Low-altitude aircraft communication decision-making method and device and readable storage medium

Abstract

The invention discloses a low-altitude aircraft communication decision-making method and device and a readable storage medium, and belongs to the technical field of network communication, and the method comprises the steps: receiving a data packet, and obtaining an identification serial number of the data packet; inquiring the identification serial number by using a counting type Bloom filter (CBF) and judging whether the identification serial number is a suspected redundant data packet or not; querying the suspected redundant data packet again by using an overtime sliding window; and calculating a redundancy success rate, a redundancy elimination timeout rate and an invalid query rate according to the timeout sliding window background record, and processing the redundancy success rate, the redundancy elimination timeout rate and the invalid query rate by adopting a multi-dimensional network state decision engine to obtain a diagnosis result of the current network state and output an optimal corresponding decision. And conventionally optimizing, shrinking or tolerating a periodic timeout sliding window under dynamic control. Through the combination of the CBF, the timeout sliding window and the multi-dimensional network state decision engine, redundancy elimination, network diagnosis and intelligent decision are realized, and the phenomena of delay, redundancy and packet loss of subsequent data packet transmission are reduced.

Description

Technical Field

[0001] The present invention relates to a low-altitude aircraft communication decision-making method, device and readable storage medium, and belongs to the technical field of network communication. Background Art

[0002] During actual flight, low-altitude aircraft often face challenges in various communication links. Specifically, (1) Communication problems of low-altitude aircraft’s radio links (such as VHF / UHF): Due to factors such as terrain and distance, the signal stability is poor; (2) Communication problems between low-altitude aircraft and ground networks (such as LTE / 5G): The signal is good in the near-ground area, but the signal weakens and the bandwidth decreases as the flight altitude increases or the coverage area becomes uneven; (3) Communication problems of low-altitude aircraft in satellite communication links (such as LEO): They provide wide coverage, but usually have high latency and limited bandwidth; therefore, the communication quality of existing low-altitude aircraft is difficult to guarantee. In these environments, if low-altitude aircraft use a single link for communication, it is often difficult to meet the requirements of low latency and high reliability. Especially when the network status changes rapidly and the link quality is unstable, frequent link switching will lead to data loss, disorder, increased latency and communication interruption, and it will not be able to adapt to the development trend of future urban low-altitude transportation.

[0003] To mitigate risks like signal interference and link interruptions, existing technologies typically employ multi-path redundant transmission strategies. For example, a single aircraft might simultaneously send the same data packet to a ground station or server via two heterogeneous links: satellite and 5G. This design significantly improves communication reliability through link redundancy, ensuring that at least one copy of critical data is successfully delivered. However, this architecture inevitably generates a large number of redundant data packets at the receiving end. Eliminating this redundancy efficiently, intelligently, and transparently to upper-layer services is key to the success of this technology approach. Summary of the Invention

[0004] The purpose of the present invention is to provide a low-altitude aircraft communication decision-making method, device and readable storage medium, aiming to solve the difficult problems of multi-path redundant transmission and insufficient adaptability to dynamic heterogeneous networks of low-altitude aircraft. By combining a counting Bloom filter and a timeout sliding window into a two-stage filtering architecture with fast screening and precision calibration functions, redundant data packets and new data packets are effectively identified, and computing and memory resources are avoided from being wasted on subsequent processing of invalid data, thereby significantly improving the effective throughput of the system. Combined with a multi-dimensional decision engine, the network status parameters within this cycle are calculated and the network is diagnosed to obtain the optimal decision for the next cycle, so as to improve the network adaptability for redundancy elimination in the next cycle, and realize that under various network conditions, the network status can be intelligently perceived and the redundancy elimination task can be completed efficiently with the optimal strategy.

[0005] In order to solve the above technical problems, the present invention is implemented by adopting the following technical solutions.

[0006] In a first aspect, the present invention provides a low-altitude aircraft communication decision-making method, comprising: The low-altitude aircraft receives the data packet; Extract the identification of the received data packets to obtain the identification number of each data packet; Use the counting Bloom filter CBF to query the identification number, obtain the identification query result and determine whether the received data packet is a suspected redundant data packet; When the received data packet is suspected to be a redundant data packet, the suspected redundant data packet is queried again using the timeout sliding window to obtain a query result; Based on the query results, data packets are discarded and the sliding window background records are modified; Calculate the background records of the timeout sliding window to obtain the redundancy success rate, redundancy elimination timeout rate, and invalid query rate within this cycle; A multi-dimensional network status decision engine is used to process the redundancy success rate, redundancy elimination timeout rate, and invalid query rate to obtain the diagnosis results of the current network status and output the optimal corresponding decision; Based on the optimal corresponding decision, the parameters of the next period timeout sliding window are adjusted, and the regular optimization, contraction or tolerance of the next period timeout sliding window is dynamically controlled.

[0007] Optionally, the low-altitude aircraft receiving the data packet further includes: Establish communication with multiple frequency band models through the QUIC protocol and build multiple QUIC communication links; Synchronously receive data packets based on multiple QUIC communication links; The data interface is used to embed the QUIC identification number into each received data packet for subsequent counting Bloom filter CBF query and sliding window verification.

[0008] Optionally, using a counting Bloom filter (CBF) to query the identification number, obtain an identification query result, and determine whether the received data packet is a suspected redundant data packet, including: If the identification number is found in the counting Bloom filter CBF, it is preliminarily determined that the data packet is suspected redundant data; If the identification number is not found in the counting Bloom filter CBF, it is preliminarily determined that the data packet is a new data packet.

[0009] Optionally, if the identification number is not found in the counting Bloom filter CBF, the data packet is preliminarily determined to be a new data packet, further comprising: When the data packet is a new data packet, the new data packet is used as valid data for delivery to the upper-layer business logic for processing; Creating a background record of the timeout sliding window based on the new data packet, wherein the background record includes the QUIC identification sequence number and arrival time of the new data; The QUIC identification number of the new data packet is hashed using the hash function of the counting Bloom filter CBF, a hash position corresponding to the new data packet is generated in the counting Bloom filter CBF, and the count value at the hash position is incremented by one.

[0010] Optionally, when the received data packet is a suspected redundant data packet, the suspected redundant data packet is queried again using the timeout sliding window to obtain a query result, including: Use the timeout sliding window to query suspected redundant data packets. If the same QUIC identification number record exists in the background of the timeout sliding window, the suspected redundant data packet is determined to be a redundant data packet. If the same QUIC identification sequence number record does not exist in the background of the timeout sliding window, the suspected redundant data packet is determined to be a new data packet.

[0011] Optionally, based on the query results, discard packets and modify the sliding window background records, including: After the suspected redundant data packet is determined to be a redundant data packet, the arrival time of the redundant data packet is extracted and the current data packet is directly discarded; The redundant delay difference of the redundant data packet is calculated based on the difference between the arrival time corresponding to the QUIC identification number recorded in the background of the timeout sliding window and the arrival time of the redundant data packet; The redundant delay difference is used as a sample to calculate the moving average of the timeout sliding window, and the updated moving average of the timeout sliding window is obtained; Calculate the timeout sliding window size based on the updated moving average value of the timeout sliding window to obtain the adjusted size of the timeout sliding window; After the size of the timeout sliding window is updated, the QUIC identification number record corresponding to the redundant data packet is deleted in the background of the timeout sliding window, and the count value at the hash position corresponding to the QUIC identification number in the counting Bloom filter CBF is simultaneously reduced by one; After the suspected redundant data packet is determined to be a new data packet, the process of querying the new data packet is counted as an invalid query event, and the new data packet is regarded as valid data.

[0012] Optionally, a multi-dimensional network status decision engine is used to process the redundancy success rate, redundancy elimination timeout rate, and invalid query rate to obtain a diagnosis result of the current network status and output the optimal corresponding decision, including: The multi-dimensional network state decision matrix is ​​used to extract the features of redundancy success rate, redundancy elimination timeout rate and invalid query rate respectively, and the features of redundancy success rate, redundancy elimination timeout rate and invalid query rate are obtained; The features of redundancy success rate, redundancy elimination timeout rate, and invalid query rate are integrated to obtain the input feature vector of the current network status; Perform feature fuzzy processing on the input feature vector of the current network state, so that each feature establishes a mapping relationship with multiple predefined fuzzy sets; Calculate multiple fuzzy sets according to predefined membership functions to obtain the membership degrees of multiple fuzzy sets; Use the pre-built fuzzy decision rule library to perform fuzzy reasoning on the real-time input feature vector to obtain the trigger strength of each rule, and aggregate the trigger strengths of all rules to obtain the final fuzzy decision result; The center of gravity method is used to calculate the final fuzzy decision result and obtain the quantitative diagnosis result of the current network status; According to the diagnosis results of the current network status, the optimal corresponding decision for adjusting the timeout sliding window of the next cycle is output.

[0013] Optionally, the timeout sliding window background record further includes regular maintenance of the timeout sliding window background record through a periodic cleaning process; When cleaning background records, first traverse all records in the timeout sliding window background and extract the survival time of each record; When the survival time of any record exceeds the tolerance time of the current timeout sliding window, the record is determined to be a timeout record, and the redundant data packet corresponding to the record has timed out; The determined timeout records are cleaned up regularly to obtain a timeout sliding window background in which the timeout records are regularly removed.

[0014] In a second aspect, the present invention provides a low-altitude aircraft communication decision-making device, comprising: Data receiving module, adaptive redundancy elimination module, network status parameter calculation module, redundancy elimination decision module; The data receiving module is used to receive data packets and extract identifications from the received data packets to obtain identification serial numbers of each data packet; The adaptive redundancy elimination module is used to use a counting Bloom filter (CBF) to query the identification number, obtain the identification query result, and determine whether the received data packet is a suspected redundant data packet; when the received data packet is a suspected redundant data packet, use the timeout sliding window to query the suspected redundant data packet again to obtain the query result; based on the query result, discard the data packet and modify the sliding window background record; The network status parameter calculation module is used to calculate the timeout sliding window background record to obtain the redundancy success rate, redundancy elimination timeout rate, and invalid query rate within this cycle; The redundancy elimination decision module is used to use a multi-dimensional network status decision engine to process the redundancy success rate, redundancy elimination timeout rate, and invalid query rate, obtain the diagnosis result of the current network status and output the optimal corresponding decision; based on the optimal corresponding decision, adjust the parameters of the next cycle timeout sliding window, and dynamically control the regular optimization, contraction or tolerance of the next cycle timeout sliding window.

[0015] In a third aspect, the present invention provides a computer-readable storage medium having a computer program / instruction stored thereon, which, when executed by a processor, implements the steps of the method described in the first aspect.

[0016] Compared with the prior art, the present invention has the following beneficial effects: (1) The present invention combines a counting Bloom filter and a timeout sliding window into a two-stage filtering architecture with fast screening and fine-tuning functions, effectively identifying redundant data packets and new data packets, avoiding wasting computing and memory resources on subsequent processing of invalid data, and significantly improving the effective throughput of the system; and then combines a multi-dimensional decision engine to calculate the network status parameters and network diagnosis within the current cycle, and obtain the optimal decision for the next cycle, so as to improve the network's adaptive ability to eliminate redundancy in the next cycle, and realize intelligent perception of network status under various network conditions and efficient completion of redundancy elimination tasks with the optimal strategy.

[0017] (2) The multi-dimensional decision engine of the present invention perceives and accurately calculates the network status changes in redundancy, delay, packet loss and other dimensions in real time, and dynamically adjusts the core algorithm parameters to adapt to the subsequent communication and achieve the optimal redundancy reduction strategy in various complex network environments.

[0018] (3) The present invention regularly cleans up the background records of the timeout sliding window through a background cleaning process, so that the timeout sliding window can adaptively adjust the window size and timeout logic under the dynamic changes of the dynamic heterogeneous network. By regularly cleaning up the timeout redundant packets, the recognition accuracy of false positive data packets can be improved, and historical data can be prevented from interfering with current decisions, while reducing memory usage.

[0019] (4) The present invention ensures the real-time and deterministic nature of the service. It diagnoses the network status through a multi-dimensional decision engine and then controls the timeout sliding window tolerance strategy and boundaries. It can not only maximize the reliability of data transmission, but also avoid infinite waiting caused by link problems through the timeout mechanism, thus ensuring the timing stability of the upper-layer service.

[0020] (5) After the timeout sliding window background record is changed, all hash positions corresponding to the counting Bloom filter are modified synchronously to achieve the synchronous increase and decrease of the status of the counting Bloom filter CBF and the sliding window, ensuring long-term operation performance.

[0021] (6) By recording and storing redundant success events, invalid query events, and redundant timeout events through a timeout sliding window, the redundant success rate, redundant timeout rate, and invalid query rate in any period are calculated, and the accuracy of network diagnosis can be quantified. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 Shown is a flow chart of the low-altitude aircraft communication decision-making method of the present invention; Figure 2 Shown is a flow chart of the two-stage filtering architecture for low-altitude aircraft of the present invention; Figure 3 Shown is a diagram showing an embodiment of the multi-dimensional decision engine network diagnosis of the present invention. DETAILED DESCRIPTION

[0023] The present invention will be further described below in conjunction with the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solutions of the present invention and are not intended to limit the scope of protection of the present invention. Example

[0024] This embodiment provides a low-altitude aircraft communication decision-making method, such as Figure 1 Shown include: The low-altitude aircraft receives the data packet; Extract the identification of the received data packets to obtain the identification number of each data packet; Use the counting Bloom filter CBF to query the identification number, obtain the identification query result and determine whether the received data packet is a suspected redundant data packet; When the received data packet is suspected to be a redundant data packet, the suspected redundant data packet is queried again using the timeout sliding window to obtain a query result; Based on the query results, data packets are discarded and the sliding window background records are modified; Calculate the background records of the timeout sliding window to obtain the redundancy success rate, redundancy elimination timeout rate, and invalid query rate within this cycle; A multi-dimensional network status decision engine is used to process the redundancy success rate, redundancy elimination timeout rate, and invalid query rate to obtain the diagnosis results of the current network status and output the optimal corresponding decision; Based on the optimal corresponding decision, the parameters of the next cycle timeout sliding window are adjusted, and the regular optimization, contraction or tolerance of the next cycle timeout sliding window is dynamically controlled to adaptively reduce the delay, redundancy and packet loss of subsequent data packet transmission according to the dynamic changes of different networks.

[0025] This embodiment utilizes an adaptive redundancy reduction engine based on intelligent perception and state synchronization. This engine is created for each redundant QUIC connection and comprises three core components: a counting Bloom filter, a sliding timeout window, and a multi-dimensional decision engine. The filter acts as a "state synchronization collaborator" and "network status sensor," providing the multi-dimensional decision engine with real-time data on redundancy, timeouts, and false positives. The decision engine uses this data to diagnose network conditions and dynamically adjust the core parameter of the sliding timeout window (W_timeout), ensuring optimal redundancy reduction strategies under various network conditions.

[0026] Optionally, the low-altitude aircraft receiving the data packet further includes: Establish communication with multiple frequency band models through the QUIC protocol and build multiple QUIC communication links; Synchronously receive data packets based on multiple QUIC communication links; The data interface is used to embed the QUIC identification number into each received data packet for subsequent counting Bloom filter CBF query and sliding window verification.

[0027] In this embodiment, after each QUIC connection is established, an independent two-stage filtering architecture is created and a unified data format and interface are defined to aggregate data from different links (LEO, 5G, and ad hoc networks) into the two-stage filtering architecture for rapid screening and precision calibration, ensuring that the data format received by multiple links is consistent.

[0028] Optionally, using a counting Bloom filter (CBF) to query the identification number, obtain an identification query result, and determine whether the received data packet is a suspected redundant data packet, including: If the identification number is found in the counting Bloom filter CBF, it is preliminarily determined that the data packet is suspected redundant data; If the identification number is not found in the counting Bloom filter CBF, it is preliminarily determined that the data packet is a new data packet.

[0029] This embodiment is a fast screening process that uses a counting Bloom filter and gives it a dual core role that goes beyond traditional filters. The CBF serves as a low-cost, high-performance "snapshot" of the "timeout sliding window." It maintains strict synchronization with the state of the sliding window.

[0030] Optionally, if the identification number is not found in the counting Bloom filter CBF, the data packet is preliminarily determined to be a new data packet, further comprising: When the data packet is a new data packet, the new data packet is used as valid data for delivery to the upper-layer business logic for processing; Create a background record of the timeout sliding window based on the new data packet, which contains the QUIC identification number and arrival time of the new data (key identifier S, current arrival time T_arrival); The QUIC identification number of the new data packet is hashed using the hash function of the counting Bloom filter CBF, a hash position corresponding to the new data packet is generated in the counting Bloom filter CBF, and the count value at the hash position is incremented by one.

[0031] Therefore, when a packet's identifier is recorded in the sliding window, the count at the corresponding location in the CBF is incremented by one. When the record is removed from the sliding window for any reason (successful redundancy elimination or timeout), the count at the corresponding location in the CBF is decremented accordingly. This synchronization mechanism ensures that CBF performance does not degrade due to the accumulation of historical data, maintaining an extremely low "false positive rate" over the long term.

[0032] CBF's "false positives" (invalid queries) are also used as a key diagnostic signal. The frequency of these false positives (the "invalid query rate") can very sensitively indicate whether there is unidirectional packet loss or severe link asymmetry in the network, thus providing a non-latency-based judgment dimension for the "multi-dimensional decision engine."

[0033] Optionally, when the received data packet is a suspected redundant data packet, the suspected redundant data packet is queried again using the timeout sliding window to obtain a query result, including: Use the timeout sliding window to query suspected redundant data packets. If the same QUIC identification number record exists in the background of the timeout sliding window, the suspected redundant data packet is determined to be a redundant data packet. If the same QUIC identification sequence number record does not exist in the background of the timeout sliding window, the suspected redundant data packet is determined to be a new data packet.

[0034] like Figure 2As shown, this embodiment uses a timeout sliding window to perform final and accurate verification of initially identified suspicious packets. Unlike traditional windows of fixed size or duration, this sliding window has a dynamic lifecycle. Its core parameter is the redundancy removal timeout window (W_timeout). The window size can be preset based on actual network latency and out-of-order conditions and can be dynamically adjusted. The sliding window size is set to 100k (100,000), the initial window validity period (e.g., T seconds) is set, and the window sliding and clearing strategies are determined. The timeout sliding window background records the unique identifiers (e.g., sequence numbers) and first arrival timestamps of recently arrived packets that have not yet completed redundancy removal. The effective lifetime of each record in the window is determined by W_timeout, which is dynamically calculated by a "multi-dimensional decision engine" based on real-time network status. This allows the window to intelligently expand or contract based on changes in network latency, achieving an optimal balance between ensuring redundancy removal success rate and conserving system resources.

[0035] If a suspected redundant packet does not find a record for S within the timeout sliding window, it is confirmed as a new packet (CBF false positive). The event is also counted, and the number of "invalid query" events is incremented by one. Finally, the new packet is forwarded for processing and the new packet processing flow is executed.

[0036] Optionally, based on the query results, discard packets and modify the sliding window background records, such as Figure 2 Shown include: After the suspected redundant data packet is determined to be a redundant data packet, the arrival time of the redundant data packet is extracted and the current data packet is directly discarded; The redundant delay difference of the redundant data packet is calculated based on the difference between the arrival time corresponding to the QUIC identification number recorded in the background of the timeout sliding window and the arrival time of the redundant data packet; The redundant delay difference is used as a sample to calculate the moving average of the timeout sliding window, and the updated moving average of the timeout sliding window is obtained; Calculate the timeout sliding window size based on the updated moving average value of the timeout sliding window to obtain the adjusted size of the timeout sliding window; After the size of the timeout sliding window is updated, the QUIC identification number record corresponding to the redundant data packet is deleted in the background of the timeout sliding window, and the count value at the hash position corresponding to the QUIC identification number in the counting Bloom filter CBF is simultaneously reduced by one; After the suspected redundant data packet is determined to be a new data packet, the process of querying the new data packet is counted as an invalid query event, and the new data packet is regarded as valid data.

[0037] The precise verification process of this embodiment is: using the key identifier S, performing a precise search in the "timeout sliding window".

[0038] If S's record is found in the window, it means that its "twin packet" has arrived before it, and it is confirmed as a redundant packet. The redundant data packet processing process is executed to discard the current redundant data packet.

[0039] Redundant packets can also be used to diagnose network latency. Based on the first arrival time (T_first_arrival) recorded in the redundant packet extraction window, the current redundant delay difference (Delta_T) is calculated as T_current - T_first_arrival. Delta_T is used as a sample to update the moving average (Avg_Delta_T) used to calculate W_timeout, thereby more accurately controlling the communication timeout strategy within the timeout window.

[0040] Remove records after obtaining Delta_T: Completely remove records about S from the sliding window because its redundant task has been completed.

[0041] Synchronize CBF state: In CBF, find all hash positions corresponding to S and decrement the count values ​​of these positions by one. Finally, count the events and add one to the number of "redundancy success" events. Synchronize the two-level filtering architecture through the above to ensure cross-node data synchronization.

[0042] Optionally, a multi-dimensional network status decision engine is used to process the redundancy success rate, redundancy elimination timeout rate, and invalid query rate to obtain an adjustment weight value for the current network status as the optimal communication decision result, including: The multi-dimensional network state decision matrix is ​​used to extract the features of redundancy success rate, redundancy elimination timeout rate and invalid query rate respectively, and the features of redundancy success rate, redundancy elimination timeout rate and invalid query rate are obtained; The features of redundancy success rate, redundancy elimination timeout rate, and invalid query rate are integrated to obtain the input feature vector of the current network status; Perform feature fuzzy processing on the input feature vector of the current network state, so that each feature establishes a mapping relationship with multiple predefined fuzzy sets; Calculate multiple fuzzy sets according to predefined membership functions to obtain the membership degrees of multiple fuzzy sets; Use the pre-built fuzzy decision rule library to perform fuzzy reasoning on the real-time input feature vector to obtain the trigger strength of each rule, and aggregate the trigger strengths of all rules to obtain the final fuzzy decision result; The center of gravity method is used to calculate the final fuzzy decision result and obtain the quantitative diagnosis result of the current network status; According to the diagnosis results of the current network status, the optimal corresponding decision for adjusting the timeout sliding window of the next cycle is output.

[0043] The multi-dimensional decision engine network diagnosis-signal detection-macro decision analysis process of this embodiment is as follows: Figure 3 As shown, this embodiment formulates different communication decisions according to different network status combinations, and can take the best response decision according to different network status combinations or dynamic changes in network status in actual applications.

[0044] In order to Figure 3 To translate macro-level response decisions into precise, executable control instructions, the multi-dimensional decision engine uses a fuzzy logic controller. This fuzzy logic controller excels at handling uncertainties and ambiguities such as network status, as follows: Step 1: Feature extraction and input vector construction. The engine first processes the raw event data periodically collected by background tasks and extracts three key performance metrics as features.

[0045] Features include the redundancy success rate (RSR), redundancy timeout rate (TR), and invalid query rate (WCR). The RSR is calculated as follows: the number of redundancy successes per unit time divided by the total number of packets arriving per unit time. The TR is calculated as follows: the number of redundancy timeouts per unit time divided by the number of new packets entering the sliding window per unit time. The WCR is calculated as follows: the number of invalid queries per unit time divided by the total number of CBF reports of "possible presence" per unit time. These three features constitute an input feature vector V_in = [RSR, TR, WCR] that reflects the current network status in real time.

[0046] The second step, fuzzification, converts precise numerical features (e.g., TR = 35%) into fuzzy language descriptions (e.g., "timeout rate is high"). The engine processes each feature in the input feature vector using a predefined membership function, mapping it to multiple fuzzy sets.

[0047] For the feature "Redundancy Timeout Rate (TR)," three fuzzy sets can be defined: {low, medium, high}. A specific TR value, such as 35%, might have membership in these three sets calculated using the membership function as {low: 0.0, medium: 0.7, high: 0.3}. This means that the current system's timeout rate is 70% medium and 30% high.

[0048] The third step is fuzzy reasoning. This step is the core of the engine. It uses a fuzzy rule base to perform logical reasoning on the fuzzified input. This rule base is a precise and formal expression of the decision matrix we previously described. The following are examples of rules: Rule 1: IF (RSR is high) AND (TR is low) AND (WCR is low) THEN (decision output is conventional optimization) Rule 2: IF (RSR is low) AND (TR is high) AND (WCR is low) THEN (decision output is tolerance increase) Rule 3: IF (RSR is low) AND (TR is high) AND (WCR is high) THEN (decision output is decisive contraction) The fuzzy reasoner calculates the triggering strength of each rule according to the membership of the input features to each fuzzy set, and obtains the final fuzzy decision result.

[0049] The fourth step is defuzzification and decision output. This step converts the fuzzy decision result into an accurate and executable value. In the present invention, this output value is a decision adjustment weight. This embodiment uses the centroid method to calculate the weight value, which is limited to other methods. The adjusted weight value directly reflects the optimal adjustment tendency corresponding to the current network state: When the network status is healthy (corresponding to Rule 1), the calculated adjustment weight value is ≈ 1.0.

[0050] When the slowpath delay increases dramatically (corresponding to Rule 2), the calculated adjustment weight value is greater than 1.0 (for example, let the adjustment weight value be γ, where the value of γ fluctuates between 1.0 and 2.0. The higher the TR, the larger the γ value).

[0051] When the link is asymmetric or packet loss occurs (corresponding to rule 3), the calculated adjustment weight value is less than 1.0 (for example, the value of γ fluctuates between 0.3 and 1.0. The higher the WCR, the smaller the γ value).

[0052] Finally, this adjusted weight value γ is used to calculate the W_timeout value for the next cycle: W_timeout_new = (Avg_Delta_T * k) * γ Where W_timeout_new is the size of the timeout sliding window for the next cycle, Avg_Delta_T is the moving average of the timeout sliding window, and k is the adjustment parameter of the timeout sliding window.

[0053] The present invention calculates the redundancy success rate, redundancy elimination timeout rate, and invalid query rate based on the total number of "redundancy success," "redundancy elimination timeout," and "invalid query" counted in the most recent cycle. These three ratios are input into a "multi-dimensional decision engine." Based on the built-in decision matrix, the engine outputs adjustment decisions for W_timeout, including conventional optimization, increase, or decrease. The value of W_timeout is updated for use in the next cycle's timeout cleanup and data processing processes. By introducing a fuzzy logic controller, the present invention goes beyond a simple "if-then" judgment and establishes a complete, smoothly transitioned, and quantified communication dynamic decision-making mechanism.

[0054] Optionally, the timeout sliding window background record further includes regular maintenance of the timeout sliding window background record through a periodic cleaning process; When cleaning background records, first traverse all records in the timeout sliding window background and extract the survival time of each record; When the survival time of any record exceeds the tolerance time of the current timeout sliding window, the record is determined to be a timeout record, and the redundant data packet corresponding to the record has timed out; The determined timeout records are cleaned up regularly to obtain a timeout sliding window background in which the timeout records are regularly removed.

[0055] In practical application, this embodiment traverses all records in the timeout sliding window backend and checks each record. If its survival time (current time minus T_arrival) exceeds the current dynamic W_timeout value, the redundant packet corresponding to the record is determined to have timed out. For each timed-out record S, it is removed from the sliding window, its count value in the CBF is decremented by one, and the number of "redundancy timeout" events is incremented by one. Example

[0056] This embodiment provides a low-altitude aircraft communication decision-making device, including: Data receiving module, adaptive redundancy elimination module, network status parameter calculation module, redundancy elimination decision module; The data receiving module is used to receive data packets and extract identifications from the received data packets to obtain identification serial numbers of each data packet; The adaptive redundancy elimination module is used to use a counting Bloom filter (CBF) to query the identification number, obtain the identification query result, and determine whether the received data packet is a suspected redundant data packet; when the received data packet is a suspected redundant data packet, use the timeout sliding window to query the suspected redundant data packet again to obtain the query result; based on the query result, discard the data packet and modify the sliding window background record; The network status parameter calculation module is used to calculate the timeout sliding window background record to obtain the redundancy success rate, redundancy elimination timeout rate, and invalid query rate within this cycle; The redundancy elimination decision module is used to use a multi-dimensional network status decision engine to process the redundancy success rate, redundancy elimination timeout rate, and invalid query rate, obtain the diagnosis result of the current network status and output the optimal corresponding decision; based on the optimal corresponding decision, adjust the parameters of the next cycle timeout sliding window, dynamically control the regular optimization, contraction or tolerance of the next cycle timeout sliding window, and reduce the delay, redundancy and packet loss of subsequent data packet transmission. Example

[0057] This embodiment provides a computer-readable storage medium having a computer program / instruction stored thereon. When the computer program / instruction is executed by a processor, the steps of the method described in Embodiment 1 are implemented: The low-altitude aircraft receives the data packet; Extract the identification of the received data packets to obtain the identification number of each data packet; Use the counting Bloom filter CBF to query the identification number, obtain the identification query result and determine whether the received data packet is a suspected redundant data packet; When the received data packet is suspected to be a redundant data packet, the suspected redundant data packet is queried again using the timeout sliding window to obtain a query result; Based on the query results, data packets are discarded and the sliding window background records are modified; Calculate the background records of the timeout sliding window to obtain the redundancy success rate, redundancy elimination timeout rate, and invalid query rate within this cycle; A multi-dimensional network status decision engine is used to process the redundancy success rate, redundancy elimination timeout rate, and invalid query rate to obtain the diagnosis results of the current network status and output the optimal corresponding decision; Based on the optimal corresponding decision, the parameters of the next period timeout sliding window are adjusted, and the regular optimization, contraction or tolerance of the next period timeout sliding window is dynamically controlled.

[0058] In summary, the present invention combines a counting Bloom filter and a timeout sliding window into a two-stage filtering architecture with fast screening and precision calibration capabilities. This architecture effectively identifies redundant and new data packets, avoids wasting computing and memory resources on subsequent processing of invalid data, and significantly improves the system's effective throughput. Furthermore, a multidimensional decision engine is used to calculate network status parameters and diagnose the network within the current cycle, obtaining the optimal decision for the next cycle. This improves the network's adaptive capability for redundancy reduction in the next cycle, enabling intelligent perception of network status and efficient implementation of the optimal redundancy reduction strategy under various network conditions. The multidimensional decision engine accurately perceives and calculates network state changes in redundancy, latency, packet loss, and other dimensions in real time, dynamically adjusting core algorithm parameters to adapt to the optimal redundancy reduction strategy for subsequent communications in various complex network environments. A background cleanup process regularly cleans up the timeout sliding window's background records, enabling the window to adaptively adjust its window size and timeout logic in response to dynamic changes in heterogeneous networks. Regularly cleaning up timeout redundant packets also improves the accuracy of false-positive packet recognition, prevents historical data from interfering with current decisions, and reduces memory usage. A multi-dimensional decision engine diagnoses network status and controls the timeout sliding window tolerance strategy and boundaries, maximizing data transmission reliability while preventing infinite waits due to link issues through a timeout mechanism, ensuring stable timing for upper-layer services. When the timeout sliding window background records changes, it simultaneously modifies all hash positions corresponding to the counting Bloom filter (CBF), ensuring synchronized increases and decreases between the counting Bloom filter and the sliding window state, guaranteeing long-term operational performance. The timeout sliding window also records and stores redundancy success events, invalid query events, and redundancy elimination timeout events. The redundancy elimination timeout rate, redundancy elimination timeout rate, and invalid query rate within any cycle are calculated, quantifying the accuracy of network diagnosis.

[0059] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0060] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0061] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0062] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0063] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present invention, ordinary technicians in this field can also make many forms without departing from the scope of protection of the purpose of the present invention and the claims, which are all protected by the present invention.

Claims

1. A low-altitude aircraft communication decision-making method, characterized by: include: The low-altitude aircraft receives the data packet; Extract the identification of the received data packets to obtain the identification number of each data packet; Use the counting Bloom filter CBF to query the identification number, obtain the identification query result and determine whether the received data packet is a suspected redundant data packet; When the received data packet is suspected to be a redundant data packet, the suspected redundant data packet is queried again using the timeout sliding window to obtain a query result; Based on the query results, data packets are discarded and the sliding window background records are modified; Calculate the background records of the timeout sliding window to obtain the redundancy success rate, redundancy elimination timeout rate, and invalid query rate within this cycle; A multi-dimensional network status decision engine is used to process the redundancy success rate, redundancy elimination timeout rate, and invalid query rate to obtain the diagnosis results of the current network status and output the optimal corresponding decision; Based on the optimal corresponding decision, the parameters of the next period timeout sliding window are adjusted, and the regular optimization, contraction or tolerance of the next period timeout sliding window is dynamically controlled.

2. The low-altitude aircraft communication decision-making method according to claim 1, characterized in that: The low-altitude aircraft receiving the data packet also includes: Establish communication with multiple frequency band models through the QUIC protocol and build multiple QUIC communication links; Synchronously receive data packets based on multiple QUIC communication links; The data interface is used to embed the QUIC identification number into each received data packet for subsequent counting Bloom filter CBF query and sliding window verification.

3. The low-altitude aircraft communication decision-making method according to claim 1, characterized in that: Use the counting Bloom filter (CBF) to query the identification number, obtain the identification query result, and determine whether the received data packet is a suspected redundant data packet, including: If the identification number is found in the counting Bloom filter CBF, it is preliminarily determined that the data packet is suspected redundant data; If the identification number is not found in the counting Bloom filter CBF, it is preliminarily determined that the data packet is a new data packet.

4. The low-altitude aircraft communication decision-making method according to claim 1, characterized in that: If the identification number is not found in the counting Bloom filter CBF, the data packet is preliminarily determined to be a new data packet, further comprising: When the data packet is a new data packet, the new data packet is used as valid data for delivery to the upper-layer business logic for processing; Creating a background record of the timeout sliding window based on the new data packet, wherein the background record includes the QUIC identification sequence number and arrival time of the new data; The QUIC identification number of the new data packet is hashed using the hash function of the counting Bloom filter CBF, a hash position corresponding to the new data packet is generated in the counting Bloom filter CBF, and the count value at the hash position is incremented by one.

5. The low-altitude aircraft communication decision-making method according to claim 1, characterized in that: When the received data packet is suspected to be a redundant data packet, the suspected redundant data packet is queried again using the timeout sliding window to obtain the query results, including: Use the timeout sliding window to query suspected redundant data packets. If the same QUIC identification number record exists in the background of the timeout sliding window, the suspected redundant data packet is determined to be a redundant data packet. If the same QUIC identification sequence number record does not exist in the background of the timeout sliding window, the suspected redundant data packet is determined to be a new data packet.

6. The low-altitude aircraft communication decision-making method according to claim 1, characterized in that: Based on the query results, packets are discarded and sliding window background records are modified, including: After the suspected redundant data packet is determined to be a redundant data packet, the arrival time of the redundant data packet is extracted and the current data packet is directly discarded; The redundant delay difference of the redundant data packet is calculated based on the difference between the arrival time corresponding to the QUIC identification number recorded in the background of the timeout sliding window and the arrival time of the redundant data packet; The redundant delay difference is used as a sample to calculate the moving average of the timeout sliding window, and the updated moving average of the timeout sliding window is obtained; Calculate the timeout sliding window size based on the updated moving average value of the timeout sliding window to obtain the adjusted size of the timeout sliding window; After the size of the timeout sliding window is updated, the QUIC identification number record corresponding to the redundant data packet is deleted in the background of the timeout sliding window, and the count value at the hash position corresponding to the QUIC identification number in the counting Bloom filter CBF is simultaneously reduced by one; After the suspected redundant data packet is determined to be a new data packet, the process of querying the new data packet is counted as an invalid query event, and the new data packet is regarded as valid data.

7. The low-altitude aircraft communication decision-making method according to claim 1, characterized in that: A multi-dimensional network status decision engine is used to process the redundancy success rate, redundancy elimination timeout rate, and invalid query rate to obtain the adjustment weight value of the current network status and use it as the optimal communication decision result, including: The multi-dimensional network state decision matrix is ​​used to extract the features of redundancy success rate, redundancy elimination timeout rate and invalid query rate respectively, and the features of redundancy success rate, redundancy elimination timeout rate and invalid query rate are obtained; The features of redundancy success rate, redundancy elimination timeout rate, and invalid query rate are integrated to obtain the input feature vector of the current network status; Perform feature fuzzy processing on the input feature vector of the current network state, so that each feature establishes a mapping relationship with multiple predefined fuzzy sets; Calculate multiple fuzzy sets according to predefined membership functions to obtain the membership degrees of multiple fuzzy sets; Use the pre-built fuzzy decision rule library to perform fuzzy reasoning on the real-time input feature vector to obtain the trigger strength of each rule, and aggregate the trigger strengths of all rules to obtain the final fuzzy decision result; The center of gravity method is used to calculate the final fuzzy decision result and obtain the quantitative diagnosis result of the current network status; According to the diagnosis results of the current network status, the optimal corresponding decision for adjusting the timeout sliding window of the next cycle is output.

8. The low-altitude aircraft communication decision-making method according to claim 1, characterized in that: The timeout sliding window background record also includes regular maintenance of the timeout sliding window background record through a periodic cleaning process; When cleaning background records, first traverse all records in the timeout sliding window background and extract the survival time of each record; When the survival time of any record exceeds the tolerance time of the current timeout sliding window, the record is determined to be a timeout record, and the redundant data packet corresponding to the record has timed out; The determined timeout records are cleaned up regularly to obtain a timeout sliding window background in which the timeout records are regularly removed.

9. A low-altitude aircraft communication decision-making device, characterized by comprising: Data receiving module, adaptive redundancy elimination module, network status parameter calculation module, redundancy elimination decision module; The data receiving module is used to receive data packets and extract identifications from the received data packets to obtain identification serial numbers of each data packet; The adaptive redundancy elimination module is used to use a counting Bloom filter (CBF) to query the identification number, obtain an identification query result, and determine whether the received data packet is a suspected redundant data packet; when the received data packet is a suspected redundant data packet, use a timeout sliding window to query the suspected redundant data packet again to obtain a query result; Based on the query results, data packets are discarded and the sliding window background records are modified; The network status parameter calculation module is used to calculate the timeout sliding window background record to obtain the redundancy success rate, redundancy elimination timeout rate, and invalid query rate within this cycle; The redundancy elimination decision module is used to use a multi-dimensional network status decision engine to process the redundancy success rate, redundancy elimination timeout rate, and invalid query rate, obtain the diagnosis result of the current network status and output the optimal corresponding decision; based on the optimal corresponding decision, adjust the parameters of the next cycle timeout sliding window, and dynamically control the regular optimization, contraction or tolerance of the next cycle timeout sliding window.

10. A computer-readable storage medium having a computer program / instruction stored thereon, characterized in that: When the computer program / instruction is executed by a processor, the steps of the low-altitude aircraft communication decision-making method described in any one of claims 1-8 are implemented.

Images