IRS-assisted cognitive unmanned aerial vehicle network perception and transmission method in low signal-to-noise ratio environment

Abstract

The invention provides a sensing and transmission method of an intelligent reflecting surface (IRS) aided cognitive unmanned aerial vehicle network (CUAVN) in a low signal-to-noise ratio environment, aiming at the pain point that a signal of a ground primary base station (PT) received by a UAV is relatively weak and is in a signal-to-noise ratio environment, so that a traditional energy detection (ED) algorithm is difficult to play a role in the aspect of spectrum sensing, and aiming at the pain point that the UAV is in the signal-to-noise ratio environment, the invention provides the sensing and transmission method of the intelligent reflecting surface (IRS) aided cognitive unmanned aerial vehicle network (CUAVN) in the low signal-to-noise ratio environment. The method comprises the steps of designing a spectrum sensing scheme based on phase difference (PD) distribution, establishing an optimization model for maximizing the throughput of CUAVNs, and developing an IRS phase shift matrix optimization algorithm in a sensing stage, an IRS phase shift matrix optimization algorithm in a transmission stage, an unmanned aerial vehicle track optimization algorithm and a sensing time joint optimization algorithm. A numerical result shows the superiority of spectrum sensing and CUAVNs communication performance under the low signal-to-noise ratio of the algorithm.

Description

Technical Field

[0001] The present invention mainly relates to the field of intelligent reflective surface-assisted cognitive drone network technology, and in particular to a low-complexity solution for cognitive drone network perception and transmission in a low signal-to-noise ratio environment. Background Art

[0002] In the post-5G era, with the massive growth of wireless devices, people are demanding increasingly higher wireless communication speeds. In recent years, drones have become increasingly popular among researchers due to their high maneuverability, ease of deployment, and ability to virtually ignore complex ground environments. However, conventional drone communications do not meet practical communication requirements. Therefore, cognitive radio technology has been introduced into drone communications to form cognitive drone networks. These networks combine the advantages of drones with the opportunistic spectrum access capabilities of cognitive radio, ensuring that interference to primary users during communication is reduced, thereby ensuring the correctness of the communication process. Traditional spectrum sensing schemes, such as energy detection (ED), perform poorly in low signal-to-noise ratio environments and fail to meet the relevant sensing performance requirements.

[0003] In recent years, intelligent reflecting surfaces (IRS) have emerged as a promising technology, attracting increasing research. Their ability to reflect the energy of incoming signals and programmatically change the angle of the reflective elements offers hope for breakthroughs in 6G technology. By strategically deploying IRSs, they can simultaneously assist in the perception and transmission of cognitive drone networks, significantly improving system throughput.

[0004] Existing technologies mostly focus on IRS-assisted drone communication solutions in normal environments, and most of these spectrum sensing solutions use ED. However, designing for more complex environments, such as those with low signal-to-noise ratios, remains an urgent challenge. Summary of the Invention

[0005] In view of this, the purpose of this invention is to provide a novel technical solution for IRS-assisted cognitive UAV networks under low signal-to-noise ratio (SNR). This solution aims to improve the system's perception performance under low SNR conditions by detecting the cosine characteristics of phase difference distribution. Furthermore, we provide an optimization solution that combines the phase shift matrix of the IRS reflection unit, perception time, and UAV trajectory, with the goal of maximizing system throughput.

[0006] In order to achieve the above-mentioned object of the invention, the present invention provides an IRS-assisted cognitive drone network perception and transmission method under low signal-to-noise ratio, comprising:

[0007] S1. Establish a system model for IRS-assisted cognitive UAV network perception and transmission methods in low signal-to-noise ratio environments;

[0008] S2. Establish a spectrum sensing scheme based on phase difference distribution to ensure excellent sensing performance under low signal-to-noise ratio;

[0009] S3. Divide the UAV's flight time into N time slots. Each time slot is divided into a perception phase and a transmission phase. The duration of the perception phase is determined by optimizing the perception time. The phase shift matrices for the perception phase and the transmission phase are optimized separately.

[0010] S4. Establish channel models, including the channel model from the transmitting base station PT to the IRS; the channel model from the IRS to the drone; the channel model from the transmitting base station PT to the drone; the channel model from the drone to the IRS; the channel model from the IRS to the primary user PU; and the channel model from the drone to the primary user PU.

[0011] S5. Develop a mathematical model to maximize system throughput by jointly optimizing the sensing and transmission phase shift matrices, UAV trajectories, and sensing time, while ensuring that the interference to the primary user (PU) is below a threshold.

[0012] S6. Solve the system throughput maximization problem through alternating iterations;

[0013] Preferably, in step S1, the cognitive drone primarily transmits when the base station PT is not using the spectrum. Therefore, in addition to the link for assisting the drone transmission, the IRS also includes a link for assisting the drone in sensing the PT's spectrum usage status. The base station PT, the drone, and the secondary user all have single antennas.

[0014] Preferably, in step S2, a cosine feature detection scheme is proposed to replace the traditional spectrum sensing scheme through the phase difference distribution function under the Rice channel; the probability density function derived from the auxiliary function between the two samples is:

[0015]

[0016] Where β represents the Rice factor, represents the average signal-to-noise ratio; based on the obvious difference in the cosine characteristics of the probability density function of the PD distribution under additive Gaussian white noise, using cosine feature recognition detection (CRD), according to the central limit theorem, the detection threshold and detection probability can be determined as: Q(x) represents the Gaussian Q function, where m is the number of samples, P fis the false alarm rate, Δ=0.05π; in practice, the value of m cannot exceed τfs, and in the present invention, m=τfs; for a given detection probability P d , P f It can be expressed as: This solution replaces the traditional spectrum sensing solution;

[0017] Preferably, in step S3, the entire flight time of the cognitive drone is set to T, which is divided into N time slots, and the time occupied by each time slot is: Each time slot is further divided into the sensing phase τ s and the transmission phase τ t , τ=τ s +τ t , where τ s On the basis of satisfying the perception performance, the optimal perception time of each time slot is obtained by bisection method; IRS adopts M=M x ×M z The UPA array of the nth time slot sensing phase phase shift matrix and the transmission phase phase shift matrix can be expressed as:

[0018]

[0019] Preferably, in step S4, all channels adopt the Rice channel model, so the h from the UAV to the IRS is ui Taking the channel as an example, the channel model is:

[0020]

[0021] Where ρ is the path loss at the reference distance d = 1m, α is the path loss factor, d ui [n] represents the distance from the UAV to the IRS in the nth time slot, and its calculation formula is: Z U and Z R Represent the distances of the UAV and IRS from the ground, q and W respectively I are the horizontal coordinates of the UAV and IRS respectively. β1 represents the Rice factor of the channel. and Indicates the line-of-sight and non-line-of-sight channel gains of the Ricean channel in the nth time slot;

[0022] Preferably, in step S5, when the PT is idle and the probability of being correctly detected is P(H0)(1-P f ), P(H0) is the probability that the PT is idle; the received signal-to-noise ratio of the kth secondary user in the nth time slot is:

[0023]

[0024] Where K represents the number of secondary users, P U represents the transmission power of the UAV, σ 2 represents the noise power. In practical applications, in order to protect the primary user PU during UAV transmission, the interference must be less than a certain threshold:

[0025]

[0026] γ PU It represents the receiving signal-to-noise ratio of the primary user PU during the UAV transmission phase; since the actual speed of the UAV is subject to certain restrictions, and the present invention assumes that the starting point q0 and the ending point q F Fixed, so there are the following restrictions:

[0027] ||q[n+1]-q[n]|| 2 ≤D 2 ,n=1,...,N-1,

[0028] ||q[N]-q F || 2 ≤D 2 ,q[1]=q0,

[0029] By combining the UAV trajectory, perception time, and the IRS phase shift matrix of the perception and transmission phases, the proposed optimization problem can be expressed as:

[0030]

[0031] P d ≥P d0 ,

[0032]

[0033] ||q[n+1]-q[n]|| 2 ≤D 2 ,n=1,...,N-1,

[0034] ||q[N]-q F || 2 ≤D 2 ,q[1]=q0,

[0035] where θ s,m [n], θ t,m [n] is the constraint on the IRS phase shift matrix, and its value range is 0 to 2π, P d ≥P d0 It is a constraint on the probability of target detection;

[0036] Preferably, in step S6, since the problem is non-convex and difficult to solve, it is decomposed into four sub-problems: optimizing the phase shift matrix in the perception stage, optimizing the phase shift matrix in the transmission stage, optimizing the trajectory, and optimizing the perception time. The four sub-problems are iterated alternately until convergence. The specific steps include:

[0037] S601. Phase shift matrix φ at a given transmission stage t , UAV trajectory Q, perception time τ s Based on this, the phase shift matrix φ in the perception stage is optimized s Since only the direct link from UAV to PT and the cascade link from UAV to IRS to PT need to be considered, the phase alignment method of the cascade link and the direct link is adopted, namely:

[0038] θ i,n =arg(H pu [n])+arg(H pi [i])-arg(H iu [i,n]),

[0039] S602. Phase shift matrix φ at a given perception stage s , UAV trajectory Q, perception time τ s Based on the optimization of the transmission phase shift matrix φ t ; Here we use a low-complexity PDDGP algorithm to solve: introduce auxiliary variables g(θ,s)=I n +sI th , the phase shift constraint in the perception stage can be rewritten as: g(θ,s)=0,s≥0; the augmented Lagrangian function can be written as:

[0040]

[0041] Given v,p, for R v,p (θ,s) finds the first-order partial derivative with respect to θ, and introduces the auxiliary variable F: F = log2(1+γ k,n );

[0042]

[0043] Since this result is too complicated, we first calculate the first-order partial derivative of the first half:

[0044]

[0045] Where A1=|h usk | 2 , C1=diag(h isk H )h ui hui H diag(h isk ); The form and Similar, no specific expression is given here; θ l Need to follow Move with a certain step size, at the i-th iteration, μ is an auxiliary variable in the iterative process; in order to obtain the next iteration θ l , which needs to be projected:

[0046]

[0047] s=max{I th -I n ,0}, alternately iterate s and θ l Until convergence, the subproblem is solved;

[0048] S603. For the third sub-problem, introduce auxiliary variables Z = {u k,n ,g n ,e n ,j n ,r k,n ,ω n ,χ n ,h n ,k n ,o n The original problem can be rewritten as:

[0049]

[0050] r k,n ≤δ T uk,n H uk,n δ uk,n ,

[0051]

[0052] 0≤χ n ≤d n δ t,n T H t,n δ t,n ,

[0053] k n ≤1-Q(h n ),

[0054] δ T p,n H p,n δ p,n ≤I th σ2 / d 1,n ,

[0055]

[0056] ||q[n+1]-q[n]|| 2 ≤D 2 ,n=1,...,N-1,

[0057] ||q[N]-q F || 2 ≤D 2 ,q[1]=q0,

[0058] in

[0059] δ T uk,n =[u k,n ,e n ],c n =(1 / K)P(H0)(τ-τ s ), H uk,n =AA H . H t,n 、H p,n and δ p,n , δ t,n With H uk,n , δ uk,n The structure is similar and can be generated in the same way; next, the continuous convex approximation method is used to perform a first-order Taylor expansion on the non-convex constraints:

[0060]

[0061] r k,n ≤-(δ uk,n (l) ) T H uk,n δ uk,n (l) +2R{(δ uk,n (l) ) T H uk,n δ uk,n (l)},

[0062] χ n ≤d n (-(δ t,n (l) ) T H t,n δ t,n (l) +2R{(δ t,n (l) )T H t,n δ t,n (l)}),

[0063] where {u k,n (l) ,g n (l) ,e n (l) ,k n (l) ,ω n (l) ,q n (l)}, δ uk,n (l) =[u k,n (l) ,e n (l) ] T , δ t,n (l) =[g n (l) ,e n (l) ] T represents the feasible point of the lth iteration, and then the problem can be rewritten as:

[0064]

[0065]

[0066] k n ≤1-Q(h n ),

[0067] δ T p,n H p,n δ p,n ≤I th σ 2 / d 1,n ,

[0068]

[0069] ||q[n+1]-q[n]|| 2 ≤D 2 ,n=1,...,N-1,

[0070] ||q[N]-q F || 2 ≤D 2 ,q[1]=q0,

[0071] This problem is a convex optimization problem and can be solved using CVX.

[0072] S604. Given φ t 、φ s , Q, since the perception time must meet the constraints: The fourth sub-problem can be rewritten as follows:

[0073]

[0074] where G[n] does not contain τ s [n], this problem can be solved using binary search.

[0075] S605. Using alternating iterations, the four sub-problems of perception phase shift matrix optimization, transmission phase shift matrix optimization, drone trajectory optimization, and perception time optimization are alternately iterated until the algorithm converges.

[0076] Compared with the prior art, the present invention has the following beneficial effects:

[0077] The present invention introduces intelligent reflective surface technology and phase difference distribution into cognitive drone networks, enabling intelligent reflective surfaces to assist both perception and communication, and enabling drone networks to have good perception and communication performance under low signal-to-noise ratios. The present invention proposes a low-complexity PDDGP algorithm to replace the traditional SDR algorithm to solve the IRS phase shift matrix, thereby reducing the overall complexity of the algorithm. Based on the existing phase difference distribution perception method, the present invention proposes a cosine feature detection algorithm under Ricean channels, combines it with drones and intelligent reflective surfaces, and maximizes the throughput of the drone network by alternately optimizing the phase shift matrix in the transmission phase, the phase shift matrix in the perception phase, the perception time, and the drone trajectory. BRIEF DESCRIPTION OF THE DRAWINGS

[0078] In order to more clearly illustrate the technical solutions of the present invention, the following briefly introduces the drawings used in the embodiments. It is obvious that the drawings only represent preferred embodiments of the present invention, and a person skilled in the art can derive other drawings based on these drawings without additional creative work. Figure 1 This is a model diagram of cognitive radio network perception and optimization assisted by intelligent reflective surfaces in low signal-to-noise ratio scenarios provided by an embodiment of the present invention;

[0079] Figure 2 This is an optimized trajectory diagram comparing a low-complexity algorithm for cognitive radio network perception and optimization assisted by intelligent reflective surfaces under low signal-to-noise ratios provided by an embodiment of the present invention with other benchmark algorithms;

[0080] Figure 3The embodiment of the present invention provides a low-complexity algorithm and energy detection algorithm for intelligent reflector-assisted cognitive radio network perception and optimization under low signal-to-noise ratio, and a graph of achievable rates in each time slot as the trajectory changes with different numbers of intelligent reflector units.

[0081] Figure 4 This is a graph showing the convergence performance of a low-complexity algorithm for cognitive radio network perception and optimization assisted by intelligent reflective surfaces under low signal-to-noise ratios, provided by an embodiment of the present invention, when deployed at different locations in an IRS.

[0082] Figure 5 This is a graph showing how the reachable rate of a drone at different locations changes with the intelligent reflective unit under a low-complexity algorithm and energy detection algorithm assisted by an intelligent reflective surface for cognitive radio network perception and optimization under low signal-to-noise ratio conditions, as provided by an embodiment of the present invention;

[0083] Figure 6 This is a graph showing the change in achievable rate of a low-complexity algorithm and energy detection algorithm for cognitive radio network perception and optimization assisted by an intelligent reflector under low signal-to-noise ratio under different PT transmission powers provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0084] The principles and features of the present invention are described below with reference to the accompanying drawings. The enumerated embodiments are only used to explain the present invention and are not intended to limit the scope of the present invention.

[0085] This embodiment provides an IRS-assisted cognitive drone network perception and communication method under low signal-to-noise ratio. The specific steps are as follows:

[0086] S1. Figure 1 As shown in the figure, a system model of a smart reflector-assisted cognitive UAV network in a low signal-to-noise ratio scenario is built. The model includes a UAV for perception and communication, k secondary users (SRs), a primary user (PU), a base station (PT), and an intelligent reflector (IRS) deployed on a high-rise building. To simulate the actual communication environment, some trees are designed to block the IRS and PT. The IRS assists in both perception and communication. To protect the primary user (PU), a direct link from the UAV to the PU, as well as cascaded links from the UAV to the IRS and from the IRS to the PU, are established.

[0087] S2. Based on the phase difference distribution function under the Ricean channel, a cosine feature detection scheme is proposed to replace the traditional spectrum sensing scheme. The probability density function derived from the auxiliary function between two samples is:

[0088]

[0089] Where β represents the Rice factor, Represents the average signal-to-noise ratio. Based on the obvious difference in the cosine characteristics of the probability density function of the PD distribution under additive Gaussian white noise, using cosine feature detection (CRD), according to the central limit theorem, the detection threshold and detection probability can be determined as: Q(x) represents the Gaussian Q function, where m is the number of samples, P f is the false alarm rate, Δ=0.05π; in practice, the value of m cannot exceed τfs, and in the present invention, m=τfs; for a given detection probability P d , P f It can be expressed as: This solution replaces the traditional spectrum sensing solution;

[0090] S3. The entire flight time of the cognitive drone is set to T, which is divided into N time slots. The time occupied by each time slot is: Each time slot is further divided into the sensing phase τ s and the transmission phase τ t , τ=τ s +τ t , where τ s On the basis of satisfying the perception performance, the optimal perception time of each time slot is obtained by bisection method. x ×M z The UPA array of the nth time slot sensing phase phase shift matrix and the transmission phase phase shift matrix can be expressed as:

[0091]

[0092] S4. All channels adopt the Rice channel model, so the distance from the UAV to the IRS is h ui Taking the channel as an example, the channel model is:

[0093]

[0094] Where ρ is the path loss at the reference distance d = 1m, α is the path loss factor, d ui [n] represents the distance from the UAV to the IRS in the nth time slot, and its calculation formula is: Z U and Z R Represent the distances of the UAV and IRS from the ground, q and W respectively I They are the horizontal coordinates of the UAV and IRS respectively; β1 represents the Rice factor of the channel, and Indicates the line-of-sight and non-line-of-sight channel gains of the Ricean channel in the nth time slot;

[0095] S5. When the PT is idle and the probability of being correctly detected is P(H0)(1-P f ), P(H0) is the probability that the PT is idle; the received signal-to-noise ratio of the kth secondary user in the nth time slot is:

[0096]

[0097] Where K represents the number of secondary users, P U represents the transmission power of the UAV, σ 2 represents the noise power. In practical applications, in order to protect the primary user PU during UAV transmission, the interference must be less than a certain threshold:

[0098]

[0099] γ PU It represents the receiving signal-to-noise ratio of the primary user PU during the UAV transmission phase; since the actual speed of the UAV is subject to certain restrictions, and the present invention assumes that the starting point q0 and the ending point q F Fixed, so there are the following restrictions:

[0100] ||q[n+1]-q[n]|| 2 ≤D 2 ,n=1,...,N-1,

[0101] ||q[N]-q F || 2 ≤D 2 ,q[1]=q0,

[0102] By combining the UAV trajectory, perception time, and the IRS phase shift matrix of the perception and transmission phases, the proposed optimization problem can be expressed as:

[0103]

[0104]

[0105] P d ≥P d0 ,

[0106]

[0107] ||q[n+1]-q[n]|| 2 ≤D 2 ,n=1,...,N-1,

[0108] ||q[N]-q F || 2≤D 2 ,q[1]=q0,

[0109] where θ s,m [n], θ t,m [n] is the constraint on the IRS phase shift matrix, and its value range is 0 to 2π, P d ≥P d0 It is a constraint on the probability of target detection;

[0110] S6. Since the proposed problem is non-convex and difficult to solve, it is decomposed into four sub-problems: optimizing the phase shift matrix in the perception phase, optimizing the phase shift matrix in the transmission phase, optimizing the trajectory, and optimizing the perception time. The four sub-problems are iterated alternately until convergence. The specific steps include:

[0111] S601. Phase shift matrix φ at a given transmission stage t , UAV trajectory Q, perception time τ s Based on this, the phase shift matrix φ in the perception stage is optimized s Since only the direct link from UAV to PT and the cascade link from UAV to IRS to PT need to be considered, the phase alignment method of the cascade link and the direct link is adopted, namely:

[0112] θ i,n =arg(H pu [n])+arg(H pi [i])-arg(H iu [i,n]),

[0113] S602. Phase shift matrix φ at a given perception stage s , UAV trajectory Q, perception time τ s Based on the optimization of the transmission phase shift matrix φ t ; Here we use a low-complexity PDDGP algorithm to solve: introduce auxiliary variables g(θ,s)=I n +sI th , the phase shift constraint in the perception stage can be rewritten as: g(θ,s)=0,s≥0; the augmented Lagrangian function can be written as:

[0114]

[0115] Given v,p, for R v,p (θ,s) finds the first-order partial derivative with respect to θ, and introduces the auxiliary variable F: F = log2(1+γ k,n ),

[0116]

[0117] Since this result is too complicated, we first calculate the first-order partial derivative of the first half:

[0118]

[0119] Where A1=|h usk | 2 , C1=diag(h isk H )h ui h ui H diag(h isk ); The form and Similar, no specific expression is given here; θ l Need to follow Move with a certain step size, at the i-th iteration, μ is an auxiliary variable in the iterative process; in order to obtain the next iteration θ l , which needs to be projected:

[0120]

[0121] s=max{I th -I n ,0}, alternately iterate s and θ l Until convergence, the subproblem is solved;

[0122] S603. For the third sub-problem, introduce auxiliary variables Z = {u k,n ,g n ,e n ,j n ,r k,n ,ω n ,χ n ,h n ,k n ,o n The original problem can be rewritten as:

[0123]

[0124] r k,n ≤δ T uk,n H uk,n δ uk,n ,

[0125]

[0126] 0≤χ n ≤d n δ t,n T Ht,n δ t,n ,

[0127] k n ≤1-Q(h n ),

[0128] δ T p,n H p,n δ p,n ≤I th σ 2 / d 1,n ,

[0129]

[0130] ||q[n+1]-q[n]|| 2 ≤D 2 ,n=1,...,N-1,

[0131] ||q[N]-q F || 2 ≤D 2 ,q[1]=q0,

[0132] where δ T uk,n =[u k,n ,e n ],c n =(1 / K)P(H0)(τ-τ s ), H uk,n =AA H ;H t,n 、H p,n and δ p,n , δ t,n With H uk,n , δ uk,n The structure is similar and can be generated in the same way; next, the continuous convex approximation method is used to perform a first-order Taylor expansion on the non-convex constraints:

[0133]

[0134] where {u k,n (l) ,g n (l) ,e n (l) ,k n (l) ,ω n (l) ,q n (l)}, δ uk,n (l) =[uk,n (l) ,e n (l) ] T , δ t,n (l) =[g n (l) ,e n (l) ] T represents the feasible point of the lth iteration, and then the problem can be rewritten as:

[0135]

[0136]

[0137] k n ≤1-Q(h n ),

[0138] δ T p,n H p,n δ p,n ≤I th σ 2 / d 1,n ,

[0139]

[0140] ||q[n+1]-q[n]|| 2 ≤D 2 ,n=1,...,N-1,

[0141] ||q[N]-q F || 2 ≤D 2 ,q[1]=q0,

[0142] This problem is a convex optimization problem and can be solved using CVX;

[0143] S604. Given φ t 、φ s , Q, since the perception time must meet the constraints: The variants are: The fourth sub-problem can be rewritten as follows:

[0144]

[0145] where G[n] does not contain τ s [n], this problem can be solved by binary search;

[0146] S605. Using alternating iterations, the four sub-problems of perception phase shift matrix optimization, transmission phase shift matrix optimization, drone trajectory optimization, and perception time optimization are alternately iterated until the algorithm converges.

[0147] Figure 2 The trajectories of the proposed scheme and other benchmark schemes are shown when T = 40s and M = 100. The drone balances the positional relationship between the IRS, SR, and PT during flight and hovers above the IRS. The energy detection scheme and the proposed scheme have similar trajectories. The scheme without IRS first flies near the SR and then completes the flight mission.

[0148] Figure 3 The rates of the proposed scheme and the energy detection scheme are given based on the optimal trajectory. It can be seen that the rate of the proposed scheme is higher than that of the energy detection scheme when far away from the PT and IRS. Only when close to the IRS do the rates of the two schemes become the same.

[0149] Figure 4 Performance graphs of the proposed low-complexity algorithm and the SDR algorithm are presented for different IRS deployment locations. The proposed algorithm converges within five iterations. When the IRS is deployed near the PT, it achieves better perception performance; when deployed near the SR, it achieves better transmission performance. However, when deployed near the PU, both perception and transmission capabilities are suppressed due to the need to protect the PU from excessive interference.

[0150] Figure 5 The variation of the achievable rate of the UAV at different locations with the number of IRSs is given. When the UAV is far away from the IRS and the PT, the transmission rate of the energy detection scheme is almost zero. When the UAV is above the IRS, the rate of the energy detection scheme is very low when the number of IRSs is small, and gradually approaches the rate of the proposed scheme as the number of IRSs increases. When the UAV is near the PT, the rates of the two schemes are also the same because the perception performance is the same at this time.

[0151] Figure 6 The results show how the proposed scheme and the energy detection scheme change with the number of IRSs at different transmit powers. The rate does not change much at the same number of IRSs, whether PT = 30mW or PT = 80mW. The rate of the energy detection scheme is much lower than that of the proposed scheme when PT = 30mW and the number of IRSs is small, but increases with the number of IRSs. When PT = 80mW, the rate of energy detection is greatly improved when the number of IRSs is small, but it is still lower than that of the proposed scheme.

[0152] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. IRS-assisted cognitive UAV network perception and transmission method in low signal-to-noise ratio environment, characterized by: The following steps are involved: S1. Establish a system model for IRS-assisted cognitive UAV network perception and transmission in low signal-to-noise ratio environments; S2. Establish a spectrum sensing scheme based on phase difference distribution to ensure excellent sensing performance under low signal-to-noise ratio; S3. Divide the UAV's flight time into N time slots. Each time slot is divided into a perception phase and a transmission phase. The perception time is optimized to determine the duration of the perception phase. The phase shift matrices for the perception phase and the transmission phase are optimized separately. S4. Establish channel models, including the channel model from the transmitting base station PT to the IRS; the channel model from the IRS to the drone; the channel model from the transmitting base station PT to the drone; the channel model from the drone to the IRS; the channel model from the IRS to the primary user PU; and the channel model from the drone to the primary user PU. S5. Develop a mathematical model to maximize system throughput by jointly optimizing the sensing and transmission phase shift matrices, UAV trajectories, and sensing time, while ensuring that the interference to the primary user (PU) is below a threshold. S6. Solve the system throughput maximization problem through alternating iterations.

2. The IRS-assisted cognitive drone network perception and transmission method in a low signal-to-noise ratio environment according to claim 1 is characterized in that: In step S1, the cognitive drone mainly transmits when the base station PT is not using the spectrum. In addition to the link that assists the drone in transmission, the IRS also includes a link that assists the drone in sensing the spectrum status of the PT. Among them, the base station PT, the drone, and the secondary user all have single antennas.

3. The IRS-assisted cognitive drone network perception and transmission method in a low signal-to-noise ratio environment according to claim 1 is characterized in that: In step S2, a cosine feature detection scheme is proposed to replace the traditional spectrum sensing scheme through the phase difference distribution function under the Rice channel; the probability density function derived from the auxiliary function between the two samples is: Where β represents the Rice factor, represents the average signal-to-noise ratio; based on the obvious difference in the cosine characteristics of the probability density function of the PD distribution under additive Gaussian white noise, using cosine feature recognition detection (CRD), according to the central limit theorem, the detection threshold and detection probability can be determined as: Q(x) represents the Gaussian Q function, where m is the number of samples, P f is the false alarm rate, Δ=0.05π; in practice, the value of m cannot exceed τfs, and in the present invention, m=τfs; for a given detection probability P d , P f It can be expressed as: This solution replaces the traditional spectrum sensing solution.

4. The IRS-assisted cognitive drone network perception and transmission method in a low signal-to-noise ratio environment according to claim 1 is characterized in that: In step S3, the entire flight time of the cognitive drone is set to T, which is divided into N time slots. The time occupied by each time slot is: Each time slot is further divided into the sensing phase τ s and the transmission phase τ t , τ=τ s +τ t , where τ s On the basis of satisfying the perception performance, the optimal perception time of each time slot is obtained through the dichotomy method; IRS adopts M=M x ×M z The UPA array of the nth time slot sensing phase phase shift matrix and the transmission phase phase shift matrix can be expressed as:

5. The IRS-assisted cognitive drone network perception and transmission method in a low signal-to-noise ratio environment according to claim 1 is characterized in that: In step S4, all channels adopt the Rice channel model, so the distance from the UAV to the IRS is h ui Taking the channel as an example, the channel model is: Where ρ is the path loss at the reference distance d = 1m, α is the path loss factor, and d ui [n] represents the distance from the UAV to the IRS in the nth time slot, and its calculation formula is: Z U and Z R Represent the distances of the UAV and IRS from the ground, q and W respectively I They are the horizontal coordinates of the UAV and IRS respectively; β1 represents the Rice factor of the channel, and Represents the line-of-sight and non-line-of-sight channel gains of the Ricean channel in the nth time slot.

6. The IRS-assisted cognitive drone network perception and transmission method in a low signal-to-noise ratio environment according to claim 1 is characterized in that: In step S5, when the PT is idle and the probability of being correctly detected is P(H0)(1-P f ), P(H0) is the probability that the PT is idle; the received signal-to-noise ratio of the kth secondary user in the nth time slot is: Where K represents the number of secondary users, P U represents the transmission power of the UAV, σ 2 represents the noise power. In practical applications, in order to protect the primary user PU during UAV transmission, the interference must be less than a certain threshold: γ PU It represents the receiving signal-to-noise ratio of the primary user PU during the UAV transmission phase. Since the actual speed of the UAV is subject to certain restrictions, and the present invention assumes that the starting point q0 and the ending point q F Fixed, so there are the following restrictions: ||q[n+1]-q[n]|| 2 ≤D 2 ,n=1,...,N-1 ||q[N]-q F || 2 ≤D 2 ,q[1]=q0, By combining the UAV trajectory, perception time, and the IRS phase shift matrix of the perception and transmission phases, the proposed optimization problem can be expressed as: where θ s,m [n], θ t,m [n] is the constraint on the IRS phase shift matrix, and its value range is 0 to 2π, P d ≥P d0 It is a constraint on the probability of target detection.

7. The IRS-assisted cognitive drone network perception and transmission method in a low signal-to-noise ratio environment according to claim 1 is characterized in that: In step S6, since the problem is non-convex and difficult to solve, it is decomposed into four sub-problems: optimizing the phase shift matrix in the perception phase, optimizing the phase shift matrix in the transmission phase, optimizing the trajectory, and optimizing the perception time. The four sub-problems are iterated alternately until convergence. The specific steps include: S601. Phase shift matrix φ at a given transmission stage t , UAV trajectory Q, perception time τ s Based on this, the phase shift matrix φ in the perception stage is optimized s Since only the direct link from UAV to PT and the cascade link from UAV to IRS to PT need to be considered, the phase alignment method of the cascade link and the direct link is adopted, namely: θ i,n =arg(H pu [n])+arg(H pi [i])-arg(H iu [i,n]), S602. Phase shift matrix φ at a given perception stage s , UAV trajectory Q, perception time τ s Based on the optimization of the transmission phase shift matrix φ t ; Here we use a low-complexity PDDGP algorithm to solve: introduce auxiliary variables g(θ,s)=I n +sI th , the phase shift constraint in the perception stage can be rewritten as: g(θ,s)=0,s≥0; the augmented Lagrangian function can be written as: Given v,p, for R v,p (θ,s) finds the first-order partial derivative with respect to θ, and introduces the auxiliary variable F: F = log2(1+γ k,n ), Since this result is too complicated, we first calculate the first-order partial derivative of the first half: Where A1=|h usk | 2 , The form and Similar, no specific expression is given here; θ l Need to follow Move with a certain step size, at the i-th iteration, μ is an auxiliary variable in the iterative process; in order to obtain the next iteration θ l , which needs to be projected: s=max{I th -I n ,0}, alternately iterate S and θ l Until convergence, the subproblem is solved; S603. For the third sub-problem, introduce auxiliary variables Z = {u k,n ,g n ,e n ,j n ,r k,n ,ω n ,χ n ,h n ,k n ,o n The original problem can be rewritten as: in H uk,n =AA H ;H t,n 、H p,n and δ p,n , δ t,n With H uk,n , δ uk,n The structure is similar and can be generated in the same way; next, the continuous convex approximation method is used to perform a first-order Taylor expansion on the non-convex constraints: where {u k,n (l) ,g n (l) ,e n (l) ,k n (l) ,ω n (l) ,q n (l) }, δ uk,n (l) =[u k,n (l) ,e n (l) ] T , δ t,n (l) =[g n (l) ,e n (l) ] T represents the feasible point of the lth iteration, and then the problem can be rewritten as: This problem is a convex optimization problem and can be solved using CVX; S604. Given φ t 、φ s , Q, since the perception time must meet the constraints: The fourth sub-problem can be rewritten as follows: where G[n] does not contain τ s [n], this problem can be solved by binary search; S605. Using alternating iteration, the four sub-problems of perception phase shift matrix optimization, transmission phase shift matrix optimization, drone trajectory optimization, and perception time optimization are alternately iterated until the algorithm converges.

Images