Hybrid vehicle system isac optimization method based on drone smart reflector

Abstract

The application relates to the field of wireless communication of unmanned aerial vehicles and Internet of Vehicles, and discloses an ISAC optimization method of a hybrid vehicle system based on an unmanned aerial vehicle intelligent reflector; first, vehicles in a road are clustered, wherein a CAV serves as a cluster head vehicle, and an HDV serves as a cluster member vehicle. The cluster head vehicle can send a sensing signal to perceive the state of the cluster member while communicating with the cluster member. Meanwhile, the CAV can also communicate with other CAVs through a virtual channel constructed by the intelligent reflector on the unmanned aerial vehicle. In order to improve the performance of the above ISAC system and solve the variable coupling problem, a joint optimization method of the CAV and the unmanned aerial vehicle intelligent reflector is designed. The method maximizes the intra-cluster and inter-cluster communication rates and the perception beam gain by jointly optimizing the precoding matrix, the unmanned aerial vehicle position and the reflector phase. The application can effectively improve the ISAC performance of the unmanned aerial vehicle intelligent reflector hybrid vehicle system.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of wireless communication, in particular to an ISAC optimization method based on unmanned aerial vehicle intelligent reflecting surface assisted hybrid intelligent vehicle system. BACKGROUND

[0002] In the future for a long time, there will be two kinds of vehicles on the road, one is intelligent communication autonomous driving vehicle (CAV), and the other is ordinary human driving vehicle (HDV). The former has much higher intelligence than the latter because it is equipped with more advanced technology, can undertake more communication and computing tasks, and can also perceive the road environment using its advanced technology hardware. However, the latter is also a large part of the traffic system, so it also needs to obtain road information, and the base station cannot cover many road details, so obtaining information from CAV will be a good solution. In order to enable HDV to obtain information from CAV in time, clustering can be considered for them, and CAV can be used as cluster head and HDV as cluster member. CAV needs to send an integrated sensing and communication signal (ISAC) to each cluster member, so as to perceive the state of the cluster member while communicating with it.

[0003] However, relying solely on a single cluster head for road perception is not enough, so multiple cluster head vehicles are needed in the road to cooperate. However, due to the blocking of the LOS channel between the cluster head vehicles in the road, and the channel capacity and reliability of the NLOS channel cannot meet the requirements. This makes it difficult for cluster head vehicles to cooperate.

[0004] Thanks to the development of intelligent reflecting surfaces, users can communicate through the virtual LOS generated by intelligent reflecting surfaces, which greatly improves the information transmission rate and stability. However, fixed reflecting surface positions often cannot adapt to the highly dynamic vehicle networking environment. And because of the high flexibility and high mobility of unmanned aerial vehicles, they can fly to most areas on land to perform tasks. Therefore, deploying intelligent reflecting surfaces on unmanned aerial vehicles is a good solution. CAVs can communicate with other CAVs through the virtual LOS generated by the intelligent reflecting surface on the unmanned aerial vehicle. Therefore, CAVs in the road need to generate intra-cluster ISAC beams and inter-cluster communication beams, and the unmanned aerial vehicle also needs to change its position and the phase of the reflecting surface according to the real-time changes in the road environment.

[0005] However, there is no solution to optimize the performance of ISAC in the joint CAV and unmanned aerial vehicle intelligent reflecting surface in the hybrid vehicle system. SUMMARY

[0006] In order to overcome the shortcomings of the prior art, for the problem of joint ISAC performance optimization of CAV and unmanned aerial vehicle intelligent reflecting surface in the hybrid vehicle system, the present application provides an effective method.

[0007] The technical solution adopted by this invention to solve its technical problem is as follows:

[0008] This invention is based on a UAV-assisted intelligent reflector hybrid intelligent vehicle system, which includes a UAV equipped with an intelligent reflector RIS. On roads served by the UAV, multiple CAVs and HDVs exist. The CAVs use their own sensing signals to detect nearby HDVs and group them into their own clusters, then inform the roadside units (RSUs) of the re-clustering results. Furthermore, the first CAV on the road wirelessly informs other CAVs of unknown information ahead.

[0009] The ISAC optimization method for UAV intelligent reflector-assisted hybrid intelligent vehicle systems includes the following steps:

[0010] 1) CAVs and HDVs enter the road. Each CAV groups nearby HDVs into its own cluster and informs the RSU of the clustering results. The UAV acquires the positions of all CAVs and calculates the channel between the CAVs and the UAV, while continuously moving within a specified period to facilitate communication between CAVs.

[0011] 2) The first CAV on the road will act as the lead vehicle, transmitting road information to the other CAVs. Each time the first CAV leaves the road, a new lead vehicle will be selected from the remaining CAVs. Simultaneously, the drone will calculate the channel between itself and each CAV based on its position, adjusting its own position in real time to assist the lead vehicle in communicating with other CAVs.

[0012] 3) The first CAV in the road generates intra-cluster ISAC beams and inter-cluster communication beams, while the remaining CAVs generate ISAC beams within their own clusters. Furthermore, intra-cluster and inter-cluster communication use different frequency bands. In addition, to measure ISAC performance, intra-cluster communication rate and... Inter-cluster communication rate and Cluster-aware beam gain and As a criterion for evaluation.

[0013] 4) Optimize the precoding matrix, UAV position, and smart reflector phase based on intra-cluster communication rate, inter-cluster communication rate, and intra-cluster beammap gain.

[0014] Furthermore, in each time slot, step 4) first optimizes the intra-cluster and inter-cluster beamforming of the lead vehicle, as well as the position and reflector phase for the UAV. Then, it optimizes the inductive beamforming of the remaining CAV clusters.

[0015] Furthermore, the optimization problem for the lead vehicle cluster and the drones is modeled as follows:

[0016] ;

[0017] Similarly, the optimization problem for the remaining CAV clusters is modeled as follows:

[0018] ;

[0019] Among them, each cluster has There are 1 HDVs, and there are a total of 100 vehicles on the road besides the lead vehicle. One CAV. These represent the intra-cluster communication rate, inter-cluster communication rate, and sensing beam gain, respectively. The weights represent the intra-cluster and inter-cluster communication rates and the sensing beam gain. These represent the vehicle's maximum transmission power, the precoding matrix of intra-cluster communication signals, the precoding matrix of inter-cluster communication signals, and the precoding matrix of sensing signals, respectively. This represents the phase shift unit of RIS, and correspondingly, This represents the RIS phase shift matrix. This indicates the current location of the drone. This indicates the drone's position at the previous moment.

[0020] In summary, equations (2) and (6) represent the vehicle's transmit power limit. Equation (3) represents the RIS phase shift range limit. Equation (4) represents the UAV's movement speed limit.

[0021] Taking the optimization problem of the lead car cluster as an example, the original optimization problem (1) is split into two sub-problems: first, optimize the precoding matrix. Then optimize the UAV position based on the precoding matrix optimization results. and intelligent reflector phase shift matrix For optimization problems of other clusters (5), the precoding matrix of inter-cluster communication signals, UAV position, and RIS phase shift are not considered. Therefore, only the precoding matrix needs to be optimized. .

[0022] Furthermore, in optimization problem (1), the intra-cluster communication rate and inter-cluster communication rate are expressed as follows:

[0023] ;

[0024] in

[0025] ;

[0026] These represent the first (i) in the cluster. The signal-to-interference-plus-noise ratio and inter-cluster density of HDV Signal-to-interference-plus-noise ratio at each CAV. Among them, Indicates CAV up to the Channel gain of HDV. and Let represent the precoding vectors of the lead vehicle transmitting signals to its HDV and other CAVs within its cluster, respectively. Therefore, the precoding matrix... , . This represents the covariance matrix of the radar. and They represent the first The HDV and the first Noise at each CAV point. Indicates the drone has reached the Channel gain of each CAV. This represents the phase shift matrix of the intelligent reflector. This indicates the number of smart reflective surface elements. This represents the channel gain from the lead vehicle to the drone.

[0027] Furthermore, in the optimization problem (1), the first cluster... The sensing beam gain in each HDV direction is expressed as:

[0028]

[0029] in Indicates azimuth. Indicates the pitch angle.

[0030] Furthermore, the optimization of the lead vehicle and the drone in step 4) specifically includes:

[0031] 4.1 Derivation about The derivative of .

[0032] 4.2 Constructing a Deep Unfolding model based on Projected Gradient Ascent (PGA). This model first unfolds each iteration of PGA into a fully connected layer in a neural network, and sets the iteration step size in PGA as a trainable parameter in the neural network. The training loss function is set as follows: The trained Deep Unfolding model can be solved within a finite number of PGA iterations. .

[0033] 4.3 Solving the UAV deployment and reflector optimization problem using the Proximal Policy Gradient (PPO) algorithm. Specifically, after the UAV acquires all CAVs and its own channels, it uses these as inputs to the PPO. The PPO is then combined with the DeepUnfolding model to output the UAV's deployment at the next time step, the reflector phase shift, and the precoding matrix of the signal transmitted from the lead vehicle.

[0034] For the remaining optimization problems of CAV, it is only necessary to use the Deep Unfolding model to solve the precoding matrix of intra-cluster synsensory signals, without considering the precoding matrix of inter-cluster communication signals or the phase shift optimization of UAV position and reflector.

[0035] The technical concept of this invention is as follows: Existing technologies still lack expertise in optimizing the performance of ISAC (Intelligent Sensor Controller) systems that combine CAV (Computer-Aided Vehicle) and UAV (Unmanned Aerial Vehicle) intelligent reflectors. Therefore, this invention focuses on this aspect, providing a solution that considers beamforming, UAV positioning, and intelligent reflector optimization. This method improves the performance of the ISAC system by optimizing the communication rate within and between clusters, as well as the sensing beam gain within the cluster.

[0036] The beneficial effects of this invention are mainly reflected in the following aspects: it fills the gap in the existing technology for ISAC performance optimization of combined CAV and UAV intelligent reflector in hybrid vehicle systems, and improves the communication rate within and between clusters and the sensing beam gain within the cluster by combining beamforming, UAV position and intelligent reflector. Attached Figure Description

[0037] Fig. 1 This is a schematic diagram of a UAV intelligent reflector-assisted hybrid intelligent vehicle system according to the method of the present invention;

[0038] Fig. 2 This is a schematic diagram illustrating the performance of the method of the present invention in different time slots;

[0039] Fig. 3 This is a schematic diagram illustrating the intra-cluster and inter-cluster communication rates and performance of the method of the present invention under different time slots. Detailed Implementation

[0040] The present invention will now be further described with reference to the accompanying drawings.

[0041] This invention first clusters vehicles on the road, with CAVs (Carrier Aerial Vehicles) acting as cluster leader vehicles and HDVs (High-Definition Vehicles) as cluster member vehicles. The cluster leader vehicle can transmit sensing signals to communicate with cluster members while simultaneously sensing their status. Simultaneously, the CAVs can also communicate with other CAVs through a virtual channel constructed by a smart reflector on an UAV. To improve the performance of the aforementioned ISAC system and address the variable coupling problem, a joint optimization method for CAVs and UAV smart reflectors is designed. This method maximizes intra-cluster and inter-cluster communication rates and sensing beam gain by jointly optimizing the precoding matrix, UAV position, and reflector phase.

[0042] Reference Figs. 1-3The ISAC optimization method for hybrid vehicle systems based on UAV intelligent reflective surfaces is implemented based on existing wireless information transmission systems. That is, the UAV intelligent reflective surface-assisted hybrid intelligent vehicle system consists of a UAV equipped with an intelligent reflective surface, multiple HDVs and multiple CAVs.

[0043] In this embodiment, the UAV provides services to CAVs (Carrier Aerial Vehicles) on the road within a specified period. It utilizes its onboard intelligent reflective surface to create a virtual Line of Spectrum (LOS), providing a communication channel between the lead vehicle and other CAVs. Simultaneously, the intelligent reflective surface can reconfigure the channel between CAVs by changing the phase of its reflected signal. Meanwhile, the CAVs on the road also send sensing signals to HDVs within their clusters, thereby perceiving their status while communicating with them. Furthermore, intra-cluster CAVs and HDVs communicate with inter-cluster CAVs on different frequency bands.

[0044] In this embodiment, the intra-cluster communication rate and the inter-cluster communication rate are respectively expressed as:

[0045] ;

[0046] in

[0047] ;

[0048] These represent the first (i) in the cluster. The signal-to-interference-plus-noise ratio and inter-cluster density of HDV Signal-to-interference-plus-noise ratio at each CAV. Among them, Indicates CAV up to the Channel gain of HDV. and Let represent the precoding vectors of the lead vehicle transmitting signals to its HDV and other CAVs within its cluster, respectively. Therefore, the precoding matrix... , . This represents the covariance matrix of the radar. and They represent the first The HDV and the first Noise at each CAV point. Indicates the drone has reached the Channel gain of each CAV. This represents the phase shift matrix of the intelligent reflector. This indicates the number of smart reflective surface elements. This represents the channel gain from the lead vehicle to the drone.

[0049] In addition, the first in the cluster The sensing beam gain in each HDV direction is expressed as:

[0050] ;

[0051] in Indicates azimuth. Indicates the pitch angle.

[0052] In this embodiment, the precoding matrix of the lead vehicle cluster, the UAV position, and the intelligent reflector phase optimization can be expressed as:

[0053] ;

[0054] The optimization of the precoding matrix for other clusters can be expressed as:

[0055] ;

[0056] The optimization problem for other clusters can be solved directly, but for the optimization problem of the lead-car cluster, the original optimization problem needs to be decomposed into two sub-problems. First, optimize the precoding matrix. Then optimize the UAV position based on the precoding matrix optimization results. and intelligent reflector phase shift matrix .

[0057] This embodiment presents an ISAC performance optimization method for a hybrid intelligent vehicle system assisted by a drone's intelligent reflector, addressing existing ISAC performance optimization methods that do not consider the joint operation of the CAV (Carrier Availability) and the drone's intelligent reflector. By jointly optimizing the precoding matrix, drone position, and intelligent reflector phase, the ISAC performance of the hybrid intelligent vehicle system is improved.

[0058] In this embodiment, CAVs and HDVs are randomly distributed along the road and continuously move forward, forming multiple clusters. The UAV moves flexibly at a position 50m above the road, facilitating communication between CAVs. Simultaneously, each CAV adjusts its precoding matrix settings in real time based on the HDV's position.

[0059] The method of this invention employs a 20-layer Deep Unfolding algorithm to solve the precoding matrix and utilizes PPO to optimize the reflector phase shift and UAV position. Comparison Method 1 uses a 20-layer Deep Unfolding algorithm to solve the precoding matrix and utilizes PPO to optimize the UAV position, while the reflector phase shift cells are randomly set. Comparison Method 2 uses the Zero Forcing (ZF) algorithm to solve the precoding matrix and utilizes PPO to optimize the reflector phase shift and UAV position. Comparison Method 3 uses a 10-layer Deep Unfolding algorithm to solve the precoding matrix and utilizes PPO to optimize the reflector phase shift and UAV position.

[0060] Fig. 2The performance of the ISAC system in each time slot of the present invention is demonstrated. This performance is evaluated using a weighted sum of intra-cluster and inter-cluster communication rates and intra-cluster sensing beam gain. It can be seen that the present invention maintains better performance than other methods across 30 consecutive time slots without significant fluctuations.

[0061] Fig. 3 The method of this invention demonstrates the communication rates within and between clusters in each time slot. It can be seen that the method of this invention can effectively improve the communication rate between clusters, exhibiting better performance compared to other comparative methods. Therefore, this invention has certain advantages.

[0062] The embodiments described in this specification are merely examples of implementations of the inventive concept and are for illustrative purposes only. The scope of protection of this invention should not be considered as limited to the specific forms stated in these embodiments. The scope of protection of this invention is also based on equivalent technical means that can be conceived by those skilled in the art based on the inventive concept.

Claims

1. An ISAC optimization method for hybrid vehicle systems based on intelligent reflective surfaces of unmanned aerial vehicles, characterized in that, The system includes a UAV equipped with a smart reflector RIS. On the road served by the UAV, there are multiple intelligent communication autonomous vehicles (CAVs) and ordinary human-driven vehicles (HDVs). The CAV uses its own sensing signals to detect nearby HDVs and classifies them into its own cluster. Then, it informs the roadside unit (RSU) of the clustering results. In addition, the first CAV on the road informs other CAVs of unknown information ahead through a wireless channel. The method includes the following steps: 1) CAVs and HDVs enter the road. Each CAV divides nearby HDVs into its own cluster and informs the RSU of the clustering results. The UAV obtains the position of all CAVs and calculates the channel between the CAVs and the UAV. At the same time, it moves continuously within a set period to assist communication between CAVs. 2) The first CAV in the road will act as the lead vehicle to transmit road information to the other CAVs. Whenever the first CAV leaves the road, a new lead vehicle will be selected from the remaining CAVs. At the same time, the UAV will calculate the channel between each CAV and itself based on the position of each CAV and adjust its own position in real time to assist the lead vehicle in communicating with other CAVs. 3) The first CAV in the road generates intra-cluster communication and sensing ISAC beams as well as inter-cluster communication beams, while the remaining CAVs generate ISAC beams within their own clusters; furthermore, intra-cluster and inter-cluster communication use different frequency bands; to measure ISAC performance, the intra-cluster communication rate and... Inter-cluster communication rate and In-cluster sensing beam gain and As a criterion for judgment; 4) Optimize the precoding matrix, UAV position, and smart reflector phase based on intra-cluster communication rate, inter-cluster communication rate, and intra-cluster beammap gain; In each time slot, in step 4), the intra-cluster and inter-cluster beamforming of the lead vehicle and the position and reflector phase of the UAV are first optimized, and then the inductive beamforming of the remaining CAVs is optimized. The modeling of the two optimization problems is as follows: The optimization problem for the lead vehicle and the drone is modeled as follows: ||L (0) -L||≤ΔL max (4) The optimization problem for the remaining CAVs is modeled as follows: Each cluster contains K HDVs, and there are a total of C CAVs on the road excluding the lead vehicle; θ represents the intra-cluster communication rate, inter-cluster communication rate, and sensing beam gain, respectively; k φ represents the azimuth angle. k χ1, χ2, χ3 represent the intra-cluster and inter-cluster communication rates and the weights of the sensing beam gain; p max ,W,W c W s These represent the vehicle's maximum transmission power, the precoding matrix of intra-cluster communication signals, the precoding matrix of inter-cluster communication signals, and the precoding matrix of sensing signals, respectively. This represents the phase shift unit of the RIS, and correspondingly, Θ represents the RIS phase shift matrix; L represents the position of the UAV; L (0) This indicates the drone's position at the previous moment; When considering the optimization problem of the head car cluster, the original optimization problem (1) is split into two sub-problems: first, optimize the precoding matrix W, W c W s Then, based on the precoding matrix optimization results, optimize the UAV position L and the intelligent reflector phase shift matrix Θ; For optimization problems of other clusters (5), the precoding matrix of inter-cluster communication signals, UAV position, and RIS phase shift are not considered. Therefore, only the precoding matrix W,W s .

2. The ISAC optimization method for hybrid vehicle systems based on UAV intelligent reflective surfaces as described in claim 1, characterized in that: In step 3), the intra-cluster communication rate and inter-cluster communication rate in each time slot are expressed as follows: in Let S and S represent the signal-to-interference-plus-noise ratio (SIR) at the k-th HDV within a cluster and the SIR at the c-th CAV between clusters, respectively. Among them, h k w represents the channel gain from CAV to the k-th HDV; k and w c Let W represent the precoding vectors of the lead vehicle transmitting signals to its HDV and other CAVs within its cluster, respectively. Therefore, the precoding matrix W = {w1, ..., w...} k }, W c ={w1,...,w c };V=W s W s H The covariance matrix of the radar; σ k 2 and σ c 2 Let represent the noise at the k-th HDV and c-th CAV, respectively; This represents the channel gain from the UAV to the c-th CAV; This represents the phase shift matrix of the intelligent reflector, where N represents the number of elements in the intelligent reflector; h cu This represents the channel gain from the lead vehicle to the drone.

3. The ISAC optimization method for hybrid vehicle systems based on UAV intelligent reflective surfaces as described in claim 1, characterized in that: The optimization of the lead vehicle and the drone in step 4) specifically includes: Step 4.1, Derivation Regarding W, W c W s The derivative; Step 4.2: Construct a Deep Unfolding model based on Projected Gradient Ascent (PGA). First, each iteration of the PGA is unfolded into a fully connected layer in the neural network, and the iteration step size in the PGA is set as a trainable parameter in the neural network. The training loss function is set as follows: After training, the Deep Unfolding model solves for W and W within a finite number of PGA iterations. c W s ; Step 4.3: Solve the UAV position deployment and reflector optimization problem using the near-end policy gradient algorithm (PPO). Specifically, after the UAV obtains all CAVs and its own channels, it uses them as input to PPO. After PPO, it will be combined with the DeepUnfolding model to output the UAV's position deployment, reflector phase shift, and precoding matrix of the signal transmitted in the lead vehicle at the next moment. Step 4.4: For the remaining CAV optimization problems, the precoding matrix is ​​solved using the Deep Unfolding model, without needing to consider the optimization of the UAV position and reflector.

Images