CN-122002232-A - Low-altitude intelligent system oriented to general sense service and deployment method thereof

CN122002232ACN 122002232 ACN122002232 ACN 122002232ACN-122002232-A

Abstract

The invention discloses a low-altitude intelligent system facing a sense-through service and a deployment method thereof, and relates to the field of unmanned aerial vehicles.A system finishes the receiving and reasoning processing of sensing data at a position close to a sense-through device by introducing an unmanned aerial vehicle as a low-altitude edge reasoning node, so that the reasoning process is changed from remote centralized processing to low-altitude edge nearby processing; the system is based on integrated sensing and communication, namely sensing data acquired by an ISAC device, and is used for quickly identifying the motion state of a target, wherein the system operation process mainly comprises an ISAC stage and a calculation stage. The system can effectively reduce the data transmission time delay before reasoning and improve the reasoning response speed, thereby meeting the application requirement of the general sense service on real-time performance.

Inventors

GUO KUN
YANG YUE
RUI BIN

Assignees

华东师范大学

Dates

Publication Date: 20260508
Application Date: 20260126

Claims (10)

1. A low-altitude intelligent system for a sense-of-passing service is characterized in that an unmanned plane is introduced to serve as a low-altitude edge inference node to finish receiving and reasoning of perception data at a position close to a sense-of-passing device, so that a reasoning process is changed from remote centralized processing to low-altitude edge nearby processing.
2. The low-altitude intelligent system for a traffic-oriented system according to claim 1, wherein in said ISAC phase said ISAC device transmits radio signals to a target to perform a detection task while transmitting received echo signals to a drone, which is solely responsible for reception, said ISAC phase employs a periodic frame structure based on radar pulse repetition intervals PRI, each PRI period duration being Divided into a sensing phase and a communication phase, each of the ISAC device collections Each sample is composed of M frequency modulated continuous wave signals FMCW.
3. The low-altitude intelligent system for the traffic-oriented service of claim 2, wherein the sensing stage comprises transmitting FMCW signals and receiving target echoes, each FMCW signal being scanned in time at a linear frequency for extracting a motion characteristic of the target, and wherein the communication stage comprises generating complex samples from the echo signals via analog-to-digital conversion and serializing the complex samples into a bit stream for real-time transmission to the unmanned aerial vehicle via the communication link.
4. The low-altitude intelligent system for the sense-of-general service according to claim 3, wherein in the calculation stage, after receiving the signal from the ISAC device, the unmanned aerial vehicle firstly performs sampling and denoising processing and converts the signal into a spectrogram through short-time fourier transform, and then inputs the generated spectrogram into a deep neural network model locally deployed by the unmanned aerial vehicle for reasoning, and the model can automatically extract the motion characteristics of the target and complete classification, so that the accurate identification of the motion state of the target is realized.
5. The low-altitude intelligent system for a traffic-oriented claim 4, wherein the low-altitude intelligent system aims to maximize overall inference throughput, and the problem modeling is as follows: communication rate by jointly optimizing drone location q and ISAC devices Maximizing the number of inferential samples for each ISAC device Is the sum of (3); the method comprises the steps of representing a set of all ISAC devices in a system, wherein k is a device index, and C1 ensures that the communication rate meets the uploading requirement of each frame of data and ensures the transmission of perceived data; representing the duration of the communication phase within a single radar pulse repetition interval period, Representing the perceived data quantity generated by each radar pulse C2 ensures the total energy consumption of the ISAC device Not exceeding its maximum available energy budget C3 guarantees the total time delay for completing sensing, communication and reasoning Not exceeding the maximum inference delay allowed by the kth device ; From constraint C2, it can be obtained Where M is the number of radar pulses required for a single inferential sample, The duration of the sensing phase within the interval period is repeated for a single radar pulse, 、 And Indicating perceived transmission power, received power and communication transmission power of the kth ISAC device respectively, Is shown in the unmanned plane position A given number of supportable inference samples determined by an energy constraint; from constraint C3, it is possible to obtain Wherein, the Representing the duration of the radar pulse repetition interval PRI, Representing the number of floating point operations required to process a single perceptual sample, Representing the computational processing power allocated to the kth ISAC device, i.e., the number of floating point operations that can be performed per second, based on which, Representing a maximum upper bound on the number of inferential samples supportable by the kth ISAC device as determined by the latency constraint; Thus, to satisfy both energy and latency constraints, the number of possible inferential samples per device is: On this basis, the optimization problem can be written as: Wherein, the Representing the equivalent communication channel gain parameter for the kth ISAC device, B represents the system bandwidth, Representing the distance between the kth ISAC device and the drone, the problem may be further translated into: in this reconstruction problem, the device set is based on And (3) with Is divided into two mutually exclusive subsets: Devices with more stringent energy constraints; devices with more stringent delay constraints.
6. A low-altitude intelligent system deployment method for a through service is characterized in that the method maximizes the number of inference samples supportable by a system under the constraint of resources, and meanwhile, in order to reduce the computational complexity introduced by device constraint type enumeration, a device constraint pre-classification strategy for unmanned aerial vehicle deployment is further provided.
7. The low-altitude intelligent system deployment method facing to the sense service according to claim 5, wherein the method comprises the following steps: Step 1, calculating a time delay constraint upper bound constant for all devices Calculating the maximum reasoning sample number supported by the time delay constraint according to the time delay constraint and the calculation resource ; Step 2, pre-classifying the equipment, and dividing the ISAC equipment into energy-limited equipment and time-delay-limited equipment according to the limiting relation of the equipment to the number of feasible reasoning samples under the energy constraint and the time-delay constraint; step 3, flexible device grouping enumeration, for flexible device collection Adopting an enumeration strategy, considering that the devices are in And Since part of the devices have been pre-classified as energy limited and must belong to Any assignment of these devices to The schemes of (a) are omitted so as to effectively reduce the search space, and therefore, only the evaluation is needed Seed grouping scheme, far less than full enumeration Seed; step 4, optimal solution selection, finally, selecting a group which enables the total target value of the system to be maximum, and outputting the corresponding UAV position Total target value Optimal number of inferential samples per device 。
8. The low-altitude intelligent system deployment method facing to the sense-through service according to claim 6, wherein in the step 1, the upper bound and the unmanned aerial vehicle position are Is irrelevant and is used for comparison analysis in the follow-up device constraint pre-classification process.
9. The low-altitude intelligent system deployment method for the traffic-oriented service according to claim 7, wherein a time delay constraint upper bound of each device is obtained After that, the step 2 analyzes the upper bound of the energy constraint Within a UAV feasible location space To determine the constraint type of each device if the maximum of a device Always less than or equal to Its performance is limited by energy constraints and the devices are classified as energy-constrained devices and fall into a collection For the rest of the devices, it And (3) with The relationship of these devices may vary with UAV position, the devices may be limited by energy constraints in some areas and by time delay constraints in other areas, such devices comprising a flexible collection of devices Its specific classification will be determined in the subsequent packet enumeration.
10. The low-altitude intelligent system deployment method facing to the sense-through service according to claim 8, wherein in each enumeration scheme in the step 3, Comprising predetermined energy-limited devices and current schemes assigned to Flexible device of (a) Then the flexible devices assigned to this set are included, said step 3 further comprising: step 3-1, after obtaining the candidate grouping scheme, fixing the set of energy-limited devices for each candidate grouping And establishes the corresponding sub-problems The sub-problem only relates to Objective functions and constraints of the medium equipment aimed at optimizing UAV position Communication rate of these devices If there is a feasible solution, it is recorded as Sub-problems The specific form of (2) is as follows: Wherein, the For easy solution, auxiliary variables are introduced Furthermore, two auxiliary variables are introduced And Based on these substitutions, the original objective function and constraints can be equivalently rewritten into a more solvable form: the non-convex constraints C4', C7 and C10 are expressed as differences between two convex functions and are therefore non-convex in nature, so that the continuous convex approximation SCA method is adopted to convert the original non-convex constraints into convex forms by performing first-order Taylor expansion on the second convex function in each constraint, so that the overall problem can be solved efficiently by using a standard convex optimization tool; Step 3-2, obtaining an optimal UAV position After that, verify Whether all devices meet the following constraints: if all constraints are satisfied, the current packet is considered a feasible packet and a total target value for the packet is calculated Corresponding to And The sum of the inference samples of all devices in the system.

Description

Low-altitude intelligent system oriented to general sense service and deployment method thereof Technical Field The invention relates to the field of unmanned aerial vehicles, in particular to a low-altitude intelligent system facing a sense-of-general service and a deployment method thereof. Background In recent years, due to the continuous development of wireless communication and intelligent sensing technologies, integrated sensing and communication technologies have been widely focused and applied in application scenes such as environment sensing and target detection. The ISAC realizes the deep coupling of the sensing function and the communication function by sharing the frequency spectrum resource and the hardware architecture, so that the system can efficiently complete the wireless transmission of sensing data while acquiring the environment and target information, thereby improving the sensing capability and the information interaction efficiency of the system on the environment state and providing important technical support for the application of the low-altitude intelligent system in the scenes of smart cities, traffic monitoring and the like. However, the conventional general sensing device is limited by factors such as device volume, power consumption and hardware cost, and the like, and generally has limited computing capacity, so that complex cognitive reasoning and decision tasks are difficult to finish locally, and further the problems of insufficient reasoning capacity and large response time delay are easily caused. For this reason, the sense-of-general service often needs to transmit the sense data back to the base station or the remote data center for centralized processing and inference analysis. However, in remote areas, disaster sites, or application scenarios where the communication infrastructure is imperfect, there are still significant limitations to relying on fixed base stations or remote data centers for centralized reasoning. The sensing data needs to be transmitted back in a long distance through the air-ground link, so that the problems of limited communication bandwidth, increased transmission delay, insufficient link stability and the like are easy to occur, the requirements of the communication service on low delay and high reliability are difficult to meet, and the overall reasoning capacity of the system is limited to a certain extent. With the rapid development of low-altitude economy, the unmanned aerial vehicle has application potential for supporting rapid reasoning in the sense-of-general service by virtue of the advantages of flexible deployment, rapid response and the like. By arranging the unmanned plane platform carrying the edge computing resources at low altitude, the reasoning task can be completed at a position close to the general sense data source, so that the data transmission distance is effectively shortened, the communication time delay is reduced, and the reasoning throughput of the system is improved. However, there are still many challenges in the unmanned aerial vehicle assisted reasoning scenario. Firstly, the spatial position of the unmanned aerial vehicle has a significant influence on the quality of an air-to-ground communication link, and the deployment position directly determines the uploading rate and the transmission delay of perceived data. Secondly, under the constraint conditions of communication power, calculation resources, time delay and the like, the limitation of communication performance further restricts the execution efficiency of the reasoning task. Furthermore, a close coupling relation exists among the unmanned aerial vehicle position, the general sense resource and the computing resource, so that the system optimization problem presents a non-convex characteristic, and the optimal reasoning performance is difficult to obtain through a simple rule or a static strategy. In view of the above challenges, existing research has proposed a number of specific solutions around the unmanned aerial vehicle assisted edge reasoning scenario. There is a collaborative edge reasoning scenario aided by a research unmanned aerial vehicle, wherein the unmanned aerial vehicle serves as an air edge server, aggregates multi-sensor features through air calculation, and completes a reasoning task. The work takes the discrimination gain as the reasoning precision index of task guidance, jointly optimizes the flight track of the unmanned aerial vehicle and the power distribution of the sensor, and considers the importance of characteristic dimensions, thereby improving the collaborative reasoning performance under the condition of limited communication. The method has the edge application of simultaneously executing AI reasoning and target tracking for the unmanned aerial vehicle, and the problem of collaborative optimization of reasoning and perception functions under the condition of resource limitation is researched. The work jointly optimizes