Short-wave ground-air communication networking link optimization method

Abstract

The invention discloses a short-wave ground-air communication networking link optimization method. The method specifically comprises the following steps: predicting the strength of a received signal; calculating noise intensity; calculating a receiving signal-to-noise ratio; carrying out site clustering analysis; and carrying out site optimization output. According to the method, the communication quality of the working frequency of each guarantee link is predicted through the historical data, the use efficiency sequence of the transmitting and receiving stations participating in guarantee is dynamically generated, and a basis is provided for link switching of ground calling users. Link optimization is based on signal-to-noise ratio prediction of a link, and dynamic factors such as distribution of receiving stations and transmitting stations, antenna pointing, positions of air users, ionosphere states and electromagnetic environments need to be comprehensively considered, so that the accuracy and reliability of the link optimization are improved. On the basis of signal-to-noise ratio prediction, ground receiving and transmitting sites are classified and aggregated through a clustering analysis technology, optimization of a ground-air link is completed through an optimization algorithm, the air-to-air optimal transmitting and receiving links can be provided for ground calling users, and one-time communication success with the maximum probability is ensured.

Description

Technical Field

[0001] The present invention relates to shortwave ground-to-air communication technology, and in particular to a shortwave ground-to-air communication networking link optimization method. Background Art

[0002] In shortwave ground-to-air communications, due to the limited power and antenna efficiency of onboard shortwave communication equipment, reliable shortwave communication requires careful selection of communication frequencies and ground communication sites. Link optimization predicts the communication quality of each supported link in real time during flight. Based on the operating frequency, it dynamically adjusts the priority of participating transceiver sites, providing a basis for link switching for ground users.

[0003] At present, the application of shortwave communication in my country's aviation communication is still mainly point-to-point and fixed-frequency communication. It has a single business and no networking application, which makes the shortwave ground-to-air communication have poor anti-interference ability and low communication connectivity rate. Summary of the Invention

[0004] In view of the problems existing in the prior art, the present invention provides a method for optimizing a shortwave ground-to-air communication network link, which specifically includes the following steps:

[0005] Step 1. Received signal strength prediction;

[0006] The calculation formula for the received signal strength is:

[0007] P r (dBm) = P t (dBm)+G r (dB)-Loss(dB) (1)

[0008] Where, P r is the signal level at the receiving end, P t is the effective radiated power of the transmitter, G r is the receiving antenna gain, LOSS is the radio wave propagation loss between the transmitting and receiving sites;

[0009] 1) Effective radiated power of the airborne station

[0010] The calculation formula for the effective radiated power of the airborne station is:

[0011] P t (dBm) = P (dBm) + G t (dB) (2)

[0012] Where P is the transmitter power, G t is the antenna gain;

[0013] 2) Ground receiving antenna gain

[0014] The ground receiving antenna gain model must be built for each site's specific antenna. Based on the antenna's radiation performance, the effects of height and elevation angle are considered to form radiation patterns at different frequencies. When calculating signal strength, the receiving antenna gain value is derived based on the operating frequency and the azimuth relative to the direction of maximum radiation.

[0015] 3) Radio wave propagation loss

[0016] The radio wave propagation loss is calculated as:

[0017] loss=prop(p t ,p r ,f,t,ssn)

[0018] Among them, p t ,p r are the locations of the transmitting and receiving points respectively, t is the loss calculation time, ssn is the number of sunspots, and f is the loss calculation time; prop(p t ,p r ,f,t,ssn) is the first shortwave propagation calculation function provided by the ITU-R P.533 model;

[0019] The propagation model provides the calculation of the highest usable frequency muf, which is written as:

[0020] muf=prop_m(p t ,p r ,t,ssn)

[0021] Among them, prop_m(p t ,p r ,t,ssn) is the second shortwave propagation calculation function provided by the ITU-R P.533 model;

[0022] The prediction method is used to correct the prediction results of ionospheric radio wave propagation loss. The short-term shortwave loss prediction method is as follows:

[0023] (1) Based on the shortwave communication detection data, the muf prediction value at the same time in the past T days is calculated;

[0024] (2) Calculate the current muf prediction value muf0 by the Kalman filter method

[0025] (3) According to the 533 propagation model, when the sunspot number ssn takes the value interval [0, A], it is considered that prop_m(.,.,.,.) is a monotonic function. According to its inverse function prop_m -1 (.,.,.,.) and the maximum passable frequency muf are used to infer the sunspot number, and the inference result is set as the pseudo sunspot number fssn∈[0,A]:

[0026] fssn=prop_m -1 (p t ,p r ,t,muf0)

[0027] (4) The propagation calculation is performed based on the pseudo sunspot number to obtain the loss prediction value loss.

[0028] loss=prop(p t ,p r ,f,t,fssn)

[0029] Among them, prop(p t ,p r ,f,t,fssn) is the first shortwave propagation calculation function provided in the ITU-R P.533 model.

[0030] step2. Calculate noise intensity;

[0031] 1) Noise intensity calculation based on historical data statistics

[0032] Conduct statistical analysis on historical spectrum monitoring data to derive the background noise distribution of each station and frequency in different time periods and seasons;

[0033] 2) Real-time noise intensity calculation;

[0034] The background noise level is calculated using the average or modified average method as follows:

[0035] The monitoring data of the same time in the previous n days are counted and the average value is calculated. If the current measured value does not change much compared with the historical average value, the average value is used as the predicted value for that time of the day. If the current measured value changes significantly compared with the historical average value, a correction value δ(t) is added to the average value to reflect the real-time nature of the regularity.

[0036] The calculation method of the correction amount δ(t) is:

[0037] (1) Calculate the mean value E and standard deviation d of the moment before the previous n days, that is, the moment t-1;

[0038] (2) Based on the actual value V at the previous moment, the historical average value E and standard deviation d of the previous n days, and the historical average value E at the current moment next , find the correction value δ at the current moment, the formula is as follows:

[0039] When V>E, δ=(VEd)*E next / E;

[0040] Otherwise, δ=-(VEd)*E next / E;

[0041] The noise prediction value V at the current moment next , is the historical average value E at the current moment next Add δ to the basis, that is:

[0042] V next =E next +δ;

[0043] Step 3. Calculate the received signal-to-noise ratio;

[0044] Based on the calculation results of steps 1 and 2, calculate the signal-to-noise ratio prediction value SNR at the next moment:

[0045] SNR = P r -V next =P t +G r -loss-V next

[0046] Among them, P r is the receiver signal strength;

[0047] For each candidate site i, the signal-to-noise ratio guaranteed by the site is a function of frequency, denoted as SNR i (f);

[0048] Step 4. Site cluster analysis;

[0049] According to the longitude and latitude of the stations, the k-means clustering method is used for clustering. After the station clusters are assigned, the station clusters are used as substitutes for the stations. The optimal guaranteed signal-to-noise ratio of all stations in each cluster is taken as the guaranteed signal-to-noise ratio record of each cluster, which is recorded as SNR. i ′(f);

[0050] The signal-to-noise ratio (SNR) is scored according to the requirements and converted into a discrete guarantee score. The guarantee score of the frequency f used by the i-th site is recorded as:

[0051] V i (f) = val(SNR i (f))

[0052] val(·) is the score function, which is calculated as follows:

[0053]

[0054] Among them, V i (f) represents the signal-to-noise ratio guarantee score of site i using frequency f, snr represents the calculated signal-to-noise ratio value, {v1,v2,...,v N} is the discrete scoring threshold of SNR, and N is the number of set scoring thresholds;

[0055] Step 5. Output site optimization;

[0056] The problem of selecting link guarantee sites can be transformed into: given sites m∈{1,2,...,M}, M is the number of known sites, and the transmission frequency of each site is assigned as {f1,f2,...,f M}, where f m is the transmission frequency of the mth station, f m = 0 means that the site is not selected for protection. The site receiving frequency is assigned to f0. Find the best frequency combination {f0,f1,f2,...,f M}, so that the total security score V is the highest;

[0057]

[0058] Among them, the guarantee score V m =V m (f m ), m>0;

[0059]

[0060] Among them, V m (f) is the security score when the mth station uses frequency f;

[0061] When the site activation cost is considered, the site activation cost G is defined as the resource cost required for site activation: Among them, G m (·) is the activation cost of the mth site, defined as: G m is the activation cost of the mth site;

[0062] Taking the total guarantee score V and resource cost G as the objective function, the nsga-II genetic algorithm is constructed, which can obtain the site optimization result.

[0063] In one embodiment of the present invention, a prediction method is used to correct the prediction result of ionospheric radio wave propagation loss; the short-term shortwave loss prediction method is as follows:

[0064] (1) Based on the shortwave communication detection data, the muf prediction values ​​at the same time in the past 7 days are calculated;

[0065] (3) According to the 533 propagation model, when the sunspot number ssn takes a value interval of [0,200], it is considered that prop_m(.,.,.,.) is a monotonic function. According to its inverse function prop_m -1 (.,.,.,.) and the maximum passable frequency muf are used to invert the sunspot number, and the inversion result is set as the pseudo sunspot number fssn∈[0,200].

[0066] In step 4 of a specific embodiment of the present invention, calculation is performed based on the assumption that the distribution range of each cluster does not exceed 100 kilometers.

[0067] The shortwave ground-to-air communication networking link optimization method of the present invention is based on the prediction of the signal-to-noise ratio of the link, comprehensively considering dynamic factors such as the distribution of receiving stations and transmitting stations, antenna pointing, the position of airborne users, ionospheric state, and electromagnetic environment. Through clustering analysis technology, ground transceiver stations are classified and aggregated, and the optimization algorithm is used to complete the optimization of the ground-to-air link. The method can provide the optimal air-to-air transmission and reception links to ground calling users, ensuring the highest probability of successful communication. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] Figure 1 Shows a block diagram of the link optimization working principle;

[0069] Figure 2 Shows the measured noise data of Ningbo;

[0070] Figure 3 Shows the comparison between the predicted and measured noise levels (Ningbo 5561KHz);

[0071] Figure 4 The comparison between the predicted and measured noise levels is shown (Ningbo 9551KHz). DETAILED DESCRIPTION

[0072] The present invention will be described in detail below with reference to the accompanying drawings.

[0073] In the link optimization calculation, the present invention mainly uses the analysis of the air-to-ground transmission and reception scenario as the basis. Under a given frequency, the signal-to-noise ratio of the air-to-ground links at the communication frequency is analyzed. At the same time, the sites (radio stations) are sorted through cluster analysis. The optimization algorithm is used to complete the optimization sorting of the ground-to-air links. The working principle of link optimization is as follows: Figure 1 shown.

[0074] Link optimization is primarily based on real-time (or near-real-time) data. Based on the current position of the aerial platform, the radio wave propagation environment, and the background electromagnetic environment, and taking into account the current site communication resource conditions, the link's signal-to-noise ratio is calculated in real time. This ratio determines the link's priority level and selects the optimal communication link (i.e., the optimal communication support site). Therefore, link optimization emphasizes real-time performance, requiring the rapid selection of the highest-performing link from multiple air-ground links.

[0075] The shortwave ground-to-air communication network link optimization method specifically includes the following steps:

[0076] 1. Received signal strength prediction

[0077] The received signal strength prediction calculation needs to consider factors such as the effective radiated power of the airborne station, the radio wave propagation loss of the signal along the propagation path, and the gain of the receiving antenna. Among them, the shortwave ionospheric propagation prediction is related to the highest usable frequency in the ionosphere.

[0078] The calculation formula for the received signal strength is:

[0079] P r (dBm) = P t (dBm)+G r (dB)-Loss(dB) (1)

[0080] In formula (1), P r is the signal level at the receiving end (dBm), P t is the effective radiated power of the transmitter (dBm), G r is the receiving antenna gain (dB), and LOSS is the radio wave propagation loss between the transmitting and receiving sites (dB).

[0081] 1) Effective radiated power of the airborne station

[0082] The effective radiated power of the airborne station is determined by the airborne station transmit power and the airborne antenna gain. The calculation formula is:

[0083] P t (dBm) = P (dBm) + G t (dB) (2)

[0084] Where P is the transmitter power (dBm), G t is the antenna gain (dB).

[0085] 2) Ground receiving antenna gain

[0086] In shortwave ground-to-air telecommunications, ground stations may employ directional antennas. The directivity of ground-mounted shortwave directional antennas is dependent on factors such as antenna height, azimuth, elevation, and operating frequency. Therefore, the ground receiving antenna gain model must be tailored to the specific antenna at each site. Based on the antenna's inherent radiation performance, the effects of height and elevation are considered to generate radiation patterns for different frequencies. When calculating signal strength, the receiving antenna gain is derived based on the operating frequency and azimuth relative to the direction of maximum radiation. Methods for determining ground receiving antenna gain are well known to those skilled in the art and will not be elaborated upon here.

[0087] 3) Radio wave propagation loss

[0088] A key factor in predicting the signal-to-noise ratio (SNR) in shortwave communications is shortwave ionospheric propagation loss. Because shortwave propagation loss involves ionospheric transmission, and the ionospheric channel is a time-varying channel, propagation loss is dependent on numerous factors, including the location of the transmitter and receiver stations, operating frequency, solar activity, date, and time. Currently, the ITU-RP.533 model is widely used internationally to calculate shortwave ionospheric propagation loss. It is a statistical model that predicts monthly average ionospheric propagation loss based on statistics such as the 11-year cycle of the sunspot number.

[0089] Because the propagation loss is related to the time-space frequency and the number of sunspots, the radio wave propagation loss can be calculated as:

[0090] loss=prop(p t ,p r ,f,t,ssn)

[0091] Among them, p t ,p r are the locations of the transmitting point and the receiving point respectively, t is the loss calculation time, ssn is the number of sunspots, and f is the loss calculation time. prop(p t ,p r ,f,t,ssn) is the first shortwave propagation calculation function provided in the ITU-R P.533 model.

[0092] The propagation model also provides the calculation of the highest usable frequency muf, which can be written as:

[0093] muf=prop_m(p t ,p r ,t,ssn)

[0094] The above formula calculation method comes from the reference document: "ITU-R P.533-14 Recommendation: Method for Predicting HF Circuit Performance" (ITU-R P.533 is a standard issued by the International Telecommunication Union (ITU), prop_m(p t ,p r ,t,ssn) is the second shortwave propagation calculation function provided in the ITU-RP.533 model.

[0095] Because the 533 model is a statistical model, it may have significant differences from real-time propagation loss. Therefore, historical data or ionospheric sounding data can be used to revise the ionospheric radio wave propagation loss prediction results using prediction methods based on different time intervals.

[0096] The short-term shortwave loss prediction method is as follows:

[0097] (1) Based on the shortwave communication detection data, the muf prediction values ​​at the same time in the past 7 days are calculated;

[0098] (2) Calculate the current muf prediction value muf0 by the Kalman filter method

[0099] (3) According to the 533 propagation model, when the sunspot number ssn takes a value interval of [0,200], it can be considered that prop_m(.,.,.,.) is a monotonic function, so its inverse function prop_m -1 (.,.,.,.) and the maximum passable frequency muf are used to infer the sunspot number, and the inverse result is set as the pseudo sunspot number fssn∈[0,200]:

[0100] fssn=prop_m -1 (p t ,p r ,t,muf0)

[0101] (4) The propagation calculation is performed based on the pseudo sunspot number to obtain the loss prediction value loss.

[0102] loss=prop(p t ,p r ,f,t,fssn)

[0103] Among them, prop(p t ,p r ,f,t,fssn) is the first shortwave propagation calculation function provided in the ITU-R P.533 model.

[0104] 2. Noise intensity calculation

[0105] Background noise consists of cosmic noise, atmospheric noise, and man-made noise. Man-made noise is the primary component of a station's background noise, making it difficult to predict and calculate background noise values ​​close to their true value. In networking systems, statistical or real-time calculations are used, depending on link optimization requirements.

[0106] 1) Noise intensity calculation based on historical data statistics

[0107] During the frequency assignment planning phase, using real-time spectrum monitoring data is not very useful. Instead, statistical analysis of historical spectrum monitoring data can be used to determine the background noise distribution for each station and frequency over different time periods and seasons, supporting signal-to-noise ratio prediction. Methods for analyzing historical data are well known to those skilled in the art and will not be detailed here.

[0108] Figure 2 An example of background noise monitoring diagram detected by spectrum receiving equipment in Ningbo station is given.

[0109] 2) Real-time noise intensity calculation

[0110] Link optimization is a quasi-real-time calculation. The background noise level can be calculated using the average value or modified average value method. The specific method is as follows.

[0111] The monitoring data for the same time period over the previous n days is collected and averaged. If the difference between the current measured value and the historical average is less than a set threshold, the average is used as the predicted value for that time period. If the difference between the current measured value and the historical average is greater than the set threshold, a correction δ(t) is added to the average to reflect the real-time nature of the pattern. This threshold can be set as a variable empirical parameter.

[0112] The calculation method of the correction amount δ(t) is:

[0113] (1) Calculate the mean value E and standard deviation d of the moment before the previous n days (i.e., moment t-1);

[0114] (2) Based on the actual value V at the previous moment, the historical average value E and standard deviation d of the previous n days, and the historical average value E at the current moment next , find the correction value δ at the current moment, the formula is as follows:

[0115] When V>E, δ=(VEd)*E next / E;

[0116] Otherwise, δ=-(VEd)*E next / E.

[0117] The noise prediction value V at the current moment next , is the historical average value E at the current moment next Add δ to the basis, that is:

[0118] V next =E next +δ.

[0119] The comparison between the predicted results of the modified average method and the measured results is as follows: Figures 3 and 4 As shown in the figure, we can see that the noise level prediction value is very close to the measured value using this method.

[0120] 3. Receive signal-to-noise ratio calculation

[0121] Based on the calculation results of steps 1 and 2, the signal-to-noise ratio prediction value SNR at the next moment can be calculated:

[0122] SNR = P r -V next =P t +G r -loss-V next

[0123] Among them, Vnext is the noise prediction value, P r is the receiver signal strength.

[0124] For link optimization calculation, the time and the sending and receiving location can generally be considered to be confirmed. In all the above calculations, the calculation time, the sunspot number and the sending and receiving location can be considered to be fixed values.

[0125] For each candidate site i, the signal-to-noise ratio guaranteed by the site is a function of frequency, denoted as SNR i (f);

[0126] 4. Site cluster analysis

[0127] Based on the longitude and latitude of the stations, the k-means clustering method is used for clustering. For example, the distribution range of each cluster is calculated as no more than 100 kilometers. After the station cluster is allocated, the station cluster is used as a replacement for the station. The optimal guaranteed signal-to-noise ratio of all stations in each cluster is taken as the guaranteed signal-to-noise ratio record of each cluster, which can be recorded as SNR i ′(f). (The k-means clustering method is well known to those skilled in the art and will not be described again.)

[0128] Score the signal-to-noise ratio (SNR) according to requirements and convert it into discrete assurance scores.

[0129] The guarantee score of the usage frequency f of the i-th site is recorded as:

[0130] V i (f) = val(SNR i (f))

[0131] val(·) is the score function, which is calculated as follows:

[0132]

[0133] Among them, V i (f) represents the signal-to-noise ratio guarantee score of site i using frequency f, snr represents the calculated signal-to-noise ratio value, {v1,v2,...,v N} is the discrete scoring threshold of SNR, and N is the number of set scoring thresholds.

[0134] 5. Site optimization output

[0135] According to the communication requirements of ground-to-air communication, the link configuration generally adopts a multi-transmit and multi-receive design, in which the multi-transmit adopts frequency diversity, and the ground shortwave equipment selects the candidate site and transmission frequency, and transmits synchronously to the airborne shortwave equipment; the multi-receive adopts space diversity, and the airborne shortwave equipment uses a single frequency for transmission, and all ground stations perform unified processing after receiving.

[0136] The problem of selecting link guarantee sites can be transformed into: given sites m∈{1,2,...,M}, M is the number of known sites, and the transmission frequency of each site is assigned as {f1,f2,...,f M}, where f m is the transmission frequency of the mth station, f m = 0 means that the site is not selected for protection. The site receiving frequency is assigned to f0. Find the best frequency combination {f0,f1,f2,...,f M}, so that the total security score V is the highest.

[0137]

[0138] Among them, the guarantee score V m =V m (f m ), m>0.

[0139]

[0140] Among them, V m (f) is the security score when the mth site uses frequency f.

[0141] When the site activation cost is considered, the site activation cost G is defined as the resource cost required for site activation: Among them, G m (·) is the activation cost of the mth site, defined as: G m is the activation cost of the mth site.

[0142] Using the guaranteed total score V and resource cost G as the objective function, we construct the NSGA-II genetic algorithm (non-dominated sorting genetic algorithm) to determine the site optimization result. The method of the non-dominated sorting genetic algorithm is well known to those skilled in the art and will not be described again.

[0143] In shortwave ground-to-air communications, due to the limited power and antenna efficiency of onboard shortwave communication equipment, reliable shortwave communication requires careful selection of communication frequencies and ground communication sites. Link optimization predicts the communication quality of each supported link in real time during flight. Based on the operating frequency, it dynamically adjusts the priority of participating transceiver sites, providing a basis for link switching for ground users.

[0144] Link optimization must first comprehensively consider various factors, especially the uncertainties introduced by shortwave ionospheric propagation characteristics, to ensure the highest probability of communication reachability. Generally speaking, factors such as the guaranteed area (route), frequency requirements, and the range of available frequency resources are deterministic factors for frequency assignment. However, ionospheric propagation characteristics, affected by the time-varying nature of the ionosphere, are uncertain factors, which significantly impact shortwave communication performance. Therefore, radio wave propagation prediction is a key factor in ensuring the credibility of link optimization results. Ionospheric radio wave propagation prediction is mainly divided into signal-to-noise ratio prediction and available frequency band range prediction. The signal-to-noise ratio prediction calculation mainly includes received signal strength (primarily ionospheric propagation loss prediction) and noise intensity prediction (or measurement).

[0145] The shortwave ground-to-air communication network link optimization method is based on link signal-to-noise ratio prediction. It comprehensively considers dynamic factors such as the distribution of receiving and transmitting stations, antenna pointing, the location of airborne users, the state of the ionosphere, and the electromagnetic environment to improve its accuracy and reliability. Based on the signal-to-noise ratio prediction, cluster analysis technology is used to classify and aggregate ground transceiver sites. An optimization algorithm is then used to optimize the ground-to-air link. This method can provide the optimal airborne transmission and reception links to ground-based users, ensuring the highest probability of successful communication.

Claims

1. A method for optimizing shortwave ground-to-air communication network links, characterized in that: The method specifically comprises the following steps: Step 1. Received signal strength prediction; The calculation formula for the received signal strength is: P r (dBm)=P t (dBm)+G r (dB)-Loss(dB) (1) Where, P r is the signal level at the receiving end, P t is the effective radiated power of the transmitter, G r is the receiving antenna gain, LOSS is the radio wave propagation loss between the transmitting and receiving sites; 1) Effective radiated power of the airborne station The calculation formula for the effective radiated power of the airborne station is: P t (dBm)=P(dBm)+G t (dB) (2) Where P is the transmitter power, G t is the antenna gain; 2) Ground receiving antenna gain The ground receiving antenna gain model must be built for each site's specific antenna. Based on the antenna's radiation performance, the effects of height and elevation angle are considered to form radiation patterns at different frequencies. When calculating signal strength, the receiving antenna gain value is derived based on the operating frequency and the azimuth relative to the direction of maximum radiation. 3) Radio wave propagation loss The radio wave propagation loss is calculated as: loss=prop(p t ,p r ,f,t,ssn) Among them, p t ,p r are the locations of the transmitting and receiving points respectively, t is the loss calculation time, ssn is the number of sunspots, and f is the loss calculation time; prop(p t ,p r ,f,t,ssn) is the first shortwave propagation calculation function provided by the ITU-R P.533 model; The propagation model provides the calculation of the highest usable frequency muf, which is written as: muf=prop_m(p t ,p r ,t,ssn) Among them, prop_m(p t ,p r ,t,ssn) is the second shortwave propagation calculation function provided by the ITU-R P.533 model; The prediction method is used to correct the prediction results of ionospheric radio wave propagation loss. The short-term shortwave loss prediction method is as follows: (1) Based on the shortwave communication detection data, the muf prediction value at the same time in the past T days is calculated; (2) Calculate the current muf prediction value muf0 by the Kalman filter method (3) According to the 533 propagation model, when the sunspot number ssn takes the value interval [0, A], it is considered that prop_m(.,.,.,.) is a monotonic function. According to its inverse function prop_m -1 (.,.,.,.) and the maximum passable frequency muf are used to infer the sunspot number, and the inference result is set as the pseudo sunspot number fssn∈[0,A]: fssn=prop_m -1 (p t ,p r ,t,muf0) (4) The propagation calculation is performed based on the pseudo sunspot number to obtain the loss prediction value loss. loss=prop(p t ,p r ,f,t,fssn) Among them, prop(p t ,p r ,f,t,fssn) is the first shortwave propagation calculation function provided in the ITU-R P.533 model. step2. Calculate noise intensity; 1) Noise intensity calculation based on historical data statistics Conduct statistical analysis on historical spectrum monitoring data to derive the background noise distribution of each station and frequency in different time periods and seasons; 2) Real-time noise intensity calculation; The background noise level is calculated using the average or modified average method as follows: The monitoring data of the same time in the previous n days are counted and the average value is calculated. If the current measured value does not change much compared with the historical average value, the average value is used as the predicted value for that time of the day. If the current measured value changes significantly compared with the historical average value, a correction value δ(t) is added to the average value to reflect the real-time nature of the regularity. The calculation method of the correction amount δ(t) is: (1) Calculate the mean value E and standard deviation d of the moment before the previous n days, that is, the moment t-1; (2) Based on the actual value V at the previous moment, the historical average value E and standard deviation d of the previous n days, and the historical average value E at the current moment next , find the correction value δ at the current moment, the formula is as follows: When V>E, δ=(VEd)*E next / E; Otherwise, δ=-(VEd)*E next / E; The noise prediction value V at the current moment next , is the historical average value E at the current moment next Add δ to the basis, that is: V next =E next +d; Step 3. Calculate the received signal-to-noise ratio; Based on the calculation results of steps 1 and 2, calculate the signal-to-noise ratio prediction value SNR at the next moment: SNR=P r -V next =P t +G r -loss-V next Among them, P r is the receiver signal strength; For each candidate site i, the signal-to-noise ratio guaranteed by the site is a function of frequency, denoted as SNR i (f); Step 4. Site cluster analysis; According to the longitude and latitude of the stations, the k-means clustering method is used for clustering. After the station clusters are assigned, the station clusters are used as substitutes for the stations. The optimal guaranteed signal-to-noise ratio of all stations in each cluster is taken as the guaranteed signal-to-noise ratio record of each cluster, which is recorded as SNR. i ′(f); The signal-to-noise ratio (SNR) is scored according to the requirements and converted into a discrete guarantee score. The guarantee score of the frequency f used by the i-th site is recorded as: V i (f)=val(SNR i (f)) val(·) is the score function, which is calculated as follows: Among them, V i (f) represents the signal-to-noise ratio guarantee score of site i using frequency f, snr represents the calculated signal-to-noise ratio value, {v1,v2,...,v N } is the discrete scoring threshold of SNR, and N is the number of set scoring thresholds; Step 5. Output site optimization; The problem of selecting link guarantee sites can be transformed into: given sites m∈{1,2,...,M}, M is the number of known sites, and the transmission frequency of each site is assigned as {f1,f2,...,f M }, where f m is the transmission frequency of the mth station, f m = 0 means that the site is not selected for protection. The site receiving frequency is assigned to f0. Find the best frequency combination {f0,f1,f2,...,f M }, so that the total security score V is the highest; Among them, the guarantee score V m =V m (f m ), m>0; Among them, V m (f) is the security score when the mth station uses frequency f; When the site activation cost is considered, the site activation cost G is defined as the resource cost required for site activation: Among them, G m (·) is the activation cost of the mth site, defined as: G m is the activation cost of the mth site; Taking the total guarantee score V and resource cost G as the objective function, the nsga-II genetic algorithm is constructed, which can obtain the site optimization result.

2. The shortwave ground-to-air communication network link optimization method according to claim 1, characterized in that: In step 1, step 3), the prediction method is used to correct the prediction results of ionospheric radio wave propagation loss; the short-term shortwave loss prediction method is as follows: (1) Based on the shortwave communication detection data, the muf prediction values ​​at the same time in the past 7 days are calculated; (3) According to the 533 propagation model, when the sunspot number ssn takes a value interval of [0,200], it is considered that prop_m(.,.,.,.) is a monotonic function. According to its inverse function prop_m -1 (.,.,.,.) and the maximum passable frequency muf are used to invert the sunspot number, and the inversion result is set as the pseudo sunspot number fssn∈[0,200].

3. The shortwave ground-to-air communication network link optimization method according to claim 1, characterized in that: In step 4, the calculation is performed based on the assumption that the distribution range of each cluster does not exceed 100 kilometers.

Images