A Modeling Method for IoT Optimization Problems Based on Intent Networks and Drone Assistance

Abstract

This invention relates to the fields of Internet of Things (IoT) and intent network technology, specifically to a modeling method for IoT optimization problems based on intent networks and drone-assisted design. The method includes: constructing an IB-IoT scenario containing base stations, users, and drones; defining three types of intent processing paths; using IBN to read historical control information authorized by users to infer potential needs; mathematically representing users' historical intents; capturing two types of correlations to infer current intents; improving the continuous crowd model to calculate mobility costs, speed, and location updates; integrating two types of channel fading to calculate capacity and latency; constructing a QoE model with a piecewise utility function; normalizing QoE; and establishing an optimization objective that maximizes the average QoE across multiple users and time slots. This addresses the problems of incomplete intent coverage, rigid mapping, and difficulty in parsing intents from long text contexts in existing intent network-based IoT technologies.

Description

Technical Field

[0001] This invention belongs to the field of Internet of Things (IoT) and intent network technology, specifically relating to a modeling method for IoT optimization problems based on intent networks and drone assistance. Background Technology

[0002] Autonomous Aerial Vehicle (AAV)-assisted Internet of Things (IoT) is a significant development trend in the current IoT field. Its core feature lies in the dynamic optimization of IoT networks and devices through AAV. In this application scenario, AAV, mobile communication base stations, and IoT devices together constitute the core nodes of the IoT network. Compared to traditional IoT networks without AAV participation, the communication modes and traffic characteristics in this scenario are more complex, making the limitations of traditional IoT architectures that rely on manually configured static rules increasingly apparent and unable to adapt to dynamically changing network requirements.

[0003] Intent-Based Networking (IBN), as a novel network paradigm, enables IoT infrastructure to transform user or device operational intentions into autonomous and self-verifiable network configurations, thereby continuously adapting to the dynamic needs of a large-scale IoT ecosystem. Leveraging the high-speed connectivity of 6G network base stations, and through the abstraction of policies and the joint orchestration of IBN intent-driven control actions, Intent-Based IoT (IB-IoT), with the assistance of Autonomous Access Controllers (AAVs), effectively improves the response speed and operational efficiency of both the user side and the underlying network structure, ultimately optimizing the user experience.

[0004] In IB-IoT systems, accurate prediction of user intentions and intention-based IBN optimization are core issues that urgently need to be addressed. Practice shows that correctly predicting user intentions and translating them into real-time network configuration tasks, combined with AAV-assisted continuous network optimization, can significantly reduce network latency and improve user experience. Currently, intention prediction and optimization solutions for IB-IoT mainly fall into the following two categories:

[0005] (1) Methods based on rule-based strategy engines and static optimization heuristics. The implementation logic of this type of method is as follows: a set of user intentions and corresponding network policy configurations are predefined, and the relationship between the predefined user intentions and network policy configurations is established through linear mapping and other methods; when a user triggers a certain preset intention, the system automatically calls the network configuration corresponding to the intention to adjust the network state. However, this type of method has obvious defects: on the one hand, the scope of intention coverage is insufficient. Due to the complexity and dynamic changes of the network environment, the predefined intentions and network policies are difficult to fully adapt to real-time requirements. Some user intentions cannot be predefined, resulting in a lack of corresponding network configuration support for these intentions; on the other hand, the intention mapping relationship is rigid. The fixed linear mapping method lacks flexibility. For the same intention triggered by a user in different time periods and different environments, the required network configuration may be different, but the fixed linear mapping cannot achieve this kind of dynamic adaptation.

[0006] (2) Methods based on deep learning (DL) and deep reinforcement learning (DRL). These methods capture the temporary evolution of intent and utilize the characteristics of DL or DRL to realize the non-linear relationship mapping between intent and network configuration, thereby overcoming to some extent the shortcomings of rule-based policy engines and static optimization heuristics. However, this type of method still has three shortcomings: First, its ability to parse the intent of long text context in edge network scenarios is limited. The intent parsing of long text interaction sequences requires accurate semantic extraction and analysis capabilities, which rely on powerful computing resources. Edge networks usually lack high-performance computing facilities, which leads to the loss of temporal coherence in long text context intent parsing and an increase in model inference latency. Second, there is an action space trap. In DRL-based methods, the combination of high-dimensional policy spaces can easily cause reinforcement learning agents to get stuck in local optima and fail to obtain globally optimal network optimization configuration strategies. Third, the noise in intent expression leads to a decrease in model performance. Since users are mostly non-professional Internet of Things (IB-IoT) personnel who are not professionals in intent-based networks, their intent expressions may be ambiguous or even erroneous, which can lead to the model training data being mislabeled, resulting in a significant decrease in model performance when facing non-training set data after deployment.

[0007] In view of this, the present invention is hereby proposed. Summary of the Invention

[0008] To address the aforementioned technical problems in existing technologies, this invention provides a modeling method for IoT optimization problems based on intent networks and drone assistance. This method solves the problems of incomplete intent coverage, rigid mapping, and difficulty in parsing intent from long text contexts in existing intent network-based IoT technologies.

[0009] To achieve the above objectives, the technical solution of the present invention is as follows:

[0010] Modeling methods for IoT optimization problems based on intent networks and drones include:

[0011] S1. Construct an intent-based Internet of Things (IoT) network scenario, wherein the IoT network scenario includes at least one communication base station, one or more users, and one or more drones; define the processing path for intents generated by user interaction with smart devices; use the intent network to read device status, use time-based historical control information, and infer potential user needs based on the historical control information;

[0012] S2. Model the intentions generated by the interaction between the user and the smart device in the IoT network scenario, and extract the user's historical intentions and infer the user's current intentions based on the device status, usage time and other historical control information.

[0013] S3. Based on the improved continuous crowd model, user mobility is modeled. After determining the user's movement cost, minimum movement direction and instantaneous speed, the user's location is updated in real time using a preset user location update method.

[0014] S4. Model the communication process between the UAV and the user, and between the UAV and the communication base station. Based on the scale fading of the wireless channel, calculate the channel capacity of the corresponding communication link in combination with the signal-to-noise ratio.

[0015] S5. Based on the channel capacity, using a preset user demand processing delay method, calculate the transmission delay of the UAV transmitting user demands to the base station and the direct processing delay of user demands by the UAV.

[0016] S6. Based on the transmission delay and demand delay, a user satisfaction function in the form of a piecewise function is used to characterize user satisfaction, a user experience quality model is constructed, and then an optimization objective function is established with the goal of maximizing the average user experience quality of multiple users and multiple time slots. Under the constraints of user action definition, user location update, UAV communication parameter calculation and user experience quality model parameters, the optimization objective function is solved to achieve intent-based IoT network optimization processing.

[0017] Furthermore, the processing path for the intent generated by the interaction between the user and the smart device includes: the drone directly processes the user intent, the drone forwards the user intent to the communication base station for processing; and each drone has a preset upper limit on the number of users it can serve, and user intents not scheduled by the drone are automatically submitted to the communication base station for processing.

[0018] Furthermore, specific information about a single action includes the time the action occurred, the corresponding device identifier, the action duration, and the control command; user historical intents. The mathematical representation formula is:

[0019]

[0020] in, For the length of historical time, For a moment Action sequence, For a moment Action sequence;

[0021]

[0022] in, For a single action;

[0023]

[0024] in, Indicates the time when the action occurs. Indicates the device that performs the action. Indicates the duration of the action. This indicates the control command corresponding to the action.

[0025] Furthermore, the user mobility cost The specific calculation formula is as follows:

[0026]

[0027] in, For the user's location, For the user's direction of movement, The maximum feasible speed function is determined by both crowd density and the user's current direction of movement. The cost of movement due to terrain obstacles, These are the weight parameters.

[0028] Furthermore, the maximum feasible velocity function is in piecewise form, specifically as follows:

[0029]

[0030] in, For population density, For population density Below The maximum feasible speed for users at that time For population density Higher than User movement speed at that time and This is the experimental calibration constant.

[0031] Furthermore, the minimum motion direction in step S3 adopts the Eikonal equation:

[0032]

[0033] in, For the direction of least cost movement, For the user's location, This is the final direction of movement;

[0034] User instantaneous speed satisfies: ; The user's movement speed in the x-axis direction. Let the user's movement speed be in the y-axis direction. The user position update formula is:

[0035]

[0036] in, for The position vector at time , for Moment Axis coordinates for Moment Axis coordinates For time intervals.

[0037] Furthermore, the coordinates of the communication base station Defined as:

[0038]

[0039] in, Let x be the coordinate of the communication base station on the x-axis. Let be the coordinates of the communication base station on the y-axis. Let Z be the coordinates of the communication base station on the z-axis.

[0040] The coordinates of the drone Defined as:

[0041]

[0042] in, Let x be the coordinates of the drone on the x-axis. Let be the y-coordinate of the drone. Let be the coordinates of the UAV on the z-axis;

[0043] The user's coordinates Defined as:

[0044]

[0045] in, For the user's x-coordinate, For the user's y-coordinate, This represents the user's coordinates on the z-axis.

[0046] 3D distance between the drone and the communication base station for:

[0047]

[0048] in, The coordinates of the drone. The coordinates of the communication base station;

[0049] The horizontal distance between the drone and the communication base station for:

[0050]

[0051] in, Let x be the coordinates of the drone on the x-axis. Let x be the coordinate of the communication base station on the x-axis. Let be the y-coordinate of the drone. Let y be the coordinate of the communication base station on the y-axis;

[0052] The elevation angle between the drone and the communication base station for:

[0053]

[0054] in, It is the arctangent function. Let be the coordinates of the UAV on the z-axis. The horizontal distance between the drone and the communication base station.

[0055] Furthermore, the large-scale fading is calculated based on path loss and shadow fading factors, and the probability of line-of-sight link fading between the UAV and the user is also considered. Represented as:

[0056]

[0057] Probability of line-of-sight link fading between drones and communication base stations Represented as:

[0058]

[0059] in, e is the natural constant. These are calibration parameters based on actual environmental measurements. The elevation angle between the drone and the communication base station. The elevation angle between the drone and the user;

[0060] Signal fading value of the line-of-sight link between the drone and the user Represented as:

[0061]

[0062] Signal fading value of the line-of-sight link between the drone and the communication base station Represented as:

[0063]

[0064] in, The 3D distance between the drone and the communication base station. 3D distance between the drone and the user The wavelength of the wireless signal. This is the path loss index. For reference distance;

[0065] Total fading between the drone and the user Represented as:

[0066]

[0067] Total fading between drones and communication base stations Represented as:

[0068]

[0069] in, For additional losses, This represents the signal fading value of the line-of-sight link between the drone and the user. This represents the signal attenuation value of the line-of-sight link between the drone and the communication base station.

[0070] Signal gain due to large-scale fading between drones and users Represented as:

[0071]

[0072] Signal gain due to large-scale fading between drones and communication base stations Represented as:

[0073]

[0074] in, and Is and Shadow decay in the environment, and These are measured values ​​under the current environmental conditions. and The log-normal distribution describes the fading pattern of shadows.

[0075] Furthermore, the small-scale fading is based on the Rician model, the model expression of which is:

[0076]

[0077] in, This is the initial phase of the UAV signal receiving antenna. This represents noise in the wireless channel, which follows a complex Gaussian distribution. That is, a complex Gaussian distribution, where 0 represents the mean of the distribution and 1 represents the variance of the distribution;

[0078] Parameters of drones and users in the Rician model Represented as Parameters of drones and communication base stations Represented as The specific formula is as follows:

[0079]

[0080]

[0081] in , These are channel measurements. express The measured value of the channel gain when it is 0. This indicates that the current environment channel gain changes with The rate of change The minimum value of the current wireless channel gain;

[0082] The value of small-scale fading in the channel is calculated as follows:

[0083]

[0084] .

[0085] Furthermore, the user experience quality model includes a satisfaction function related to user latency tolerance and a satisfaction function without drone assistance, specifically including:

[0086] The satisfaction function related to user latency tolerance is denoted as: Based on the probability density function of the β distribution, the following is calculated: when the actual response time t satisfies hour, ,when or hour, ;

[0087] in, Let be the probability density function of the β distribution. , It is a Beta function and , It is a gamma function and , This represents the user's maximum tolerance for transmission latency. Set to a fixed value This is a hyperparameter that controls the growth of the function.

[0088] Compared with existing technologies, the above-mentioned IoT optimization problem modeling method based on intent networks and drones provided by this invention includes: constructing an IB-IoT scenario containing communication base stations, users, and drones; defining three types of intent processing paths; using IBN to read historical control information authorized by users to infer potential needs; representing users' historical intents mathematically; capturing two types of correlations to infer current intents; improving the continuous crowd model; calculating mobility costs, speed, and location updates; integrating two types of channel fading; calculating capacity and time delay; constructing a QoE model containing a piecewise utility function; normalizing QoE; and establishing an optimization objective that maximizes the average QoE across multiple users and time slots. This solves the problems of incomplete intent coverage, rigid mapping, and difficulty in parsing intents from long text contexts in existing intent network-based IoT technologies. Attached Figure Description

[0089] Figure 1 A flowchart illustrating the IoT optimization problem modeling method provided in this embodiment of the invention. Detailed Implementation

[0090] The technical solution of the present invention will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are not all embodiments of the present invention. All other embodiments obtained by those skilled in the art without creative effort are within the protection scope of the present invention.

[0091] It should be noted that, unless otherwise specifically stated, the relative arrangement and numerical expressions of the components and steps described in these embodiments should not be construed as limiting the scope of the invention.

[0092] The following description of exemplary embodiments is merely illustrative and is not intended to limit the invention or its application or use in any way. Techniques, methods, and apparatus known to those skilled in the art may not be discussed in detail herein, but where applicable, such techniques, methods, and apparatus should be considered part of this specification.

[0093] Example

[0094] See Figure 1 , Figure 1 The specific steps of the IoT optimization problem modeling method based on intent networks and UAV assistance proposed in this invention may include:

[0095] S1. Construct an intent-based Internet of Things (IoT) network scenario, which includes at least one communication base station, one or more users, and one or more drones; define the processing path for intents generated by user interaction with smart devices; use the intent network to read device status, use time-based historical control information, and infer potential user needs based on the historical control information;

[0096] A general scenario for IB-IoT is constructed, assuming that the IB-IoT network consists of one communication base station. , individual users , One drone Composition. At any given moment Each user may trigger a certain intent. The intention to interact with a smart device might be processed by a drone, denoted as... It may also be forwarded by the drone to the communication base station for processing, denoted as... In this scenario, each drone's capabilities can only serve... Because there are multiple users, some user requests may be unprocessable by drones. These user requests will be automatically submitted to the communication base station. , recorded as .

[0097] Typically, users neither directly describe nor submit their intentions, as they are usually non-professional consumers who lack the ability to accurately describe their intentions. Instead, users authorize IBN to access historical control information, enabling IBN to infer the user's potential needs. Historical control information includes device status and usage time. IBN extracts this information to infer user intentions and proactively prepares for the next step, thereby mitigating misunderstandings caused by users' inability to accurately describe their intentions.

[0098] S2. Model the intentions generated by the interaction between the user and the smart device in the IoT network scenario, and extract the user's historical intentions and infer the user's current intentions based on the device status, usage time and other historical control information.

[0099] For example, when a user is controlling laundry-related devices (washing machine and dryer), they may be controlling other devices (TV and water purifier) ​​at the same time, because controlling laundry-related devices does not occupy all of the user's time and energy during this time slot.

[0100] Therefore, the sequence of actions within a time slot This can lead to unstable intents. Therefore, IBN needs to accurately infer the user's current intent from historical intents. Intents at any future moment often have multiple dependencies on historical intents; for example, the frequency of previous device actions significantly influences the likelihood of subsequent actions (e.g., turning on and off the TV); furthermore, time patterns are important, such as the probability of turning off lights at dawn being far greater than at dusk. Therefore, IBN collectively captures both "intent-intent" and "time-intent" correlations.

[0101] This invention will obtain users from IBN The historical intention is expressed as:

[0102] , (Formula 1)

[0103] in Indicates the length of historical time.

[0104] (Formula 2)

[0105] This represents each action sequence. To represent each action,

[0106] , (Formula 3)

[0107] in Indicates the time when the action occurs. Indicates the device that performs the action. Indicates the duration of the action. This indicates the control command corresponding to the action.

[0108] S3. Based on the improved continuous crowd model, user mobility is modeled. After determining the user's movement cost, minimum movement direction and instantaneous speed, the user's location is updated in real time using a preset user location update method.

[0109] In IBN, user locations are neither static nor uniformly distributed around BS. This invention models user mobility based on a continuous crowd model, enabling accurate evaluation of AAV trajectory planning. In the continuous crowd model, three functions constrain the user's movement trajectory, as described in Equations 5 to 7 below. First, the user's location is defined as...

[0110] , (Formula 4)

[0111] in, , , The three-dimensional coordinates represent the user's initial position, where 0 indicates that only the user is considered to be on a horizontal plane.

[0112] When users move, there are certain costs involved. For example, the cost of moving is higher when the terrain is obstructed than when it is unobstructed, and the cost of moving is higher when the crowd is dense than when the crowd is sparse. The continuous crowd model defines the user's movement cost as...

[0113] , (Formula 5)

[0114] in Indicates the user's location. Indicates the user's movement direction. This represents the maximum feasible speed function determined by both crowd density and the user's current direction of movement. This indicates the cost of movement due to terrain obstacles. This represents the weighting parameter.

[0115] This invention applies to continuous population models The specific calculation method has been redefined as follows:

[0116] (Formula 6)

[0117] Compared with the original continuous population model In comparison, this invention proposes The goal is to make changes in user movement speed smoother, which is more in line with the reality of human movement, that is, the user's speed will not change abruptly, but will change gradually from slow to fast.

[0118] In formula 6, This indicates that when the population density is lower than At that time, the user's maximum speed limit; This indicates the user's movement speed during congestion. Regarding crowd density, this invention considers that when congestion exceeds a certain threshold... At this point, the user's movement speed can no longer be increased, i.e., it becomes a constant value. . and This is the experimental calibration constant.

[0119] For a given Assuming that users will choose to move in the direction with the lowest cost when moving, this invention uses Equation 7, namely the Eikonal equation, to solve for a given... The direction of minimum motion is denoted as . (Formula 7)

[0120] when Once determined (i.e., once the Far East direction with minimum cost for the user is determined), the user's current instantaneous velocity is updated according to Formula 6. Become (Formula 8)

[0121] Finally, the user's location information was updated as follows:

[0122] (Formula 9)

[0123] in, Indicates the length of the time slot.

[0124] S4. Model the communication process between the UAV and the user, and between the UAV and the communication base station. Based on the large-scale fading and small-scale fading of the wireless channel, calculate the channel capacity of the corresponding communication link in combination with the signal-to-noise ratio.

[0125] Drones inevitably encounter complex communication environments and various interference factors during their movement. This invention proposes a communication model for drone movement that comprehensively considers both large-scale and small-scale wireless channel fading. The coordinates of the communication base station are defined as follows: ,

[0126] The coordinates of the drone are defined as follows ,in It is a fixed value, meaning the drone's altitude is stable.

[0127] The user's coordinates are ;

[0128] Based on the above definition, the 3D distance between the drone and the communication base station Represented as:

[0129] (Formula 10)

[0130] The horizontal distance between the drone and the communication base station Represented as:

[0131] (Formula 11)

[0132] The elevation angle between the drone and the communication base station Represented as:

[0133] (Formula 12)

[0134] 3D distance between drone and user Represented as:

[0135] (Formula 13)

[0136] Horizontal distance between drone and user Represented as:

[0137] (Formula 14)

[0138] Angle of elevation between the drone and the user Represented as:

[0139] (Formula 15)

[0140] S41. Modeling the impact of large-scale fading on wireless channel gain: For large-scale fading, this invention comprehensively considers path loss and shadow fading factors to calculate the fading value.

[0141] Path loss consists of line-of-sight (LoS) fading and non-line-of-sight (LOS) fading. The probability of line-of-sight (LoS) fading is related to the wireless signal strength between the UAV and the user. Represented as:

[0142] (Formula 16)

[0143] Probability of line-of-sight link fading between drones and communication base stations Represented as:

[0144]

[0145] in, e is the natural constant. These are calibration parameters based on actual environmental measurements. The elevation angle between the drone and the communication base station. The elevation angle between the drone and the user;

[0146] Signal fading value of the line-of-sight link between the drone and the user Represented as:

[0147] (Formula 18)

[0148] Signal fading value of the line-of-sight link between the drone and the communication base station Represented as:

[0149]

[0150] in, The 3D distance between the drone and the communication base station. 3D distance between the drone and the user The wavelength of the wireless signal. This is the path loss index. For reference distance;

[0151] Total fading between the drone and the user Represented as:

[0152]

[0153] Total fading between drones and communication base stations Represented as:

[0154] (Formula 21)

[0155] in, For additional losses, This represents the signal fading value of the line-of-sight link between the drone and the user. This represents the signal attenuation value of the line-of-sight link between the drone and the communication base station.

[0156] Signal gain due to large-scale fading between drones and users Represented as:

[0157]

[0158] Signal gain due to large-scale fading between drones and communication base stations Represented as:

[0159]

[0160] in, and Is and Both shadow fading in the environment and shadow fading follow a log-normal distribution. and The measured values ​​for the current environment represent the variance of the log-normal distribution; and To describe the log-normal distribution that governs shadow fading, and It is obtained by sampling from the log-normal distribution that describes the fading pattern of shadows.

[0161] S42. Modeling the impact of small-scale fading on wireless channel gain: For small-scale fading scenarios, this invention uses the Rician model for modeling, with the specific formula as follows:

[0162] (Formula 24)

[0163] in, This is the initial phase of the UAV signal receiving antenna. This represents noise in the wireless channel, which follows a complex Gaussian distribution. That is, a complex Gaussian distribution, where 0 represents the mean of the distribution and 1 represents the variance of the distribution. Generally speaking, 0 and 1 are more commonly used, but when conditions permit, the measured values ​​of the current scenario can also be used.

[0164] Parameters of drones and users in the Rician model Represented as Parameters of drones and communication base stations Represented as ,

[0165] (Formula 25)

[0166] (Formula 26)

[0167] in, , These are channel measurements. express The measured value of the channel gain when it is 0. This indicates that the current environment channel gain changes with The rate of change The minimum value of the current wireless channel gain.

[0168] The value of small-scale fading in the channel is calculated as follows:

[0169] (Formula 27)

[0170]

[0171] S43. Modeling the combined impact of large-scale fading and small-scale fading: After comprehensively considering the channel gain and noise power of large-scale fading and small-scale fading, a method for calculating the total signal-to-noise ratio of the channel is proposed, and the channel capacity of the channel is calculated according to Shannon's formula.

[0172] This invention addresses the IB-IoT paradigm, where user intentions are transformed into action sequences, which in turn are converted into a series of tasks. This invention uses task transmission latency to represent user satisfaction with IB-IoT; lower latency equates to higher satisfaction, while higher latency results in lower satisfaction. Therefore, based on the ratio of task size to channel capacity, the task transmission latency can be obtained, thus reflecting the user's satisfaction with the task execution.

[0173] S431. The modeling method for signal-to-noise ratio is as follows:

[0174] The linear signal-to-noise ratio (SNR) between the drone and the communication base station is expressed as: The linear signal-to-noise ratio (SNR) of the wireless channel between the drone and the user is expressed as: The specific formula is as follows:

[0175] (Formula 28)

[0176] (Formula 29)

[0177] in, It is transmission power; and Antenna gain; It is the noise power spectral density; It is the system bandwidth, measured in Hertz.

[0178] S432, The channel capacity modeling method is as follows:

[0179] The channel capacity between the drone and the communication base station is expressed as The channel capacity between the drone and the user is expressed as Based on Shannon's formula, we can obtain:

[0180] (Formula 30)

[0181] (Formula 31)

[0182] S433. The processing delay of user tasks is represented as follows:

[0183] (Formula 32)

[0184] (Formula 33)

[0185] in, This indicates the transmission latency from the drone transmitting user requests to the communication base station. This indicates that the drone directly handles the latency that the user is aware of.

[0186] S5. Based on the channel capacity, using a preset user demand processing delay method, calculate the transmission delay of the UAV transmitting user demands to the communication base station and the delay of the UAV directly processing user demands.

[0187] Based on the aforementioned IBN network modeling, user intent modeling, user mobility modeling, and UAV communication modeling methods, this section unifies the various models involved in the system model into a single user QoE model and further clarifies the optimization objective of this invention. This invention is based on a common, general QoE model, with the specific formula as follows:

[0188] (Formula 34)

[0189] in, Indicates user Intention The actual response time, which represents the communication delay referred to in Formulas 32 and 33. Indicates user Intent The utility function represents the user's satisfaction with task processing. This function is a time-decreasing function, meaning that user satisfaction decreases over time. This invention proposes a solution based on the idea of ​​the sigmoid function. The specific definition of .

[0190] This invention is defined as being based on a common general QoE model, with the specific formula being:

[0191] (Formula 35)

[0192] in, This is the first inflection point of QoE; before this, QoE declines slowly. This indicates that the user communicates directly with the communication base station without going through the drone, meaning the drone is unable to capture the user's intent, therefore... Set to a larger fixed value This marks the second turning point in QoE, after which QoE bottoms out.

[0193] and For pre-set hyperparameters, Determines the rate of decrease of the first segment of the function OoE. Control the rate of decrease of QoE in the second function segment. Indicates the delay of the task. and The user's maximum QoE value between these values. Calculation method , and It is a weighting coefficient used to map the value of Formula 35 to the range of 0-1, and thus represent user satisfaction.

[0194] This represents the user's satisfaction level when their needs are met within a specific time period, and its value ranges from 0 to 1. This indicates the upper limit of the user's tolerance for transmission latency. At that time, user satisfaction was not affected by whether or not drones assisted. The specific calculation method is based on the probability density function of the β distribution. calculate,

[0195] (Formula 36)

[0196] (Formula 37)

[0197] (Formula 38)

[0198] therefore, (Formula 39)

[0199] in, It is the probability density function of the beta distribution. and It is a hyperparameter that controls the growth of the function. Representing users at time Subsequently, when the QoE is met directly by the communication base station without the assistance of drones, this invention employs a Sigmoid curve design. The calculation method takes values ​​between 0 and 1:

[0200] (Formula 40)

[0201] in, This determines the steepness of the function, that is, the rate at which the function grows.

[0202] S6. Based on the aforementioned delay, a user satisfaction function in the form of a piecewise function is used to characterize user satisfaction, a user experience quality model is constructed, and then an optimization objective function is established with the goal of maximizing the average user experience quality across multiple users and time slots. Under the constraints of user action definition, user location update, UAV communication parameter calculation, and user experience quality model parameters, the optimization objective function is solved to achieve intent-based IoT network optimization processing.

[0203] Based on formula 34, and in conjunction with formulas 35, 39, and 40, this invention determines... The range of values ​​is This yields the following conversion formula:

[0204] (Formula 41)

[0205] However, Since only a single user's single intent is considered, for scenarios with multiple users and multiple intents, this invention proposes the following final model optimization objective function:

[0206] (Formula 42)

[0207]

[0208] in, It is a time slot A set of.

[0209] For the drone-assisted Internet of Things (IB-IoT) optimization scenario based on intent networks, this invention obtains the optimal solution of formula 42 under the constraints of formulas 3, 9, 10-32, and 34-40. This enables the optimal intent prediction and conversion, as well as the maximization of average user satisfaction, thereby achieving efficient optimization of IB-IoT.

[0210] In summary, the present invention has the following advantages:

[0211] 1. By reading historical control information through IBN, user needs can be inferred, thus solving the problems of incomplete intent coverage and noise in user expression;

[0212] 2. Improve the continuous crowd model by constructing a movement cost function that integrates crowd density, terrain obstacles, etc., to achieve smooth transitions in speed segments, dynamically update user locations, and adapt to drone trajectory planning.

[0213] 3. By combining large-scale and small-scale fading of wireless channels with Shannon's formula to calculate channel capacity, the latency calculation is more accurate and reduces errors;

[0214] 4. The QoE model uses piecewise functions to simulate changes in user psychology. The optimization objective is to maximize the average QoE of multiple users and multiple time slots. It clearly defines multi-dimensional constraints, avoids local optima, and is adapted to large-scale networks.

[0215] The above specific embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to examples, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for modeling IoT optimization problems based on intent networks and UAV assistance, characterized in that, include: S1. Construct an intent-based Internet of Things (IoT) network scenario, wherein the IoT network scenario includes at least one communication base station, one or more users, and one or more drones; define the processing path for intents generated by user interaction with smart devices; The intent network is used to read device status and use time-based historical control information, and the user's potential needs are inferred based on the historical control information. S2. Model the intentions generated by the interaction between the user and the smart device in the IoT network scenario, and extract the user's historical intentions and infer the user's current intentions based on the device status, usage time and other historical control information. S3. Based on the improved continuous crowd model, user mobility is modeled. After determining the user's movement cost, minimum movement direction and instantaneous speed, the user's location is updated in real time using a preset user location update method. S4. Model the communication process between the UAV and the user, and between the UAV and the communication base station. Based on the large-scale fading and small-scale fading of the wireless channel, calculate the channel capacity of the corresponding communication link in combination with the signal-to-noise ratio. S5. Based on the channel capacity, using a preset user demand processing delay method, calculate the transmission delay of the UAV transmitting user demands to the base station and the direct processing delay of user demands by the UAV. S6. Based on the transmission delay and demand delay, a user satisfaction is characterized using a piecewise function, and a user experience quality model is constructed. Then, an optimization objective function is established to maximize the average user experience quality across multiple users and time slots. Under constraints of user action definition, user location update, UAV communication parameter calculation, and user experience quality model parameters, the optimization objective function is solved to achieve intent-based IoT network optimization. The specific optimization objective function is as follows: in, Indicates user Intent Utility function; It indicates the satisfaction level of a user's needs being met within a specific time period; Representing users at time Then, the user experience quality when the communication base station directly meets its needs without the assistance of drones.

2. The method for modeling IoT optimization problems based on intent networks and UAV assistance as described in claim 1, characterized in that, The processing path for user intents generated by interaction with smart devices includes: the drone directly processes the user intent, the drone forwards the user intent to the communication base station for processing; and each drone has a preset upper limit on the number of users it can serve, and user intents not scheduled by the drone are automatically submitted to the communication base station for processing.

3. The method for modeling IoT optimization problems based on intent networks and UAV assistance as described in claim 1, characterized in that, Specific information for a single action includes the time the action occurred, the corresponding device identifier, the action duration, and the control command; user's historical intents. The mathematical representation formula is: in, For the length of historical time, For a moment Action sequence, For a moment Action sequence; in, For a single action; in, Indicates the time when the action occurs. Indicates the device that performs the action. Indicates the duration of the action. This indicates the control command corresponding to the action.

4. The method for modeling IoT optimization problems based on intent networks and UAV assistance as described in claim 1, characterized in that, User mobility costs The specific calculation formula is as follows: in, For the user's location, For the user's direction of movement, The maximum feasible speed function is determined by both crowd density and the user's current direction of movement. The cost of movement due to terrain obstacles, These are the weight parameters.

5. The method for modeling IoT optimization problems based on intent networks and UAV assistance according to claim 4, characterized in that, The maximum feasible velocity function is in piecewise form, specifically as follows: in, For population density, For population density Below The maximum feasible speed for users at that time For population density Higher than User movement speed at that time and This is the experimental calibration constant.

6. The method for modeling IoT optimization problems based on intent networks and UAV assistance according to claim 1, characterized in that, The minimum motion direction in step S3 adopts the Eikonal equation: in, For the direction of least cost movement, For the user's location, The final direction of movement; User instantaneous speed satisfies: ; The user's movement speed in the x-axis direction. Let the user's movement speed be in the y-axis direction. The user position update formula is: in, for The position vector at time , for Moment Axis coordinates for Moment Axis coordinates For time intervals.

7. The method for modeling IoT optimization problems based on intent networks and UAV assistance as described in claim 1, characterized in that, The coordinates of the communication base station Defined as: in, Let x be the coordinate of the communication base station on the x-axis. Let be the coordinates of the communication base station on the y-axis. Let Z be the coordinates of the communication base station on the z-axis. The coordinates of the drone Defined as: in, Let x be the coordinates of the drone on the x-axis. Let be the y-coordinate of the drone. Let be the coordinates of the UAV on the z-axis; The user's coordinates Defined as: in, For the user's x-axis coordinates, For the user's y-coordinate, This represents the user's coordinates on the z-axis. The 3D distance between the drone and the communication base station for: in, The coordinates of the drone, The coordinates of the communication base station; The horizontal distance between the drone and the communication base station for: in, Let x be the coordinates of the drone on the x-axis. Let x be the coordinate of the communication base station on the x-axis. Let be the y-coordinate of the drone. The coordinates of the communication base station on the y-axis; The elevation angle between the drone and the communication base station for: in, It is the arctangent function. Let be the coordinates of the UAV on the z-axis. The horizontal distance between the drone and the communication base station.

8. The method for modeling IoT optimization problems based on intent networks and UAV assistance according to claim 1, characterized in that, The large-scale fading is calculated based on path loss and shadow fading factors, representing the probability of line-of-sight link fading between the UAV and the user. Represented as: Probability of line-of-sight link fading between drones and communication base stations Represented as: in, e is the natural constant. These are calibration parameters based on actual environmental measurements. The elevation angle between the drone and the communication base station. The elevation angle between the drone and the user; Signal fading value of the line-of-sight link between the drone and the user Represented as: Signal fading value of the line-of-sight link between the drone and the communication base station Represented as: in, The 3D distance between the drone and the communication base station. 3D distance between the drone and the user The wavelength of the wireless signal. This is the path loss index. For reference distance; Total fading between the drone and the user Represented as: Total fading between drones and communication base stations Represented as: in, For additional losses, This represents the signal fading value of the line-of-sight link between the drone and the user. This represents the signal attenuation value of the line-of-sight link between the drone and the communication base station. Signal gain due to large-scale fading between drones and users Represented as: Signal gain due to large-scale fading between drones and communication base stations Represented as: in, and Is and Shadow decay in the environment, and These are measured values ​​under the current environmental conditions. and The log-normal distribution describes the fading pattern of shadows.

9. The method for modeling IoT optimization problems based on intent networks and UAV assistance according to claim 1, characterized in that, The small-scale fading is based on the Rician model, and the model expression is: in, This is the initial phase of the UAV signal receiving antenna. This represents noise in the wireless channel, which follows a complex Gaussian distribution. That is, a complex Gaussian distribution, where 0 represents the mean of the distribution and 1 represents the variance of the distribution; Parameters of drones and users in the Rician model Represented as Parameters of drones and communication base stations Represented as The specific formula is as follows: in , These are channel measurements. express The measured value of channel gain when it is 0. This indicates that the current environment channel gain changes with The rate of change The minimum value of the current wireless channel gain; The value of small-scale fading in the channel is calculated as follows: 。 10. The method for modeling IoT optimization problems based on intent networks and UAV assistance according to claim 1, characterized in that, The user experience quality model includes a satisfaction function related to user latency tolerance and a satisfaction function without drone assistance, specifically including: The satisfaction function related to user latency tolerance is denoted as: Based on the probability density function of the β distribution, the following is calculated: when the actual response time t satisfies hour, ,when or hour, ; in, Let be the probability density function of the β distribution. , It is a Beta function and , It is a gamma function and , This represents the user's maximum tolerance for transmission latency. Set to a fixed value This is a hyperparameter that controls the growth of the function.

Images