and encapsulated to generate a high-guarantee transmission stream. At the same time, this embodiment triggers a strict threshold interception mechanism, forcibly stripping overloaded data segments exceeding the network carrying capacity threshold (e.g., high-definition video cache of the in-vehicle entertainment system or log synchronization packets of background software) from the backbone transmission channel and marking them as degraded transmission streams. Through the above physical isolation and stripping actions, this embodiment successfully obtains data streams of different transmission levels at the transport layer, completely eliminating the queuing and blocking effect of large-capacity low-priority services on critical vehicle network short messages.

[0027] Finally, this embodiment performs independent segmentation processing on the data streams of different transmission levels obtained from the above processing, according to a preset payload granularity matching each transmission level, thereby ultimately obtaining a service data slice set composed of multiple data slices. For the high-guarantee transmission stream, this embodiment calls a minimal preset payload granularity (e.g., setting a small payload block with a fixed length of 300 bytes for each slice) to perform high-density fragmentation segmentation, so as to maximize the adaptation to the discrete fragmented bandwidth generated by multiple heterogeneous links at the underlying layer during severe fading, ensuring that critical data can be transmitted at extremely high speeds. For the degraded transmission stream, this embodiment matches it with a larger preset payload granularity (e.g., setting a single slice size of 1500 bytes, the standard Ethernet maximum transmission unit size) for overall segmentation, so as to significantly reduce the additional signaling overhead caused by frequent protocol header encapsulation. Through this dynamic decoupling and segmentation mechanism based on the differences in service level and carrying capacity, the service data slice set generated in this embodiment not only perfectly matches the inherent quality of service requirements of the data to be transmitted, but also achieves cross-layer mapping with the predicted network physical carrying topology, laying a highly structured data carrier foundation for subsequent multimodal optimization matching and concurrent network transmission.

[0028] Furthermore, the specific implementation process of step 400 is as follows: This embodiment, based on the predicted link quality topology sequence, performs communication mode optimization matching on each data slice in the service data slice set, thereby determining the target communication mode mapping relationship corresponding to each data slice, such as... Figure 3 As shown, this embodiment first establishes a multi-dimensional utility maximization objective function and simultaneously acquires the predicted link quality topology sequence derived from the previous steps. Subsequently, the system enters a multi-dimensional traversal evaluation and loop judgment mechanism: using the objective function, it calculates the transmission utility value of each data slice in the service data slice set under each available communication mode, and determines in real time whether the maximum utility value of all data slices has been found; if not, a loop optimization instruction is triggered to continue evaluating the next data slice or candidate mode; if yes, the traversal loop is exited, and the candidate mode corresponding to the maximum transmission utility value is strictly selected as the target communication mode, thereby comprehensively establishing the mapping relationship between each data slice and the corresponding target communication mode, thus completing the entire optimization and matching process.

[0029] Specifically, this embodiment first establishes a multidimensional utility maximization objective function, which is configured with throughput gain, latency penalty, and handover overhead penalty as multidimensional optimization parameters. These three key indicators characterize the comprehensive performance of heterogeneous networks in terms of bandwidth supply, data queuing time, and signaling loss caused by underlying radio frequency link reselection and handover. By incorporating the above multidimensional parameters into a unified evaluation framework, this embodiment can provide accurate performance calibration criteria across the physical and network layers for each data slice when facing up to four or more parallel available communication modes (such as cellular vehicular networks, dedicated short-range communications, satellite links, and wireless LANs).

[0030] In the quantitative evaluation process, this embodiment combines the predicted link quality topology sequence and uses the multidimensional utility maximization objective function to evaluate the transmission utility value of each data slice in the service data slice set under each available communication mode. For the throughput gain index, this embodiment extracts the predicted throughput gain quantization value of the data slice under a specific available communication mode. This quantization value characterizes the transmission gain capability relative to the system's underlying preset baseline bandwidth. To eliminate computational overflow caused by different dimensions, this embodiment normalizes the predicted throughput gain quantization value by dividing it by the maximum possible value of the corresponding system index (i.e., the network physical upper limit). For example, if the baseline bandwidth threshold is 10 Mbps, the absolute bandwidth prediction value of a certain slice under a specific mode is 50 Mbps, and the theoretical extreme value of that mode (i.e., the maximum possible value) is set to 100 Mbps, then this embodiment maps it to a dimensionless positive gain value between zero and one through normalization, ensuring that the physical bandwidth characteristics are smoothly integrated into the evaluation function.

[0031] For the aforementioned latency penalty index and handover overhead penalty index, this embodiment employs a negative reduction mechanism for data processing and quantification. This embodiment analyzes the expected transmission latency of a specific data slice under a specific available communication mode and the potential cross-modal reconfiguration overhead loss, and normalizes these by dividing them by the system's preset maximum tolerable latency limit and maximum allowable handover loss extreme value (i.e., the maximum possible value of each corresponding index). In actual vehicle-road cooperative scenarios, this embodiment strictly configures the maximum tolerable latency limit to 100 milliseconds and limits the maximum allowable handover overhead penalty extreme value to 20 milliseconds. When a slice incurs a 15-millisecond transmission queuing latency if residing in the current mode, and incurs a 5-millisecond physical layer interface handover signaling overhead due to cross-modal handover, this embodiment, through the aforementioned normalization action, transforms the microsecond-level physical time consumption into a dimensionless negative penalty factor in the system utility evaluation, accurately quantifying the degree of erosion of real-time services by the highly dynamic network environment.

[0032] After completing the normalization and quantization of the multidimensional optimization parameters, this embodiment introduces a service-aware weighted fusion calculation mechanism into the multidimensional utility maximization objective function to calculate the final utility value. This embodiment configures a positive weight coefficient for the normalized throughput gain index, and negative weight coefficients for the normalized latency penalty index and handover overhead penalty index. The sum of these three weighting coefficients is strictly constrained to a constant value of 1. This embodiment dynamically allocates coefficient ratios according to the service priority level of the current data slice. For example, for a high-priority autonomous driving control slice, this embodiment increases the negative weight coefficient of latency penalty to 0.6, and sets the positive throughput weight and handover penalty weight to 0.2 respectively. Then, the positive weighted throughput gain is subtracted from the negative weighted latency penalty and weighted handover penalty. Through this linear algebraic fusion operation, the transmission utility value of each data slice under each available communication mode is finally aggregated and output.

[0033] Finally, this embodiment performs a comprehensive decision-making process at the system level to maximize multi-dimensional utility based on the set of all transmission utility values ​​obtained through traversal calculations. For each individual data slice in the service data slice set, this embodiment horizontally scans and compares its calculation results under different candidate communication modes, and strictly selects the available communication mode with the largest transmission utility value as the target communication mode for that specific slice. By performing the above optimization allocation action on each of the complete service data slice set containing, for example, 50 discrete slices, this embodiment successfully establishes the target communication mode mapping relationship between each data slice and its corresponding target communication mode, guided by maximizing the overall system utility. This mapping relationship is solidified in the memory controller in the form of a low-level routing policy configuration table, providing globally optimal transmission protocol reconfiguration instructions for customized message encapsulation and physical layer concurrent distribution implemented in subsequent steps.

[0034] Specifically, the expression for the multidimensional utility maximization objective function is: Wherein, if the final selected mode for each slice i is denoted as The corresponding optimal choice is: ;in, The set of indices for the data slice set; Let i be the set of possible modes for the i-th slice; The throughput gain of the i-th slice in mode m (the quantized value of the gain relative to the baseline, normalized to...) Or the upper limit of a broader area); The time delay penalty for the i-th slice in mode m; The switching overhead penalty for the i-th slice in mode m; These represent the maximum possible values ​​for the corresponding indicators (used for normalization). U represents the total utility at the system level. The optimal mode selection is for the i-th slice.

[0035] Furthermore, this embodiment combines the predicted link quality topology sequence with the multidimensional utility maximization objective function to evaluate the transmission utility value of each data slice in the service data slice set under each available communication mode. Specifically, this embodiment first parses the predicted link quality topology sequence using an in-vehicle protocol analysis component. The predicted link quality topology sequence objectively records the dynamic evolution trajectory of the heterogeneous physical channel state in the time dimension. This embodiment extracts network feature data from the sequence within a consecutive 50-millisecond span for the current evaluation period through a preset time slice alignment mechanism, thereby accurately obtaining the expected physical layer bandwidth and predicted link transmission delay of the available communication mode. The expected physical layer bandwidth reflects the theoretical maximum data throughput that the underlying modem can provide, while the predicted link transmission delay characterizes the expected combined time consumption of electromagnetic wave signal propagation in space and baseband digital processing. For example, this embodiment measures a peak bandwidth of up to 200 megabits per second for a certain high-frequency available mode. These fundamental parameters provide the initial physical layer foundation support for subsequent utility index quantification.

[0036] After obtaining the underlying basic parameters, this embodiment maps the expected physical layer bandwidth to a network layer throughput characterization value to quantify the throughput gain index. Since the physical layer channel bandwidth inevitably includes underlying signaling overhead such as forward error correction codes and synchronization pilot signals, this embodiment introduces a fixed overhead attenuation coefficient during data processing. For example, it actively deducts 15% of the physical layer encapsulation and verification losses, mapping the remaining pure payload carrying capacity to the actual network layer throughput characterization value. Subsequently, this embodiment performs a ratio calculation between this network layer throughput characterization value and a preset benchmark throughput within the system (e.g., fixed as a reference standard value of 100 megabits per second). By calculating the expansion ratio or attenuation ratio of the currently available mode relative to the benchmark bandwidth, the throughput gain index is accurately quantified. This index ultimately presents the relative gain advantage of the selected communication mode in terms of data concurrency throughput capability in the form of a positive scalar value, objectively and intuitively.

[0037] Simultaneously, this embodiment, targeting the high timeliness characteristics of in-vehicle services, extracts the delay deviation between the predicted link transmission delay and the corresponding data slice's tolerance delay constraint to quantify the delay penalty index. This embodiment reads the tolerance delay constraint declared in the protocol header of the currently processed data slice and performs a numerical difference evaluation with the predicted link transmission delay parsed in the first step. If the predicted link transmission delay exceeds the tolerance delay constraint (for example, for a certain autonomous driving braking slice, its tolerance limit is strictly set at 20 milliseconds, while the predicted transmission time of the current mode reaches 25 milliseconds), this embodiment defines the excess 5 milliseconds absolute time difference as the delay deviation. This embodiment further transforms the delay deviation into a high negative penalty value through a preset nonlinear amplification mapping function, thereby completing the quantification configuration of the delay penalty index. This ensures that candidate communication modes that cannot meet the timeliness baseline requirements receive a significant weight downgrade in the final utility evaluation.

[0038] To address the signaling oscillation loss caused by dynamic switching between heterogeneous networks, this embodiment continuously monitors the camping status of the underlying RF interface of the On-Board Unit (OBU). This embodiment uses high-frequency polling to read the status flag bits of the Medium Access Control (MAC) layer routing register to track the RF mode affiliation of the service data stream to which the target data slice belongs over the past two consecutive historical transmission cycles. When it is confirmed that the available communication mode has undergone cross-mode link reselection relative to the historical transmission cycle (i.e., it is determined that the current data slice is about to be scheduled to a completely different physical RF antenna interface than the previous cycle), this embodiment immediately triggers a switching loss assessment mechanism, collecting RF interface reconfiguration delay parameters to quantify the switching overhead penalty indicator. The RF interface reconfiguration delay parameter covers the total physical time consumed by protocol stack context unbinding, target RF phase-locked loop resynchronization, and network authentication handshake. In this embodiment, this baseline hardware reconfiguration overhead is measured as 50 milliseconds. This embodiment directly maps and amplifies this overhead time to the switching overhead penalty indicator, thereby strongly suppressing high-frequency and disordered "ping-pong" link blind switching behavior during the assessment phase.

[0039] Finally, this embodiment performs normalization and fusion processing on the quantified throughput gain index, latency penalty index, and switching overhead penalty index using the multidimensional utility maximization objective function, aggregating and outputting the transmission utility value of the data slice under the available communication mode. During the data computation for normalization and fusion processing, this embodiment first uses the theoretical extreme value limits of each index at the system level to perform division compression constraints on the three major indices, ensuring that their numerical results are uniformly standardized within the dimensionless standard range of 0 to 1. Subsequently, this embodiment dynamically assigns three adaptive weighting factors to the current data slice based on its specific service type, multiplying the normalized throughput gain index by a positive gain weight, and subtracting the results of multiplying the normalized latency penalty index and switching overhead penalty index by their respective negative loss weights. Through this normalized weighted summation processing mechanism based on multidimensional link characteristics, this embodiment finally aggregates and calculates a scalar result that comprehensively characterizes the degree of transmission adaptability of the current mode, namely the transmission utility value. This value provides a unique and accurate basis for subsequently establishing the globally optimal target communication mode mapping relationship.

[0040] Furthermore, the specific implementation process of step 500 is as follows: This embodiment calculates the dynamic protocol stack encapsulation parameters corresponding to each data slice based on the target communication mode mapping relationship. Specifically, this embodiment first reads the forwarding table entries and status registers at the bottom layer of the vehicular network to parse the transmission attributes of the heterogeneous physical links pointed to by the target communication mode mapping relationship. These transmission attributes at least cover two key underlying physical environment evaluation indicators: maximum transmission unit (MPU) and expected bit error rate (BER). The MPU characterizes the maximum data frame length that a specific physical link can carry without network layer splitting, and its value directly determines the upper limit of the effective payload size of the top-level message encapsulation. The BER objectively reflects the transmission reliability benchmark of the link in the current time-varying radio electromagnetic environment. For example, for a 5G cellular vehicular network link in a high-speed mobile scenario, this embodiment parses and obtains its MPU benchmark value as 1500 bytes, and quantizes its BER to the order of 10⁻⁵ based on the channel state information fed back from the baseband chip. These precise physical transmission attributes provide a solid data foundation for the subsequent adaptive adjustment of protocol stack control parameters.

[0041] After extracting the underlying transmission attributes, this embodiment calculates the fragment length control factor based on the maximum transmission unit. Since the candidate transmission path set for a specific data slice in a multimodal concurrent transmission architecture of the vehicle network typically contains multiple heterogeneous link nodes connected in series or parallel, this embodiment first traverses and scans all physical links involved in the selected path set for the data slice, and rigorously extracts the available minimum and maximum transmission unit values ​​for each link. This embodiment defines this selected minimum value as the effective maximum transmission unit on the candidate path of the data slice. Its core function is to take the minimum load extreme value of each link in the path based on the "barrel effect" principle, thereby setting an absolutely safe upper limit for the packet encapsulation size on the entire transmission path, fundamentally preventing forced secondary fragmentation or direct packet loss by routers caused by load exceeding the limit at heterogeneous network junction nodes. For example, if a composite vehicle network data routing path contains two independent series segments with maximum transmission units of 1500 bytes and 1460 bytes respectively, this embodiment will force the use of 1460 bytes as the effective maximum transmission unit on the global path of the slice.

[0042] Furthermore, this embodiment introduces environmental constraints based on the effective maximum transmission unit (EMU) and calculates the final fragment length control factor. This embodiment extracts the predicted composite quality of the path containing the data slice as an adjustment parameter for the multiplication operation to control the dynamic expansion and contraction of the fragment length within a safe range. The predicted composite quality is a dimensionless evaluation factor strictly normalized to a value between 0 and 1, and its value is positively correlated with the available bandwidth margin and channel anti-interference stability of the path. This embodiment directly multiplies the aforementioned confirmed EMU value by the predicted composite quality to output the fragment length control factor for that specific data slice. Through this multiplication mapping mechanism, when the predicted overall quality is extremely good and approaches 1.0, the calculated fragment length control factor will infinitely approach the upper limit of the physical load of the link, thereby maximizing the transmission throughput efficiency. Conversely, when severe weather causes the predicted overall quality to drop to 0.4, this embodiment actively shrinks the fragment length control factor, significantly reducing the overhead of large-scale physical layer retransmission caused by severe channel fading by greatly reducing the transmission volume of a single message.

[0043] To address network congestion avoidance control in high-concurrency scenarios, this embodiment calculates the congestion window backoff coefficient in parallel based on the expected bit error rate and a preset packet loss recovery threshold. Specifically, this embodiment observes in real time the queuing delay increment and retransmission request frequency of candidate transmission paths within the current time slot, and weights and fuses these underlying observations with the expected bit error rate to generate a congestion index characterizing the severity of the current network bottleneck. This congestion index is designed as a dimensionless value, and its magnitude dynamically changes as the network queuing congestion worsens. To prevent the protocol stack from exhibiting overly conservative window contraction or response lag when dealing with transient congestion, this embodiment pre-codes the minimum and maximum boundaries of the backoff coefficient in the underlying protocol stack. In typical vehicle-mounted cooperative driving service transmission scenarios, this embodiment strictly sets the minimum boundary of the backoff coefficient to 0.5 to ensure that the protocol stack can maintain at least half of its basic throughput under extreme congestion; simultaneously, it constrains the maximum boundary to 0.9 to prevent unnecessary severe transmission degradation triggered by slight network jitter.

[0044] After clarifying the congestion indicators and upper and lower boundary constraints, this embodiment calculates the target backoff coefficient using a mapping function, combining the above factors. This backoff factor is used to dynamically adjust the congestion window size when a congestion event occurs. This embodiment uses the aforementioned boundary thresholds to smooth and normalize the dynamically generated congestion indicators. When a real congestion event is detected at the physical layer, the current congestion window size is directly multiplied by the calculated backoff coefficient for backoff updating, thus outputting the updated available congestion window threshold. For example, if the current congestion window size is at full capacity (64 segments), and the backoff coefficient calculated based on the real-time congestion indicators is 0.75, this embodiment will forcibly reduce and update the congestion window to 48 segments in the next control cycle. After these two logically independent data processing branches, this embodiment finally combines the calculated fragment length control factor and the congestion window backoff coefficient to form the dynamic protocol stack encapsulation parameters. These two parameters form a complete protocol configuration instruction package, providing precise quantitative control input for subsequent adaptive reconstruction and customized encapsulation of packets at the lower level of the communication protocol stack.

[0045] Specifically, the expression for the fragment length control factor is: ; Where i is the slice index. Let be the set of candidate paths for the i-th slice. The maximum transmission unit of link l, The minimum MTU available in the selected path for this slice. Let be the slice length of the i-th slice; The effective MTU on the candidate path of this slice is taken from the minimum MTU of each link in the path; The predicted overall quality of this slice path is determined by a value ranging from [value range missing]. This is used to control the expansion and contraction of the fragment length; Let be the set of candidate paths for the i-th slice.

[0046] The expression for the congestion window backoff coefficient is: If a congestion event occurs, the window will be updated to... With congestion indicators And change, This is the congestion index (dimensionless, derived from a combination of observations such as queuing delay and packet loss) for the i-th path in the current time slot. These represent the minimum and maximum boundaries of the retreat coefficient, respectively. It is used as a backoff coefficient to dynamically adjust the congestion window size during congestion; This represents the current congestion window size.

[0047] Furthermore, the specific implementation process of step 600 is as follows: This embodiment obtains the pointer to the protocol stack transmission control block preset in the kernel space of the vehicle unit, and replaces the default packet fragmentation threshold of the communication protocol stack with the fragmentation length control factor in the dynamic protocol stack encapsulation parameters. Specifically, this embodiment overwrites the fragmentation control register of the underlying network layer through a memory mapping mechanism. For example, it forcibly erases and overwrites the 1500-byte default fragmentation limit fixed in the standard network protocol with a 1280-byte fragmentation length control factor calculated based on link prediction in the early stage. Simultaneously, this embodiment listens to the congestion control context of the transport layer state machine and replaces the default retransmission window coefficient with the congestion window backoff coefficient in the dynamic protocol stack encapsulation parameters. This embodiment replaces the constant backoff multiplier fixed in the underlying source code of the protocol stack with a dynamically generated precise value. For example, it replaces the default halved backoff coefficient of 0.50 with a congestion window backoff coefficient of 0.75 calculated based on the expected bit error rate of the channel, thereby completing the update of the control parameters. This physical memory-level parameter overwrite action completely breaks the static parameter rigidity defect of the traditional vehicle communication protocol stack, giving the underlying data transmission engine the ability to adaptively perceive and actively defend against the highly dynamic physical channels of the vehicle network.

[0048] After overwriting the underlying control parameters, this embodiment uses the updated communication protocol stack to generate a protocol stack header with a new control header structure. This embodiment allocates a contiguous direct memory access buffer in the vehicle memory bus and calls the protocol stack core encapsulation function to sequentially burn customized protocol identifier bits and routing control fields into this buffer. Unlike conventional static fixed-length basic headers, this embodiment, based on the previously determined target communication mode mapping relationship, dynamically appends customized cross-modal routing addressing tags and service level indication bits to the end of the standard header data stream. For example, this embodiment allocates and writes a 4-byte customized extended control field immediately after the conventional 20-byte basic network layer header, thereby generating a protocol stack header with a total length of 24 bytes in physical memory. This protocol stack header with a new control header structure not only fully retains the handshake capability of backward compatibility with the underlying standard protocol, but also enables the underlying multi-mode radio frequency transceiver hardware to achieve ultra-fast hardware-level routing and forwarding without packet disassembly by embedding a compact physical mode feature routing field, significantly reducing the microsecond-level processing time caused by cross-layer parsing.

[0049] Finally, in this embodiment, the corresponding data slices are encapsulated as payloads into the protocol stack header to generate a reconstructed message, thereby completing the adaptive optimization configuration of the multimodal communication protocol. This embodiment utilizes the direct memory access controller hardware within the vehicle-mounted processing unit to directly move and seamlessly splice the data slices pre-residing in the application layer cache queue to the physical tail address space of the protocol stack header via a zero-memory copy mechanism. Through this low-level physical address pointer mapping and high-speed data block splicing, this embodiment strictly adheres to the physical size boundaries constrained by the fragment length control factor, generating a fixed-length reconstructed message with a total length of, for example, precisely 1304 bytes. This reconstructed message perfectly matches the current actual physical carrying capacity limit of the target heterogeneous underlying link in terms of physical structure and payload volume, completely eliminating the phenomenon of forced secondary fragmentation or tail truncation caused by volume overload after the message is sent to the media access control layer. Thus, this embodiment completes the entire execution chain from underlying parameter reshaping to physical message finalization, ensuring that each frame of critical vehicle networking service data can be dispatched to the matching radio frequency antenna in the optimal protocol form for deterministic low-latency transmission.

[0050] This embodiment also provides an OBU multimodal communication protocol optimization system, including: Data acquisition module 1 is used to acquire the multimodal network environment parameters and service data to be transmitted from the on-board unit (OBU). The timing state deduction module 2 is used to input the multimodal network environment parameters into a preset spatiotemporal correlation channel prediction model to perform timing state deduction and obtain a predicted link quality topology sequence. The transport layer decoupling and segmentation module 3 is used to perform transport layer decoupling and segmentation on the service data to be transmitted based on the service data to be transmitted and the predicted link quality topology sequence, through a dynamic protocol stack orchestration mechanism, to obtain a service data slice set composed of multiple data slices. The communication mode optimization and matching module 4 is used to perform communication mode optimization and matching on each of the data slices in the service data slice set based on the predicted link quality topology sequence, and determine the target communication mode mapping relationship corresponding to each of the data slices; The encapsulation parameter calculation module 5 is used to calculate the dynamic protocol stack encapsulation parameters corresponding to each data slice based on the target communication mode mapping relationship. The protocol configuration and customized encapsulation module 6 is used to update the control parameters of the communication protocol stack of the on-board unit (OBU) according to the dynamic protocol stack encapsulation parameters, and combine the target communication mode mapping relationship to perform customized message encapsulation on the corresponding data slice in the service data slice set using the communication protocol stack with updated control parameters, so as to complete the adaptive optimization configuration of the multimodal communication protocol.

[0051] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0052] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. Furthermore, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. An optimization method for OBU multimodal communication protocol, characterized in that, include: Obtain the multimodal network environment parameters and service data to be transmitted from the on-board unit (OBU); The multimodal network environment parameters are input into a preset spatiotemporal correlation channel prediction model to perform time-series state deduction, thereby obtaining a predicted link quality topology sequence. Based on the service data to be transmitted and the predicted link quality topology sequence, the service data to be transmitted is decoupled and segmented at the transport layer through a dynamic protocol stack orchestration mechanism to obtain a service data slice set composed of multiple data slices. Based on the predicted link quality topology sequence, communication mode optimization matching is performed on each data slice in the service data slice set to determine the target communication mode mapping relationship corresponding to each data slice. Based on the target communication mode mapping relationship, calculate the dynamic protocol stack encapsulation parameters corresponding to each data slice; The control parameters of the communication protocol stack of the on-board unit (OBU) are updated according to the dynamic protocol stack encapsulation parameters. Combined with the target communication mode mapping relationship, the communication protocol stack with updated control parameters is used to perform customized message encapsulation on the corresponding data slice in the service data slice set to complete the adaptive optimization configuration of the multimodal communication protocol.

2. The OBU multimodal communication protocol optimization method according to claim 1, characterized in that, The multimodal network environment parameters include: Real-time channel status and vehicle spatiotemporal trajectory data of multiple heterogeneous physical links.

3. The OBU multimodal communication protocol optimization method according to claim 2, characterized in that, The real-time channel status of the multiple heterogeneous physical links includes: real-time signal-to-noise ratio, received signal strength indication, and available bandwidth of the link; the vehicle spatiotemporal trajectory data includes: the current geographical coordinates, driving speed, and driving heading angle of the vehicle mounted on the on-board unit (OBU).

4. The OBU multimodal communication protocol optimization method according to claim 1, characterized in that, The step of inputting the multimodal network environment parameters into a preset spatiotemporal correlation channel prediction model for time-series state deduction to obtain a predicted link quality topology sequence includes: The spatiotemporal feature extraction network in the spatiotemporal correlation channel prediction model is used to perform feature mapping on the multimodal network environment parameters to obtain the spatiotemporal correlation feature vector of the channel; The channel spatiotemporal correlation feature vector is input into the long short-term memory network node in the spatiotemporal correlation channel prediction model to perform temporal evolution deduction, and the link quality state probability matrix for multiple consecutive time steps in the future is obtained. Based on the link quality state probability matrix, the state node with the highest probability at each time step is extracted to obtain the predicted link quality topology sequence.

5. The OBU multimodal communication protocol optimization method according to claim 1, characterized in that, Based on the service data to be transmitted and the predicted link quality topology sequence, the service data to be transmitted is decoupled and segmented at the transport layer through a dynamic protocol stack orchestration mechanism to obtain a service data slice set composed of multiple data slices, including: Extract the service priority and tolerable latency constraints attached to the service data to be transmitted, and parse the predicted link quality topology sequence to obtain the network carrying capacity threshold; The network carrying capacity threshold is used as the upper limit of resource allocation for the transmission queue at the bottom layer of the protocol stack. Within the upper limit of resource allocation, data segments that meet the constraints of high service priority and low latency tolerance in the data to be transmitted are extracted to generate a high-guarantee transmission stream. Data segments that exceed the network carrying capacity threshold are stripped into a degraded transmission stream to obtain data streams of different transmission levels. The data streams of the different transmission levels are independently segmented according to a preset payload granularity that matches each of the transmission levels to obtain the service data slice set.

6. The OBU multimodal communication protocol optimization method according to claim 1, characterized in that, The step of performing communication mode optimization matching on each data slice in the service data slice set based on the predicted link quality topology sequence to determine the target communication mode mapping relationship corresponding to each data slice includes: A multidimensional utility maximization objective function is established, wherein the multidimensional utility maximization objective function is configured to use throughput gain index, latency penalty index and switching overhead penalty index as multidimensional optimization parameters; Based on the predicted link quality topology sequence, the transmission utility value of each data slice in the service data slice set under each available communication mode is evaluated using the multidimensional utility maximization objective function; The available communication mode with the largest transmission utility value is selected as the target communication mode to establish the target communication mode mapping relationship between each data slice and the corresponding target communication mode.

7. The OBU multimodal communication protocol optimization method according to claim 6, characterized in that, The step of combining the predicted link quality topology sequence and using the multidimensional utility maximization objective function to evaluate the transmission utility value of each data slice in the service data slice set under each available communication mode includes: The predicted link quality topology sequence is analyzed to obtain the expected physical layer bandwidth and predicted link transmission delay of the available communication modes; The expected physical layer bandwidth is mapped to a network layer throughput characterization value to quantify the throughput gain index, and the delay deviation between the predicted link transmission delay and the corresponding data slice's tolerance delay constraint is extracted to quantify the delay penalty index. The stationary status of the underlying radio frequency interface of the vehicle unit (OBU) is detected. When it is confirmed that the available communication mode has undergone cross-mode link reselection relative to the historical transmission cycle, the radio frequency interface reconfiguration delay parameter is collected to quantify the handover overhead penalty index. The quantized throughput gain index, latency penalty index, and switching overhead penalty index are normalized and fused using the multidimensional utility maximization objective function, and the transmission utility value of the data slice under the available communication mode is aggregated and output.

8. The OBU multimodal communication protocol optimization method according to claim 1, characterized in that, The step of calculating the dynamic protocol stack encapsulation parameters corresponding to each data slice based on the target communication mode mapping relationship includes: The transmission attributes of the heterogeneous physical link pointed to by the target communication mode mapping relationship are analyzed, wherein the transmission attributes include: maximum transmission unit and expected bit error rate; Calculate the fragment length control factor based on the maximum transmission unit; Calculate the congestion window backoff coefficient based on the expected bit error rate and the preset packet loss recovery threshold; The combination of fragment length control factor and congestion window backoff coefficient constitutes the dynamic protocol stack encapsulation parameters.

9. The OBU multimodal communication protocol optimization method according to claim 8, characterized in that, The step of updating the control parameters of the communication protocol stack of the on-board unit (OBU) according to the dynamic protocol stack encapsulation parameters, and combining the target communication mode mapping relationship, using the communication protocol stack with updated control parameters to perform customized message encapsulation on the corresponding data slice in the service data slice set, in order to complete the adaptive optimization configuration of the multimodal communication protocol, includes: The default message fragmentation threshold of the communication protocol stack is replaced with the fragmentation length control factor in the dynamic protocol stack encapsulation parameters, and the default retransmission window coefficient is replaced with the congestion window backoff coefficient in the dynamic protocol stack encapsulation parameters to complete the update of the control parameters. The communication protocol stack with updated control parameters is used to generate a protocol stack header with a new control header structure. The corresponding data slice is encapsulated as a payload into the protocol stack header to generate a reconstructed message, thereby completing the adaptive optimization configuration.

10. An OBU multimodal communication protocol optimization system, characterized in that, include: The data acquisition module is used to acquire the multimodal network environment parameters and the service data to be transmitted from the on-board unit (OBU). The timing state deduction module is used to input the multimodal network environment parameters into a preset spatiotemporal correlation channel prediction model to perform timing state deduction and obtain a predicted link quality topology sequence. The transport layer decoupling and segmentation module is used to decouple and segment the service data to be transmitted based on the service data to be transmitted and the predicted link quality topology sequence through a dynamic protocol stack orchestration mechanism, so as to obtain a service data slice set composed of multiple data slices. The communication mode optimization and matching module is used to perform communication mode optimization and matching on each of the data slices in the service data slice set based on the predicted link quality topology sequence, and determine the target communication mode mapping relationship corresponding to each of the data slices; The encapsulation parameter calculation module is used to calculate the dynamic protocol stack encapsulation parameters corresponding to each of the data slices based on the target communication mode mapping relationship. The protocol configuration and customized encapsulation module is used to update the control parameters of the communication protocol stack of the on-board unit (OBU) according to the dynamic protocol stack encapsulation parameters, and combine the target communication mode mapping relationship to perform customized message encapsulation on the corresponding data slice in the service data slice set using the communication protocol stack with updated control parameters, so as to complete the adaptive optimization configuration of the multimodal communication protocol.

Images