Online joint power distribution and task scheduling method for low earth orbit satellite network

Abstract

The invention discloses an online joint power distribution and task scheduling method for a low earth orbit satellite network, and belongs to the technical field of spatial information. The method comprises the following steps: acquiring a ground station set and low-orbit satellite network parameters, constructing an accessible ground station set, constructing a to-be-unloaded data set, a to-be-unloaded data backlog queue and an excess total energy consumption queue, predicting a channel gain of a satellite-ground link, calculating an optimal unloading transmitting power of a low-orbit satellite, and selecting an optimal ground station to perform data unloading. Updating the to-be-unloaded data backlog queue, the excess total energy consumption queue and the average channel gain; a two-layer optimization method is used for removing coupling of power distribution and task scheduling problems, a classic User Confidence Bound algorithm is used for solving task scheduling in outer-layer optimization, and an optimal solution of power distribution is obtained through a classification method in inner-layer optimization. The complexity of the resource management problem in the low earth orbit satellite network is greatly simplified, and the solving efficiency and quality of the problem can be effectively guaranteed.

Description

Technical Field

[0001] The present invention belongs to the field of space information technology, and in particular relates to an online joint power allocation and task scheduling method for a low-orbit satellite network, which can allocate power and schedule tasks for the low-orbit satellite network. Background Art

[0002] Low-orbit satellite networks are widely used in emergency communications, environmental monitoring, disaster relief, military, and media entertainment. The rapid increase in the number of low-orbit satellites and the globalization of information demand have led to an explosive growth in space data in low-orbit satellite networks. Massive amounts of space data urgently need to be unloaded from low-orbit satellites to ground stations for further processing and analysis. This makes the data unloading service of low-orbit satellite networks in short supply. Unlike ground networks, the high-speed movement of low-orbit satellites gives low-orbit satellite networks the characteristics of line-of-sight communication, which greatly limits the network's data unloading capabilities. Therefore, it is urgent to design an efficient data unloading solution to improve the resource utilization efficiency of low-orbit satellite networks and alleviate the contradiction between the explosive growth of space data and limited intermittent communication resources.

[0003] Previous research on data offloading in low-orbit satellite networks is mainly divided into two categories: offline data offloading strategies and online data offloading strategies. In offline data offloading strategies, all network resource information and network task information are assumed to be known in advance and fixed. However, in actual low-orbit satellite networks, not only are network resources time-varying, but network task information is also uncertain. Specifically, the network topology is highly dynamic. The channel state between the satellite and the ground station changes in real time, which makes it impossible to accurately obtain the channel state between the satellite and the ground station in real time. The data volume and arrival time of network tasks are affected by multiple factors such as random events, application fields, and business types, making network task information unpredictable and uncertain. Therefore, offline data offloading strategies are difficult to adapt to the high dynamics of low-orbit satellite networks, cannot handle the random arrival of space data, and are difficult to efficiently utilize time-varying network resources. In order to meet these challenges, the second type of research work aims to solve the problem of online data offloading for low-orbit satellite networks, so as to capture the random arrival of space data in real time and avoid the unknown dynamic environment of low-orbit satellite networks. Unfortunately, these research works ignore the energy consumption constraints of low-orbit satellite networks, which greatly limits their scope of application. Therefore, it is urgent to propose an online scheduling method for network resources under energy consumption constraints that is suitable for the high dynamic characteristics of low-orbit satellite networks. Summary of the invention

[0004] In view of the shortcomings of the existing technology, an online joint power allocation and task scheduling method for low-orbit satellite networks is provided to achieve joint real-time optimization of low-orbit satellite power allocation and data offloading task scheduling, thereby ensuring efficient utilization of low-orbit satellite network resources.

[0005] To achieve the above technical objectives, the present invention discloses an online joint power allocation and task scheduling method for a low-orbit satellite network. For the data unloading scenario of the low-orbit satellite network that has been studied, including a low-orbit satellite and multiple ground stations connected to the low-orbit satellite signal, in the process of data unloading and scheduling between the low-orbit satellite and the ground station, the Lyapunov optimization framework is used to separate the tightly coupled power allocation problem and the task scheduling problem. Due to the communication delay of the satellite-to-ground link, the channel gain of the satellite-to-ground link is difficult to obtain in real time, and thus the current data unloading amount cannot be obtained in real time. Therefore, the Upper Confidence Bound (UCB) algorithm is used to predict the delayed channel gain, calculate the network data unloading amount in real time, and provide real-time feedback.

[0006] The specific steps are as follows:

[0007] S1. Obtain the network parameters of low-orbit satellites and all ground stations, including the number of ground stations G and the ground station generation set Average energy threshold of low-orbit satellites and all ground stations Channel bandwidth B c , the minimum transmission power P of low-orbit satellites to transmit data to all ground stations min And the maximum transmission power P max ; Initialize the average channel gain of each ground station g represents the ground station index, and the time axis is divided into T time slots with a time slot length of τ at equal intervals, thereby obtaining the time slot index set

[0008] S2. Use the Satellite Tool Kit software to calculate the visible time window between the low-orbit satellite and each ground station, and determine the time slot Whether it is within the visible time window between the ground station and the low-orbit satellite, the set of accessible ground stations in each time slot t is constructed based on the judgment result

[0009] S3, construct a set of data to be unloaded that arrives at the low-orbit satellite network within time slot t, construct a backlog queue Q(t) of data to be unloaded, and construct an excess total energy consumption queue Z(t) for calculating the optimal power;

[0010] S4. In each time slot t, predict all current satellite-to-ground link channel gains of all accessible ground stations of the low-orbit satellite.

[0011] S5. In each time slot t, according to the current satellite-to-ground link channel gains of all accessible ground stations of the low-orbit satellite, The Lyapunov optimization framework is used to calculate the offload transmit power cutoff point to calculate the optimal offload transmit power of the low-orbit satellite.

[0012] S6. In each time slot t, the optimal unloading transmission power of the low-orbit satellite is With the goal of maximizing the amount of data offloading, the optimal ground station is selected to generate the optimal data offloading strategy for data offloading x g,t ;

[0013] S7. Due to the delay in the satellite-to-ground link channel gain, it is impossible to obtain the optimal power allocation and the real-time channel gain information required by the ground station. Therefore, the UCB algorithm is used in each time slot t to predict the satellite-to-ground link channel gain through historical channel gain information, calculate the current time slot data unloading amount and the current time slot energy consumption, and provide real-time feedback based on the calculation results, and then update the backlog queue of data to be unloaded, the excess total energy consumption queue and the average channel gain. The average energy consumption of satellite unloading data is controlled within the energy consumption constraint.

[0014] Furthermore, the method for constructing the set of accessible ground stations in each time slot t is as follows:

[0015] The ephemeris of the low-orbit satellite and The latitude and longitude information of each ground station in the satellite toolkit STK software is imported to calculate the visible time window between the low-orbit satellite and each ground station, and determine whether the time slot t is within the ground station

[0016] If the ground station g is within the visible time window of the low-orbit satellite, then add the ground station g to the accessible ground station set. Gather the ground stations one by one By judging all ground stations in the 's construction.

[0017] Furthermore, the specific process of constructing a set of data to be unloaded that arrives at the low-orbit satellite network within time slot t, constructing a backlog queue of data to be unloaded, and constructing an excess total energy consumption queue is as follows:

[0018] 3a. Construct a data set to be unloaded a = {a(1), a(2), ..., a(T)}, where 1, 2 ... T represent time slot indexes, T is the largest time slot index number, and a(t) represents the amount of data arriving at the low-orbit satellite network in time slot t. Specifically, at the beginning of time slot t, the data acquisition device in the low-orbit satellite acquires the data amount a(t) of the task and stores the data in a cache;

[0019] 3b. Construct a backlog queue Q(t) of data to be unloaded, which is the total amount of data to be unloaded in the low-orbit satellite at the beginning of time slot t. The value of data to be unloaded is the sum of the amount of unloaded data stored in the low-orbit satellite before the t-th time slot and the amount of data to be unloaded generated in the t-th time slot. At time slot 0, the value of data to be unloaded is Q(0)=0;

[0020] 3c. Construct the excess total energy consumption queue Z(t), initialize Z(0) = 0, and calculate the excess total energy consumption value by judging whether the difference between the energy consumption of the low-orbit satellite in time slot t-1 and the average energy threshold e plus the energy consumption value of Z(t-1) is greater than 0. If so, assign it to Z(t), otherwise assign 0 to Z(t).

[0021] Furthermore, the channel gain process of the satellite-to-ground link is predicted as follows:

[0022] 4a. Construct the total number of ground station unloading function n g (t), that is, the ground station before and including time slot t The total number of times selected by low-orbit satellites for data offloading, when n g When (t) = 0, it indicates that the ground station Never been chosen by LEO satellites to offload data;

[0023] 4b. In each time slot t, predict the channel gain of the satellite-to-ground link formed between the low-orbit satellite and the ground station Using the formula Computing ground station The predicted value of the channel gain corresponding to time slot t If n g (t-1) = 0, set Will 0 Defined as the default value for the channel gain, set to a constant determined empirically.

[0024] Furthermore, the optimal unloaded transmission power of the low-orbit satellite is calculated as follows:

[0025] 5a. In each time slot t, use the formula: According to the predicted value of channel gain Calculate the offload transmit power cutoff point Where Q(t) is the backlog queue of data to be unloaded. The value of is divided into the following three cases to calculate the optimal unloading transmission power of the low-orbit satellite

[0026] 5b. If the transmit power demarcation point is uninstalled It is within the transmission power range of all ground stations transmitting data, that is, P min Represents the minimum transmission power of the satellite, P max Indicates the maximum transmission power of the satellite and the optimal transmission power under construction conditions And use the formula Calculate the optimal transmit power based on the conditions The value of is divided into the following three cases to calculate the optimal unloading transmission power of the low-orbit satellite

[0027] 5b1. If Then set the optimal unloading power

[0028] 5b2. If Then set the optimal unloading power

[0029] 5b3. If Then set the optimal unloading power

[0030] 5c. If Optimal transmit power for construction conditions And use the formula Calculate, then according to The value of is divided into the following three cases to calculate the optimal unloading transmission power of the low-orbit satellite

[0031] 5c1. If Then set the optimal unloading power

[0032] 5c2, if Then set the optimal unloading power

[0033] 5c3. If Then set the optimal unloading power

[0034] 5d. If Then set the optimal unloading power

[0035] Furthermore, the specific steps for selecting the optimal ground station for data unloading are as follows:

[0036] 6a. In each time slot t, calculate the distance between the low-orbit satellite and the ground station in time slot t. The rate of unloaded data R g (t), according to Shannon's theorem, in For all predicted accessible ground stations The channel gain, To optimize the unloading transmission power;

[0037] 6b. Use the following formula to calculate the distance between the low-orbit satellite and the ground station in time slot t. Time consumed by uninstalling data d g (t):

[0038]

[0039] 6c. Calculate the optimal ground station for linking with the LEO satellite:

[0040] Build function: Where V is a constant set based on experience,

[0041] Using the formula: Calculate the function U g (t) The ground station with the minimum value The ground station with the minimum value is selected as the optimal ground station g for satellite data unloading in time slot t. * , if there are multiple different ground stations such that U g (t) takes the minimum value, then randomly select one from it;

[0042] 6d. Construct data offloading strategy x in time slot t g,t , data offloading strategy x g,t Indicates whether the low-orbit satellite is sending data to the ground station in time slot t. Uninstall data, x g,t is a zero-one variable, when x g,t =1, indicating that the low-orbit satellite chooses to unload data to the ground station g in time slot t, and x g,t =0, indicating that the low-orbit satellite does not choose to unload data to the ground station g in time slot t;

[0043] 6e. Through the optimal ground station g * Determine the optimal data offloading strategy x g,t The value of

[0044] Furthermore, the update process of the backlog queue of data to be offloaded, the excess total energy consumption queue, and the average channel gain are calculated using the UCB algorithm as follows:

[0045] 7a. In each time slot t, first use the formula: Through the optimal data offloading strategy x g,t Calculate the unloading rate of the low-orbit satellite in time slot t, and then use the formula: Calculate the unloading consumption time d(t) of the low-orbit satellite in time slot t, and finally use the formula: b(t) = R(t)d(t) to calculate the data unloading amount of the low-orbit satellite in time slot t;

[0046] 7b. According to the formula Calculate the unloading energy consumption e(t) of the low-orbit satellite in time slot t;

[0047] 7c. Update the backlog queue Q(t) of data to be unloaded according to the formula: Q(t+1)=max[Q(t)-b(t),0]+a(t);

[0048] 7d. Use the formula: Update the excess total energy consumption queue Z(t);

[0049] 7e. Use the formula: Function of uninstall times n g (t) is updated using the formula: Average channel gain Update, where Ψ g (t) is the distance from the low-orbit satellite to the ground station in time slot t The channel gain obtained when offloading data;

[0050] 7f. Determine whether the preset termination condition is met. If so, output the final calculation result. Otherwise, repeat the contents of 7a-7e.

[0051] A computer device comprises a processor and a memory, wherein the processor is electrically connected to the memory, the memory is used to store instructions and data, and the processor is used to execute an online joint power allocation and task scheduling method for a low-orbit satellite network.

[0052] Compared with the prior art, the present invention has the following advantages:

[0053] 1) This method improves network performance by jointly optimizing power allocation and task scheduling online. Compared with traditional methods, it takes into account the average energy consumption constraints of low-orbit satellites and efficiently controls the energy consumption of low-orbit satellites from a time-averaged perspective, which greatly improves the efficiency of energy management and has high practicality.

[0054] 2) The present invention proposes an online learning algorithm, which can not only perform real-time evaluation of the channel gain between the low-orbit satellite and the ground station, solving the problem that the satellite-to-ground link channel state cannot be predicted in real time under large time and space, but also adapt to the arrival of random space data and time-varying satellite-to-ground link resources, thereby realizing real-time updating of network resources and real-time scheduling of space tasks, and improving the utilization rate of low-orbit satellite network resources.

[0055] 3) The present invention iteratively optimizes the data offloading task scheduling strategy and the power allocation strategy through a two-layer optimization method, thereby obtaining an approximate optimal solution. Specifically, by using the Lyapunov optimization framework and the UCB algorithm, the tightly coupled power allocation problem and the task scheduling problem are separated, and the two-layer optimization method is used to remove the coupling between the power allocation and task scheduling problems, thereby obtaining a high-quality solution. The UCB algorithm is used in the outer optimization to solve the task scheduling, and the classification method is used in the inner optimization to obtain the optimal solution to the power allocation problem. By converting the joint online optimization problem into a set of sub-problems for each time slot, the complexity of the resource management problem in the network is greatly simplified. On this basis, the optimal solution to the power allocation is obtained for each sub-problem, which improves the efficiency and quality of problem solving. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 It is a flow chart of an online joint power allocation and task scheduling method for a low-orbit satellite network of the present invention;

[0057] Figure 2 is a schematic diagram of a scenario used in an embodiment of the present invention;

[0058] Figure 3 is a schematic diagram of the communication situation between the satellite and the ground station changing over time in the use scenario of an embodiment of the present invention;

[0059] Figure 4 is a simulation result diagram of the average data arrival rate and the total amount of unloaded data in an embodiment of the present invention;

[0060] Figure 5 It is a simulation result diagram of average data arrival rate and average energy consumption in an embodiment of the present invention. DETAILED DESCRIPTION

[0061] The embodiments of the present invention are further described below in conjunction with the accompanying drawings.

[0062] The invention discloses an online joint power allocation and task scheduling method for a low-orbit satellite network.

[0063] Figure 1The steps included in the online joint power allocation and task scheduling method for low-orbit satellite networks of the present invention are shown. This example starts from the data offloading scenario in the low-orbit satellite network to illustrate the implementation process of the present invention.

[0064] The data offloading task scheduling strategy and power allocation strategy are iteratively optimized through a two-layer optimization method to obtain an approximate optimal solution. Specifically, the Lyapunov optimization method is used to transform the online joint power allocation and task scheduling problem into a single-slot joint optimization problem. The two-layer optimization method is further used to remove the coupling of the power allocation and task scheduling problems to obtain a high-quality solution. The UCB algorithm is used in the outer layer optimization to solve the task scheduling, and the optimal solution to the power allocation problem is obtained through the classification method in the inner layer optimization.

[0065] like Figure 2 and Figure 3 As shown, the example implementation steps of the present invention are as follows:

[0066] Step 1: Obtain the ground station set and low-orbit satellite network parameters, and initialize the average channel gain: Construct the ground station set Among them, G is the total number of ground stations. Specifically, according to the working environment of the ground station, select the ground station that meets the conditions and add its index to the ground station set. The low-orbit satellite network parameters that need to be obtained are: the average energy threshold of the low-orbit satellite Channel bandwidth B c , time slot length τ, minimum transmission power P of low-orbit satellite min and the maximum transmit power P max , initialize each ground station The average channel gain

[0067] See also Figure 3 , the number of ground stations is 3, so initialize the ground station set

[0068] Step 2: In each time slot t, construct a set of accessible ground stations Import the ephemeris of the low-orbit satellite and the longitude and latitude information of each ground station into the satellite toolkit STK software, calculate the visible time window of the low-orbit satellite and each ground station, and determine whether the time slot t is within the ground station If the ground station g is within the visible time window of the low-orbit satellite, then the ground station g is added to the accessible ground station set. Gather the ground stations one by one By judging all ground stations in the The construction of

[0069] See also Figure 3, the number of time slots is 3, so the time slot set According to the communication situation between the low-orbit satellite and the ground station, we can get

[0070] Step 3: Build the data set to be unloaded, the backlog queue of data to be unloaded, and the excess total energy consumption queue:

[0071] (3a) constructing a set of data to be unloaded a = {a(1), a(2), ..., a(T)}, where a(t) represents the amount of data arriving at the low-orbit satellite network in time slot t. Specifically, at the beginning of time slot t, the data acquisition device in the low-orbit satellite acquires the amount of data a(t) of the task and stores the data in the cache;

[0072] (3b) constructing a backlog queue Q(t) of data to be unloaded, which is the total amount of data to be unloaded in the low-orbit satellite at the beginning of time slot t. Its value is the sum of the amount of unloaded data stored in the low-orbit satellite before the t-th time slot and the amount of data to be unloaded generated in the t-th time slot. In particular, its value is Q(0) = 0 at time slot 0;

[0073] (3c) Construct the excess total energy consumption queue Z(t) and initialize Z(0) = 0. The numerical calculation process is to determine the energy consumption of the low-orbit satellite in time slot t-1 and the average energy threshold Is the difference plus the energy consumption value of Z(t-1) greater than 0? If so, assign it to Z(t); otherwise, assign 0 to Z(t);

[0074] Step 4: In each time slot t, predict the accessible ground stations Channel gain

[0075] (4a) Construct the total number of ground station unloading function n g (t), that is, the ground station before and including time slot t The total number of times selected by low-orbit satellites for data offloading, for n g (t) = 0, indicating that the ground station Never been chosen by LEO satellites to offload data;

[0076] (4b) Predict the channel gain of the satellite-to-ground link Using the formula Computing ground station The predicted value of the channel gain corresponding to time slot t In particular, if n g (t-1) = 0, set Among them Ψ 0 is the default value of channel gain, which can be set as a constant based on experience;

[0077] Step 5: In each time slot t, according to the channel gain predicted in step 4 Calculating the Optimal Unloaded Transmit Power for Low Earth Orbit Satellites That is, in time slot t, Optimal unloading transmission power of low-orbit satellites performing data unloading:

[0078] (5a) Based on the predicted value of the channel gain obtained in step 4 Calculate the offload transmit power cutoff point That is Where Q(t) is the backlog queue of data to be unloaded constructed in step (3b), according to The value of is divided into the following three cases to calculate the optimal unloading transmission power of the low-orbit satellite

[0079] (5b) If Optimal transmit power for construction conditions And use the formula Calculate, based on The value of is divided into the following three cases to calculate the optimal unloading transmission power of the low-orbit satellite

[0080] (5b1) If Then set the optimal unloading power

[0081] (5b2) If Then set the optimal unloading power

[0082] (5b3) If Then set the optimal unloading power

[0083] (5c) If Optimal transmit power for construction conditions And use the formula Calculate, based on The value of is divided into the following three cases to calculate the optimal unloading transmission power of the low-orbit satellite

[0084] (5c1) If Then set the optimal unloading power

[0085] (5c2) If Then set the optimal unloading power

[0086] (5c3) If Then set the optimal unloading power

[0087] (5d) If Then set the optimal unloading power

[0088] Step 6: In each time slot t, select the optimal ground station for data unloading: According to the optimal unloading transmission power of the low-orbit satellite obtained in (5), Select the best ground station to generate the best data offloading strategy:

[0089] (6a) Calculate the distance between the low-orbit satellite and the ground station in time slot t The rate of unloaded data R g (t), according to Shannon's theorem, in This is the accessible ground station predicted in step 4 The channel gain, This is the optimal unloading transmission power obtained in step 5;

[0090] (6b) Calculate the distance between the low-orbit satellite and the ground station in time slot t Time consumed by uninstalling data d g (t), specifically, according to the formula

[0091] (6c) Calculate the optimal ground station. Specifically, first construct the function Where V is the constant set in step 5, and then calculate the function U g (t) The ground station with the minimum value The optimal ground station g selected by the satellite for data offloading in time slot t * ,Right now If there are multiple different ground stations such that U g (t) takes the minimum value, then randomly select one from it;

[0092] (6d) Construct the data offloading strategy x in time slot t g,t , specifically, x g,t is a zero-one variable indicating whether the low-orbit satellite is sending signals to the ground station in time slot t. Unload data, where x g,t =1 means that the low-orbit satellite chooses to unload data to the ground station g in time slot t, and x g,t =0 means that the low-orbit satellite does not choose to unload data to the ground station g in time slot t;

[0093] (6e) Calculate the unloading strategy based on the optimal ground station obtained in (6c). Specifically, according to g * Determine the decision variable x constructed in (6d) g,t The value of

[0094] Step 7: In each time slot t, calculate the data unloading amount and energy consumption of the current time slot, update the backlog queue of data to be unloaded, the excess total energy consumption queue and the average channel gain. Specifically:

[0095] (7a) Calculate the data unloading amount b(t) in the current time slot. Specifically, first, according to the unloading strategy x calculated in step 6, g,t Calculate the unloading rate of the low-orbit satellite in time slot t Secondly, the unloading time consumed by the low-orbit satellite in time slot t is calculated as On this basis, the data unloading amount of the low-orbit satellite in time slot t is calculated b(t)=R(t)d(t);

[0096] (7b) According to the formula Calculate the unloading energy consumption e(t) of the low-orbit satellite in time slot t;

[0097] (7c) Update Q(t) according to the formula Q(t+1)=max[Q(t)-b(t),0]+a(t);

[0098] (7d) Using the formula Update the excess total energy consumption queue Z(t);

[0099] (7e) Update the average channel gain. Specifically, first use the formula Function of uninstall times n g (t) is updated. Further, the formula The average of the observed channel gains Update, where Ψ g (t) is the distance from the low-orbit satellite to the ground station in time slot t The channel gain obtained when offloading data;

[0100] (7f) Determine whether the termination condition is met, if so, terminate, otherwise return to step 4. The termination condition can be set to determine whether the time slot t reaches the upper limit of the number of time slots.

[0101] The technical effect of the present invention is further illustrated by the following simulation.

[0102] 1. Simulation conditions

[0103] The following is a scenario involving 100 ground stations as an example for verification to illustrate the advantages of the present invention.

[0104] Assume that the time slot length τ = 1s, and there are 100,000 time slots in total. Due to the dynamic nature of the satellite-to-ground link, the low-orbit satellite can select at most one of 20 ground stations for data unloading in each time slot. A data unloading task is generated at the beginning of each time slot. The amount of data to be unloaded for each task follows the Poisson distribution. The parameters of the Poisson distribution are the average arrival rate of the data to be unloaded and the channel bandwidth B. c =960GHz, average energy consumption constraint e=100J, constant V=3×10 7 , the minimum transmission power P of the low-orbit satellite min =0W, maximum transmission power P max =200W.

[0105] There are four power allocation and task scheduling schemes used in the simulation: one is the scheme of the present invention; the second is the optimal scheme, that is, the scheduling scheme obtained by using the optimal power allocation method in step 5 when the channel gains of all ground stations in each time slot are known; the third is the random scheme, that is, the power allocation and ground station selection of each time slot are generated by a random algorithm; the fourth is the DAC scheme proposed by the author Cheng in the article "Space / aerial-assisted computing offloading for IoT applications: A learning-based approach".

[0106] 2. Simulation content and results

[0107] Simulation 1: The above four schemes are used to simulate and compare the total amount of unloaded data of low-orbit satellites. The results are as follows: Figure 4 .from Figure 4 It can be seen that as the scale of unloading task data increases, the total amount of data unloaded by the random scheme and the DAC algorithm remains basically unchanged, indicating that the above two schemes cannot adapt to changes in the scale of task data. In comparison, the total amount of data unloaded by the scheme of the present invention and the optimal scheme increases with the increase in the scale of task data, indicating that the scheme of the present invention and the optimal scheme can adapt to scenarios with different scales of task data. In particular, compared with the random scheme and the DAC algorithm, the total amount of data unloaded by the scheme of the present invention greatly exceeds that of these two schemes, and is very close to the optimal scheme. This shows that the scheme of the present invention is more accurate in predicting channel gain and has higher efficiency in scheduling and power allocation of unloading tasks.

[0108] Simulation 2: The average energy consumption of low-orbit satellites is simulated and compared using the above four schemes. The results are as follows: Figure 5 .from Figure 5It can be seen that as the scale of offloaded task data increases, the average energy consumption of the random scheme remains basically unchanged, while the average energy consumption of the DAC algorithm fluctuates within a certain range, which again shows that these two schemes cannot adapt to changes in the scale of task data. In comparison, the average energy consumption of the scheme of the present invention and the optimal scheme increases with the increase in the scale of task data, indicating that the scheme of the present invention and the optimal scheme can adapt to scenarios with different scales of task data. In particular, the average energy consumption of the scheme of the present invention no longer increases after reaching a certain level, indicating that the power allocation of the scheme of the present invention takes into account the average energy consumption constraint. Compared with the DAC algorithm, the average energy consumption of the scheme of the present invention is less. Combined with Figure 4 The scheme of the present invention can approach the optimal scheme well under the premise of consuming a small amount of additional energy, which shows that the scheme of the present invention has high efficiency in power allocation and task scheduling of data offloading tasks.

[0109] The above description is only a specific example of the present invention. Obviously, for professionals in this field, after understanding the content and principles of the present invention, it is possible to make various modifications and changes in form and details without departing from the principles and structures of the present invention. However, these modifications and changes based on the ideas of the present invention are still within the scope of protection of the claims of the present invention.

Claims

1. An online joint power allocation and task scheduling method for low-orbit satellite networks, characterized in that: For the data unloading scenario of the low-orbit satellite network that has been studied, including a low-orbit satellite and multiple ground stations connected to the low-orbit satellite signal, the Lyapunov optimization framework is used to separate the tightly coupled power allocation problem and task scheduling problem during the data unloading and scheduling process between the low-orbit satellite and the ground station. Due to the communication delay in the satellite-to-ground link, it is difficult to obtain the channel gain of the satellite-to-ground link in real time, and thus it is impossible to obtain the current data unloading amount in real time. Therefore, the Upper Confidence Bound (UCB) algorithm is used to predict the delayed channel gain, so as to calculate the network data unloading amount in real time and provide real-time feedback. The specific steps are as follows: S1. Obtain the network parameters of low-orbit satellites and all ground stations, including the number of ground stations G and the ground station generation set Average energy threshold of low-orbit satellites and all ground stations Channel bandwidth B c , the minimum transmission power P of low-orbit satellites to transmit data to all ground stations min And the maximum transmission power P max ; Initialize the average channel gain of each ground station g represents the ground station index, and the time axis is divided into T time slots with a time slot length of τ at equal intervals, thereby obtaining the time slot index set S2. Use the Satellite Tool Kit software to calculate the visible time window between the low-orbit satellite and each ground station, and determine the time slot Whether it is within the visible time window between the ground station and the low-orbit satellite, the set of accessible ground stations in each time slot t is constructed based on the judgment result S3, construct a set of data to be unloaded that arrives at the low-orbit satellite network within time slot t, construct a backlog queue Q(t) of data to be unloaded, and construct an excess total energy consumption queue Z(t) for calculating the optimal power; S4. In each time slot t, predict all current satellite-to-ground link channel gains of all accessible ground stations of the low-orbit satellite. S5. In each time slot t, according to the current satellite-to-ground link channel gains of all accessible ground stations of the low-orbit satellite, The Lyapunov optimization framework is used to calculate the offload transmit power cutoff point to calculate the optimal offload transmit power of the low-orbit satellite. S6. In each time slot t, the optimal unloading transmission power of the low-orbit satellite is With the goal of maximizing the amount of data offloading, the optimal ground station is selected to generate the optimal data offloading strategy for data offloading x g,t ; S7. Due to the delay in the satellite-to-ground link channel gain, it is impossible to obtain the optimal power allocation and the real-time channel gain information required by the ground station. Therefore, the UCB algorithm is used in each time slot t to predict the satellite-to-ground link channel gain through historical channel gain information, calculate the current time slot data unloading amount and the current time slot energy consumption, and provide real-time feedback based on the calculation results, and then update the backlog queue of data to be unloaded, the excess total energy consumption queue and the average channel gain. The average energy consumption of satellite unloading data is controlled within the energy consumption constraint.

2. The online joint power allocation and task scheduling method for low-orbit satellite networks according to claim 1 is characterized in that: The accessible ground station set in each time slot t is constructed as follows: The ephemeris of the low-orbit satellite and The latitude and longitude information of each ground station in the satellite toolkit STK software is imported to calculate the visible time window between the low-orbit satellite and each ground station, and determine whether the time slot t is within the ground station If the ground station g is within the visible time window of the low-orbit satellite, then add the ground station g to the accessible ground station set. Gather the ground stations one by one By judging all ground stations in the 's construction.

3. The online joint power allocation and task scheduling method for low-orbit satellite networks according to claim 1 is characterized in that: The specific process of constructing a set of data to be unloaded that arrives at the low-orbit satellite network within time slot t, constructing a backlog queue of data to be unloaded, and constructing an excess total energy consumption queue is as follows: 3a. Construct a data set to be unloaded a = {a(1), a(2), ..., a(T)}, where 1, 2 ... T represent time slot indexes, T is the largest time slot index number, and a(t) represents the amount of data arriving at the low-orbit satellite network in time slot t. Specifically, at the beginning of time slot t, the data acquisition device in the low-orbit satellite acquires the data amount a(t) of the task and stores the data in a cache; 3b. Construct a backlog queue Q(t) of data to be unloaded, which is the total amount of data to be unloaded in the low-orbit satellite at the beginning of time slot t. The value of data to be unloaded is the sum of the amount of unloaded data stored in the low-orbit satellite before the t-th time slot and the amount of data to be unloaded generated in the t-th time slot. At time slot 0, the value of data to be unloaded is Q(0)=0; 3c. Construct the excess total energy consumption queue Z(t), initialize Z(0) = 0, and calculate the excess total energy consumption value by judging the energy consumption of the low-orbit satellite in time slot t-1 and the average energy threshold. Is the difference plus the energy consumption value of Z(t-1) greater than 0? If so, assign it to Z(t); otherwise, assign 0 to Z(t).

4. The online joint power allocation and task scheduling method for low-orbit satellite networks according to claim 3 is characterized in that: The process of predicting the channel gain of the satellite-to-ground link is as follows: 4a. Construct the total number of ground station unloading function n g (t), that is, the ground station before and including time slot t The total number of times selected by low-orbit satellites for data offloading, when n g When (t) = 0, it indicates that the ground station Never been chosen by LEO satellites to offload data; 4b. In each time slot t, predict the channel gain of the satellite-to-ground link formed between the low-orbit satellite and the ground station Using the formula Computing ground station The predicted value of the channel gain corresponding to time slot t If n g (t-1) = 0, set Ψ0 is defined as the default value of the channel gain, which is set to a constant determined empirically.

5. The online joint power allocation and task scheduling method for low-orbit satellite networks according to claim 1 is characterized in that: The process of calculating the optimal unloaded transmission power of a low-orbit satellite is as follows: 5a. In each time slot t, use the formula: According to the predicted value of channel gain Calculate the offload transmit power cutoff point Where Q(t) is the backlog queue of data to be unloaded. The value of is divided into the following three cases to calculate the optimal unloading transmission power of the low-orbit satellite 5b. If the transmit power demarcation point is uninstalled It is within the transmission power range of all ground stations transmitting data, that is, P min Represents the minimum transmission power of the satellite, P max Indicates the maximum transmission power of the satellite and the optimal transmission power under construction conditions And use the formula Calculate the optimal transmit power based on the conditions The value of is divided into the following three cases to calculate the optimal unloading transmission power of the low-orbit satellite 5b1. If Then set the optimal unloading power 5b2. If Then set the optimal unloading power 5b3. If Then set the optimal unloading power 5c. If P0 g (t)∈(P max ,∞), construct the conditional optimal transmission power P g 0 (t), and use the formula Calculate, then according to The value of is divided into the following three cases to calculate the optimal unloading transmission power of the low-orbit satellite 5c1. If Then set the optimal unloading power 5c2, if Then set the optimal unloading power 5c3. If Then set the optimal unloading power 5d. If Then set the optimal unloading power 6. The online joint power allocation and task scheduling method for low-orbit satellite networks according to claim 1 is characterized in that: The specific steps for selecting the optimal ground station for data offloading are as follows: 6a. In each time slot t, calculate the distance between the low-orbit satellite and the ground station in time slot t. The rate of unloaded data R g (t), according to Shannon's theorem, in For all predicted accessible ground stations The channel gain, To optimize the unloading transmission power; 6b. Use the following formula to calculate the distance between the low-orbit satellite and the ground station in time slot t. Time consumed by uninstalling data d g (t): 6c. Calculate the optimal ground station for linking with the LEO satellite: Build function: Where V is a constant set based on experience, Using the formula: Calculate the function U g (t) The ground station with the minimum value The ground station with the minimum value is selected as the optimal ground station g for satellite data unloading in time slot t. * , if there are multiple different ground stations such that U g (t) takes the minimum value, then randomly select one from it; 6d. Construct data offloading strategy x in time slot t g,t , data offloading strategy x g,t Indicates whether the low-orbit satellite is sending data to the ground station in time slot t. Uninstall data, x g,t is a zero-one variable, when x g,t =1, indicating that the low-orbit satellite chooses to unload data to the ground station g in time slot t, and x g,t =0, indicating that the low-orbit satellite does not choose to unload data to the ground station g in time slot t; 6e. Through the optimal ground station g * Determine the optimal data offloading strategy x g,t The value of 7. The online joint power allocation and task scheduling method for low-orbit satellite networks according to claim 6 is characterized in that: The update process of the backlog queue of data to be offloaded, the excess total energy consumption queue and the average channel gain using the UCB algorithm is as follows: 7a. In each time slot t, first use the formula: Through the optimal data offloading strategy x g,t Calculate the unloading rate of the low-orbit satellite in time slot t, and then use the formula: Calculate the unloading consumption time d(t) of the low-orbit satellite in time slot t, and finally use the formula: b(t) = R(t)d(t) to calculate the data unloading amount of the low-orbit satellite in time slot t; 7b. According to the formula Calculate the unloading energy consumption e(t) of the low-orbit satellite in time slot t; 7c. Update the backlog queue Q(t) of data to be unloaded according to the formula: Q(t+1)=max[Q(t)-b(t),0]+a(t); 7d. Use the formula: Update the excess total energy consumption queue Z(t); 7e. Use the formula: Function of uninstall times n g (t) To update, use the formula: Average channel gain Update, where Ψ g (t) is the distance from the low-orbit satellite to the ground station in time slot t The channel gain obtained when offloading data; 7f. Determine whether the preset termination condition is met. If so, output the final calculation result. Otherwise, repeat the contents of 7a-7e.

8. A computer device, characterized in that: It includes a processor and a memory, the processor is electrically connected to the memory, the memory is used to store instructions and data, and the processor is used to execute the online joint power allocation and task scheduling method for a low-orbit satellite network as described in any one of claims 1-7.

Images