Search

CN-121013141-B - 5G network optimization method and system based on intelligent dynamic multi-band self-adaptive switching system

CN121013141BCN 121013141 BCN121013141 BCN 121013141BCN-121013141-B

Abstract

The invention relates to the technical field of vehicle-mounted communication, and provides a 5G network optimization method and system based on an intelligent dynamic multi-band self-adaptive switching system, wherein the method comprises the steps of obtaining network signal quality data, environment perception data and a communication task demand set; the method comprises the steps of extracting dynamic barrier characteristics based on environment perception data, outputting signal shielding prediction through a dual-branch mixed deep learning model, screening candidate link construction candidate sets which are qualified in inspection, generating dynamic weights by utilizing a deep Q network, calculating candidate link comprehensive scores, determining a main communication link and generating communication strategy instructions under a hard constraint condition, carrying out dynamic fragmentation and multi-path distribution on data streams according to instruction types of the communication strategy instructions, carrying out verification, recombination and intelligent delivery on receiving ends, and collecting feedback data packets for network closed-loop feedback. The invention combines multi-mode sensing and signal shielding prediction to construct a self-adaptive switching and cooperative transmission mechanism of multi-band communication.

Inventors

  • SHENG XIAOFEI
  • SHEN DI
  • REN PENG
  • BAI HAIMING
  • HE WENHAO

Assignees

  • 芜湖辛巴网络科技有限公司

Dates

Publication Date
20260512
Application Date
20250812

Claims (9)

  1. 1. The 5G network optimization method based on the intelligent dynamic multi-band self-adaptive switching system is characterized by comprising the following steps of: acquiring real-time network signal quality data, environment awareness data and a communication task demand set; extracting a dynamic obstacle feature vector set based on environment sensing data, integrating the dynamic obstacle feature vector set based on the environment sensing data and the dynamic obstacle feature vector set, inputting the dynamic obstacle feature vector set into a parallel double-branch mixed deep learning model for signal shielding prediction, and outputting a signal shielding prediction result of a multi-dimensional parameter; sequentially executing signal link feasibility test and network resource availability test on the obtained candidate links, and screening out qualified candidate links to construct a candidate set of the multi-dimensional performance profile; Judging whether to activate a communication decision process or not by combining a dual-mode decision trigger mechanism based on real-time network signal quality data and signal shielding prediction results, generating a dynamic weight factor by using a depth Q network according to a state vector of a current communication environment if the communication decision process is activated, and calculating a comprehensive score of each candidate link according to the dynamic weight factor and a multi-dimensional performance profile of each candidate link; The dual-mode decision triggering mechanism is activated when any one of the following conditions is met, if the real-time network signal quality data shows that the key performance index of the current using link is lower than the preset minimum availability threshold value, which indicates that the service quality is unacceptably deteriorated, the passive response triggering mechanism is activated; according to the hard constraint conditions of the communication service demands, carrying out hard constraint condition inspection on the candidate links ordered according to the comprehensive score descending order one by one until the candidate links meeting the hard constraint conditions are found to be the main communication links, and carrying out decision between a switching strategy instruction and a cooperative strategy instruction according to the overall condition of the current network resource to generate a final communication strategy instruction; The switching strategy instruction comprises a strategy instruction type 'HANDOVER', and finally the frequency band parameter of the selected main communication link and the triggering reason of the current link switching are recorded; The method comprises the steps of judging the instruction type according to a communication strategy instruction, executing dynamic fragmentation on a data stream to generate a data packet capable of being transmitted concurrently if the communication strategy instruction is a cooperative strategy, sequentially executing cyclic redundancy check, disordered recombination and intelligent delivery on the data packet at a receiving end to restore the data packet to a complete and ordered data stream, collecting and packaging a feedback data packet containing key performance indexes of the instruction after the communication strategy instruction is executed, and using the feedback data packet for iterative training of a deep Q network to update decision weights so as to realize self-adaptive closed loop feedback.
  2. 2. The 5G network optimization method based on the intelligent dynamic multi-band adaptive switching system according to claim 1, wherein the method for outputting the signal shielding prediction result comprises: Carrying out multi-stage processing on the dynamic obstacle data to generate a dynamic obstacle characteristic vector set containing morphology and motion parameters; And integrating static environment data, vehicle self dynamic data and dynamic obstacle feature vector sets, inputting the integrated static environment data, the vehicle self dynamic data and the dynamic obstacle feature vector sets into a parallel double-branch mixed deep learning model, respectively extracting a spatial feature vector and a time feature vector, fusing the spatial feature vector and the time feature vector to construct a space-time fusion feature vector, decoding the space-time fusion feature vector by combining a prediction solution terminal, and outputting a signal shielding prediction result comprising a predicted affected frequency band, a predicted signal attenuation value, a predicted starting time of a shielding event and a predicted duration of the shielding event.
  3. 3. The 5G network optimization method based on the intelligent dynamic multi-band adaptive switching system of claim 2, wherein the generating of the dynamic obstacle feature vector set of morphology and motion parameters includes morphology feature extraction and trajectory feature extraction; carrying out morphological feature extraction by adopting a convex hull algorithm based on dynamic obstacle data, determining the length, the width and the height of the dynamic obstacle, further dividing data points into subcubes, carrying out weighted summation according to the data point density of each subcubes, and estimating the total volume of the obstacle; And extracting track characteristics of the centroids of the same obstacle identified in the continuous frames by adopting a Kalman filtering algorithm, and estimating the instantaneous speed and the instantaneous movement direction angle of the obstacle by calculating the position change of the centroids between continuous time steps.
  4. 4. The method for optimizing a 5G network based on an intelligent dynamic multi-band adaptive switching system according to claim 1, wherein the method for generating the communication policy instruction comprises: Acquiring a candidate link list and a network load rate, performing signal link feasibility inspection on the candidate link list based on real-time network signal quality data, performing network resource availability inspection on the candidate link list based on the network load rate, constructing a candidate set by all the remaining qualified candidate links together after the feasibility and availability inspection is completed, and generating a multidimensional performance profile containing a signal quality index, a network load index and a distance index based on each candidate link in the candidate set; Judging whether to activate a communication decision process or not by combining a dual-mode decision trigger mechanism based on real-time network signal quality data and signal shielding prediction results, generating a dynamic weight factor by using a depth Q network according to a state vector of a current communication environment if the communication decision process is activated, and calculating a comprehensive score of each candidate link according to the dynamic weight factor and a multi-dimensional performance profile of each candidate link; According to the hard constraint condition of the communication service requirement, carrying out hard constraint condition test on the first candidate link ranked in the candidate set which is ordered in descending order according to the comprehensive score, if the candidate link meets the hard constraint condition, determining the candidate link as a main communication link, if the candidate link does not meet the hard constraint condition, automatically iterating to the next candidate link in the candidate set which is ordered in descending order, continuing to carry out hard constraint condition test until the candidate link which meets the hard constraint condition is found, and carrying out decision between a switching strategy instruction and a cooperative strategy instruction according to the overall condition of the current network resource, so as to generate a final communication strategy instruction; the hard constraint condition of the communication service requirement is the performance requirement of the key data flow of the data flow task with the highest service priority in the communication service requirement, including the minimum guarantee width and the maximum transmission delay.
  5. 5. The method for optimizing a 5G network based on an intelligent dynamic multi-band adaptive switching system according to claim 4, wherein the method for choosing between the switching policy instruction and the cooperative policy instruction comprises: When the system judges that the current network environment is relatively tense, namely, the quality of other candidate links is generally poor except the selected main communication link, or the main purpose of the decision is to avoid the deterministic failure risk of the current link, the system starts a switching strategy instruction generation flow; When the system judges that the current network environment has abundant resources, namely, a plurality of other high-quality links exist in the candidate set besides the determined main communication link, the system can start a collaborative strategy instruction generation flow.
  6. 6. The method for optimizing a 5G network based on an intelligent dynamic multi-band adaptive switching system according to claim 5, wherein the generating process of the collaborative policy instruction includes: Distributing the control task with the highest priority level to the determined main communication link, and writing the control task as the first key value pair in the parameter dictionary; creating two temporary resource pools of a task pool to be handled and an available link pool, wherein the task pool to be handled comprises other service flows except for the allocated key service in the communication service demand and is ordered according to the service priority from high to low, and the available link pool comprises other high-quality candidate links except for the occupied main communication link in the candidate set and is ordered according to the comprehensive score from high to low; and processing the service with the highest priority from the to-be-handled task pool, and traversing the available link pool in turn to find the best matching link for the service.
  7. 7. The intelligent dynamic multi-band adaptive switching system-based 5G network optimization method of claim 6, wherein the best matching link comprises three types of scenarios: A single link meets a scene, if the performance of the link with the highest comprehensive score in the available link pool of the current business to be handled meets the core requirement of the business, the link is allocated to the business, corresponding key value pairs are added in a parameter dictionary, and meanwhile, the business and the link are removed from respective resource pools, so that the resources are ensured not to be repeatedly allocated; In a link aggregation scenario, if any single link in the available link pool can meet the current high-bandwidth service requirement, the system further judges whether the service requirement can be met by aggregating the bandwidths of a plurality of links; in the link multiplexing scenario, after the link allocation of the task with high priority is completed, the system continues to process the task with lower priority, and in addition to considering the unused link resources in the available link pool, it also intelligently evaluates whether there are allocated links but still have residual capacity for multiplexing.
  8. 8. The method for optimizing a 5G network based on an intelligent dynamic multi-band adaptive switching system according to claim 1, wherein the method for implementing adaptive closed-loop feedback comprises: according to the instruction type of the communication strategy instruction, if the instruction type is a cooperative strategy, dynamically calculating the optimal slicing size based on the real-time available bandwidth of each link and a target transmission time window, and executing dynamic slicing on the data stream according to the optimal slicing size to generate a data packet capable of carrying out multilink parallel transmission; Performing cyclic redundancy check on the received data packet, storing the checked data packet into an annular buffer area according to the serial number of the data packet for disordered recombination, and adopting a bidirectional scanning recombination algorithm to perform intelligent decoding and delivery on the data packet in the annular buffer area in sequence so as to restore the received data packet into a complete and ordered data stream; after the communication strategy instruction is executed, the execution effect of the communication strategy instruction is evaluated, the key performance index acquired in the evaluation is packaged into a feedback data packet, and the feedback data packet is used as a training sample for iterative training of a deep Q network to update decision weights, so that self-adaptive closed-loop feedback is realized.
  9. 9. The 5G network optimization system based on the intelligent dynamic multi-band self-adaptive switching system is used for realizing the 5G network optimization method based on the intelligent dynamic multi-band self-adaptive switching system, and is characterized by comprising a multi-source heterogeneous data acquisition and sensing module, an intelligent signal shielding prediction module, a communication strategy instruction generation module and a closed loop feedback optimization module; The multi-source heterogeneous data acquisition and perception module is used for acquiring real-time network signal quality data, environment perception data and a communication task demand set; The intelligent signal shielding prediction module extracts a dynamic obstacle feature vector set based on environment sensing data, integrates the dynamic obstacle feature vector set based on the environment sensing data and the dynamic obstacle feature vector set, inputs the integrated dynamic obstacle feature vector set into a parallel double-branch mixed deep learning model to conduct signal shielding prediction, and outputs a signal shielding prediction result of multi-dimensional parameters; The communication strategy instruction generation module is used for sequentially executing signal link feasibility test and network resource availability test on the obtained candidate links, and screening out qualified candidate links to construct a candidate set of the multidimensional performance profile; judging whether to activate a communication decision process based on real-time network signal quality data and a signal shielding prediction result, combining a dual-mode decision trigger mechanism, generating a dynamic weight factor by using a deep Q network according to a state vector of a current communication environment and calculating a comprehensive score of each candidate link according to the dynamic weight factor and the multi-dimensional performance profile of each candidate link after the communication decision process is activated, wherein the dual-mode decision trigger mechanism is activated when any one of the following conditions is met, and activates a passive response trigger mechanism when the real-time network signal quality data shows that a key performance index of a currently used link is lower than a preset minimum availability threshold value, and shows that unacceptable deterioration of service quality occurs, and activates a passive response trigger mechanism when the signal shielding prediction result shows that the current communication link is about to be subjected to high probability and high attenuation degree signal shielding event in a short time in the future, and activates an active prevention trigger mechanism; The closed loop feedback optimization module is used for judging the instruction type according to the communication strategy instruction, if the communication strategy instruction is a cooperative strategy, executing dynamic fragmentation on the data stream to generate a data packet capable of being transmitted concurrently, sequentially executing cyclic redundancy check, disordered recombination and intelligent delivery on the data packet at a receiving end to restore the data packet to a complete and ordered data stream, acquiring and packaging a feedback data packet containing key performance indexes of the instruction after the communication strategy instruction is executed, and using the feedback data packet for iterative training of a deep Q network to update decision weights so as to realize self-adaptive closed loop feedback.

Description

5G network optimization method and system based on intelligent dynamic multi-band self-adaptive switching system Technical Field The invention relates to the technical field of vehicle-mounted communication, in particular to a 5G network optimization method and system based on an intelligent dynamic multi-band self-adaptive switching system. Background With the deep application of 5G technology in the fields of intelligent transportation and automatic driving, vehicle remote control has become a core technology for realizing high-level automatic driving and unmanned operation. In a complex and changeable electromagnetic environment, how to provide continuous, stable and low-delay communication guarantee for remote control command and high-definition video feedback, effectively overcomes the defects of easy interruption, high delay and the like of the traditional single-path communication scheme, and becomes an important subject to be solved in the large-scale commercial process of pushing the vehicle remote control technology. The Chinese patent application with the publication number of CN120264315A provides a network slice optimization method of a 5G private network, which comprises the steps of constructing a time sequence performance model in the slice operation process, monitoring small changes including end-to-end time delay, throughput and packet loss rate, extracting a short-period fluctuation mode of an index, comparing a gradient with a historical stability threshold, judging whether the current response area of the slice is in a performance critical state by identifying continuous jitter and not-default index drift behaviors, recording unequal response including jitter improvement and delay decline brought about after each resource adjustment, carrying out trend fitting on data, judging whether a performance platform period or a critical jump point exists or not, defining three intervals of resource investment in the model, namely an invalidity area, a high-sensitivity area with rapid change of performance and an ascending saturation area with stable performance and continuous resources, identifying the current response area of the slice through the model, determining that the regulation strategy is not to be input as a core by the equal proportion, and selecting the optimal response area as a resource allocation interval. However, the current technology still faces many challenges. The 5G high-frequency band signal has limited propagation capability at the physical level, particularly in dense areas of urban high buildings or areas with complex terrains, has weak diffraction and penetration capability, is extremely easy to be blocked by buildings or mountains, and thus leads to frequent interruption of communication links. The existing system mostly adopts a passive link switching mechanism, and the backup link is usually triggered only after the current signal quality is seriously deteriorated, so that the future channel state is lack of prospective prediction, and switching delay and even failure are caused. Meanwhile, the switching decision logic is simple, real-time quality and business differentiation requirements of each candidate link, such as time delay and bandwidth, cannot be dynamically and comprehensively weighed, global optimal scheduling is difficult to realize, and further remote control instruction loss or video feedback blocking is caused, so that serious threat is formed on driving safety. Disclosure of Invention In order to achieve the above purpose, the present invention provides a 5G network optimization method based on an intelligent dynamic multi-band adaptive switching system, and the specific technical scheme is as follows: A5G network optimization method and system based on an intelligent dynamic multi-band self-adaptive switching system comprises the following steps: Based on vehicle-mounted integrated multi-band equipment, a multi-source sensor network, a positioning system and an application perception recognition module, synchronously acquiring real-time network signal quality data P sig, environment perception data D env and a communication task demand set R task; Based on the environment sensing data D env, extracting a dynamic obstacle feature vector set F obs, integrating based on the environment sensing data D env and the dynamic obstacle feature vector set F obs, inputting the integrated dynamic obstacle feature vector set F obs into a parallel double-branch mixed deep learning model M DL for signal shielding prediction, and outputting a signal shielding prediction result O pred of the multidimensional parameter; Sequentially executing signal link feasibility test and network resource availability test on the obtained candidate links, screening out a candidate set C set of qualified candidate links to construct a multi-dimensional performance profile, generating a dynamic weight factor by utilizing a deep Q network when a communication decision flow is activat