An Unauthorized Spectrum Access Method in a UAV MEC Communication Network

Abstract

The present invention relates to a method for license-free spectrum access in a UAV MEC communication network, belonging to the field of wireless communication. The method includes: the UAV simultaneously carries a base station and an MEC server to call the license-free spectrum to provide services for mobile users with different requirements, and divides each time slot of the UAV into a real-time transmission stage and a non-real-time power control transmission stage. In the real-time transmission stage, the UAV simultaneously provides services for uplink computing users and downlink real-time communication users. In the non-real-time power control stage, some downlink non-real-time power control users outside the WiFi AP area can transmit simultaneously and on the same frequency as the WiFi device. Based on the designed access scheme, jointly considering the duty cycle allocation, flight trajectory, power allocation of power control users, and bandwidth allocation of users in the real-time transmission stage, an optimization problem with the maximum total offloading amount of computing users is established, and a multi-variable iterative algorithm based on the block coordinate descent and successive convex approximation methods is proposed for solution. The present invention can effectively improve the total offloading amount and license-free spectrum efficiency.

Description

Technical Field

[0001] The present invention belongs to the field of wireless communications and relates to a method for unlicensed spectrum access in an unmanned aerial vehicle (UAV) MEC communication network. Background Art

[0002] Nowadays, with the rapid development of the fifth generation mobile communication technology (5G) and mobile edge computing (MEC), more and more services and applications have emerged and are widely used by smart mobile devices, such as live streaming and virtual reality. However, due to the limited computing power of smart mobile devices, they cannot complete their own computing-intensive tasks within the tolerable delay. At the same time, the existing cellular network cannot meet the experience quality of all users with high traffic requirements. Although MEC allows users to offload tasks to edge servers for computing, the coverage of MEC servers is limited, which cannot effectively solve the problem of surging computing offloading demand in hot spots such as stadiums or concert venues. The most direct way to solve this problem is to deploy MEC servers in more locations on the ground, but this method is very costly and not flexible enough. Due to the flexibility of drones and the advantages of high-possible strong line-of-sight communication links, drones can carry MEC servers and small base stations at the same time to serve mobile users with different needs in hot spots.

[0003] In response to the scarcity of licensed spectrum resources, companies such as Qualcomm, Samsung, and Verizon have proposed using unlicensed spectrum (LTE Advanced in Unlicensed Spectrum, LTE-U) technology to increase available bandwidth. LTE-U technology aims to apply LTE technology to unlicensed spectrum while maintaining the original Long Term Evolution (LTE) protocol specifications as much as possible. By deploying small base stations in unlicensed spectrum, and relying on carrier aggregation technology to allow LTE to work in unlicensed and licensed spectrum, the purpose of increasing cellular system capacity and improving spectrum utilization in unlicensed spectrum can be achieved. Therefore, expanding LTE technology to unlicensed spectrum has become a research direction for 5G communication standards and a research hotspot for 6G communication standards.

[0004] LTE cellular mobile communication is centrally scheduled by base stations, while the WiFi system is based on a contention-based approach. If cellular mobile users are directly connected to the unlicensed band, it will seriously affect the performance of the WiFi system. According to different coexistence methods, the Listen Before Talk (LBT)-based LAA technology and the Duty-Cycle (DC)-based Carrier Sense Adaptive Transmission technology are two main unlicensed band access technologies.

[0005] Existing research mainly focuses on wireless communication networks and mobile edge computing networks with separate UAV cooperation. However, there is little research on the coexistence of wireless communication and mobile edge computing networks with UAV cooperation, and currently, the combination of LTE-U technology and MEC communication networks with UAV cooperation has not been considered. UAV unlicensed communication has received increasing research attention in many applications, such as caching, virtual reality, and disaster recovery. However, so far, there has not been much research on using unlicensed spectrum for UAV communication services. Most research considers the combination of static UAV base stations and LTE-U technology, without considering the mobility of UAV base stations. Their main strategy is to optimize the channel occupancy time of UAVs and the deployment of UAV positions through machine learning algorithms using the duty cycle method. To improve the efficiency of unlicensed spectrum and make full use of the mobility of UAVs to serve more users with different needs, it is necessary to study unlicensed spectrum access technology in UAV-assisted MEC communication networks. Summary of the Invention

[0006] In view of this, considering the scenario of MEC communication networks with UAV cooperation, the present invention designs a new unlicensed spectrum access scheme. On the premise of meeting the different transmission rates of all mobile users, jointly considering duty cycle allocation, UAV flight trajectory, transmission power allocation for downlink non-real-time power control users, and bandwidth allocation for uplink and downlink users in the real-time phase, an optimization problem of maximizing the total offloading volume of uplink computing users is constructed, and a multi-variable iterative optimization algorithm based on the block coordinate descent and successive convex approximation methods is proposed. The access scheme proposed by the present invention can allow more mobile users to normally access the unlicensed channel, significantly improve the spectrum efficiency, and further increase the total offloading volume of computing users. To achieve the above object, the present invention provides the following technical solutions:

[0007] An unlicensed spectrum access method in a UAV MEC communication network, the method comprising:

[0008] As Figure 2As shown in the figure, the rotor UAV carries a small base station and an MEC server at the same time, and calls the unlicensed spectrum to provide services for users in the hot spot area periodically. In the present invention, each time slot of the UAV is divided into a real-time communication phase and a non-real-time power control phase. There are M downlink real-time communication users in the area communicating with the UAV in the real-time communication phase. These users are randomly distributed in the coverage area of the UAV and have high requirements for the real-time performance of data transmission. In addition, V downlink users are distributed outside the coverage area of the WiFi AP. These users have low requirements for the communication link quality and can use the same unlicensed frequency band as the WiFi device in the non-real-time power control phase. Therefore, these users will cause serious interference to the WiFi device and are called downlink non-real-time power control (NPC) users. In addition, K users in the area need to continuously perform computationally intensive tasks and need to offload the tasks to the UAV MEC server with strong computing power for calculation, such as network live broadcast and virtual reality. Therefore, the K uplink users in the present invention are called uplink computing users. There is a ground WiFi in the coverage area of the UAV. During the flight of the UAV, it always covers all users in the area, including WiFi users, and continuously provides services for the users.

[0009] As Figure 3 shown, the task execution period T of the UAV is used as the flight period. Each flight period is divided into N time slots, and each time slot is divided into a real-time transmission phase and a non-real-time power control transmission phase. Furthermore, K computing users and M downlink real-time communication users can share the channel through the FDMA technology in the real-time transmission phase, and V downlink NPC users transmit on the same channel as the WiFi device in the non-real-time power control transmission phase.

[0010] Furthermore, considering the flight trajectory of the UAV, the length δ of each time slot t = T / N. In the nth time slot, the position coordinates of the UAV can be expressed as Q u [n] = (X u [n], Y u [n]). Assuming that the maximum flight speed of the UAV is V max , therefore, the trajectory of the UAV should satisfy the following constraints:

[0011]

[0012] Considering that the UAV returns to the starting position after a task cycle T, therefore, the following constraints should be satisfied:

[0013] Q u [0] = Q u [N]. (2)

[0014] The channel power gains from the uplink computing user to the UAV and from the UAV to the downlink user are respectively expressed as:

[0015]

[0016]

[0017]

[0018] where \(w_0\) is the channel power gain at a distance of 1 m, \(H\) is the fixed altitude of the UAV, \(C\) k and \(D\) m 、\(D\) v respectively represent the coordinates \((X\) k ,Y\) k ) and \((X\) m ,Y\) m )、\((X\) v ,Y\) v ) of the uplink computing user, the downlink real-time communication user, and the downlink NPC user.

[0019] If the cellular mobile user uses the unlicensed band of the WiFi device, it is necessary to ensure that the WiFi device will not be interfered by the non-real-time power control user. To ensure the balance of the transmission of users in the non-real-time power control transmission stage and the real-time transmission stage in each time slot, the following constraints should be met:

[0020] \(F\) min \(\leq b[n]\leq F\) max (6)

[0021] where \(b[n]\) represents the time occupancy ratio of the real-time communication stage in the \(n\)th time slot, \(F\) min and \(F\) max respectively represent the minimum time occupancy ratio and the maximum time occupancy ratio allocated in the real-time transmission stage.

[0022] Furthermore, considering the real-time stage transmission model, in the real-time transmission stage, the UAV uses the FDMA technology to allocate bandwidth resources to provide communication services for \(M\) real-time communication users and \(K\) computing users on the ground at the same time, that is, the bandwidth allocated to the \(m\)th downlink real-time communication user in the \(n\)th time slot is \(B\) m [n], and the \(k\)th uplink computing user uses the bandwidth \(B\) k [n] to unload tasks to the UAV at the same time. The bandwidth should meet the following constraints:

[0023]

[0024] The UAV provides high-rate services for real-time communication users. Therefore, the minimum throughput of \(M\) real-time users in each flight time slot should meet the following formula:

[0025]

[0026] In formula (10), p m [n] represents the power allocated to the m-th user in the n-th time slot during the real-time transmission phase of the UAV, N0 represents the noise power spectral density, represents the minimum transmission throughput of the downlink real-time communication user in each flight time slot. For each real-time user, the communication link quality is similar and can be set as a constant. Therefore, in this paper, p m [n] is set as a constant without optimization, that is, p m [n] = P u_max / M, P u_max is the maximum transmission power of the UAV during the real-time transmission phase. During the calculation of the user offloading phase, each user uses the allocated bandwidth B k [n] to perform offloading simultaneously. Assuming that their powers are the same and constant, the offloading amount of the k-th uplink computing user in the n-th time slot during the real-time transmission phase is where p is the offloading power of each uplink computing user. Considering that the uplink users have been performing local computing, assuming that the local computing ability of the k-th user in the n-th time slot is f k [n], and using C k to represent the number of CPU revolutions required for a user to calculate 1 bit of task, the amount of tasks that the k-th uplink user can locally process in the n-th time slot can be expressed as:

[0027]

[0028] In the present invention, we assume that the local computing ability of the k-th uplink user in the n-th time slot is a random number between 0.4 - 0.5 GHz. To meet the minimum computing requirements of each user, the following constraints need to be satisfied:

[0029]

[0030] Furthermore, considering the non-real-time power control phase transmission model, in the non-real-time power control transmission phase, the UAV selects V NPC users outside the coverage area of the WiFi AP and WiFi devices to simultaneously access the unlicensed channel. Assuming that each NPC user is allocated equal bandwidth, the unlicensed spectrum is reused by jointly optimizing the transmission power of the NPC users, the proportion of the transmission time in the non-real-time power control phase, and the UAV movement trajectory. Assuming P max is the maximum average power allocated by the UAV to each NPC user v during the non-real-time power control transmission phase, and P is the maximum transmission power of the UAV during the non-real-time power control transmission, so it should satisfy:

[0031]

[0032]

[0033] where p v [n] is the power allocated by the UAV to user v in the nth time slot. In the actual scenario, the WiFi devices located in the WiFi AP coverage area maintain connections with the UAV base station simultaneously. Therefore, it is assumed that the WiFi devices can actively or be required to inform the UAV of the status of the WiFi system, such as the number and location of the WiFi devices. Therefore, to ensure the communication quality of the WiFi devices, we need to ensure that the sum of the average interference powers received by each WiFi device does not exceed the threshold Γ l , that is, the following constraints are satisfied:

[0034]

[0035] Meanwhile, we need to meet the minimum average communication rate requirements of each NPC user, and the following constraints need to be satisfied:

[0036]

[0037] where represents the minimum average communication rate (bits / s / Hz) of each user v, and N n represents the value of Gaussian white noise.

[0038] Furthermore, considering the established system optimization model, the optimization objective of the present invention is to optimize the maximum task volume that can be calculated in the UAV MEC unlicensed communication system under the constraints of system resources and multi-class user experience quality, while eliminating the interference between multi-class users and improving the computing performance. This optimization problem can be expressed as:

[0039]

[0040] s.t. (1), (2), (6), (7), (8), (10), (11), (12), (13) and (14)

[0041]

[0042] Furthermore, considering the solution of the entire optimization problem, since the objective function is non-concave and there are variable couplings in Constraints (8), (10), (13), and (14), the original optimization problem is a non-convex optimization problem with multi-variable couplings and is difficult to handle. To effectively solve the problem, the present invention proposes a multi-variable iterative alternating optimization algorithm based on the Block Coordinate Descent (BCD) and Successive Convex Approximation (SCA) methods. First, the above original problem is decomposed into four sub-problems: duty cycle allocation (P1), trajectory optimization (P2), bandwidth allocation (P3), and power allocation (P4). When optimizing each sub-problem, the optimization variables of other sub-problems are fixed, and then the SCA method is used to convert the non-convex sub-problems into convex ones. Through the process of iterating the four sub-problems and alternating optimization, the accuracy is finally satisfied, and the final solution of the original problem is found. Among the four sub-problems of the present invention, both bandwidth allocation and power allocation are standard convex optimization problems, while duty cycle allocation is a linear programming problem. Therefore, they can all be accurately solved using the standard convex optimization tool CVX, while the trajectory optimization sub-problem is non-convex and can first use the SCA method to approximate its non-convex to convex. Further, the flight trajectory sub-problem is solved. For the given duty cycle allocation {b}, bandwidth allocation {B}, and power allocation {P}, this sub-optimization problem is to maximize the total offloading by optimizing the UAV flight trajectory. Therefore, we can rewrite the original problem as Problem (P2):

[0043]

[0044] s.t. (1), (2), (8), (10), (13), and (14).

[0045] The objective function of this problem is non-concave, and Constraints (8), (10), (13), and (14) are all non-convex constraints. Next, slack variables are introduced and the SCA method is used to solve them sequentially.

[0046] (a) We first introduce the non-negative slack variable x m [n] as the lower bound of the left side of Constraint (8). For the left side expression, let γ = p m [n] / (N0B m [n]), then Constraint (8) can be expressed as:

[0047]

[0048] The left side of the above inequality is non-convex with respect to Q u [n], but is convex with respect to ||Q u [n] - Dm || 2 is convex. Thus, for ||Q u [n] - D m || 2 perform a first-order Taylor approximation as the lower bound of the convex function. Assume that the local value at the i-th iteration is given as The lower bound of the left side of the above inequality can be transformed into:

[0049]

[0050] where

[0051] Therefore, constraint (8) can be replaced by:

[0052]

[0053]

[0054] For the transformed constraints (8.1) and (8.2) above, the left side of the inequality of constraint (8.2) is a concave function with respect to Q u [n] and the right side is a linear function. According to the criterion, it can be obtained that the convex constraint is satisfied, and constraint (8.1) is a linear constraint.

[0055] (b) Similarly, the objective function and constraint (10) can be transformed into the following forms respectively:

[0056]

[0057]

[0058]

[0059] where

[0060] For the transformed objective function, constraint (10.1) and constraint (10.2), they all satisfy the properties of convex functions and belong to convex constraints.

[0061] (c) For constraint (13), we first introduce a slack variable t v [n] is the lower bound of ||Q u [n] - W l ||, that is, constraint (13) can be transformed into:

[0062] ||Q u [n] - W l || 2 ≥ t v [n], (13.1)

[0063]

[0064] For the above constraints (13.1) and (13.2), constraint (13.2) is a convex constraint. For the non-convex constraint (13.1), we adopt the SCA method and can transform it into a convex constraint (13.3) as follows:

[0065]

[0066] Therefore, constraint (13) is equivalent to constraint (13.2) and constraint (13.3), and has been transformed into convex constraints so far.

[0067] (d) Constraint (14) can be directly processed using the SCA method, similar to the processing methods of constraints (8) and (10). The processing results are as follows:

[0068]

[0069]

[0070] where

[0071] Therefore, constraint (14) is transformed into convex constraints (14.1) and (14.2), and the processing is completed.

[0072] Therefore, the sub-problem (P2) can be transformed into:

[0073]

[0074] s.t. (1), (2), (8.1), (8.2), (10.1), (10.2), (13.2), (13.3), (14.1) and (14.2).

[0075] Through a series of transformations, the above sub-problem (P2) regarding trajectory optimization is transformed into a standard convex problem (P2.1), which can be directly solved using the convex optimization tool CVX in Matlab.

[0076] Advantages and beneficial effects of the present invention

[0077] The advantages of the present invention are as follows: Existing research mainly focuses on wireless communication networks and mobile edge computing networks with separate UAV collaborations. However, there is little research on the coexistence of wireless communication and mobile edge computing networks with UAV collaborations, and the combination of LTE-U technology with the MEC communication network with UAV collaborations has not been considered so far. So far, there have not been many studies on using unlicensed spectrum for UAV communication services. Most studies consider the combination of static UAV base stations and LTE-U technology, without considering the mobility of UAV base stations. Therefore, the present invention first considers the unlicensed spectrum access scheme in the MEC communication network with UAV collaborations, makes full use of the mobility of UAVs, maximizes the total offloading volume of uplink computing users, and significantly improves the spectrum efficiency.

[0078] The beneficial effects of the present invention are as follows: The present invention provides a method for unlicensed spectrum access in a UAV MEC communication network. Based on the designed access scheme, on the premise of meeting the different transmission rates of uplink computing users and downlink communication users, it ensures the normal transmission of Wi-Fi devices in the non-real-time power control stage. By jointly considering duty cycle allocation, UAV flight trajectories, transmission power allocation for downlink non-real-time power control users, and bandwidth allocation for uplink and downlink users in the real-time stage, an optimization problem with the maximum total offloading volume of uplink computing users is established. Next, a multi-variable iterative optimization algorithm based on BCD-SCA is proposed to solve the complex optimization problem. Through the access scheme designed by the present invention, the total offloading volume of uplink computing users can be maximized, and at the same time, more mobile users can be connected to the unlicensed channel, further improving the spectrum efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0079] In order to make the objectives, technical solutions, and beneficial effects of the present invention clearer, the following drawings are provided for illustration:

[0080] Figure 1 It is a schematic flowchart of an embodiment of the present invention;

[0081] Figure 2 It is a scenario diagram of a UAV-collaborated MEC unlicensed communication network according to an embodiment of the present invention;

[0082] Figure 3 It is a diagram of an unlicensed spectrum access scheme in a UAV MEC communication network according to an embodiment of the present invention; DETAILED DESCRIPTION OF THE EMBODIMENTS

[0083] The preferred embodiments of the present invention will be described in detail below with reference to the drawings.

[0084] As Figure 1 、 Figure 2 and Figure 3As shown in the figure, the present invention provides an unlicensed spectrum access method in a drone MEC communication network, including:

[0085] Step 1: Construct a drone-assisted MEC communication network and design a coexistence scheme for unlicensed spectrum access in the drone MEC communication network. As Figure 3 shown, take the task execution cycle T of the drone as the flight cycle. Each flight cycle is divided into N time slots, and each time slot is divided into a real-time transmission and a non-real-time power control transmission stage. Furthermore, K computing users and M downlink real-time communication users can share the channel through the FDMA technology in the real-time transmission stage, and V downlink NPC users transmit on the same channel as the WiFi device in the non-real-time power control transmission stage.

[0086] Step 2: Maximize the amount of computing tasks in the drone MEC unlicensed communication system under the constraints of system resources and the quality of experience of multiple types of users, and improve the computing performance: Maximize the total offloading amount of the uplink computing users by jointly optimizing the transmission power of the downlink NPC users, the bandwidth allocation between the uplink computing users and the downlink real-time communication users, and the allocation of the drone trajectory and duty cycle:

[0087]

[0088] s.t. C1~C11

[0089] Among them, constraint C1 is indicating the constraint that the trajectory of the drone should satisfy, Q u [n]=(X u [n], Y u [n]) is the position coordinate of the drone, V max is the maximum flight speed of the drone, and n is the nth time slot; constraint C2 is Q u [0]=Q u [N]. indicating that the drone returns to the starting position after a short task cycle T; C3 is F min ≤b[n]≤F max to ensure the balance of the transmission of users in the non-real-time power control transmission stage and the real-time transmission stage in each time slot. b[n] is the time occupancy ratio of the real-time communication stage in the nth time slot, F min and F max respectively represent the minimum time occupancy ratio and the maximum time occupancy ratio allocated to the real-time transmission stage; C4 is indicating that in the real-time transmission stage, the drone uses the FDMA technology to allocate bandwidth resources to provide communication services for M downlink real-time communication users and K computing users on the ground at the same time, that is, the bandwidth allocated to the mth downlink real-time communication user in the nth time slot is B m [n], and the bandwidth used by the kth uplink computing user in the nth time slot is Bk [n] The constraint that the bandwidth should satisfy when offloading tasks to the UAV; C5 is represents the minimum throughput that M real-time users should satisfy in each time slot, p m [n] represents the power allocated to the m-th user by the UAV in the n-th time slot during the real-time transmission phase, N0 represents the noise power spectral density, represents the minimum throughput of the downlink real-time communication users in each time slot. Since the communication link quality of each user is similar, p m [n] is set to a constant, that is, p m [n] = P u_max / M, P u_max is the maximum transmission power of the UAV during the real-time phase; C6 is represents meeting the minimum computing requirements of each computing user, is the amount of tasks that the k-th computing user can process locally in the n-th time slot, is the offloading amount of the k-th uplink computing user in the n-th time slot during the real-time transmission phase, p is the offloading power of each uplink computing user, f k [n] is the local computing ability of the k-th user in the n-th time slot. In the present invention, we assume that f k [n] is a random number between 0.4 - 0.5 GHz, C k is the number of revolutions of the CPU required for the user to compute 1 bi of tasks; C7 is C8 is where P max is the maximum average power allocated by the UAV to each NPC user v during the non-real-time power control transmission phase, P is the maximum transmission power of the UAV during non-real-time power control transmission, p v [n] is the power allocated by the UAV to the downlink NPC user v in the n-th time slot; C9 is represents ensuring that the sum of the average interference powers received by each WiFi device does not exceed the threshold Γ l , W l is the coordinate of the WiFi device, w0 represents the channel power gain at a distance of 1 m, H is the fixed flight altitude of the UAV; C10 is the constraint that needs to meet the minimum average communication rate requirement of each downlink NPC user: Among them, is the minimum average communication rate (bits / s / Hz) of each downlink NPC user v, N n represents Gaussian white noise; the constraint C11 is and

[0090] Step 3: For the complex optimization problem established in Step 2, an efficient multi-variable iterative optimization algorithm based on BCD-SCA is proposed to solve the established optimization model. First, based on the BCD algorithm, the above original problem is decomposed into four sub-problems: duty cycle allocation (P1), trajectory optimization (P2), bandwidth allocation (P3), and power allocation (P4). Among the four sub-problems of the present invention, both bandwidth allocation and power allocation are standard convex optimization problems, while duty cycle allocation is a linear programming problem, and they can all be accurately solved by using the standard convex optimization tool CVX. The flight trajectory sub-problem is non-convex and non-concave, and the SCA method can be used to transform it into an approximate problem (P2.1) to obtain the optimal solution. Therefore, the present invention decomposes the above original problem into four independent sub-problems (P1, P2, P3, and P4), and then uses the SCA method to transform the non-convex sub-problem (P2) into a convex problem (P2.1). When optimizing each sub-problem, the optimization variables of other sub-problems are fixed, and through the process of iterating the four sub-problems and alternately optimizing, the accuracy is finally satisfied to find the final solution of the original problem.

[0091] Furthermore, the SCA method is used to solve the non-convex sub-problem of the trajectory, and there is:

[0092]

[0093] s.t.C1,C2,C5,C6,C9,C10.

[0094] (a) First, introduce a non-negative slack variable x m [n] as the lower bound of the left side of constraint C5. For the left side of the equation, let γ = p m [n] / (N0B m [n]), then constraint C5 can be expressed as:

[0095]

[0096] The left side of the above inequality is non-convex with respect to Q u [n], but is convex with respect to ||Q u [n] - D m || 2 Therefore, for ||Q u [n] - D m || 2 perform a first-order Taylor approximation as the lower bound of the convex function. Assume that the local value of the i-th iteration is given as The lower bound of the left side of the above inequality can be transformed into:

[0097]

[0098] where

[0099] Therefore, the constraint C5 can be replaced by:

[0100]

[0101]

[0102] For the transformed constraints (5.1) and (5.2) above, the left - hand side of the inequality of constraint (5.2) is a concave function with respect to Q u [n], and the right - hand side is a linear function. According to the criterion, it can be obtained that the convex constraint is satisfied, and constraint (5.1) is a linear constraint;

[0103] (b) Similarly, the objective function and constraint C6 can be transformed into the following forms respectively:

[0104]

[0105]

[0106]

[0107] where For the transformed objective function, constraint (6.1) and constraint (6.2), they all satisfy the properties of convex functions and belong to convex constraints;

[0108] (c) For constraint C9, first introduce a slack variable t v [n] is the lower bound of ||Q u [n] - W l ||, that is, constraint C9 can be transformed into:

[0109] ||Q u [n] - W l || 2 ≥t v [n], (9.1)

[0110]

[0111] For the above constraints (9.1) and constraint (9.2), constraint (9.2) is a convex constraint, and for the non - convex constraint (9.1), using the SCA method, it can be transformed into a convex constraint (9.3) as follows:

[0112]

[0113] Therefore, constraint C9 is equivalent to constraint (9.2) and constraint (9.3), and has been transformed into convex constraints so far;

[0114] (d) The constraint C10 can be directly processed using the SCA method, similar to the processing methods of constraints C5 and C6. The processing results are as follows:

[0115]

[0116]

[0117] Among them Therefore, the constraint C10 is transformed into convex constraints (10.1) and (10.2), and the processing is completed;

[0118] The sub-problem (P2) can be transformed into:

[0119]

[0120] s.t. C1, C2, (5.1), (5.2), (6.1), (6.2), (9.2), (9.3), (10.1) and (10.2).

[0121] Through a series of transformations, the above sub-problem (P2) regarding trajectory optimization is transformed into a standard convex problem (P2.1), which can be directly solved using the convex optimization tool CVX in Matlab.

[0122] Therefore, the present invention decomposes the above original problem into four independent sub-problems (P1, P2, P3, and P4), and then uses the SCA method to transform the non-convex sub-problem (P2) into a convex problem (P2.1). When optimizing each sub-problem, the optimization variables of other sub-problems are fixed. Through the process of iterating the four sub-problems and alternately optimizing, the accuracy is finally satisfied, and the final solution of the original problem is found.

[0123] Finally, it should be noted that the above preferred embodiments are only used to illustrate the technical solutions of the present invention and are not restrictive. Although the present invention has been described in detail through the above preferred embodiments, those skilled in the art should understand that various changes can be made in form and details without departing from the scope defined by the claims of the present invention.

Claims

1. A method for license-free spectrum access in a UAV MEC communication network, characterized in that: The method includes the following steps: Step 1: For the UAV MEC communication network, construct an unlicensed spectrum access method using time-division multiplexing and frequency-division multiplexing technologies, and allocate different time-frequency resources for different types of users: Use the task execution cycle T of the UAV as the flight cycle. Each flight cycle is divided into N time slots, and each time slot is divided into a real-time transmission and a non-real-time power control transmission phase. Furthermore, K computing users and M downlink real-time communication users can share the channel through FDMA technology during the real-time transmission phase, and V downlink NPC users transmit on the same channel as the WiFi device during the non-real-time power control transmission phase; Step 2: Based on the method constructed in Step 1, while satisfying the different transmission rates of the uplink computing users and the downlink communication users, jointly consider the duty cycle allocation in the real-time transmission phase and the non-real-time power control phase, the UAV flight trajectory, the transmission power allocation of the downlink non-real-time power controlled (NPC) users, and the bandwidth allocation of the uplink and downlink users in the real-time phase, and construct an optimization problem to maximize the total offloading amount of the uplink computing users: (P)： s.t.C1~C11 Among them, the constraint C1 is the constraint that the trajectory of the UAV should satisfy, Q u [n] = (X u [n], Y u [n]) is the position coordinate of the UAV, V max is the maximum flight speed of the UAV, and n is the nth time slot; the constraint C2 is Q u [0] = Q u [N] . , indicating that the UAV returns to the starting position after a short mission cycle T; C3 is F min ≤b[n]≤F max , ensuring the balance of the transmission of users in the non-real-time power control transmission stage and the real-time transmission stage in each time slot. b[n] is the time ratio of the real-time communication stage in the nth time slot, F min and F max respectively represent the minimum time ratio and the maximum time ratio allocated to the real-time transmission stage; C4 is indicating that in the real-time transmission stage, the UAV uses the FDMA technology to allocate bandwidth resources to provide communication services for M real-time communication users and K computing users on the ground at the same time. That is, the bandwidth allocated to the mth downlink real-time communication user in the nth time slot is B m [n], and the kth uplink computing user uses the bandwidth B k [n] to unload tasks to the UAV at the same time. This is the constraint that the bandwidth should satisfy; C5 is indicating the minimum throughput that M real-time users should satisfy in each flight time slot. p m [n] represents the power allocated to the mth user by the UAV in the nth time slot in the real-time transmission stage. N0 represents the noise power spectral density, represents the minimum transmission throughput of the downlink real-time communication user in each flight time slot. Since the communication link quality of each real-time user is similar, p m [n] is set as a constant without optimization, that is, p m [n] = P uav_max / M, P uav_max is the maximum transmission power of the UAV in the real-time transmission stage, and g um [n] is the channel gain between the UAV and the real-time user; C6 is indicating the constraint to meet the minimum computing requirements of each user, is the amount of tasks that the kth uplink user can process locally in the nth time slot, is the offloading amount of the kth uplink computing user in the nth time slot in the real-time transmission stage. p is the offloading power of each uplink computing user. It is f k [n] the local computing ability of the kth user in the nth time slot, C k is the number of revolutions of the CPU required for the user to calculate a 1-bit task, g ku [n] is the channel gain between the uplink user and the UAV; C7 is C8 is where P max is the maximum average power allocated by the UAV to each NPC user v in the non-real-time power control transmission phase, P is the maximum transmission power of the UAV during non-real-time power control transmission, p v [n] is the power allocated by the UAV to user v in the nth time slot; C9 is indicates that the sum of the average interference powers received by each WiFi device does not exceed the threshold Γ l , W l is the WiFi device coordinate, w0 is the channel power gain at a distance of 1 m, H is the fixed height of the UAV, Q u [n] is the coordinate of the UAV at time n; C10 is the constraint that the minimum average communication rate requirement of each NPC user needs to be met: where, is the minimum average communication rate (bits / s / Hz) of each user v, N n represents Gaussian white noise; the constraint C11 is and Step 3: For the complex optimization problem established in Step 2, propose an efficient multi-variable iterative optimization algorithm based on BCD-SCA to solve the established optimization model: First, according to the four parameters of the duty cycle, trajectory, bandwidth, and power that need to be optimized, based on the BCD algorithm, decompose the original problem into four sub-problems: duty cycle allocation (P1), trajectory optimization (P2), bandwidth allocation (P3), and power allocation (P4); Among the four decomposed sub-problems, bandwidth allocation and power allocation are both standard convex optimization problems, while duty cycle allocation is a linear programming problem, and they can all be accurately solved using the standard convex optimization tool CVX. The flight trajectory sub-problem is non-convex and non-concave, and can be transformed into an approximate problem through the SCA method to obtain the optimal solution. Furthermore, the original problem is decomposed into four independent sub-problems. When optimizing each sub-problem, fix the optimization variables of other sub-problems, and then use the SCA method to transform the non-convex sub-problem from non-convex to convex. Through the process of iterating the four sub-problems and alternately optimizing, finally meet the accuracy and find the final solution of the original problem.

2. The license-free spectrum access method in a drone MEC communication network according to claim 1, wherein For the non-convex and non-concave sub-problem of the trajectory in Claim 1, the SCA method is used for solution, and there is: (P2)： s.t.C1, C2, C5, C6, C9, C10. (a) First, introduce the non - negative slack variable x m [n] as the lower bound of the left - hand side of constraint C5. For the left - hand side expression, let γ = p m [n] / (N0B m [n]), then constraint C5 can be expressed as: where \(D_m\) is the geographical location of the downlink real-time communication user, and the left side of the above inequality with respect to \(Q\) u [n] is non-convex, but with respect to \(\|Q\) u [n] - D m \| 2 is convex. Therefore, for \(\|Q\) u [n] - D m \| 2 a first-order Taylor approximation is made as the lower bound of the convex function. Assuming that the local value at the \(i\)-th iteration is given as the lower bound of the left side of the above inequality can be transformed into: Among them Therefore, constraint C5 can be replaced by: For the above transformed constraints (5.1) and (5.2), the left - hand side of the inequality of constraint (5.2) is a concave function with respect to Q u [n], and the right - hand side is a linear function. According to the criterion, it can be obtained that the convex constraint is satisfied, and constraint (5.1) is a linear constraint; (b) Reintroduce the non - negative slack variable x k [n] As the lower bound of C6, the objective function P2 and the constraint C6 can be transformed into the following forms respectively: where Ck is the geographical coordinate of the uplink user, For the transformed objective function, constraint (11.1), and constraint (11.2), they all satisfy the properties of convex functions and belong to convex constraints; (c) For constraint (9), first introduce a slack variable t v [n] is the lower bound of ||Q u [n] - W l ||, that is, constraint C9 can be transformed into: ||Q u [n]-W l || 2 ≥t v [n], (9.1) For the above constraints (9.1) and (9.2), constraint (9.2) is a convex constraint, and for the non-convex constraint (9.1), using the SCA method, it can be transformed into a convex constraint (9.3) as follows: Therefore, constraint C9 is equivalent to constraint (9.2) and constraint (9.3), and at this point, it has been transformed into a convex constraint; (d) Constraint C10 can be directly processed using the SCA method, similar to the processing methods of constraint C5 and constraint C6, and the processing results are as follows: Among them represents the rate of the downlink user v at time n. Therefore, the constraint C10 is transformed into the convex constraints (10.1) and (10.2), and the processing is completed; Sub-problem (P2) can be transformed into: (P2.1)： s.t. C1, C2, (5.1), (5.2), (6.1), (6.2), (9.2), (9.3), (10.1), (10.2). Through a series of transformations, the above sub-problem (P2) regarding trajectory optimization is transformed into a standard convex problem (P2.1), which can be directly solved using the convex optimization tool CVX in Matlab.

Images