A method and system for dynamic computing power optimization of multi-band communication in high-speed rail

Abstract

This application provides a method and system for dynamic computing power optimization in multi-band communication for high-speed rail. It is applied in the field of communication technology. By dynamically sensing the channel status and computing resource status, it determines the computing power required for emergency recovery tasks of the target frequency band and adjusts the resource allocation to meet the demand. This has the advantages of being able to quickly respond to sudden channel degradation, dynamically adjust computing power allocation, and ensure communication reliability.

Description

Technical Field

[0001] This application relates to the field of communication technology, and in particular to a method and system for dynamic computing power optimization of multi-band communication in high-speed rail. Background Technology

[0002] High-speed rail onboard communication systems connect to terrestrial networks via multiple wireless frequency bands (including operator cellular network bands, railway-dedicated bands, and satellite communication bands) during high-speed operation to ensure train operation safety and passenger service quality. These frequency bands carry both critical and non-critical services, and different frequency bands and service types have varying requirements for heterogeneous computing units in the onboard computing resource pool. For example, critical services rely on general-purpose processors and encryption / decryption accelerators, high-volume services rely on data compression / decompression accelerators and general-purpose processors, and deteriorating channel quality increases the demand for channel encoding / decoding accelerators.

[0003] However, the high-speed operation of high-speed rail leads to rapid and dynamic changes in the wireless environment. Although generally predictable, sudden, localized, and severe interference events may occur in certain areas, causing a rapid deterioration in the channel quality of specific frequency bands within a very short time. This includes a decrease in signal-to-noise ratio, an increase in bit error rate, and a higher packet loss rate. This degradation is localized and non-gradual. In such situations, the onboard communication system will activate emergency handling mechanisms, such as increasing the number of automatic retransmissions, enabling forward error correction coding, rapid channel switching or frequency hopping, and protocol parameter renegotiation. These tasks cause a sudden surge in demand for onboard computing resources, with the demand concentrated on specific types of computing units. However, the onboard computing resource pool has limited and diverse heterogeneous resources. If a certain type of computing resource is limited or heavily occupied, it will become a bottleneck.

[0004] Traditional computing power allocation and scheduling strategies, based on total load or static rules, are ill-suited to handle sudden, unbalanced surges in computing power demand. This can lead to slow or continued deterioration of communication links in affected frequency bands, impacting communication reliability and even threatening the security of critical services. Furthermore, the lack of refined and coordinated computing power adjustment strategies may negatively affect the performance of other normal frequency bands, reducing overall communication efficiency.

[0005] Therefore, in the high-speed, multi-band communication environment of high-speed rail, how to quickly respond to sudden local interference events, accurately perceive the instantaneous demand of emergency handling tasks on specific types of computing power in the limited heterogeneous computing power resources on board, and carry out dynamic, fine-grained, and collaborative computing power allocation and scheduling are key challenges to ensure the reliability and efficiency of high-speed rail communication.

[0006] To address the aforementioned issues, existing technologies urgently need improvement. Summary of the Invention

[0007] In view of the shortcomings of the prior art, this application provides a method and system for dynamic computing power optimization of high-speed rail multi-band communication, which has the advantages of being able to quickly respond to sudden channel degradation, dynamically adjust computing power allocation, and ensure communication reliability.

[0008] Firstly, a dynamic computing power optimization method for multi-band communication in high-speed rail is provided, applied to high-speed rail, wherein the high-speed rail is equipped with a multi-band communication system and an on-board heterogeneous computing resource pool, and the method includes the following steps:

[0009] S1: Obtain the channel status parameters of each frequency band in the multi-band communication system, and the computing resource status parameters of each type of computing resource in the vehicle-mounted heterogeneous computing resource pool;

[0010] S2: Based on the channel state parameters, determine the target frequency band in each frequency band where instantaneous channel degradation occurs;

[0011] S3: In response to the target frequency band, determine the type of target computing resources required to perform the emergency recovery task of the target frequency band, and the corresponding target computing requirements;

[0012] S4: Based on the computing resource status parameters, adjust the resource allocation of the target computing resource type to meet the target computing demand.

[0013] This application proposes a dynamic computing power optimization method for multi-band communication in high-speed rail, which can quickly respond to sudden channel degradation, dynamically adjust computing power allocation, and ensure communication reliability.

[0014] Furthermore, step S1 includes:

[0015] S11: Generate a unified data collection trigger;

[0016] S12: Based on the unified acquisition trigger, obtain the channel state parameters and the computing resource state parameters;

[0017] S13: Pair the channel state parameters and computing resource state parameters corresponding to the same unified acquisition trigger to form a synchronization state snapshot, wherein the channel state parameters in the synchronization state snapshot are used to determine the target frequency band.

[0018] This application proposes a dynamic computing power optimization method for multi-band communication in high-speed rail, which improves the accuracy and real-time performance of status perception by synchronously collecting status data.

[0019] Furthermore, step S2 includes:

[0020] S21: Obtain the operating status of the high-speed rail;

[0021] S22: Based on the operating state, obtain the predicted channel state corresponding to the operating state;

[0022] S23: Based on the predicted channel state and the channel state parameters, determine the unexpected change in the channel state parameters;

[0023] S24: When the unexpected change meets the preset conditions, the frequency band is determined as the target frequency band.

[0024] This application proposes a dynamic computing power optimization method for multi-band communication in high-speed rail, which combines operating status and predicted channel status to more accurately identify unexpected channel degradation.

[0025] Furthermore, step S23 includes:

[0026] S231: Obtain the prediction confidence level corresponding to the predicted channel state;

[0027] S232: Generate a dynamic reference range for the channel state parameters based on the predicted confidence level;

[0028] S233: The deviation of the channel state parameter from the dynamic reference interval is determined as the unexpected change.

[0029] This application proposes a dynamic computing power optimization method for multi-band communication in high-speed rail, which uses prediction confidence to generate a dynamic benchmark interval, making the judgment of unexpected changes more flexible and accurate.

[0030] Furthermore, step S232 includes:

[0031] S2321: Obtain the section characteristics of the currently operating high-speed railway line;

[0032] S2322: Determine the interval generation parameters for generating the dynamic reference interval based on the segment characteristics;

[0033] S2323: Generate a dynamic reference interval for the channel state parameters based on the interval generation parameters, the predicted channel state, and the predicted confidence level.

[0034] Furthermore, step S3 includes:

[0035] S31: Obtain the service type carried in the target frequency band, and the service protection level corresponding to the service type;

[0036] S32: Determine the emergency recovery resource requirements corresponding to each of the service types based on the instantaneous channel degradation of the target frequency band and the service protection level;

[0037] S33: Based on the emergency recovery resource requirements, generate the target computing resource type and the target computing requirement.

[0038] Furthermore, step S32 includes:

[0039] S321: Obtain the service assurance level of services carried in non-target frequency bands;

[0040] S322: Based on the service assurance level of the non-target frequency band, determine the resource assurance requirements of the non-target frequency band;

[0041] S323: Under the premise of meeting the resource guarantee requirements of the non-target frequency band, determine the emergency recovery resource requirements corresponding to each of the service types based on the instantaneous channel degradation of the target frequency band and the service guarantee level.

[0042] Furthermore, step S4 includes:

[0043] S41: Obtain the service assurance level corresponding to the target computing demand;

[0044] S42: Prioritize the target computing requirements according to the business assurance level;

[0045] S43: Based on the priority sorting results and the computing resource status parameters, adjust the resource allocation of the target computing resource type.

[0046] Furthermore, step S43 includes:

[0047] S431: Obtain the status parameters of each computing resource instance within the target computing resource type;

[0048] S432: Obtain the target computing demand for the computing resource instance;

[0049] S433: Select a computing resource instance from the target computing resource type based on the status parameters and the requirements;

[0050] S434: Allocate resources to the selected computing resource instance.

[0051] Secondly, a high-speed rail multi-band communication dynamic computing power optimization system is provided for implementing the method described in any of the above claims, the system comprising:

[0052] Acquisition module: Acquires channel status parameters of each frequency band in the multi-band communication system, and computing resource status parameters of each type of computing resource in the vehicle-mounted heterogeneous computing resource pool;

[0053] Determination module: Based on the channel state parameters, determine the target frequency band in each frequency band where instantaneous channel degradation occurs;

[0054] Response module: In response to the target frequency band, determines the type of target computing resources required to perform the emergency recovery task of the target frequency band, and the corresponding target computing requirements;

[0055] Allocation module: Based on the computing resource status parameters, adjust the resource allocation of the target computing resource type to meet the target computing demand.

[0056] Beneficial effects: The dynamic computing power optimization method and system for high-speed rail multi-band communication proposed in this application determines the computing power required for emergency recovery tasks of the target frequency band by dynamically sensing the channel status and computing resource status, and adjusts the resource allocation to meet the demand. Thus, it has the advantages of being able to quickly respond to sudden channel degradation, dynamically adjust computing power allocation, and ensure communication reliability. Attached Figure Description

[0057] Figure 1 This is a flowchart of a dynamic computing power optimization method for high-speed rail multi-band communication proposed in this application.

[0058] Figure 2 This is a structural diagram of a dynamic computing power optimization system for multi-band communication in high-speed rail proposed in this application.

[0059] Figure 3 This is an architecture diagram of a dynamic computing power optimization system for multi-band communication in high-speed rail proposed in this application.

[0060] Labeling Explanation: 201, Acquisition Module; 202, Determination Module; 203, Response Module; 204, Allocation Module. Detailed Implementation

[0061] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. The components of the embodiments of this application described and marked in the accompanying drawings can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0062] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0063] Please refer to Figure 1 A dynamic computing power optimization method for multi-band communication in high-speed rail is proposed and applied to high-speed rail, which is equipped with a multi-band communication system and an onboard heterogeneous computing resource pool. The method includes the following steps:

[0064] S1: Obtain the channel status parameters of each frequency band in the multi-band communication system, as well as the computing resource status parameters of each type of computing resource in the vehicle-mounted heterogeneous computing resource pool;

[0065] S2: Based on channel state parameters, determine the target frequency band where instantaneous channel degradation occurs in each frequency band;

[0066] S3: In response to the target frequency band, determine the type of target computing resources required to perform the emergency recovery task for the target frequency band, and the corresponding target computing requirements;

[0067] S4: Based on the computing resource status parameters, adjust the resource allocation of the target computing resource type to meet the target computing demand.

[0068] Among them, channel state parameters refer to indicators that describe the quality of wireless channels, such as signal-to-noise ratio, bit error rate, packet loss rate, channel gain, and fading coefficient. They can be obtained by physical layer measurement, data link layer statistics, network layer probing, etc., with the aim of monitoring the communication quality of each frequency band in real time.

[0069] Computing resource status parameters refer to the availability, load, and performance indicators of various computing resources in the vehicle-mounted heterogeneous computing resource pool, such as CPU utilization, memory usage, accelerator queue length, and task completion time. These parameters can be obtained through operating system interfaces, virtualization platform monitoring, hardware sensors, etc., with the aim of understanding the real-time availability of vehicle-mounted heterogeneous computing resources.

[0070] Instantaneous channel degradation refers to a rapid and unexpected decline in the quality of a wireless channel within a short period of time. It can be judged based on the unexpected changes in channel state parameters, and its purpose is to quickly identify specific frequency bands affected by sudden interference.

[0071] The target frequency band refers to one or more specific wireless frequency bands that have been identified as experiencing transient channel degradation. The purpose is to focus subsequent emergency response and computing power adjustments on the affected frequency bands.

[0072] Emergency recovery tasks refer to processing tasks initiated to maintain communication connections or minimize data loss when instantaneous channel degradation occurs. Examples include enhanced forward error correction, increased retransmission, and rapid channel switching. The purpose is to address communication interruptions or performance degradation caused by channel degradation.

[0073] The target computing resource type refers to the specific type of heterogeneous computing resources required to perform emergency recovery tasks, such as general-purpose processors or specific hardware accelerators. Its purpose is to identify the specific heterogeneous computing power requirements of emergency tasks.

[0074] The target computing requirement refers to the specific computing power or resource quantity of a particular type of computing resources required to perform an emergency recovery task. Its purpose is to accurately assess the specific computing power requirements of an emergency task.

[0075] Adjusting resource allocation refers to dynamically modifying the allocation strategy for specific types of computing resources based on the real-time status of computing resources and the target computing demand. This could involve things like increasing priority, increasing the allocation ratio, or rescheduling tasks to available instances. The goal is to ensure that emergency recovery tasks can obtain the necessary computing power support in a timely manner.

[0076] The core innovation of this application lies in combining real-time channel status awareness of a multi-band communication system with real-time resource status awareness of an on-board heterogeneous computing resource pool. Based on the instantaneous degradation of the channel, it accurately assesses the demand for specific types of computing power for emergency recovery tasks, thereby dynamically adjusting the allocation of heterogeneous computing resources. This solves the problem that when sudden local interference causes instantaneous degradation of some frequency band channels, it is difficult to respond quickly and dynamically and finely adjust the allocation of heterogeneous computing power resources to prioritize communication recovery tasks. This achieves the effects of rapid response to channel degradation, accurate matching of computing power demand, priority for critical tasks, and improved system communication reliability.

[0077] This method achieves its functionality through dynamic feedback and adjustment. First, the system continuously or periodically acquires channel state parameters for each frequency band in the multi-band communication system, as well as computing resource state parameters for each type of computing resource in the vehicle-mounted heterogeneous computing resource pool. This data forms the basis for subsequent decision-making.

[0078] Next, based on the acquired channel state parameters, the system analyzes the communication quality of each frequency band, identifies the specific frequency band experiencing instantaneous channel degradation, and designates it as the target frequency band. Once the target frequency band is determined, the system immediately responds, assessing the specific types of heterogeneous computing resources required to perform the emergency recovery task and their specific computational load based on information such as the degradation status and the type of services carried by the target frequency band. This step transforms the communication problem into a demand for specific computing power.

[0079] Finally, the system dynamically adjusts and allocates target computing resources based on the acquired computing resource status parameters and the assessed target computing demand. This adjustment is real-time and targeted, aiming to prioritize emergency recovery tasks in the target frequency band and ensure that it can obtain the necessary computing power support in a timely manner.

[0080] The entire process forms a closed loop: sensing the status, identifying problems, assessing needs, and adjusting resources, thereby achieving a rapid and effective computing power response to sudden channel degradation.

[0081] As an exemplary implementation, this solution can be specifically implemented as follows: During the high-speed operation of a high-speed train,

[0082] A certain frequency band in the vehicle-mounted communication system encountered strong interference, causing a rapid decrease in the signal-to-noise ratio (SNR) and an increase in the bit error rate (BER). The system periodically reads the SNR and BER of each frequency band as channel status parameters through the physical layer interface of the vehicle-mounted communication module; and obtains the utilization rate and idle queue length of each CPU core, GPU, and FPGA accelerator as computing resource status parameters through the resource monitoring agent of the vehicle-mounted computing platform. This data is then aggregated in the vehicle-mounted resource management unit.

[0083] The onboard resource management unit analyzes channel state parameters and finds that the bit error rate (BER) of this frequency band exceeds a preset threshold for a short period of time, and deviates significantly from the BER predicted based on train position and speed. This indicates that the frequency band has experienced instantaneous degradation and is identified as the target frequency band. Based on the service type carried by this target frequency band and the degree of channel degradation, it is determined that enhanced forward error correction coding is needed to reduce the BER. The enhanced error correction coding task mainly relies on the onboard FPGA accelerator. The computational load required for enhanced coding is assessed, and the target computational resource type is determined to be an FPGA accelerator, with the target computational requirement defined as a specific unit of computational power.

[0084] The onboard resource management unit checks the computing resource status parameters of the FPGA accelerator and finds that the current FPGA utilization is high, but there are still some idle resources. Based on the target computing demand determined by the assessment, the task scheduling strategy of the FPGA accelerator is adjusted, the priority of the enhanced error correction task in this frequency band is increased, and the required FPGA computing resource instances or computing time slices are allocated to it to ensure that the task can be executed in a timely manner. If FPGA resources are insufficient, resource scheduling may need to be carried out according to the service guarantee level.

[0085] Through the above scheme, this application can quickly identify the target frequency band in the high-speed rail multi-band communication system where the channel is momentarily degraded, accurately assess the specific type of heterogeneous computing resources and their computing requirements required to perform the emergency recovery task of the target frequency band, and dynamically adjust the allocation of target computing resource types based on the real-time status of the on-board heterogeneous computing resources, thereby prioritizing the communication recovery task of the damaged frequency band, improving the reliability of high-speed rail communication under sudden local interference environment, and optimizing the utilization efficiency of on-board heterogeneous computing resources.

[0086] Furthermore, step S1 includes:

[0087] S11: Generate a unified data collection trigger;

[0088] S12: Based on unified acquisition triggering, obtain channel status parameters and computing resource status parameters;

[0089] S13: Pair the channel state parameters and computing resource state parameters corresponding to the same unified acquisition trigger to form a synchronization state snapshot, wherein the channel state parameters in the synchronization state snapshot are used to determine the target frequency band.

[0090] Among them, unified acquisition triggering refers to a signal, event or instruction used to synchronously start data acquisition operations of different system modules or processes. It can be implemented by timer interrupt, specific event signal or centralized control instruction.

[0091] Channel state parameters refer to data that reflect the quality and performance of wireless communication links, and may include indicators such as signal-to-noise ratio, bit error rate, packet loss rate, throughput, and latency.

[0092] Computing resource status parameters refer to data that reflects the availability and load of various computing resources in the vehicle-mounted heterogeneous computing resource pool. These parameters may include indicators such as CPU utilization, memory utilization, specific accelerator load, and task queue length.

[0093] Synchronization state snapshot refers to a data structure formed by logically associating or packaging the set of channel state parameters and the set of computing resource state parameters collected under the same unified acquisition trigger. It can be implemented using data records with timestamps or trigger IDs, associative arrays, or structures.

[0094] This application introduces a unified acquisition trigger mechanism to ensure that the acquisition actions of channel state parameters and computing resource state parameters are highly consistent in time. Specifically, when a unified acquisition trigger is generated, the channel state acquisition module and the computing resource state acquisition module within the system respond to the trigger simultaneously, initiating their respective data acquisition processes. Since the acquisition is based on the same trigger, the acquired channel state parameters and computing resource state parameters can reflect the true and synchronized state of the system at the trigger moment.

[0095] Subsequently, these synchronously acquired parameters are paired to form a synchronous state snapshot. This snapshot provides a view of the relationship between the channel environment and computing resource availability at a specific point in time. In this way, subsequent channel degradation judgments based on channel state parameters can directly utilize computing resource status information synchronized with the channel state, enabling emergency computing power allocation decisions based on the concurrent computing resource status when channel degradation is identified. This synchronous state awareness capability provides accurate basic data reflecting the overall system status for subsequent judgments and decisions, avoiding decision-making biases caused by asynchronous state information.

[0096] In one embodiment, the unified acquisition trigger can be generated by a periodically running timer, for example, the timer generates a trigger signal every fixed time interval (e.g., 100 milliseconds). Both the channel state acquisition unit and the vehicle computing resource management unit in the vehicle communication system listen to this timer signal. When a trigger signal is received, the channel state acquisition unit immediately acquires the channel state parameters for each frequency band, and the computing resource management unit immediately acquires the computing resource state parameters for each type of computing resource. The acquired set of channel state parameters and the set of computing resource state parameters are marked with the same timestamp or trigger ID and sent to a state processing module. Upon receiving the two sets of parameters with the same timestamp or trigger ID, the state processing module associates them to form a synchronous state snapshot data structure, such as a record containing a channel state list and a computing resource state list, and stores or passes it to subsequent processing units for channel degradation judgment and resource allocation. The set of channel state parameters in the synchronous state snapshot is then used to perform channel degradation judgment to determine the target frequency band.

[0097] Furthermore, step S2 includes:

[0098] S21: Obtain the operating status of high-speed trains;

[0099] S22: Based on the operating state, obtain the predicted channel state corresponding to the operating state;

[0100] S23: Determine the unexpected changes in the channel state parameters based on the predicted channel state and the channel state parameters;

[0101] S24: When the unexpected change meets the preset conditions, the frequency band is determined as the target frequency band.

[0102] The above technical solution is explained in detail. Step S21 refers to acquiring the current wireless environment-related status information of the high-speed train, such as its geographical location, operating speed, and specific line section. This information forms the basis for predicting the expected wireless channel state at the current location and speed.

[0103] Step S22 refers to predicting the expected channel quality level for the frequency band under the acquired operational status information using historical data. This predicted channel status provides a benchmark for measuring whether the actual channel status is abnormal.

[0104] Step S23 refers to comparing the actual measured channel state parameters, such as signal-to-noise ratio and bit error rate, with the predicted channel state and determining the degree of difference or deviation. This difference reflects the extent to which the actual channel state deviates from normal expectations and is a key indicator for identifying sudden and unexpected degradation.

[0105] Step S24 refers to the condition that a sudden, instantaneous channel degradation is considered to have occurred in the frequency band only when the unexpected change in the actual channel state relative to the predicted value reaches or exceeds a preset threshold or condition, requiring emergency handling. This preset condition is used to filter out normal, predictable environmental fluctuations or gradual degradation, ensuring that only genuine sudden degradation is identified.

[0106] As a specific implementation method, the above method can be implemented as follows:

[0107] First, the high-speed rail's current geographical location, speed, and the section of the line it is on are obtained through the onboard GPS module, speed sensor, and pre-stored line information. These constitute the high-speed rail's operating status.

[0108] Then, a pre-trained channel prediction model based on location and velocity can be used to predict the expected signal-to-noise ratio (SNR) of the frequency band at the current location and velocity, based on the acquired operational status. For example, the prediction model may be based on historical measurement data, having learned the average SNR and its fluctuation range at different locations and velocities.

[0109] Next, the current signal-to-noise ratio (SNR) of the frequency band, actually measured by the wireless communication module, is compared with the predicted SNR, and the difference between the two is calculated. This difference represents the unexpected change in the channel state parameters.

[0110] Finally, a preset condition is set. For example, if the actual signal-to-noise ratio (SNR) is lower than the predicted SNR by more than a certain threshold (e.g., 10 dB), then the unexpected change is considered to meet the preset condition, and the frequency band is identified as the target frequency band. In this way, frequency bands with actual SNRs far below the expected level can be identified, as these frequency bands are likely to have suffered sudden interference.

[0111] Furthermore, step S23 includes:

[0112] S231: Obtain the prediction confidence level corresponding to the predicted channel state;

[0113] S232: Generate a dynamic reference range for channel state parameters based on the predicted confidence level;

[0114] S233: The deviation of the channel state parameters from the dynamic reference interval is determined as an unexpected change.

[0115] Among them, prediction confidence refers to the quantitative assessment of the reliability of the predicted channel state, which can be achieved by using the confidence interval, probability value, or reliability score calculated based on historical data and current environmental characteristics output by the prediction model.

[0116] The dynamic reference interval refers to the interval that is dynamically adjusted based on the reliability of the prediction and is used to define the normal fluctuation range of the channel state parameters. It can be implemented by calculating the upper and lower limits based on the predicted value and the prediction confidence, for example, the predicted value ± a range factor related to the confidence.

[0117] Deviation refers to the distance or degree of difference between the actual channel state parameters and the dynamic reference interval. It can be achieved by calculating the minimum distance from the actual value to the interval boundary, the standardized difference between the actual value and the interval midpoint, or a Boolean judgment on whether the actual value falls outside the interval.

[0118] The proposed solution optimizes the process of determining unexpected changes in channel state parameters. First, it obtains the prediction confidence level associated with the predicted channel state, providing information for assessing the reliability of the prediction result. Then, based on this prediction confidence level, instead of using a single predicted value or a fixed threshold as the judgment benchmark, it generates a dynamically adjusted benchmark interval. A higher prediction confidence level indicates a more reliable prediction, allowing for a narrower benchmark interval and thus stricter requirements on actual channel state fluctuations; conversely, a lower prediction confidence level indicates greater prediction uncertainty, allowing for a wider benchmark interval and a greater range of fluctuations in the actual channel state. This dynamically adjusted benchmark interval more reasonably reflects the normal fluctuation range of channel state parameters under the current prediction reliability level.

[0119] Finally, the unexpected changes are determined by calculating the deviation between the actual channel state parameters and this dynamic reference interval. Only when the actual channel state parameters significantly exceed this normal fluctuation range dynamically determined based on prediction reliability is a large unexpected change considered to exist. This method fully considers the uncertainty of the prediction itself, avoiding misjudgments or omissions caused by prediction errors or normal fluctuations, making the determination of unexpected changes more accurate and robust. Using these determined unexpected changes to subsequently determine whether a transient channel degradation has occurred can significantly improve the reliability of the judgment, thereby more effectively triggering subsequent emergency recovery tasks and computing power adjustment processes.

[0120] In one embodiment, determining the unexpected changes in channel state parameters can be achieved as follows: First, obtain the predicted channel state and the prediction confidence level simultaneously output by the prediction model. For example, the prediction confidence level can be a value between 0 and 1, with a higher value indicating a more reliable prediction. Then, based on the prediction confidence level, generate a dynamic baseline interval for the channel state parameters. Specifically, a basic fluctuation range can be set, and this range can be scaled according to the prediction confidence level. For example, the width of the baseline interval can be proportional to the reciprocal of the prediction confidence level or a function related to the confidence level, or the upper and lower limits of the interval can be determined based on the confidence level by looking up a table. The higher the prediction confidence level, the narrower the interval; the lower the prediction confidence level, the wider the interval.

[0121] Finally, the actual acquired channel state parameters are compared with this dynamic reference interval, and the deviation is calculated. For example, if the actual channel state parameters fall within the interval, the deviation can be set to zero or a very small value; if the actual channel state parameters fall outside the interval, the deviation can be calculated as the distance from the actual value to the nearest interval boundary. The calculated deviation is used as the unexpected change in the channel state parameters.

[0122] Furthermore, step S232 includes:

[0123] S2321: Obtain the section characteristics of the currently operating high-speed rail line;

[0124] S2322: Determine the interval generation parameters used to generate the dynamic benchmark interval based on the segment characteristics;

[0125] S2323: Generate a dynamic reference interval for channel state parameters based on interval generation parameters, predicted channel state, and predicted confidence level.

[0126] Among them, the segment characteristics refer to the environmental attributes of the section of the high-speed train currently in operation. For example, the segment may be an urban area, a tunnel, a bridge, a mountainous area, a plain, or near a specific signal interference source. These characteristics reflect the typical patterns of wireless signal propagation and attenuation in the area, as well as potential interference risks.

[0127] Interval generation parameters are numerical values ​​or functions used to calculate the dynamic reference interval range and / or center offset. The values ​​of these parameters depend on the characteristics of the current segment. For example, in tunnel segments, due to rapid signal attenuation and large fluctuations, the interval generation parameters may be set to larger values ​​to allow for a larger normal fluctuation range; while in open plain segments, where the signal is stable, the parameters may be set to smaller values.

[0128] The dynamic reference interval refers to a numerical range around the predicted channel state value. This interval represents the normal fluctuation range of the channel state parameters expected based on the prediction information in the current segment environment. Actual measured values ​​of the channel state parameters falling within this interval are generally considered to be normal fluctuations, while values ​​exceeding this interval may indicate unexpected changes.

[0129] In one specific embodiment, the segment characteristics of the high-speed rail's current operating line can be obtained by acquiring the train's real-time location information through an onboard positioning system (such as GPS or BeiDou). Then, a pre-stored route map database is queried. This database records the geographical and environmental features of each segment along the route; for example, the database can mark which latitude and longitude ranges correspond to tunnels and which correspond to densely populated urban areas. Based on the segment type of the current location (e.g., the current location is in a "tunnel" segment), interval generation parameters for generating the dynamic reference interval are determined. This can be a lookup table process; for example, for a "tunnel" segment, the interval generation parameters might be set to a specific set of values, such as a large fluctuation range factor and a specific offset. Then, this set of interval generation parameters determined for the "tunnel" segment is combined with the predicted channel state value (e.g., predicted signal-to-noise ratio) and the corresponding predicted confidence value of the current frequency band obtained from a prediction model. Using a preset calculation formula or algorithm, a dynamic reference interval for that frequency band in the current "tunnel" segment is generated. For example, the calculation formula could be to take the predicted signal-to-noise ratio as the center, determine a basic width based on the predicted confidence level, then amplify the basic width based on the fluctuation range factor in the interval generation parameters, and fine-tune the center based on the offset, ultimately obtaining a dynamic reference interval that better matches the channel fluctuation characteristics of the tunnel environment.

[0130] Furthermore, step S3 includes:

[0131] S31: Obtain the types of services carried in the target frequency band, and the service protection level corresponding to the service types;

[0132] S32: Determine the emergency recovery resource requirements for each service type based on the instantaneous channel degradation and service assurance level of the target frequency band;

[0133] S33: Generate the target computing resource type and target computing requirement based on the emergency recovery resource needs.

[0134] Among them, service type refers to different types of data streams or services transmitted within the target frequency band, such as train control services, passenger internet services, video surveillance services, etc.

[0135] Business assurance level refers to the importance or priority of different businesses in the emergency recovery process. It can be divided according to factors such as the criticality of the business, real-time requirements, and tolerance for data loss. For example, it can be divided into high assurance level, medium assurance level, and low assurance level.

[0136] Transient channel degradation refers to a rapid decline in the quality of the wireless channel in a target frequency band within a short period of time, which can manifest as a decrease in signal-to-noise ratio, an increase in bit error rate, and an increase in packet loss rate. Emergency recovery resource requirements refer to the amount of computing resources needed to cope with transient channel degradation in the target frequency band, maintain communication connections, or minimize data loss. This can include the need for different types of computing resources such as general-purpose processors and dedicated accelerators (e.g., channel codec accelerators, encryption / decryption accelerators).

[0137] The target computing resource type refers to the specific type of computing resources required to perform emergency recovery tasks in the target frequency band, which may include general-purpose processors, graphics processors, field-programmable gate arrays, application-specific integrated circuits, etc.

[0138] The target computing requirement refers to the total amount of specific types of computing resources required to perform emergency recovery tasks in the target frequency band. It can be expressed in terms of the number of computing cores, processing power units, memory capacity, etc.

[0139] This application provides a detailed approach to determining the computational power requirements for emergency recovery tasks in a target frequency band. In one embodiment, it is assumed that the target frequency band carries two services: train control services and passenger internet services. In step S31, the system identifies that the target frequency band carries both train control services and passenger internet services, and obtains their corresponding protection levels, for example, level A for train control services and level C for passenger internet services. In step S32, the system detects that the instantaneous channel degradation of the target frequency band is severe. Based on a preset strategy or model, for level A train control services under severe degradation conditions, it is determined that their emergency recovery requires, for example, general-purpose processor cores and encryption / decryption accelerator instances; for level C passenger internet services under severe degradation conditions, it is determined that their emergency recovery requires, for example, general-purpose processor cores and data compression accelerator instances. In step S33, the system summarizes these requirements. The total general-purpose processor requirement is the sum of the train control service requirement and the passenger internet service requirement; the encryption / decryption accelerator requirement is the train control service requirement; and the data compression accelerator requirement is the passenger internet service requirement. Based on these requirements, the system generates target computing resource types as general-purpose processors, encryption / decryption accelerators, and data compression accelerators, with corresponding target computing requirements of aggregated general-purpose processor, encryption / decryption accelerator, and data compression accelerator requirements, respectively. These requirements can be further adjusted according to resource granularity.

[0140] By acquiring the types of services carried within the target frequency band and their protection levels, and combining this with the degree of instantaneous channel degradation, a refined assessment of the emergency recovery resource requirements for different services can be conducted. Based on this refined demand assessment, more accurate target computing resource types and requirements can be generated. This helps to prioritize the computing power needs of critical services based on their importance, especially when onboard heterogeneous computing resources are limited. This improves the communication reliability of high-speed rail multi-band communication under sudden channel degradation events.

[0141] Furthermore, step S32 includes:

[0142] S321: Obtain the service assurance level of services carried in non-target frequency bands;

[0143] S322: Determine the resource guarantee requirements for non-target frequency bands based on the service guarantee level of non-target frequency bands;

[0144] S323: Under the premise of meeting the resource guarantee requirements of non-target frequency bands, determine the emergency recovery resource requirements corresponding to each service type based on the instantaneous channel degradation and service guarantee level of the target frequency band.

[0145] Among them, the service assurance level refers to the degree of service requirements for communication reliability, real-time performance, bandwidth, etc. It can be represented by preset level classification or numerical indicators, such as critical services, important services, and ordinary services.

[0146] The resource guarantee requirements for non-target frequency bands refer to the minimum amount of computing resources or guarantee level required to maintain the normal operation of services carried on non-target frequency bands. These requirements can be determined by looking up tables, calculating, or configuring based on factors such as the service type, service guarantee level, and current service load of the non-target frequency band.

[0147] Under the premise of meeting the resource guarantee needs of non-target frequency bands, it means that when making resource allocation decisions, priority should be given to ensuring that the resources required for non-target frequency bands are reserved or allocated. The remaining resources, or emergency recovery resources for target frequency bands, should be considered only after this premise is met.

[0148] This solution further refines and improves the process of determining the computational resource requirements for emergency recovery tasks in the target frequency band. Given the limited heterogeneous computing resources in vehicles, considering only the resource requirements of the target frequency band experiencing instantaneous channel degradation may encroach on the resources needed for other normally functioning non-target frequency bands, impacting the overall performance of the communication system.

[0149] To address this issue, this solution incorporates consideration of resource assurance requirements for non-target frequency bands when determining the emergency recovery resource requirements for the target frequency band. By obtaining the service assurance level of services carried in non-target frequency bands, the system can understand the importance of services operating on other frequency bands that are currently functioning normally. Next, based on the service assurance level of non-target frequency bands, the system determines the resource assurance requirements for those bands. This means the system calculates the minimum amount of resources or assurance level required to maintain the normal operation of services in other frequency bands, based on their importance. Finally, while meeting the resource assurance requirements of non-target frequency bands, the system determines the emergency recovery resource requirements for each service type based on the instantaneous channel degradation degree of the target frequency band and the service assurance level of the services carried in that band.

[0150] This process ensures that when allocating emergency recovery resources for the damaged target frequency band, the resource guarantee of other normal frequency bands is not excessively sacrificed. Thus, with limited resources, emergency recovery of the target frequency band is carried out while ensuring the normal operation of non-target frequency bands, improving the balance and efficiency of overall resource allocation.

[0151] By combining this solution with the aforementioned scheme for determining the types of resources required for emergency missions in the target frequency band, this solution forms a more comprehensive and balanced resource demand assessment mechanism. This ensures both the rapid recovery of the damaged target frequency band and the maintenance of the communication performance of other normally functioning frequency bands, all within the constraints of limited onboard computing resources.

[0152] For example, suppose a vehicle-mounted communication system includes frequency band A and frequency band B. Frequency band A is the target frequency band experiencing instantaneous channel degradation and carries critical services, while frequency band B is a non-target frequency band carrying ordinary services. First, the system obtains the service assurance level of the ordinary services carried by frequency band B, for example, determining it to be a medium assurance level. Next, based on the medium assurance level, the system determines that frequency band B requires a certain amount of computing resources, such as general-purpose processor computing power for data processing. Subsequently, ensuring this computing power is guaranteed, the system calculates the additional computing resources required for emergency processing of frequency band A, such as enhanced error correction and retransmission, based on the severe degradation of the target frequency band A and the high assurance level of the critical services it carries. This additional computing power could be channel codec accelerator computing power. The final determined emergency recovery resource requirements are the result of comprehensively considering the emergency needs of the target frequency band and the assurance needs of the non-target frequency band.

[0153] Furthermore, step S4 includes:

[0154] S41: Obtain the business assurance level corresponding to the target computing requirements;

[0155] S42: Prioritize the target computing requirements based on the business assurance level;

[0156] S43: Based on the priority sorting results and the computing resource status parameters, adjust the resource allocation of the target computing resource type.

[0157] Among them, business assurance level refers to the indicator that measures the importance or criticality of a specific business. It can be represented by a preset level classification system, such as numerical level, text description or priority queue identifier.

[0158] Priority sorting refers to the process of arranging multiple items to be processed in order according to preset rules. It can be implemented using comparison-based sorting algorithms or group-based priority queue management mechanisms.

[0159] Adjusting resource allocation refers to dynamically modifying the allocation scheme of computing resources based on current needs and resource status. This can be done through operations such as resource reservation, resource scheduling, task migration, or resource release.

[0160] To address the aforementioned issues, this application proposes an improved resource allocation method. Instead of simply allocating resources, this method first obtains the service assurance level associated with each target computing demand. Then, based on the obtained service assurance level, all target computing demands are prioritized. Services with higher assurance levels are assigned higher priority for their corresponding computing demands. This prioritization mechanism provides a clear decision-making basis when multiple computing demands compete for limited resources.

[0161] Finally, the system strictly sorts the results according to the established priority and, in conjunction with the current computing resource status parameters, performs resource allocation for the target computing resource type. This means that the system will first attempt to satisfy the highest priority computing needs, and then process lower priority needs in turn, while ensuring that the allocation decision is based on the actual available resources.

[0162] In this way, the method of this application ensures that, in the event of a surge in computing power demand and resource scarcity due to sudden channel degradation, the most critical and urgent emergency recovery tasks can obtain the necessary computing resources first, thereby improving the reliability and resilience of the entire communication system. This priority allocation strategy based on service assurance levels effectively supplements and optimizes the basic resource allocation scheme, making resource allocation more intelligent and targeted.

[0163] In one specific embodiment, when channel degradation occurs simultaneously in multiple frequency bands, or when degradation in a single frequency band triggers multiple emergency recovery tasks for different services, resulting in multiple target computing demands, the system first obtains the service assurance level corresponding to each target computing demand. For example, an emergency task for train control signal processing might be assigned the highest level 1, an emergency task for security monitoring video transmission might be assigned level 2, and an emergency task for passenger internet access might be assigned level 3. The system can maintain a mapping table between service types and assurance levels, or receive this information from the upper-level service management system. After obtaining these levels, the system prioritizes all target computing demands according to these levels. For example, all level 1 demands are ranked first, followed by level 2 demands, and finally level 3 demands. Within the same level, secondary rules such as first-come-first-served or demand size can be used for sorting. After sorting, the system will check the status parameters of the target computing resource types in the current onboard heterogeneous computing resource pool according to this priority order, such as the available computing power of a certain type of accelerator. The system prioritizes allocating resources to the highest priority computing needs. If available resources can meet the need, allocation is made; otherwise, all available resources are allocated, and the need is marked as partially met or pending. The system then processes the next priority need. This process continues until all needs are processed or resources are fully allocated. In this way, even with limited resources, the most critical business operations are ensured to receive priority.

[0164] Furthermore, step S43 includes:

[0165] S431: Obtain the status parameters of each computing resource instance within the target computing resource type;

[0166] S432: Obtain the computing resource instance requirements for the target computing demand;

[0167] S433: Select a computing resource instance from the target computing resource type based on the status parameters and requirements;

[0168] S434: Allocate resources to the selected computing resource instance.

[0169] Among them, the status parameters of computing resource instances refer to the real-time operating information of each independent computing unit within a specific type of computing resource in the computing resource pool. These parameters can be represented by information such as load rate, available capacity, temperature, error rate, and the busy / idle status of specific functional units.

[0170] The target computing demand for computing resource instances refers to the specific preferences or necessary conditions of a particular computing task for the computing resource instances that execute the task in terms of capabilities, configuration, or status. These can be manifested in the form of required instruction set support, minimum memory capacity, specific accelerator functions, exclusive access permissions, etc.

[0171] Selecting a computing resource instance refers to the process of identifying and determining one or more computing resource instances that best meet the current target computing requirements from multiple computing resource instances that conform to a specific type, based on a preset matching strategy or algorithm.

[0172] Resource allocation refers to the process of associating selected computing resource instances with specific target computing demands and configuring the necessary access permissions and resource isolation so that the demand can be executed on the selected instances.

[0173] This resource allocation method works by first acquiring real-time status parameters of each computing resource instance within the target computing resource type. This allows the system to grasp fine-grained information such as the current load and availability of each specific computing unit, overcoming the limitation of relying solely on the total capacity of the type. Simultaneously, it obtains the specific requirements of the target computing demand on computing resource instances, understanding the different preferences of different tasks for the capabilities of resource instances or the environment.

[0174] Next, based on the acquired resource instance status parameters and the requirements for resource instances, the system selects the most suitable computing resource instance from the target computing resource types. This selection process comprehensively considers the real-time status of the resource and the specific requirements of the requirements for resource capabilities, intelligently matching task requirements with the capabilities and status of resource instances, thereby selecting the most suitable computing resource instance for executing the current task. This status- and requirement-based selection avoids assigning tasks to unsuitable or poorly functioning resources.

[0175] Finally, resource allocation is performed on the selected computing resource instances. After determining the computing resource instances that best match the task requirements, specific resource allocation operations are executed, assigning the selected resource instances to the corresponding target computing demands. Because the previous selection process fully considered the characteristics of resources and demands, this allocation can more effectively ensure the execution of high-priority emergency tasks and optimize the overall use of resources.

[0176] By introducing awareness of the individual state of computing resource instances and understanding of the specific requirements of demand on these instances, and by making refined selections and allocations based on this information, the shortcomings of allocating resources solely based on the overall state of resource types are overcome. This enables more precise and effective allocation of resources to high-priority emergency tasks with specific needs in limited and heterogeneous onboard computing resource environments. This fine-grained resource allocation method, combined with prioritizing target computing demands based on business assurance levels, allows the system to not only allocate the required types of resources to high-priority emergency tasks after identification, but also to find the most suitable and available specific resource instances within that type, thereby further improving the execution efficiency and communication assurance capabilities of emergency tasks.

[0177] In one specific embodiment, assume the target computing resource type is a channel codec accelerator in an onboard heterogeneous computing resource pool. When a transient channel degradation occurs in a certain frequency band, a high-priority emergency recovery task is generated. This task requires enhanced forward error correction processing, thus creating a target computing demand on the channel codec accelerator. The system first obtains the status parameters of all channel codec accelerator instances within this type. For example, accelerator instance A is currently at 90% load, accelerator instance B is currently at 20% load, accelerator instance C supports enhanced LDPC codes and is at 30% load, and accelerator instance D does not support enhanced LDPC codes and is at 10% load. Simultaneously, the system obtains the requirements of the emergency recovery task for the computing resource instances. For example, the task requires support for enhanced LDPC code processing and requires the allocated instances to be currently at less than 50% load.

[0178] Next, the system makes a selection based on these status parameters and requirements. Instance A has too high a load and does not meet the requirements. Instance B has a load that meets the requirements, but it is uncertain whether it supports enhanced LDPC codes. Instance C supports enhanced LDPC codes and its load meets the requirements. Instance D does not support enhanced LDPC codes and does not meet the requirements. Therefore, the system selects instance C.

[0179] Finally, the system assigns this high-priority emergency recovery task to channel codec accelerator instance C for processing. This ensures that emergency tasks requiring specific functions and high real-time performance are assigned to the computing resource instance with the appropriate capabilities and currently optimal state.

[0180] Please refer to Figure 2 , Figure 3 A dynamic computing power optimization system for multi-band communication in high-speed rail, used to implement the method of the above embodiments, the system includes:

[0181] Acquisition module 201: Acquires channel status parameters of each frequency band in the multi-band communication system, and computing resource status parameters of each type of computing resource in the vehicle-mounted heterogeneous computing resource pool;

[0182] Determination module 202: Based on channel state parameters, determine the target frequency band in each frequency band where instantaneous channel degradation occurs;

[0183] Response module 203: In response to the target frequency band, determines the type of target computing resources required to perform the emergency recovery task of the target frequency band, and the corresponding target computing requirements;

[0184] Allocation module 204: Based on computing resource status parameters, adjusts the resource allocation of the target computing resource type to meet the target computing demand.

[0185] Among them, the acquisition module 201 refers to the unit responsible for collecting system operating status data, which can be implemented by means of data interface, sensor interface or software agent.

[0186] The determination module 202 refers to the unit responsible for analyzing the collected data and identifying specific events or states, which can be implemented by means of a data processor, analysis algorithm unit or logic judgment circuit.

[0187] The response module 203 is a unit responsible for assessing the required resources or taking action based on the identified events or states. It can be implemented by means of an assessment calculation unit, a strategy generation unit, or demand forecasting logic.

[0188] The allocation module 204 is the unit responsible for adjusting the allocation of system resources based on the evaluation results. It can be implemented using a resource scheduler, configuration manager, or resource control logic.

[0189] The system, through the collaborative work of its constituent modules, implements the specific functions of the aforementioned method. The acquisition module 201 is responsible for continuously collecting channel status parameters of each frequency band within the multi-band communication system and computing resource status parameters of various types of computing resources within the vehicle-mounted heterogeneous computing resource pool, providing basic data for subsequent processing.

[0190] The determination module 202 receives the channel state parameters provided by the acquisition module, and analyzes and judges based on these parameters to identify the target frequency band where instantaneous channel degradation occurs.

[0191] The response module 203 receives the target frequency band information identified by the determination module and assesses the type of computing resources and specific computing requirements required to perform the target frequency band emergency recovery task based on the information.

[0192] The allocation module 204 receives the computing resource status parameters provided by the acquisition module and the resource requirement information determined by the response module, and dynamically adjusts the resource allocation of the target computing resource type based on this information to meet the computing power requirements of the emergency recovery task.

[0193] The entire process forms a closed loop of data acquisition, problem identification, requirement assessment, and resource scheduling. The various modules of the system interact through data flow and control signals, transforming method steps into executable functional units. This ensures the rapid and accurate execution of the method in a high-speed, dynamic environment and effective collaboration between modules. As the concrete implementation carrier of the method, the system enables the method to run efficiently and reliably on the hardware platform, overcoming implementation efficiency and collaboration issues that cannot be solved by simply relying on abstract method descriptions.

[0194] In this document, relational terms such as first and second are used only to distinguish one entity or operation from another entity or operation, without necessarily requiring or implying any such actual relationship or order between these entities or operations.

[0195] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A dynamic computing power optimization method for multi-band communication in high-speed rail, applied to high-speed rail, wherein the high-speed rail is equipped with a multi-band communication system and an onboard heterogeneous computing resource pool, characterized in that, The method includes the following steps: S1: Obtaining the channel status parameters of each frequency band in the multi-band communication system, and the computing resource status parameters of each type of computing resource in the vehicle-mounted heterogeneous computing resource pool; S2: Based on the channel state parameters, determine the target frequency band in each frequency band where instantaneous channel degradation occurs; Step S2 includes: S21: Obtain the operating status of the high-speed rail; S22: Based on the operating state, obtain the predicted channel state corresponding to the operating state; S23: Based on the predicted channel state and the channel state parameters, determine the unexpected change in the channel state parameters; S24: When the unexpected change meets the preset conditions, the frequency band is determined as the target frequency band; Step S23 includes: S231: Obtain the prediction confidence level corresponding to the predicted channel state; S232: Generate a dynamic reference range for the channel state parameters based on the predicted confidence level; S233: The deviation of the channel state parameter from the dynamic reference interval is determined as the unexpected change. S3: In response to the target frequency band, determine the type of target computing resources required to perform the emergency recovery task of the target frequency band, and the corresponding target computing requirements; S4: Based on the computing resource status parameters, adjust the resource allocation of the target computing resource type to meet the target computing demand.

2. The method for dynamic computing power optimization of high-speed rail multi-band communication according to claim 1, characterized in that, Step S1 includes: S11: Generate a unified data collection trigger; S12: Based on the unified acquisition trigger, obtain the channel state parameters and the computing resource state parameters; S13: Pair the channel state parameters and computing resource state parameters corresponding to the same unified acquisition trigger to form a synchronization state snapshot, wherein the channel state parameters in the synchronization state snapshot are used to determine the target frequency band.

3. The method for dynamic computing power optimization of high-speed rail multi-band communication according to claim 1, characterized in that, Step S232 includes: S2321: Obtain the section characteristics of the currently operating high-speed railway line; S2322: Determine the interval generation parameters for generating the dynamic reference interval based on the segment characteristics; S2323: Generate a dynamic reference interval for the channel state parameters based on the interval generation parameters, the predicted channel state, and the predicted confidence level.

4. The method for dynamic computing power optimization of high-speed rail multi-band communication according to claim 1, characterized in that, Step S3 includes: S31: Obtain the service type carried in the target frequency band, and the service protection level corresponding to the service type; S32: Determine the emergency recovery resource requirements corresponding to each of the service types based on the instantaneous channel degradation of the target frequency band and the service protection level; S33: Based on the emergency recovery resource requirements, generate the target computing resource type and the target computing requirement.

5. The method for dynamic computing power optimization of high-speed rail multi-band communication according to claim 4, characterized in that, Step S32 includes: S321: Obtain the service assurance level of services carried in non-target frequency bands; S322: Based on the service assurance level of the non-target frequency band, determine the resource assurance requirements of the non-target frequency band; S323: Under the premise of meeting the resource guarantee requirements of the non-target frequency band, determine the emergency recovery resource requirements corresponding to each of the service types based on the instantaneous channel degradation of the target frequency band and the service guarantee level.

6. The method for dynamic computing power optimization of high-speed rail multi-band communication according to claim 1, characterized in that, Step S4 includes: S41: Obtain the service assurance level corresponding to the target computing demand; S42: Prioritize the target computing requirements according to the business assurance level; S43: Based on the priority sorting results and the computing resource status parameters, adjust the resource allocation of the target computing resource type.

7. The method for dynamic computing power optimization of high-speed rail multi-band communication according to claim 6, characterized in that, Step S43 includes: S431: Obtain the status parameters of each computing resource instance within the target computing resource type; S432: Obtain the target computing demand for the computing resource instance; S433: Select a computing resource instance from the target computing resource type based on the status parameters and the requirements; S434: Allocate resources to the selected computing resource instance.

8. A dynamic computing power optimization system for multi-band communication in high-speed rail, characterized in that, The system for implementing the method according to any one of claims 1-7 comprises: Acquisition module: Acquires channel status parameters of each frequency band in the multi-band communication system, and computing resource status parameters of each type of computing resource in the vehicle-mounted heterogeneous computing resource pool; Determination module: Based on the channel state parameters, determine the target frequency band in each frequency band where instantaneous channel degradation occurs; Response module: In response to the target frequency band, determines the type of target computing resources required to perform the emergency recovery task of the target frequency band, and the corresponding target computing requirements; Allocation module: Based on the computing resource status parameters, adjust the resource allocation of the target computing resource type to meet the target computing demand.

Images