CN113872666A - Unmanned aerial vehicle deployment method based on Backhaul capacity constraint in dense urban area - Google Patents
Unmanned aerial vehicle deployment method based on Backhaul capacity constraint in dense urban area Download PDFInfo
- Publication number
- CN113872666A CN113872666A CN202111102927.2A CN202111102927A CN113872666A CN 113872666 A CN113872666 A CN 113872666A CN 202111102927 A CN202111102927 A CN 202111102927A CN 113872666 A CN113872666 A CN 113872666A
- Authority
- CN
- China
- Prior art keywords
- unmanned aerial
- aerial vehicle
- user
- base station
- link
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B7/00—Radio transmission systems, i.e. using radiation field
- H04B7/14—Relay systems
- H04B7/15—Active relay systems
- H04B7/185—Space-based or airborne stations; Stations for satellite systems
- H04B7/18502—Airborne stations
- H04B7/18504—Aircraft used as relay or high altitude atmospheric platform
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W16/00—Network planning, e.g. coverage or traffic planning tools; Network deployment, e.g. resource partitioning or cells structures
- H04W16/18—Network planning tools
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W16/00—Network planning, e.g. coverage or traffic planning tools; Network deployment, e.g. resource partitioning or cells structures
- H04W16/22—Traffic simulation tools or models
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D30/00—Reducing energy consumption in communication networks
- Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Physics & Mathematics (AREA)
- Astronomy & Astrophysics (AREA)
- Aviation & Aerospace Engineering (AREA)
- General Physics & Mathematics (AREA)
- Mobile Radio Communication Systems (AREA)
Abstract
The invention provides a unmanned aerial vehicle deployment method based on Backhaul capacity constraint in a dense urban area. In the method, the shielding of a building and the movement of a user need to be considered, and the three-dimensional motion track model of the unmanned aerial vehicle is established through a reinforcement learning algorithm to improve the system and the speed. The ground base station and the unmanned aerial vehicle base station establish a Backhaul link, the unmanned aerial vehicle base station selects a user to Access and establishes an Access link by considering the capacity of the Backhaul link, and judges a line-of-sight link and a non-line-of-sight link through a Fresnel zone ray model. The first step is a reinforcement learning deployment algorithm, namely an unmanned aerial vehicle finds a correct initial position according to the positions of a building and a user; and the second step is a reinforcement learning movement algorithm, which allows the user to move and continuously iterate the position of the unmanned aerial vehicle, and finds the optimal motion track to adapt to the complex user change.
Description
Technical Field
The invention relates to the technical field of wireless communication, in particular to deployment design of Unmanned Aerial Vehicle (UAV) base station groups based on Backhaul capacity constraint in dense urban areas.
Background
In recent years, the rapid development of 5G communication brings great convenience to our lives. As part of the 5G proposal, drone assisted cellular network communications are also increasingly known. Under the environment of multiple unmanned aerial vehicle base stations, it is a hot topic that an unmanned aerial vehicle is used as a base station to improve the signal quality of a ground user. Furthermore, drones are extensively studied by communication industry researchers due to their high line-of-sight link and high mobility. This is because, the problem that ground basic station can't be solved can be solved to the unmanned aerial vehicle basic station:
problem of insufficient supply of ground base station: with the development of science and technology, people have more and more electronic devices and are used more and more frequently. Therefore, the amount of data exchanged by people in the fields of entertainment, life, etc. is increasing, and terrestrial base stations have not been able to meet the rapidly increasing capacity demands. However, the cost of expanding a new terrestrial base station is too high, and a large amount of terrestrial resources are occupied. The unmanned aerial vehicle base station has the advantages of low cost, small volume, flexible action, no occupation of ground resources and the like; the problem of poor signal in hot spots: with the improvement of the living standard of people, the number of people in sports stadiums and concerts is increased, but the signal problem inside the stadiums is still suffered from scaling. The crowd suddenly gathers in a short time, and great communication pressure is caused to the ground base station. If the ground base station is saturated, a plurality of unmanned aerial vehicles are moved to a relevant area to serve as the unmanned aerial vehicle base station to share the pressure of the ground base station, so that the effects of improving the communication quality and avoiding communication overload can be achieved; building shading problems; with the development of the urbanization process, the density of buildings is highly increased year by year, so that a line-of-sight (LOS) link of a ground station is blocked, and the communication quality is reduced. The unmanned aerial vehicle has the characteristic of a high LOS link, can avoid shielding of a building, and improves communication of users.
In the scene of intensive urban area, unmanned aerial vehicle base station has the characteristics of nimble quick deployment and high-line-of-sight transmission, however along with the increase of building density, there is the dual challenge of collision risk and the sheltering from the link of building in unmanned aerial vehicle's removal in urban area. Because the buildings are dense in urban areas, the unmanned aerial vehicle needs to move to avoid the buildings, collision is avoided, and due to the increase of the number of the buildings and users, the shielding of links is more serious, which can cause unprecedented challenges to the path planning of the unmanned aerial vehicle. In theoretical analysis, the interference between unmanned aerial vehicle base stations can be reduced by traditional methods such as random geometry and greedy algorithm, and the number of unmanned aerial vehicle base stations is reduced. Therefore, an Artificial Intelligence (AI) -based algorithm has been widely used in recent years because it has an advantage of a high calculation speed in solving a complicated problem. The K mean value clustering algorithm combines users to enable the users to communicate with the unmanned aerial vehicle relay, and the problem of resource allocation based on power minimization is solved; reinforcement Learning (RL) algorithms have been used in drone deployment scenarios as the strongest branch of artificial intelligence, but the connection between the user and the drone is fixed. Therefore, artificial intelligence based backhaul optimization deployment that takes building blockage into account is a problem to be solved.
Disclosure of Invention
The invention considers the joint optimization deployment method of the access link, the return link and the building shelter in the dense urban area, and provides service for ground users. Specifically, a three-dimensional motion scene of a plurality of unmanned aerial vehicle base stations of a single ground macro base station is considered, and the distribution of buildings follows the parameters of density, height and number in a dense urban area scene; the unmanned aerial vehicle base stations jointly provide communication services for users in the area, and dynamically move in the area according to a certain sequence and a certain direction, so that the constraint limitation that only one unmanned aerial vehicle base station is arranged in one partition in the traditional method is eliminated, the flexibility of the unmanned aerial vehicle base stations is greatly improved, and the system is more stable; the user can freely move in the area and follow a random walk model (RWP), namely, the moving direction, the moving distance and the stopping probability are randomly selected;
the unmanned aerial vehicle base station deployment method comprises the following steps:
and 200, modeling the building position, the movement of the user and the movement of the UAV according to the scene of the dense urban area, and selecting the user to carry out Access link connection according to the Backhaul link capacity constraint.
The UAV obtains scene parameters of dense urban areas: coordinates (x) of terrestrial macro base stationss,ys,zs) Position of building (x)b,yb,zb) And length byWidth bxHigh bhHas a rectangular parallelepiped shape, the length and width of which followwmin＜bx＜by＜wmaxThe uniform distribution in the range, the height follows the Rayleigh distribution with the parameter gamma, and the density is the density parameter beta of the dense urban building. Deploying the unmanned aerial vehicle base station in an area, and discretizing the three-dimensional space into a small cube g with granularity of gx×gy×gzWherein g ═ gx＝gy＝gz. And then, carrying out Access connection between the user and the unmanned aerial vehicle base station, considering backhaul capacity constraint, and allowing the user to Access the network to the maximum extent: all users select the unmanned aerial vehicle base station with the highest signal intensity to request access, and form a first group to-be-accessed group of the unmanned aerial vehicle base station j
Then the unmanned aerial vehicle base station slave
Selecting users to be accessed according to the sequence of signal intensity from large to small, and when the unmanned aerial vehicle accesses the average signal intensity of the link
Less than the signal strength of the backhaul link
Then the user is granted access, where NjThe number of the users that the unmanned aerial vehicle j has accessed is the number of the users that the unmanned aerial vehicle j has accessed, and after all the unmanned aerial vehicles select the users to access, the remaining users that are not accessed continue to select the unmanned aerial vehicle with the second highest signal intensity to carry out the next round of access process until the user coverage rate reaches xi or the number of the access rounds is larger than the total number V of the unmanned aerial vehicles.
Judging the standard reference Fresnel zone theory of LOS and NLOS, if a building exists in the minimum Fresnel zone, judging the LOS: the specific method is that a connecting line l from the unmanned aerial vehicle base station to the user is divided into n points lii ∈ 1, 2.. n, and calculate liPoint minimum fresnel zone radius
Wherein d isi1,di2Is aiPoint to drone and user distance, λ is wavelength, then liThe radius is R with the point as the circle centeriWhen n → ∞ is reached, n circles approach to a rotational ellipsoid formed by the minimum fresnel zone, 4 × n points are taken at a uniform orientation on the n circles, the points at the same orientation are connected to form 4 curves, and whether the curves intersect with a building or not is judged to determine whether the curves are line-of-sight links.
Finally, generating a problem:
wherein C in (1a) and (1b)uAnd CvIs the sum rate of the Access link and the Backhaul link, and (1c) represents that joint optimization makes the capacity of the Access link restricted by the capacity of the Backhaul link. (1d) The system and rate take the average of the Access link. (1e) Height is limited for the drone, (1f) one user can only be served by one drone.
And step 210, determining the moving sequence of the unmanned aerial vehicle base station according to a reinforcement learning deployment algorithm, and finding the optimal initial position. The reinforcement learning is a self-learning model, global information does not need to be known, and only the external environment needs to be gradually explored according to the current environment. The agent takes action in the current environment and finds the path that can receive the greatest reward.
Q-learning is a value function-based Reinforcement Learning (RL) algorithm, and in a state of a certain time s, the expression Q (s, a) of the profit can be obtained by taking an action a, the environment gives a reward value r to the action, and then the next optimal action is selected through a Q matrix and iteration is continued until the optimal value is converged finally. The five elements of the RL parameter are defined as follows:
● agent: and the unmanned aerial vehicle changes the position of the unmanned aerial vehicle through the state selection model and the action selection model.
● State: the position of each unmanned aerial vehicle is defined as a three-dimensional coordinate(xUAV,yUAV,zUAV) The values are respectively xUAV∈{0,1,2,…,xmax},yUAV∈{0,1,2,…,ymax},zUAV∈{hmin,…,hmaxThe position of each user is defined as a two-dimensional coordinate (x)user,yuser) Is valued as
Wherein xmax,ymaxRepresenting the boundary of the scene, hmin,hmaxRepresenting the flight height limitations of the drone. The state space also includes the position of the building and is dispersed into a three-dimensional array, and the value is 0. When the unmanned aerial vehicle moves, whether the value of the moving direction of the unmanned aerial vehicle is 0 or not is judged.
● action: at every moment, unmanned aerial vehicle all can select a direction of motion, and only a kind of unmanned aerial vehicle can select the direction of movement at every moment, and other unmanned aerial vehicles are static. Simplifying the motion direction of the unmanned aerial vehicle into six motion directions and static from top to bottom, left to right, front to back:
action[x][y][z]∈[000,001,010,100,00-1,0-10,-100] (2)
● reward value: the size of the reward value affects the convergence speed of the algorithm, especially in large scenarios, which affects the final performance of the algorithm. The prize values followed by the invention are:
wherein, beta0And beta1Is a correlation coefficient of the prize value, CnextRepresenting the system and rate at the next time, CcurrentRepresents the system and rate at the current time, so that the next time the system and rate increase, the reward value is positive, when decreasing, it is negative, when not increasing or decreasing, the reward value is beta1。
● policy: there are two strategies that can let the drone act at the next moment of selection. The first is the optimal strategy:
the direction that maximizes overall system and velocity is selected:
the second strategy is to randomly choose the direction of movement to explore the largest space.
The reinforcement learning deployment algorithm is derived as follows:
inputting:
the state is as follows: user position U [0] at time t ═ 0](xuser[0],yuser[0]) Building position and length, width and height agents: unmanned aerial vehicle position V [0] at time t ═ 0](xuav[0],yuav[0],zuav[0])
Q table: q table Q0 (x 0, y 0, z 0, 7) at time t ═ 0
Strategy epsilon: probability of exploring the external environment
And (3) outputting:
the intelligent agent: t is tstopPosition vt of unmanned aerial vehicle at time](xuav[t],yuav[t],zuav[t])
Q table: t is tstopTime Q meter Q [ t ]](x[t],y[t],z[t],7)
The method comprises the following steps: judging the unmanned aerial vehicle v to move at the moment t:
i.e. by calculating the sum velocity of each drone and then selecting the smallest drone to move first.
Step two: selecting the direction a [ t ] in which drone v moves at time t:
that is, a random number r of 0-1 is generated, if r is larger than epsilon, the Q value is selected to be the largest in 7 directions, if r is smaller than epsilon, the direction is randomly selected, and the external environment is continuously explored. This is a process of exploring and using dynamic trade-offs, where the value of epsilon decreases gradually as the time t of operation increases.
Step three: change drone according to action a and update state s t + 1.
Step four: calculating the reward value for action a based on equation (2)
Step five: updating the Q table:
step six: t is t +1, when t > tstopAnd finishing the algorithm, otherwise, returning to the step one.
The algorithm is specifically as follows: and inputting environment state information, and initializing an agent position and a Q table. And enabling the unmanned aerial vehicle with the minimum sum rate to move, selecting one action option from the action options through an action selection algorithm, changing the environment state, and obtaining a new system and a new rate. And then determining the reward value according to the comparison of the front, back and speed, and further updating the Q table. Reciprocating cycle tstopAnd then, completing the deployment of the unmanned aerial vehicle base station.
And step 220, determining the movement track of the unmanned aerial vehicle base station following the user according to a reinforcement learning movement algorithm, and obtaining the unmanned aerial vehicle initial position and a Q table based on the algorithm. Increasing user movement based on step six
At the moment, the unmanned aerial vehicle needs to learn the movement track information of the user, move along with the user, continuously update the Q table and keep the best state. The specific algorithm is similar to the reinforcement learning deployment algorithm, but adds to the movement of the user.
Advantageous effects
The invention provides a method for dynamically deploying unmanned aerial vehicles based on a reinforcement learning algorithm, aiming at the problems of high building density and poor signal caused by frequent movement of users in dense urban areas. By utilizing the reinforcement learning algorithm, the limitation of the traditional algorithm in solving the large-scale scene problem is effectively solved; determining users connected with the unmanned aerial vehicle base station according to Backhaul capacity constraint; determining the optimal initial position of the unmanned aerial vehicle according to the position of the building; and determining the moving track of the unmanned aerial vehicle according to the moving track of the user. The method has guiding significance for deployment of the unmanned aerial vehicle base station in an actual urban scene.
The entire scene is initially modeled by different building locations and the initial location of the user. When the unmanned aerial vehicle base station serves, due to the change of the blocking condition of the obstacles and the frequent switching of line-of-sight (LOS) and non-line-of-sight (NLOS) links, the volatility of the links can be increased. Therefore, there is a need to make the connection of the user to the drone more stable, minimizing the fluctuations. Therefore, a reinforcement learning deployment algorithm is introduced, and the unmanned aerial vehicle base station finds a proper initialization position.
Modeling is carried out on user movement, a reinforcement learning movement algorithm is introduced, the movement track of the unmanned aerial vehicle is determined according to the movement track of the user, the system and the speed are maximized, and the optimal deployment of the unmanned aerial vehicle base station group is realized.
Drawings
FIG. 1 is a schematic diagram of a deployment model of unmanned aerial vehicles based on Backhaul capacity constraint in dense urban areas;
FIG. 2 is a flow chart of an algorithm implementation of the present invention;
FIG. 3 is a graph comparing the effects of three algorithms as the user moves;
FIG. 4 is a graph comparing the performance of three algorithms for different user densities and numbers of drones;
fig. 5 is a comparison graph of three algorithms for different user densities and drone transmit powers.
Detailed Description
The invention provides a deployment method of dynamic 3D movement of an unmanned aerial vehicle base station aiming at the characteristics of high building density and high user mobility in dense urban areas, wherein an uplink connection model is as shown in the attached drawing 1: in a dense urban area scene, a ground macro base station is arranged in the area and Backhaul connection is established between the ground macro base station and a plurality of unmanned aerial vehicle base stations; the user selects the unmanned aerial vehicle base station with the highest signal intensity to carry out an Access request, and the unmanned aerial vehicle base station selectively accesses the user according to the Backhaul capacity limit of the user to form an Access link; a certain number of buildings exist in the area, an Access link shielded by the buildings is called an NLOS link, and other links are LOS links; the deployment height and position of the unmanned aerial vehicle base station are dynamically adjusted along with the movement of the user. A Time Division Duplex (TDD) mode is used on the Access link, so the interference of the drone comes from the interference of a non-connected user to the drone in the same time slot.
When the unmanned aerial vehicle base station serves, due to the fact that the user continuously moves, the unmanned aerial vehicle moves along with the user, the change of the blocking condition of the obstacle is caused, line of sight (LOS) and non-line of sight (NLOS) links are frequently switched, and the volatility of the links is increased. Therefore, there is a need to reasonably match the connection of the user and the drone, minimizing the fluctuations. In a considered scene, on the premise that a user is motionless, a reasonable initial position of an unmanned aerial vehicle base station is found by applying a reinforcement learning deployment algorithm according to the initial position and the distribution of buildings, and an initialization Q table is generated; then, let the user begin to move, unmanned aerial vehicle carries out reasonable change to the position according to user's movement track to constantly update Q table.
The algorithm flow of this case is shown in fig. 2, and the specific implementation steps are as follows:
And 310, introducing a reinforcement learning deployment algorithm to find the optimal initial position of the unmanned aerial vehicle. Firstly, determining the moving sequence of the unmanned aerial vehicle base station, moving the unmanned aerial vehicle with the lowest capacity utilization rate first, and reasonably distributing resources. The Q table is then updated based on the reward value and the maximum system and rate are gradually approached until the Q table is substantially stable.
And 320, introducing a reinforcement learning movement algorithm, modeling the movement of the user, determining the movement track of the unmanned aerial vehicle base station along with the movement of the user, and keeping the optimal state.
The simulation results are shown in fig. 3-5. Wherein, part of parameters are set as: v. ofmin＝0.1，vmax＝1，fu＝fv＝2GHz，pu＝26dBm，pv＝46dBm，xmax＝1000，ymax＝1000hmin＝20，hmax＝100，g＝1，β0＝10000，β1＝1。
Figure 3 is a graph of the variation of sum rate with increasing user density at different drone powers. As the user density increases, the curve value increases, but the trend gradually slows down, indicating that as the number of users increases, the drone is overloaded and cannot provide high quality service for so many users. The network has a saturation point, and when the number of users reaches a certain threshold, the number of unmanned aerial vehicles needs to be increased.
FIG. 4 is a comparison of a reinforcement learning algorithm, a greedy algorithm, and a K-means algorithm. As the number of drones increases, the system and rate tend to increase and then decrease. The primary reason for this increase is that there are fewer drones in the early stages and fewer drones that the user can select. As the number of drones increases, the number of drones available to the user also increases, resulting in increased systems and rates. As the number of drones increases, which leads to a reduction in the sum rate, since the number of drones peaks, continuing to increase the number of drones leads to high interference between drones, which leads to a drop. We can also find that when the user density is different, the optimal number of drones is also different. When the user density is low, the optimal number of unmanned aerial vehicles is small. When the user density is higher, the optimal number of drones increases, which also indicates that it is better not to have more drones.
Figure 5 the user starts moving according to the initial position of the drone obtained by the reinforcement learning deployment algorithm. As can be seen from the figure, the reinforcement learning algorithm has a process of decreasing first and then increasing, which indicates that the unmanned aerial vehicle still learns the motion rules of the user in the initial stage, and the system and the speed start to increase after the unmanned aerial vehicle learns the motion rules. The greedy algorithm is a process of increasing first and then decreasing, that is, pursuit of high profit does not last for a long time, and is easy to fall into local optimization, resulting in poor performance. However, the K-means clustering algorithm has obvious fluctuation, which indicates that the algorithm is not suitable for a user moving scene and has low convergence speed.
Claims (7)
1. A unmanned aerial vehicle deployment method based on Backhaul capacity constraint under dense urban areas is characterized by comprising the following steps: building position, user movement and UAV movement modeling, and establishing an unmanned aerial vehicle three-dimensional motion track model through a reinforcement learning algorithm to improve the system and speed; the method comprises the steps that a background link is established between a ground base station and an unmanned aerial vehicle base station, the unmanned aerial vehicle base station selects a user to Access and establishes an Access link by considering the capacity of the background link, the line-of-sight and non-line-of-sight are judged on the link through a Fresnel region ray model, and different channel models are used; specifically, a two-step reinforcement learning method is proposed: the first step is a reinforcement learning deployment algorithm, namely, the unmanned aerial vehicle finds the correct initial position according to the positions of the building and the user; and the second step is a reinforcement learning movement algorithm, which allows the user to move and continuously iterate the position of the unmanned aerial vehicle, and finds the optimal motion track to adapt to the complex user change.
2. The method of claim 1, wherein the three-dimensional space is discretized into small cubes g of granularity gx×gy×gzWherein g ═ gx＝gy＝gzBuilding modeled as having a length byWidth bxHigh bhHas a length and width of wmin＜bx＜by＜wmaxThe uniform distribution in the range, the height follows the Rayleigh distribution with the parameter of gamma, and the density is the density parameter beta of the dense urban building; the user movement follows a random walk model, namely the user randomly selects a movement direction from 0 to 2 pi, randomly selects a movement distance and selects whether to reside according to the probability of tau; the unmanned aerial vehicle moves in three-dimensional space and is quantized into six parts, namely an upper part, a lower part, a left part, a right part, a front part and a rear partAnd moving the building by a unit length g at each moment, reselecting the moving direction when the building exists in the grid in the moving direction, and using a foldback strategy when the moving direction exceeds a specified moving range.
3. The method of claim 1, wherein one ground base station establishes Backhaul links with multiple drone base stations based on
Calculating backhaul link capacity, where B is total bandwidth, V is number of drones, SNRvIs the signal-to-noise ratio of the drone to base station link.
4. Method according to claim 1 or 3, characterized in that all users select the drone base station with the highest signal strength to request access and form the first group to be accessed of drone base station j
Then the unmanned aerial vehicle base station slave
Selecting users to be accessed according to the sequence of signal intensity from large to small, and when the unmanned aerial vehicle accesses the average signal intensity of the link
Less than the signal strength of the backhaul link
Then the user is granted access, where NjThe number of the users that the unmanned aerial vehicle j has accessed is the number of the users that the unmanned aerial vehicle j has accessed, and after all the unmanned aerial vehicles select the users to access, the remaining users that are not accessed continue to select the unmanned aerial vehicle with the second highest signal intensity to carry out the next round of access process until the user coverage rate reaches xi or the number of the access rounds is larger than the total number V of the unmanned aerial vehicles.
5. The method according to claim 1, wherein based on the fresnel zone theory, if there is architectural obstruction in the minimum fresnel zone between the drone base station and the user, the link is determined to be non-line-of-sight, specifically, the connection line l from the drone base station to the user is divided equally into n points lii ∈ 1, 2.. n, and calculate liPoint minimum fresnel zone radius
Wherein d isi1,di2Is aiPoint to drone and user distance, λ is wavelength, then liThe radius is R with the point as the circle centeriWhen n → ∞ is reached, n circles approach to a rotational ellipsoid formed by the minimum fresnel zone, 4 × n points are taken at a uniform orientation on the n circles, the points at the same orientation are connected to form 4 curves, and whether the curves intersect with a building or not is judged to determine whether the curves are line-of-sight links.
6. The method of claim 1, wherein to maximize system and speed Max ∑ C, the drone with the smallest sum speed is first selected for movement, the probability of the direction of movement having ∈ looks at the direction in the Q-table based on the current position Q value being the largest, the probability of 1- ∈ randomly selects the direction, updates the Q-table based on the reward value after movement, and recalculates the system and speed.
7. The method according to claim 1 or 6, characterized in that in the reinforcement learning deployment algorithm, the user does not move, and the drone base station moves to the initial position with the maximum velocity; in the reinforcement learning mobile algorithm, the unmanned aerial vehicle base station moves along with the user, learns information such as a user moving track and the like, and keeps a state of maximizing the speed.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111102927.2A CN113872666B (en) | 2021-09-15 | 2021-09-15 | Unmanned aerial vehicle deployment method based on Backhaul capacity constraint in dense urban area |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111102927.2A CN113872666B (en) | 2021-09-15 | 2021-09-15 | Unmanned aerial vehicle deployment method based on Backhaul capacity constraint in dense urban area |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113872666A true CN113872666A (en) | 2021-12-31 |
| CN113872666B CN113872666B (en) | 2023-01-31 |
Family
ID=78992923
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111102927.2A Active CN113872666B (en) | 2021-09-15 | 2021-09-15 | Unmanned aerial vehicle deployment method based on Backhaul capacity constraint in dense urban area |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113872666B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110381444A (en) * | 2019-06-24 | 2019-10-25 | 广东工业大学 | A kind of unmanned plane track optimizing and resource allocation methods |
| CN110392350A (en) * | 2019-08-06 | 2019-10-29 | 北京邮电大学 | A kind of user-center unmanned plane group construction method based on the perception of Backhaul capacity |
| CN111970713A (en) * | 2020-08-14 | 2020-11-20 | 北京邮电大学 | Unmanned aerial vehicle base station deployment method and three-dimensional parameter setting for dense urban area |
| CN112511197A (en) * | 2020-12-01 | 2021-03-16 | 南京工业大学 | Unmanned aerial vehicle auxiliary elastic video multicast method based on deep reinforcement learning |
-
2021
- 2021-09-15 CN CN202111102927.2A patent/CN113872666B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110381444A (en) * | 2019-06-24 | 2019-10-25 | 广东工业大学 | A kind of unmanned plane track optimizing and resource allocation methods |
| CN110392350A (en) * | 2019-08-06 | 2019-10-29 | 北京邮电大学 | A kind of user-center unmanned plane group construction method based on the perception of Backhaul capacity |
| CN111970713A (en) * | 2020-08-14 | 2020-11-20 | 北京邮电大学 | Unmanned aerial vehicle base station deployment method and three-dimensional parameter setting for dense urban area |
| CN112511197A (en) * | 2020-12-01 | 2021-03-16 | 南京工业大学 | Unmanned aerial vehicle auxiliary elastic video multicast method based on deep reinforcement learning |
Non-Patent Citations (1)
| Title |
|---|
| JIANGHUI LIU等: ""Deployment Optimization of UAV-Aided Networks Through a Dynamic Tunable Model"", 《IEEE COMMUNICATIONS LETTERS》, vol. 25, 17 March 2021 (2021-03-17), XP011865181, DOI: 10.1109/LCOMM.2021.3066617 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113872666B (en) | 2023-01-31 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Challita et al. | Cellular-connected UAVs over 5G: Deep reinforcement learning for interference management | |
| CN112929866B (en) | Unmanned aerial vehicle deployment method for adaptively optimizing network coverage of urban disaster area | |
| CN112511197B (en) | Unmanned aerial vehicle auxiliary elastic video multicast method based on deep reinforcement learning | |
| CN110049566B (en) | Downlink power distribution method based on multi-unmanned-aerial-vehicle auxiliary communication network | |
| CN114268903B (en) | Geographic information assisted unmanned aerial vehicle relay position deployment and power distribution method | |
| CN111836271B (en) | 3D position deployment method of multi-unmanned aerial vehicle communication base station | |
| CN111970713B (en) | Unmanned aerial vehicle base station deployment method and three-dimensional parameter setting for dense urban area | |
| CN112367111A (en) | Unmanned aerial vehicle relay deployment method and system, computer equipment and application | |
| CN115103409B (en) | Resource allocation method for multi-beam unmanned aerial vehicle cooperative communication | |
| CN113395706B (en) | Particle swarm algorithm-based heterogeneous unmanned aerial vehicle three-dimensional space deployment method and device | |
| CN111818535B (en) | Wireless local area network three-dimensional optimization deployment method fusing multi-population optimization algorithm | |
| Liu et al. | Multiobjective optimization for improving throughput and energy efficiency in UAV-enabled IoT | |
| CN115499921A (en) | Three-dimensional trajectory design and resource scheduling optimization method for complex unmanned aerial vehicle network | |
| Parvaresh et al. | A continuous actor–critic deep Q-learning-enabled deployment of UAV base stations: Toward 6G small cells in the skies of smart cities | |
| Nasr-Azadani et al. | Single-and multiagent actor–critic for initial UAV’s deployment and 3-D trajectory design | |
| CN116600316A (en) | Air-ground integrated Internet of things joint resource allocation method based on deep double Q networks and federal learning | |
| CN113556750B (en) | Unmanned equipment content cooperation realization method based on alliance formed game | |
| Sazak et al. | UAV-BS trajectory optimization under coverage, backhaul and QoS constraints using Q-learning | |
| CN113872666B (en) | Unmanned aerial vehicle deployment method based on Backhaul capacity constraint in dense urban area | |
| Lou et al. | Terrain-based UAV deployment: Providing coverage for outdoor users | |
| Fonseca et al. | Adaptive height optimization for cellular-connected UAVs: A deep reinforcement learning approach | |
| CN117221948A (en) | Network selection and power control method oriented to dense scene | |
| CN117058929A (en) | DDPG-based air safety data acquisition and resource allocation method | |
| CN113993099B (en) | Three-dimensional space-oriented mobile unmanned aerial vehicle user switching parameter configuration method | |
| Zhuang et al. | GA-MADDPG: A Demand-Aware UAV Network Adaptation Method for Joint Communication and Positioning in Emergency Scenarios |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |