Multi-unmanned aerial vehicle auxiliary sensing integrated trajectory planning and resource allocation method

Abstract

The invention relates to a multi-unmanned aerial vehicle auxiliary sensing integrated trajectory planning and resource allocation method, which comprises the following steps that a sensing integrated system comprising a plurality of unmanned aerial vehicles, a plurality of ground nodes and a plurality of mobile terminals is established, sensing integrated signals emitted by the unmanned aerial vehicles are received and reflected by the mobile terminals, and the sensing integrated signals are transmitted to the unmanned aerial vehicles; the reflected communication and inductance integrated signal is received and processed by a ground node; constructing a Cramer-Rao bound model of the mobile terminal at the target position of the mobile terminal according to the path loss of the flux-inductance integrated signal; establishing a joint optimization problem based on set constraint conditions by taking minimization of the maximum Cramer-Rao bound of all the mobile terminals at the target positions as a target; and solving the joint optimization problem to obtain the moving track and the transmitting power of the unmanned aerial vehicle and the corresponding ground node. According to the method, perception performance optimization under the constraint of communication performance is realized, and communication and calculation burden of data fusion is reduced on the premise of ensuring the perception performance.

Description

Technical Field

[0001] The present invention relates to the technical field of synaesthesia integration, and in particular to a trajectory planning and resource allocation method for multi-unmanned aerial vehicle-assisted synaesthesia integration. Background Art

[0002] Drones are highly maneuverable, intelligent, and flexible. Utilizing the high viewing angle and wide coverage of drones, drone-assisted communication can provide a larger coverage range and better channel quality for communications. In addition, drone-assisted perception can also be used in applications such as measurement, photography, mapping, and positioning. In order to improve the efficiency of electromagnetic spectrum use, the Integrated Sensing and Communication (ISAC) technology, which aims to integrate communication and radar perception in waveform and hardware, has received widespread attention from industry and academia. Compared with single-function drones, drone-assisted ISAC technology can achieve a lighter payload. In addition, the communication and perception functions on drones can achieve mutual benefit.

[0003] However, there are still some challenges in the current UAV-assisted ISAC technology. First, it is necessary to design a trajectory planning method suitable for the UAV-assisted ISAC system. Since the communication channel gain is affected by the distance between the UAV and the user, and the perception performance is also affected by the UAV geometry and channel gain, a trajectory planning method that considers both communication performance and perception performance is needed. Second, the UAV-assisted ISAC system needs to adopt a suitable resource allocation method. Since multi-station radars are designed to cooperate to improve perception performance, and multi-station communication needs to avoid interference to improve communication performance, there is a trade-off in the resource allocation of communication and perception. It is necessary to design a suitable resource allocation scheme to maximize perception performance and manage communication interference. Finally, for the networked UAV-assisted ISAC system, data fusion between multiple nodes will increase the communication and computational burden between nodes while introducing performance gains. In order to ensure performance, a suitable node selection method also needs to be considered. Most importantly, trajectory optimization, power control, and node selection problems are often coupled with each other, so it is challenging to jointly optimize these parameters.

[0004] In addition, most of the existing joint optimization strategies only consider scenarios with static targets, but do not consider scenarios with multiple moving targets. In addition, the positioning information of the targets is predicted in the optimization algorithm. These assumptions may not be applicable to actual scenarios, and waste the perception resources of multiple UAVs. Summary of the invention

[0005] The technical problem to be solved by the present invention is to provide a multi-UAV assisted synaesthesia integrated trajectory planning and resource allocation method, which can achieve perception performance optimization under the constraint of communication performance and reduce the communication and computing burden of data fusion under the premise of ensuring perception performance.

[0006] The technical solution adopted by the present invention to solve the technical problem is: to provide a multi-UAV assisted synaesthesia integrated trajectory planning and resource allocation method, comprising the following steps:

[0007] Building a synaesthesia integration system including multiple drones, multiple ground nodes and multiple mobile terminals, wherein the synaesthesia integration signal emitted by the drone is received and reflected by the mobile terminal, and the reflected synaesthesia integration signal is received and processed by the ground node;

[0008] According to the path loss of the synaesthesia integrated signal from the UAV to the mobile terminal and then to the ground node, a Cramer-Rao bound model of the mobile terminal at its target position is constructed, and the target position is the position of the mobile terminal in the next time slot;

[0009] With the goal of minimizing the maximum Cramer-Rao bound of all the mobile terminals at their target locations, a joint optimization problem of ground node selection, UAV trajectory planning and resource allocation is established based on set constraints;

[0010] Solving the joint optimization problem obtains the movement trajectory and transmission power of each UAV and the corresponding ground node selection.

[0011] Furthermore, the set constraints include total communication spectrum efficiency constraints, minimum and maximum perception performance constraints, UAV position constraints, UAV speed constraints, UAV collision avoidance constraints, UAV motion constraints, transmission power constraints, and ground node selection vector constraints.

[0012] Furthermore, the solution of the joint optimization problem to obtain the moving trajectory and transmission power of each UAV and the corresponding ground node selection is achieved by dividing the optimization parameters of the joint optimization problem into a ground node selection vector, a transmission power vector and a UAV trajectory vector and performing alternating optimization.

[0013] Furthermore, the ground node selection vector, the transmission power vector and the drone trajectory vector are alternately optimized by the following method:

[0014] Using Schur's complement theorem, the objective function is transformed into a convex problem for the ground node selection vector, a convex problem for the transmission power vector, and a convex problem for the UAV trajectory vector;

[0015] The convex optimization algorithm is used to iteratively solve the three convex problems obtained in the previous step to obtain the receiving ground node, transmission power and moving trajectory of each UAV in the next time slot.

[0016] Furthermore, the synaesthesia integration signal includes a synaesthesia integration phase signal transmitted by multiple drones in a time division mode and a communication-only phase signal transmitted by multiple drones simultaneously.

[0017] Furthermore, the step of constructing a Cramer-Rao bound model of the mobile terminal at its target location according to the path loss of the synaesthesia integrated signal from the drone to the mobile terminal and then to the ground node includes:

[0018] predicting the target location;

[0019] According to the model of the synaesthesia integrated signal transmitted by the UAV and the path loss from the UAV to the mobile terminal and then to the ground node, a pilot signal model of the synaesthesia integrated signal received and processed by the ground node is constructed;

[0020] A likelihood function with the path loss and the target position as parameters to be estimated is derived according to the pilot signal model, and then the Fisher information matrix of the likelihood function is inverted to obtain a Cramer-Rao bound model of the mobile terminal at its target position.

[0021] Furthermore, the target position is predicted based on a uniform motion model.

[0022] Furthermore, the joint optimization problem is expressed as

[0023]

[0024] Where η is the objective function, is the set of mobile terminals receiving communication signals in the nth time slot, R k (n) is the downlink spectrum efficiency of the kth mobile terminal, R min is the minimum overall spectral efficiency, and They are the mobile terminal position x k (n) and y k (n) of the Cramer-Rao bound, and It is nth t The position of the drone, X min and X max They are The minimum and maximum values ​​of Y min and Y max They are The minimum and maximum values ​​of V max Drone speed The maximum value, dmin is the minimum distance between drones, ΔT is the frame length of the synaesthesia integrated signal, and the transmission power vector P max is the UAV transmission power The maximum value of b(n) is the ground node selection vector If and only if the nth r When there are ground nodes, It is equal to 1, N r is the number of ground nodes.

[0025] Furthermore, the downlink spectrum efficiency R k (n) is expressed as

[0026]

[0027] in, It is nth t The channel gain between the kth UAV and the kth mobile terminal in the nth time slot is: It is nth t The transmission power of the UAV in the nth time slot, σ 2 is the noise variance.

[0028] Furthermore, the Cramer-Rao bound model is expressed as

[0029]

[0030] Function A k,11 ,A k,22 and A k,12 Defined as

[0031]

[0032] The parameter β is defined as l=1,2,…,L is the pilot number, L is the number of pilots, k l is the wave number, is the processed noise variance, It is nth t The channel gain between the kth UAV and the kth mobile terminal in the nth time slot is: It is nth t The transmission power of the drone in the nth time slot, and are the propagation distances of the synaesthesia integrated signal from the UAV to the mobile terminal and from the mobile terminal to the ground node, respectively. and It is nth r The location of the ground nodes, and is the predicted target location of the mobile terminal, Nt and N r are the number of UAVs and ground nodes respectively.

[0033] Beneficial Effects

[0034] Due to the adoption of the above technical scheme, the present invention has the following advantages and positive effects compared with the prior art: the present invention builds a multi-UAV assisted multi-user ISAC system, in which the UAV provides communication services for mobile terminals (ISAC users) and acts as a transmitter of the MIMO radar system, and the ground node (GN) receives the reflected signal and fuses the data to locate the ISAC user, which makes full use of the maneuverability, flexibility and intelligence of the UAV. At the same time, the introduction of GN avoids the complex full-duplex design and reduces the communication and calculation burden. Based on the above system, the present invention further proposes a real-time joint trajectory planning and resource allocation scheme, which predicts the target position of the next time slot according to the positioning result of the previous time slot. Based on the Pareto criterion, a joint GN selection, transmission power control and trajectory design problem is constructed. The optimization problem aims to jointly optimize the GN selection, the transmission power of the UAV and the trajectory planning under the constraint of minimizing the total spectrum efficiency of multiple ISAC users with the maximum positioning Cramer-Rao bound (CRB), thereby realizing the optimization of perception performance under the constraint of communication performance. At the same time, the selection of GN reduces the communication and computational burden of data fusion while ensuring the perception performance. The present invention also approximates and simplifies the CRB expression, and then alternately optimizes the variables based on the block coordinate descent method (BCD technology), and rewrites the optimization problem by introducing equivalent transformation, penalty function and continuous convex approximation SCA technology, thereby finally realizing the solution to the above joint optimization problem. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 is a flow chart of an embodiment of the present invention;

[0036] Figure 2 Schematic diagram of a system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0037] The present invention will be further described below in conjunction with specific embodiments. It should be understood that these embodiments are only used to illustrate the present invention and are not intended to limit the scope of the present invention. In addition, it should be understood that after reading the content taught by the present invention, those skilled in the art can make various changes or modifications to the present invention, and these equivalent forms fall within the scope limited by the appended claims of the application equally.

[0038] The embodiments of the present invention relate to a multi-UAV assisted synaesthesia integrated trajectory planning and resource allocation method, such as Figure 1 As shown, the following steps are included:

[0039] S0 builds an ISAC system including multiple unmanned aerial vehicles (UAVs), multiple ground nodes (GNs) and multiple mobile terminals (UEs, i.e., ISAC users), wherein the ISAC signals emitted by the UAVs are received and reflected by the mobile terminals, and the reflected signals are received and processed by the ground nodes;

[0040] S1 constructs a Cramer-Rao bound model of the mobile terminal at its target location based on the path loss of the synaesthesia integrated signal from the UAV to the mobile terminal and then to the ground node;

[0041] S2 aims to minimize the maximum Cramer-Rao bound of all mobile terminals at their target locations and establishes a joint optimization problem of ground node selection, UAV trajectory planning and resource allocation based on the set constraints;

[0042] S3 solves the joint optimization problem to obtain the movement trajectory and transmission power of the UAV and the corresponding ground node selection.

[0043] The present implementation is further described below in conjunction with the ISAC system.

[0044] (1) ISAC system model and ISAC signal frame structure transmitted by drones

[0045] like Figure 2 As shown, in the nth time slot, the system consists of N t Located in n t =1,…,N t UAVs, including H T is the altitude of the drone, N r,tot Located in n r =1,…,N r,tot The UAVs are composed of K ground nodes located at x k (n)=(x k (n),y k (n),0) T ,k=1,…,K mobile ISAC users, and locate ISAC users at the same time. The positioning information can be used for target tracking, trajectory optimization and resource scheduling of drones. The drone transmits an orthogonal frequency division multiplexing (OFDM) signal as the ISAC signal, and the GN receives the reflected signal reflected by the ISAC user. The introduction of GN avoids the complex full-duplex design. In order to reduce the communication and computing burden of data fusion, only N is selected here. r Each GN is used as an ISAC receiving station.

[0046] Figure 2The structure of the ISAC frame is also shown. The ISAC frame consists of two parts, namely the interawareness phase and the communication-only phase. This structure is adopted because the duration of perception does not need to be too long (a few milliseconds) and the structure can reduce data redundancy. In order to maximize the perception performance, the UAV transmits signals in a time-division mode during the interawareness phase. During the communication phase, multiple UAVs transmit signals simultaneously. In our proposed real-time trajectory planning and resource allocation scheme, based on the positioning results of the previous moment, the system predicts the location of the UE at the next ISAC frame transmission moment, selects a certain number of GNs to receive and process the reflected ISAC signals, and determines the transmission power of the UAV for interference management in the communication phase and CRB optimization of the target location, and plans the trajectory of the UAV to further optimize the CRB. The UAV then moves along the predetermined trajectory and transmits the downlink signal, while the selected GN receives and preprocesses the reflected signal. The fusion center collects and processes the signal to locate the ISAC user. The system will repeat the above steps to provide communication services and continuously manage location information.

[0047] (2) Perception Signal Model

[0048] No.n t The signal transmitted by a UAV in the nth time slot can be expressed as

[0049]

[0050] where i is the OFDM symbol index, It is nth t The transmission power of the UAV in the nth ISAC time slot, f c is the carrier frequency, m is the subcarrier number, M is the number of subcarriers, Δf is the subcarrier spacing, T s is the duration of a single OFDM symbol, and rect(t) is a rectangular window function that has non-zero values ​​1 only in t∈[0,1]. Then in the nth time slot, the nth r The GN receives the t The signal of a UAV can be expressed as:

[0051]

[0052] Where k is the ISAC user serial number, Indicates that from the nth t The UAV goes to the kth UE and then to the nth UE. r The signal strength change caused by the path loss of each GN, represents the propagation delay in the above path, is satisfied complex Gaussian white noise.

[0053] According to the radar equation, satisfy: Where Γ0 is the reference channel gain at 1m, defined here

[0054] Afterwards, after sampling and OFDM demodulation at the receiving end, the signal can be converted from the time domain to the frequency domain and located using the pilot. The received pilot subcarrier signal is multiplied by the conjugate of the corresponding pilot symbol, and the pilot signals of I OFDM symbols are averaged. The processed signal can be modeled as

[0055]

[0056] Where l = 1, 2, ..., L is the pilot number, L is the number of pilots, f l is the lth pilot subcarrier frequency, k l is to satisfy k l =2pif l / c wave number, is the signal propagation distance, are the propagation distance of the incident signal and the propagation distance of the reflected signal respectively. is the processed noise, which can be proved to be complex Gaussian white noise, and the variance is reduced to To simplify the expression, define

[0057] (3) Communication channel model

[0058] Assuming that the communication channel is dominated by the line-of-sight (LoS) component, according to the free space path loss model, the nth t The channel gain between the kth UAV and the kth user in the nth time slot can be written as

[0059]

[0060] Will Denoted as the UAV serial number associated with the k-th UE, the downlink SE of the k-th UE can be modeled as

[0061]

[0062] (4) Derivation of the Cramer-Rao lower bound

[0063] According to the properties of Gaussian white noise, the parameters to be estimated can be defined as in The expression of the likelihood function of the received signal can be derived

[0064]

[0065] where C is a constant that can be ignored in subsequent steps, and s(d) is defined as s(d) = [exp(-jk1d), exp(-jk2d), ..., exp(-jk L d)] T .

[0066] To model the GN selection problem, we define the GN selection vector Only when n is selected r The element is equal to 1 only when the GNth element is equal to 1. The likelihood function can be rewritten as a form related to b(n)

[0067]

[0068] It is worth noting that, without loss of generality and for simplicity of expression, we will use 1, 2, …, N after GN is selected. r For the selected N r GNs were renumbered.

[0069] CRB can be obtained by the inverse of the Fisher information matrix F, that is,

[0070]

[0071] MSE i )≥CRB(u i )=[F -1 ] ii .

[0072] Since CRB is related to the true value of the estimated parameter u(n), it is impossible to obtain an accurate CRB before perception. Therefore, we use the predicted target position and the mathematical expectation of the path loss to calculate CRB. The prediction of the target position adopts the simplest uniform motion model, that is,

[0073] Assuming that the targets are far apart and the mutual influence between targets can be ignored, after a series of derivations, the CRB of the kth target can be approximately written as

[0074]

[0075]

[0076] Among them A k,11 ,A k,22 and A k,12 Defined as

[0077]

[0078] The parameter β is defined as

[0079] (5) Construction and solution of optimization problems

[0080] Formulate the joint optimization problem according to the Pareto criterion:

[0081]

[0082] in Defined as the set of UEs that receive communication signals in the nth time slot, the transmit power vector p(n) is defined as The objective function η is the maximum target position CRB, V max , d min is the maximum speed of the UAV and the minimum distance between UAVs. The constraints of the optimization problem are communication and rate constraints, minimum and maximum perception performance constraints, UAV position constraints, UAV maximum speed constraints, UAV collision avoidance constraints, UAV motion constraints, transmission power constraints, and GN selection vector constraints.

[0083] Using the BCD technique, the optimization parameters are divided into three groups, namely, the GN selection vector, the transmission power vector, and the UAV trajectory are alternately optimized.

[0084] In the GN selection vector optimization problem, the CRB constraint can be rewritten using the Schur complement theorem to write the constraint as a convex positive definite constraint. At the same time, the penalty function method is used to solve the 0-1 constraint of the GN selection vector, and the auxiliary variable is introduced. And guarantee The penalty function is defined as

[0085]

[0086] Where ∈ is the penalty coefficient, which decreases as the number of iterations increases. The GN selection vector optimization problem is thus rewritten as alternating optimization b(n), The problem, yes The optimization problem can be directly expressed in closed form:

[0087]

[0088] When given When b(n) is optimized, the optimization problem can be written as the following convex problem

[0089]

[0090] Where N r,tot dimensional vector, The nth r The elements are defined as

[0091]

[0092]

[0093] For the transmit power control problem, similarly, the CRB-related constraints can be rewritten using the Schur complement theorem. For the communication constraints, the SCA technique can be used to rewrite them as a convex lower bound. The transmit power optimization problem can be written as:

[0094]

[0095] Where N t Dimensional vector The nth t The elements are defined as

[0096]

[0097] Defined as

[0098]

[0099] For trajectory planning problems, auxiliary parameters are introduced And satisfy the constraints Using auxiliary parameters Rewrite the communication constraints and use SCA technology to combine communication constraints, CRB constraints, and collision avoidance constraints with Rewritten in convex form. The trajectory planning problem can be rewritten as:

[0100]

[0101] in, Defined as

[0102]

[0103] Defined as

[0104]

[0105] Respectively represent the parameters x k ,y k The approximate CRB of and their corresponding one-section Taylor expansion.

[0106] For the above three sets of problems, mature convex optimization solving tools can be used to solve them, and convergent suboptimal solutions can be obtained by iterative solution.

Claims

1. A multi-UAV assisted synaesthesia integrated trajectory planning and resource allocation method, characterized in that: The following steps are involved: Building a synaesthesia integration system including multiple drones, multiple ground nodes and multiple mobile terminals, wherein the synaesthesia integration signal emitted by the drone is received and reflected by the mobile terminal, and the reflected synaesthesia integration signal is received and processed by the ground node; According to the path loss of the synaesthesia integrated signal from the UAV to the mobile terminal and then to the ground node, a Cramer-Rao bound model of the mobile terminal at its target position is constructed, and the target position is the position of the mobile terminal in the next time slot; With the goal of minimizing the maximum Cramer-Rao bound of all the mobile terminals at their target locations, a joint optimization problem of ground node selection, UAV trajectory planning and resource allocation is established based on set constraints; Solving the joint optimization problem obtains the movement trajectory and transmission power of each UAV and the corresponding ground node selection.

2. The method according to claim 1, characterized in that The set constraints include total communication spectrum efficiency constraints, minimum and maximum perception performance constraints, UAV position constraints, UAV speed constraints, UAV collision avoidance constraints, UAV motion constraints, transmission power constraints, and ground node selection vector constraints.

3. The method according to claim 2, characterized in that The solution to the joint optimization problem to obtain the moving trajectory and transmission power of each UAV and the corresponding ground node selection is achieved by dividing the optimization parameters of the joint optimization problem into a ground node selection vector, a transmission power vector and a UAV trajectory vector and performing alternating optimization.

4. The method according to claim 3, characterized in that The ground node selection vector, the transmission power vector and the drone trajectory vector are alternately optimized by the following method: Using Schur's complement theorem, the objective function is transformed into a convex problem for the ground node selection vector, a convex problem for the transmission power vector, and a convex problem for the UAV trajectory vector; The convex optimization algorithm is used to iteratively solve the three convex problems obtained in the previous step to obtain the receiving ground node, transmission power and moving trajectory of each UAV in the next time slot.

5. The method according to claim 1, characterized in that The synaesthesia integration signal includes a synaesthesia integration phase signal transmitted by multiple drones in a time division mode and a communication-only phase signal transmitted by multiple drones simultaneously.

6. The method according to claim 1, characterized in that The Cramer-Rao bound model of the mobile terminal at its target location is constructed according to the path loss of the synaesthesia integrated signal from the drone to the mobile terminal and then to the ground node, include: predicting the target location; According to the model of the synaesthesia integrated signal transmitted by the UAV and the path loss from the UAV to the mobile terminal and then to the ground node, a pilot signal model of the synaesthesia integrated signal received and processed by the ground node is constructed; according to the pilot signal model, a likelihood function with the path loss and the target position as the parameters to be estimated is derived, and then the Fisher information matrix of the likelihood function is inverted to obtain the Cramer-Rao bound model of the mobile terminal at its target position.

7. The method according to claim 1, characterized in that The target position is predicted based on a uniform motion model.

8. The method according to claim 1, characterized in that The joint optimization problem is expressed as st 1 T b(n)=N r Where η is the objective function, is the set of mobile terminals receiving communication signals in the nth time slot, R k (n) is the downlink spectrum efficiency of the kth mobile terminal, R min is the minimum overall spectral efficiency, and They are the mobile terminal position x k (n) and y k (n) Cramer Rao bound, and It is nth t The position of the drone, X min and X max They are The minimum and maximum values ​​of Y min and Y max They are The minimum and maximum values ​​of V max Drone speed The maximum value, d min is the minimum distance between drones, ΔT is the frame length of the synaesthesia integrated signal, and the transmission power vector P max is the UAV transmission power The maximum value of b(n) is the ground node selection vector If and only if the nth r When the ground node It is equal to 1, N r is the number of ground nodes.

9. The method according to claim 8, characterized in that The downlink spectrum efficiency R k (n) is expressed as in, It is nth t The channel gain between the kth UAV and the kth mobile terminal in the nth time slot is: It is nth t The transmission power of the UAV in the nth time slot, σ 2 is the noise variance.

10. The method according to claim 8, characterized in that The Cramer-Rao bound model is expressed as Function A k,11 ,A k,22 and A k,12 Defined as The parameter β is defined as l=1,2,…,L is the pilot number, L is the number of pilots, k l is the wave number, is the processed noise variance, It is nth t The channel gain between the kth UAV and the kth mobile terminal in the nth time slot is: It is nth t The transmission power of the UAV in the nth time slot, and are the propagation distances of the synaesthesia integrated signal from the UAV to the mobile terminal and from the mobile terminal to the ground node, respectively. and It is nth r The location of the ground nodes, and is the predicted target location of the mobile terminal, N t and N r are the number of UAVs and ground nodes respectively.

Images