Millimeter wave communication system with adaptive frequency modulation

Abstract

The invention discloses an adaptive frequency modulation millimeter wave communication system, which comprises a channel quality evaluation module for measuring a link signal-to-noise ratio, multipath time delay spread and a Doppler frequency shift phase noise index in real time; the dynamic spectrum analysis module monitors a frequency occupancy rate through a spectrum sensing array and constructs a spectrum state matrix, and the user priority judgment module calculates a comprehensive service priority according to a service type weight, a demand level and a user mobility index; and the frequency modulation decision module generates a real-time frequency modulation strategy based on a pre-distortion matching algorithm. According to the method, channel state real-time sensing and phase error calibration are fused through a double-time-scale cross measurement architecture, a time-frequency joint response model is constructed, and multipath fading estimation errors are reduced; and based on a geometric weighted multi-dimensional user priority differential decision model, the conflict problem between the service emergency degree and the mobility is solved, so that the bandwidth allocation efficiency of the high-priority user is improved.

Description

Technical Field

[0001] The present invention belongs to the field of communications, and more specifically, particularly relates to a millimeter-wave communication system with adaptive frequency modulation. Background Art

[0002] Millimeter-wave communication technology has become a core enabling technology for 5G and future 6G networks due to its characteristics of ultra-large bandwidth and ultra-high speed. Traditional static channel compensation mechanisms are difficult to maintain stable link quality in dynamic propagation environments. Especially in dense urban areas and mobile scenarios, the high-speed mobility of user terminals causes severe jitter in the instantaneous channel state. Existing systems adopt a fixed-period channel measurement and update strategy, resulting in a delay window of more than 500 ms in channel quality assessment and being unable to capture transient channel deterioration events.

[0003] Current spectrum management schemes are mostly based on static spectrum allocation databases and are difficult to adapt to the complex dynamic interference environment in the millimeter-wave band. The broadcast spectrum sensing mechanism will generate spectrum sensing conflicts in scenarios with dense base station deployments, resulting in a deviation of more than 15% in interference temperature calculation.

[0004] In terms of service quality guarantee, existing user priority determination models one-sidedly rely on service type labels and lack the quantitative integration of user mobility characteristics and channel quality evolution laws. When high-speed mobile users (speed ≥ 80 km / h) access the network, the fixed resource allocation strategy will result in a bandwidth resource mismatch of about 30%, leading to a sharp increase in the transmission delay of emergency services. In addition, traditional frequency modulation decision algorithms mostly select carrier frequencies through exhaustive optimization. It takes more than 50 ms to complete a spectrum search within a 200 MHz bandwidth, making it difficult to meet the 10 ms-level delay requirements of URLLC services;

[0005] Based on the above, we propose a millimeter-wave communication system with adaptive frequency modulation to specifically solve the above problems. Summary of the Invention

[0006] The purpose of the present invention is to solve the deficiencies in the prior art, and a millimeter-wave communication system with adaptive frequency modulation is proposed. By fusing real-time channel state perception and phase error calibration through a dual-time-scale cross-measurement architecture, a time-frequency joint response model is constructed to reduce the multipath fading estimation error to below 3 dB; a dynamic spectrum prediction-correction mechanism driven by an LSTM network is deployed to achieve accurate prediction of the interference situation at the millisecond level, with improved prediction accuracy compared to traditional models; a multi-dimensional user priority differential decision model based on geometric weighting is used to solve the conflict between service urgency and mobility, improving the bandwidth allocation efficiency for high-priority users.

[0007] To achieve the above object, the present invention provides the following technical solutions:

[0008] A millimeter-wave communication system with adaptive frequency modulation, comprising:

[0009] Channel quality assessment module, which measures the signal-to-noise ratio γ of the link, the multipath delay spread Δτ, and the Doppler frequency shift f in real time d , and the phase noise index θ PN ;

[0010] Dynamic spectrum analysis module, which monitors the frequency occupancy rate O through a spectrum sensing array f and constructs a spectrum state matrix S f = [s ij N×M , where N is the total number of transmitter nodes deployed in the system, M is the number of spectrum monitoring points, i is the transmitter index and satisfies 1 ≤ i ≤ N, and j is the monitoring point index and satisfies 1 ≤ j ≤ M;

[0011] User priority determination module, which calculates the comprehensive service priority according to the service type weight ωt, the demand level Q level and the user mobility index ν;

[0012] Frequency modulation decision module, which generates a real-time frequency modulation strategy based on the predistortion matching algorithm, and the optimal carrier frequency selection calculation formula is:

[0013]

[0014] where η se is the spectrum efficiency function, is the frequency fitness factor;

[0015] ΨI(f k ) = ∑ m I m g(f k - f m ) is the aggregated interference function, f bw is the adjustable bandwidth reference value, B k is the channel coherence bandwidth at f k , and the entire process adopts a sliding window spectrum prediction mechanism.

[0016] Preferably, in the constructed spectrum state matrix S f , where α is the environmental attenuation coefficient, is the effective radiated power of the transmitter, and its calculation method is P PA is the output power of the power amplifier; G ant is the antenna gain, which is inversely proportional to the beam width; L feed is the feeder loss; d ij is the spatial distance between nodes, and β is the path loss exponent.

[0017] ​Preferably, the calculation formula of the spectral efficiency function is as follows:

[0018]

[0019] where γ is the signal-to-noise ratio, Δτ is the multipath delay spread, c is the fitting coefficient, R 0 is the reference value, and R curve(f) is the frequency response curve, which is defined as:

[0020]

[0021] In the formula, σf is the standard deviation of the transmitter frequency response, and f c is the center frequency.

[0022] Preferably, the calculation formula of the frequency fitness Γ(f k ) is as follows:

[0023]

[0024] In the formula, represents the influence function of the nth environmental factor on the frequency f k , ω n is the weight of the nth environmental parameter, indicating the importance of this parameter in the calculation of frequency fitness. SINR hist (f k ) represents the average signal-to-noise ratio at the frequency f k within the historical time window. SINR th is the threshold of the signal-to-noise ratio, used to normalize the historical signal-to-noise ratio. ∈ is a constant to avoid the denominator being zero;

[0025] When constructing the dynamic spectrum state matrix, the interference temperature model is introduced, and the expression is:

[0026]

[0027] In the formula, τ is the environmental coherence time, Ψf is the interference spectrum matrix predicted by the LSTM network, and Ψf(t + Δt) represents the state value of the spectrum after the time interval Δt. When the time width of Δt is larger, the value of e -τΔt tends to be closer to 0, indicating that the update of the spectrum state mainly depends on the predicted value Ψf.

[0028] Preferably, the user priority determination module adopts a three-layer decision mechanism, specifically:

[0029] The first layer discriminates the service urgency (E):

[0030] E = [1 + exp(-0.5(Q level – 3))] -1

[0031] In the formula, Q level represents the urgency level of the service, which is determined by the service type or user requirements. The larger the value, the more urgent the service. exp is an exponential function used to convert the linear input into a non-linear output. This layer maps the urgency level of the service to an urgency value within the range of [0, 1] through a logical function. The closer the value is to 1, the more urgent the service;

[0032] The second layer calculates the mobility factor (M):

[0033]

[0034] In the formula, ν represents the moving speed of the user, and ν th represents the threshold of the moving speed, which is used to standardize the mobility factor. This layer calculates the mobility factor through the ratio of the moving speed to the threshold. The closer the value is to 1, the higher the mobility of the user;

[0035] The third layer synthesizes the priority (P U ):

[0036]

[0037] In the formula, E is the service urgency, which comes from the calculation of the first layer, and M is the mobility factor, which comes from the calculation of the second layer. represents the geometric mean of the service urgency and the mobility factor, which is used to balance the influence of the two on the priority. SNR hist represents the historical data of the signal-to-noise ratio of the user, which reflects the communication quality of the user. ω1 and ω2 are weight coefficients, which are used to adjust the influence of the service urgency, the mobility factor, and the SNR hist historical data on the synthesized priority. This layer comprehensively considers the service urgency, the mobility factor, and the SNR hist historical data through weighted summation to generate the final priority of the user.

[0038] Preferably, the spectrum state matrix S introduces a prediction term:

[0039] S(t + Δt) = e -τΔt ·S(t) + (1 - e -τΔt )·Ψ(t + Δt);

[0040] Among them, τ is the environmental coherence time, Ψ(t + Δt) represents the predicted interference spectrum matrix through the LSTM network, and e -τΔt ·S(t) represents the decay term of the current spectrum state, which reflects the inertial change of the spectrum state. That is, within the time interval Δt, the current spectrum state S(t) will decay according to the exponential law.

[0041] Preferably, the channel quality assessment module is configured with a dual-time scale measurement mechanism, specifically including:

[0042] Fast time scale (Δt = 1ms), used to quickly update the signal-to-noise ratio γ and the delay spread Δτ;

[0043] Slow time scale (T = 1s), used to calibrate the phase noise index σ φ , the expression is:

[0044]

[0045] In the formula, σ φ represents the phase noise index, f 0 represents the carrier frequency, P n represents the noise power P sig represents the signal power, BER target represents the target bit error rate, erfc -1 represents the inverse complementary error function, used to convert the target bit error rate into the corresponding signal-to-noise ratio threshold.

[0046] Preferably, the frequency modulation decision module adopts a multi-scale sliding window mechanism. When an emergency spectrum occupancy change event is detected, the policy evaluation period is automatically shortened to 1 / 5 of the original time window width, and it resumes to the normal update rhythm under the spectrum stable state.

[0047] Preferably, the dynamic spectrum analysis module integrates a blind source separation unit, which identifies the dominant components in the mixed interference signals through unsupervised learning and uses the dominant interference source type as the decision basis for the weight allocation of the spectrum state matrix.

[0048] Technical effects and advantages of the present invention: The adaptive frequency modulation millimeter-wave communication system provided by the present invention, compared with the prior art, the present invention adopts a multi-dimensional channel state real-time coupling analysis mechanism. Through a dual-time scale cross-measurement architecture, a time-frequency joint response surface model is constructed, and the coupling effects of the delay spread amount and the phase noise parameters are incorporated into the channel quality evaluation index system, reducing the multi-path fading estimation error and improving the phase tracking accuracy;

[0049] Through real-time channel quality assessment and spectrum state prediction, the system can automatically sense multi-path delay, Doppler frequency shift dynamic interference, and complete carrier frequency switching and parameter optimization within milliseconds, ensuring continuous and reliable millimeter-wave link transmission in complex mobile scenarios, effectively reducing the communication interruption probability. Based on real-time interference temperature modeling and spectrum state matrix analysis, the system intelligently avoids high-interference frequency bands and dynamically allocates idle spectra, realizing the efficient reuse of spectrum space-time two-dimensional resources, and at the same time supporting multi-node collaborative frequency modulation, significantly improving the spectrum utilization rate;

[0050] Through a three - layer priority decision - making mechanism, the system can accurately identify the special needs of emergency services and high - speed mobile users, and dynamically adjust the bandwidth allocation strategy to ensure that high - priority users obtain stable quality of service in latency - sensitive scenarios. By adopting a sliding - window dynamic adjustment mechanism, when spectrum mutation or sudden interference is detected, the system can instantaneously shorten the decision - making cycle and quickly reconstruct communication parameters to meet the millisecond - level response requirements of emergency services;

[0051] Integrating blind source separation technology and a multi - scale interference analysis model, the system can actively identify the dominant components in mixed interference and optimize the spectrum weights accordingly, maintaining the noise tolerance of the communication link in scenarios of dense base - station deployment or sudden strong interference. Through dynamic optimization of power amplifier efficiency and adaptive adjustment strategies for antenna gain, the system can intelligently adjust the transmit power according to the real - time service load while ensuring communication quality, reducing the energy consumption in high - frequency communication scenarios. Brief Description of the Drawings

[0052] Figure 1 It is an architecture diagram of the millimeter - wave communication system with adaptive frequency modulation according to the present invention. Detailed Embodiments

[0053] In order to make the objectives, technical solutions and advantages of the present invention clearer and more understandable, the following further details the present invention with reference to specific embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts fall within the scope of protection of the present invention.

[0054] The present invention provides an adaptive - frequency - modulation millimeter - wave communication system as shown in Figure 1 which uses a pre - distortion compensation algorithm to correct the frequency response curve in advance at the transmitting end, combines an iterative algorithm to transform the non - linear optimization problem into a system of linear equations for solution, and obtains the globally optimal carrier frequency in real - time through matrix eigenvalue decomposition. This mechanism realizes the three - dimensional matching of channel characteristics, interference distribution and spectral efficiency, maintaining the optimal solution of spectral utilization under high - signal - to - noise ratio constraints in a dynamic environment. At the same time, the pre - distortion correction module dynamically updates the beamforming weights according to the real - time feedback frequency - offset parameters, forming a closed - loop frequency - modulation compensation chain;

[0055] The system includes:

[0056] A channel quality assessment module that measures the link signal - to - noise ratio γ, multipath delay spread Δτ, Doppler frequency shift f d , phase - noise index θ PN ;

[0057] A dynamic spectrum analysis module that monitors the frequency occupancy O f through a spectrum sensing array and constructs a spectrum status matrix Sf = [s ij N×M , where N is the total number of transmitter nodes deployed in the system, M is the number of spectrum monitoring points, i is the transmitter index and satisfies 1 ≤ i ≤ N, and j is the monitoring point index and satisfies 1 ≤ j ≤ M; the constructed spectrum status matrix S f in where α is the environmental attenuation coefficient, is the effective radiated power of the transmitter, and its calculation method is P PA is the output power of the power amplifier; G ant is the antenna gain, which is inversely proportional to the beam width; L feed is the feeder loss; d ij is the spatial distance between nodes, and β is the path loss exponent.

[0058] Furthermore, the calculation formula of the spectrum efficiency function is:

[0059]

[0060] where γ is the signal-to-noise ratio, Δτ is the multipath delay spread, c is the fitting coefficient, R 0 is the reference value, R curve(f) is the frequency response curve, which is defined as:

[0061]

[0062] In the formula, σf is the standard deviation of the transmitter frequency response, f c is the center frequency.

[0063] The dynamic spectrum analysis module integrates a blind source separation unit, which identifies the dominant components in the mixed interference signal through unsupervised learning and uses the type of dominant interference source as the decision basis for weight allocation of the spectrum status matrix.

[0064] The user priority determination module calculates the comprehensive service priority according to the service type weight ωt, the demand level Q level and the user mobility index ν; the user priority determination module adopts a three-layer decision mechanism, specifically:

[0065] The first layer discriminates the service urgency (E):

[0066] E = [1 + exp(-0.5(Q level – 3))] -1

[0067] In the formula, Q level ​Indicates the urgency level of the service, which is determined by the service type or user requirements. The larger the value, the more urgent the service. exp is an exponential function used to convert linear input into non-linear output. This layer maps the service urgency level to an urgency value within the range of [0, 1] through a logical function. The value closer to 1 indicates the more urgent the service;

[0068] The second layer calculates the mobility factor (M):

[0069]

[0070] In the formula, ν represents the user's moving speed, ν th represents the threshold of the moving speed, which is used to standardize the mobility factor. This layer calculates the mobility factor through the ratio of the moving speed to the threshold. The value closer to 1 indicates the higher the user's mobility;

[0071] The third layer synthesizes the priority (P U ):

[0072]

[0073] In the formula, E is the service urgency, which comes from the calculation of the first layer, and M is the mobility factor, which comes from the calculation of the second layer. represents the geometric mean of the service urgency and the mobility factor, which is used to balance the influence of the two on the priority. SNR hist represents the historical data of the user's signal-to-noise ratio, which reflects the user's communication quality. ω1 and ω2 are weight coefficients, which are used to adjust the influence of the service urgency, the mobility factor, and the SNR hist historical data on the synthesized priority. This layer comprehensively considers the service urgency, the mobility factor, and the SNR hist historical data through weighted summation to generate the user's final priority.

[0074] The frequency modulation decision module generates a real-time frequency modulation strategy based on the predistortion matching algorithm. The calculation formula for its optimal carrier frequency selection is:

[0075]

[0076] In the formula, where η se is the spectrum efficiency function, is the frequency fitness factor;

[0077] ΨI(f k ) = ∑ m I m g(f k -f m ) is the aggregation interference function, f bw is the adjustable bandwidth reference value, B k is f kThe coherence bandwidth of the channel, and a sliding window spectrum prediction mechanism is adopted throughout the process;

[0078] It should be noted that the frequency fitness Γ(f k ) is calculated as follows:

[0079]

[0080] In the formula, represents the influence function of the nth environmental factor on the frequency f k , ω n is the weight of the nth environmental parameter, indicating the importance of this parameter in the calculation of frequency fitness, and SINR hist (f k ) represents the average signal-to-noise ratio at the frequency f k within the historical time window, SINR th is the threshold of the signal-to-noise ratio, used to normalize the historical signal-to-noise ratio, and ∈ is a constant to avoid the denominator being zero;

[0081] In addition, the frequency modulation decision module adopts a multi-scale sliding window mechanism. When an emergency spectrum occupancy change event is detected, the policy evaluation period is automatically shortened to 1 / 5 of the original time window width and restored to the normal update rhythm under the spectrum stable state;

[0082] As described above, the channel quality evaluation module is configured with a dual-time scale measurement mechanism, specifically including:

[0083] Fast time scale (Δt = 1ms), used to quickly update the signal-to-noise ratio γ and the delay spread Δτ;

[0084] Slow time scale (T = 1s), used to calibrate the phase noise index σ φ , and the expression is:

[0085]

[0086] In the formula, σ φ represents the phase noise index, f 0 represents the carrier frequency, P n represents the noise power P sig represents the signal power, BER target represents the target bit error rate, and erfc -1 represents the inverse complementary error function, used to convert the target bit error rate into the corresponding signal-to-noise ratio threshold.

[0087] When constructing the dynamic spectrum state matrix, the interference temperature model is introduced, and the expression is:

[0088]

[0089] where τ is the environmental coherence time, Ψf is the interference spectrum matrix predicted by the LSTM network, Ψf(t + Δt) represents the state value of the spectrum after the time interval Δt. When the time width of Δt is larger, the value of e -τΔt tends to approach 0, indicating that the update of the spectrum state mainly depends on the predicted value Ψf;

[0090] The prediction term is introduced into the spectrum state matrix S:

[0091] S(t + Δt) = e -τΔt ·S(t) + (1 - e -τΔt )·Ψ(t + Δt);

[0092] where τ is the environmental coherence time, Ψ(t + Δt) represents the interference spectrum matrix predicted by the LSTM network, and e -τΔt ·S(t) represents the attenuation term of the current spectrum state. This attenuation term reflects the inertial change of the spectrum state, that is, within the time interval Δt, the current spectrum state S(t) will decay according to the exponential law.

[0093] In summary, the present invention adopts a multi-dimensional channel state real-time coupling analysis mechanism. Through a dual-time scale cross-measurement architecture, a time-frequency joint response surface model is constructed. The coupling effects of the delay spread and phase noise parameters are incorporated into the channel quality evaluation index system, reducing the multi-path fading estimation error and improving the phase tracking accuracy;

[0094] Through real-time channel quality assessment and spectrum state prediction, the system can automatically sense multi-path delay and Doppler frequency shift dynamic interference, and complete carrier frequency switching and parameter optimization within milliseconds, ensuring continuous and reliable millimeter-wave link transmission in complex mobile scenarios, effectively reducing the probability of communication interruption. Based on real-time interference temperature modeling and spectrum state matrix analysis, the system intelligently avoids high-interference frequency bands and dynamically allocates idle spectra, realizing the efficient reuse of spectrum space-time two-dimensional resources, and at the same time supporting multi-node cooperative frequency modulation, significantly improving the spectrum utilization rate;

[0095] Through a three-layer priority decision mechanism, the system can accurately identify the special needs of emergency services and high-speed mobile users, and dynamically adjust the bandwidth allocation strategy to ensure that high-priority users obtain stable service quality in delay-sensitive scenarios. By adopting a sliding window dynamic adjustment mechanism, when spectrum mutation or sudden interference is detected, the system can instantaneously shorten the decision-making cycle and quickly reconstruct communication parameters to ensure the millisecond-level response requirements of emergency services;

[0096] Integrating the blind source separation technology and the multi-scale interference analysis model, the system can actively identify the dominant components in the mixed interference and optimize the spectrum weights accordingly, maintaining the anti-noise tolerance of the communication link in scenarios of dense base station deployment or sudden strong interference. Through the dynamic optimization of power amplifier efficiency and the adaptive adjustment strategy of antenna gain, the system can intelligently adjust the transmission power according to the real-time service load on the premise of ensuring communication quality, reducing the energy consumption in high-frequency communication scenarios.

[0097] Finally, it should be noted that the above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions recorded in the foregoing embodiments or perform equivalent replacements for some of the technical features. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. An adaptive frequency modulation millimeter wave communication system, characterized in that: include: Channel quality assessment module, real-time measurement of link signal-to-noise ratio γ, multipath delay spread Δτ, Doppler frequency shift f d , phase noise index θ PN ; Dynamic spectrum analysis module, monitoring frequency occupancy through spectrum sensing array f And construct the spectrum state matrix S f =[s ij ] N×M , where N is the total number of transmitter nodes deployed in the system, M is the number of spectrum monitoring points, i is the transmitter index and satisfies 1≤i≤N, and j is the monitoring point index and satisfies 1≤j≤M; User priority determination module, based on the service type weight ωt, demand level Q level and user mobility index ν to calculate the comprehensive service priority; The frequency modulation decision module generates a real-time frequency modulation strategy based on the pre-distortion matching algorithm. The optimal carrier frequency selection calculation formula is: In the formula, η se is the spectrum efficiency function, is the frequency fitness factor; ΨI(f k )=∑ m I m g(f k -f m ) is the aggregate interference function, f bw is the adjustable bandwidth reference value, B k f k The channel coherence bandwidth is determined, and the whole process adopts a sliding window spectrum prediction mechanism.

2. The adaptive frequency modulation millimeter wave communication system according to claim 1, characterized in that: The constructed spectrum state matrix S f middle, Where α is the environmental attenuation coefficient, is the effective radiated power of the transmitter, which is calculated as follows: P PA is the output power of the power amplifier; G ant is the antenna gain, which is inversely proportional to the beam width; L feed is the feeder loss; d ij is the spatial distance between nodes, and β is the path loss exponent.

3. The adaptive frequency modulation millimeter wave communication system according to claim 1, characterized in that: The calculation formula of the spectrum efficiency function is: Where γ is the signal-to-noise ratio, Δτ is the multipath delay spread, c is the fitting coefficient, R0 is the reference value, and R curve(f) is the frequency response curve, defined as: Where σf is the standard deviation of the transmitter frequency response, f c is the center frequency.

4. The adaptive frequency modulation millimeter wave communication system according to claim 1, characterized in that: The frequency adaptability The calculation formula is as follows: In the formula, Indicates the effect of the nth environmental factor on the frequency f k The influence function of n is the weight of the nth environmental parameter, indicating the importance of this parameter in the frequency fitness calculation, SINR hist (f k ) indicates that within the historical time window, the frequency f k Average signal-to-noise ratio, SINR th is the threshold of the signal-to-noise ratio, which is used to normalize the historical signal-to-noise ratio, ∈ is a constant to avoid the denominator being 0; The interference temperature model is introduced when constructing the dynamic spectrum state matrix, and the expression is: In the formula, τ is the environmental coherence time, Ψf is the interference spectrum matrix predicted by the LSTM network, and Ψf(t+Δt) represents the state value of the spectrum after the time interval Δt. -τΔt The closer the value of is to 0, the more the update of the spectrum state depends on the predicted value Ψf.

5. The adaptive frequency modulation millimeter wave communication system according to claim 1, characterized in that: The user priority determination module adopts a three-layer decision-making mechanism, specifically: The first level determines the urgency of the business (E): E=1+exp(-0.5(Q level –3))] -1 ; In the formula, Q level Indicates the urgency level of the business, which is determined by the business type or user demand. The larger the value, the more urgent the business. exp is an exponential function, which is used to convert linear input into nonlinear output. This layer maps the business urgency level to an urgency value in the range of [0,1] through a logical function. The closer the value is to 1, the more urgent the business is. The second layer calculates the mobility factor (M): In the formula, ν represents the user's moving speed, ν th The threshold value of the moving speed is used to normalize the mobility factor. This layer calculates the mobility factor by the ratio of the moving speed to the threshold value. The closer the value is to 1, the higher the mobility of the user. The third layer comprehensive priority (P U ): Where E is the service urgency, which comes from the first layer calculation, and M is the mobility factor, which comes from the second layer calculation. Represents the geometric mean of service urgency and mobility factor, used to balance the impact of the two on priority, SNR hist Represents the historical data of the user's signal-to-noise ratio, reflecting the user's communication quality. ω1 and ω2 are weight coefficients, which are used to adjust the service urgency and mobility factors and SNR respectively. hist The impact of historical data on the comprehensive priority. This layer takes into account the service urgency, mobility factor and SNR through weighted summation. hist Historical data,generates the final priority of users.

6. The adaptive frequency modulation millimeter wave communication system according to claim 4, characterized in that: The spectrum state matrix S introduces the prediction term: S(t+Δt)=e -τΔt ·S(t)+(1-e -τΔt )·Ψ(t+Δt); Where τ is the environmental coherence time, Ψ(t+Δt) represents the interference spectrum matrix predicted by the LSTM network, and e -τΔt S(t) represents the attenuation term of the current spectrum state, which reflects the inertial change of the spectrum state. That is, within the time interval Δt, the current spectrum state S(t) will decay according to the exponential law.

7. The adaptive frequency modulation millimeter wave communication system according to claim 1, characterized in that: The channel quality assessment module is configured with a dual time scale measurement mechanism, specifically including: Fast time scale (Δt = 1ms), used to quickly update the signal-to-noise ratio γ and delay spread Δτ; Slow time scale (T = 1s), used to calibrate the phase noise index σ φ , the expression is: In the formula, σ φ represents the phase noise index, f0 represents the carrier frequency, P n Represents the noise power P sig Indicates signal power, BER target Indicates the target bit error rate, erfc -1 represents the inverse complementary error function, which is used to convert the target bit error rate into the corresponding signal-to-noise ratio threshold.

8. The adaptive frequency modulation millimeter wave communication system according to claim 1, characterized in that: The frequency modulation decision module adopts a multi-scale sliding window mechanism. When an emergency spectrum occupancy change event is detected, the strategy evaluation cycle is automatically shortened to 1 / 5 of the original time window width, and the normal update rhythm is restored when the spectrum is stable.

9. The adaptive frequency modulation millimeter wave communication system according to claim 2, characterized in that: The dynamic spectrum analysis module is integrated with a blind source separation unit, which identifies the dominant component in the mixed interference signal through unsupervised learning, and uses the type of the dominant interference source as the decision basis for the weight allocation of the spectrum state matrix.

Images