Physical layer security-oriented single-user photoelectric fusion edge computing system resource allocation method

Abstract

The invention belongs to the technical field of communication, and discloses a physical layer security-oriented single-user photoelectric fusion edge computing system resource allocation method, which comprises the following steps of: establishing a queue model of each step of task unloading in a photoelectric fusion system, and deducing a queue length dynamic expression; performing problem modeling by taking maximization of an average processing rate on all frames as a target; a physical layer security resource transmission allocation strategy based on a photoelectric dual-channel channel is provided, and under the condition of considering the security of a physical layer, the optimization problem is solved by using a Lyapunov optimization theory by taking the throughput of a system as a target. The method has a good effect on reasonable resource allocation of single-user communication, the eavesdropping effect of the eavesdropping end is reduced, and the resource utilization rate and the communication quality of the whole communication network architecture are improved.

Description

Technical Field

[0001] The present invention belongs to the field of communication technology, and in particular relates to a resource allocation method for a single-user optoelectronic fusion edge computing system oriented to physical layer security. Background Art

[0002] With the development of Mobile Edge Computing (MEC) technology, more and more terminal devices access the network through base stations, offloading computing-intensive tasks to edge computing servers, thereby achieving higher computing efficiency and low-latency experience. However, the existing MEC system still has some shortcomings in the process of information transmission and processing.

[0003] First, existing wireless communication systems face challenges in information security. Wireless communication channels are vulnerable to eavesdropping attacks, especially during uplink and downlink transmission, where the risk of information leakage is high. Although there are some secure communication technologies based on encryption algorithms, these technologies usually rely on high computing resources and complex key management mechanisms, making them difficult to effectively apply on resource-constrained edge devices.

[0004] Secondly, the resource allocation efficiency of existing MEC systems is not ideal. Due to the limited resources of edge computing, how to reasonably allocate computing resources and communication bandwidth among different users to maximize the system's processing rate and reduce task delays is a difficulty in current research. Existing resource allocation methods usually adopt fixed scheduling strategies and cannot be flexibly adjusted according to the dynamic changes of task loads, resulting in low resource utilization, especially in multi-user environments, where the system's processing capacity and throughput are limited.

[0005] In addition, the fusion technology of optical communication and electrical communication has not been fully applied. Although optical communication has the advantages of rich spectrum resources and high security, its stability is poor under severe weather conditions, while the spectrum resources of electrical communication are limited and its security is not as good as that of optical communication. Therefore, how to effectively integrate optical communication and electrical communication to dynamically switch under different conditions and further improve the security of information transmission and the throughput of the system is an urgent problem to be solved by existing technologies. Summary of the invention

[0006] The purpose of the present invention is to propose a resource allocation method for a single-user optoelectronic fusion edge computing system with good security and high transmission rate for physical layer security, so as to meet the information transmission processing and service quality requirements that are difficult to meet with traditional MIMO systems.

[0007] In order to achieve the above technical objectives, the present invention adopts the following technical solutions:

[0008] A resource allocation method for a single-user optoelectronic fusion edge computing system for physical layer security is proposed. The method includes establishing a queue model for each step of task offloading in the optoelectronic fusion system, deriving a dynamic expression for the queue length, and modeling the problem with the goal of maximizing the average processing rate on all frames. A physical layer security resource transmission allocation strategy based on an optoelectronic dual-channel channel is proposed. Taking physical layer security into consideration, the optimization problem is solved using the Lyapunov optimization theory with the goal of maximizing the system throughput.

[0009] Furthermore, the modeling of the queue model includes establishing two queue models, namely, an optical queue and an electrical queue. The queue model includes an unloading sending queue Q up , local computing queue Q lc , edge computing queue Q ec , edge computing results return queue Q down .

[0010] Furthermore, the queue model sets a power limit and establishes an optimization model with the maximum average throughput as the goal.

[0011] Furthermore, the Lyapunov optimization theory is used to establish the optimization problem solution, which is to use the Lyapunov optimization theory algorithm to minimize the upper bound of drift plus penalty at each moment, give the expression of drift plus penalty, solve the optimal uplink communication power, MEC server computing resources, downlink power, and time slot resources, and obtain a stable resource allocation method with maximum throughput.

[0012] Furthermore, the optoelectronic fusion system integrates wireless communication and optical communication systems, including: a transmitting end user, a receiving end base station, an edge computing system, and takes into account an illegal eavesdropping end; wherein the transmitting end includes a wireless queue transmitting end and an optical queue transmitting end.

[0013] Furthermore, the terminal information is offloaded to the edge computing system for computation and processing, and the queue model is considered between the terminal and the edge computing server:

[0014] The terminal maintains two queues locally, namely the local computing queue Q lc and unload queue Q up , where the unload queue Q up It is an optical queue or an electrical queue. When a task is generated at the terminal, it is necessary to decide whether to calculate the task locally or offload the calculation. If it is decided to calculate locally, then the task is placed in the calculation queue; if it is decided to offload the calculation, then the task is placed in the sending queue. First, establish the model of the local calculation queue and the sending queue. Whether the sending queue is selected as an optical queue or an electrical queue is determined by the subsequent channel selection.

[0015] Furthermore, after receiving the uplink data from the terminal, the base station immediately sends it to the edge computing server. The edge computing server maintains two queues, namely edge computing queue Q ec And the result is returned to queue Q down , where the results are returned to queue Q down In the same optical or electrical queue, the task first enters the edge computing queue Q ec , the edge computing server provides a certain amount of computing power to process the edge computing queue Q ec The tasks in the queue and the results generated enter the result return queue Q down , the result is returned to queue Q down The choice of optical queue or electrical queue is also determined by the subsequent channel selection. Finally, the execution result of the task is sent to the terminal through the base station to complete the entire offloading process.

[0016] Furthermore, a time division duplex communication mode is selected, the time of the duplex system is divided into frames of duration Δ, and then uplink time slots and downlink time slots are divided in each frame, and the proportion of uplink time slots and downlink time slots in each frame is configured in a dynamic time slot manner, where the proportion of uplink time slots in each frame is τ up , τ up ∈[0,1], the downlink time slot ratio is τ down , and there are also The length of the uplink time slot in frame t is The downlink time slot duration is The data volume of a single task is λ, in bits or bytes. The computational density of the task, that is, the amount of computation required per unit data volume is ρ, and the unit is CPU computation cycle per bit or Byte; the ratio of the amount of data returned by the task to the amount of data of the task itself is υ, that is, the amount of computation required to execute each task is ρ·λ, and the amount of data generated by the execution is υ·λ. The system model assumes that tasks cannot be divided at the same time, that is, all the data of the task either enters the local computation queue Q lc , or all enter the unloading queue Q up However, downlink reception and local calculation can be performed at the same time. As long as the current time slot is a downlink time slot, the terminal can perform downlink reception. When it is an uplink time slot, the terminal must suspend the processing of the local calculation queue.

[0017] Furthermore, the system model is established as follows:

[0018] (1) Task offloading queue model

[0019] In frame t, the amount of data generated by the terminal is d t Modeled as a random variable that follows a 0-1 distribution, the probability of generating a task is denoted by P task , if a task is generated, then dt =λ, otherwise d t =0, recorded as In frame t, the local computation queue Q lc The amount of data entered into the queue is x t ·d t , unload queue Q up The amount of data to be entered is (1-x t )·d t ,x t ∈{0,1};

[0020] With the help of the power-frequency scheduling algorithm, the local computing power p of the terminal is given lc and computing power f lc The relationship is shown in formula (1):

[0021]

[0022] Among them, f lc represents the computing frequency of the chip, that is, the number of computing cycles that can be executed per unit time. If the duration of local computing in frame t is Then in this frame, the local calculation queue Q lc The amount of data dequeued As shown in formula (2):

[0023]

[0024] in is the time proportion of local calculation in frame t. Since the terminal cannot perform local calculation and uplink transmission at the same time, τ lc ≤1-τ up ;

[0025] The terminal's local computing queue Q lc The amount of backlog data (i.e. queue length) The dynamic expression of is shown in formula (3):

[0026]

[0027] x t ∈{0,1} represents the execution strategy of the task, x t =1 means putting the task into the local execution queue, x t =0 means putting the task into the unloaded send queue. The formula shows the length of the queue in the next frame. is the current length Add column data amount x t ·d t , reduce the amount of column data And it is stipulated that the amount of data dequeued cannot be greater than the queue length, that is,

[0028] Similar to the local calculation queue model, in frame t, according to the uplink timeslot length of the frame And the communication rate You can get the unloaded send queue Q up The amount of data dequeued As shown in formula (4):

[0029]

[0030] Given the terminal's offload send queue Q up The amount of backlog data The dynamic expression of is shown in formula (5):

[0031]

[0032] That is, the length of the queue in the next frame is the current length Add column data amount (1-x t )d t , reduce the amount of column data And it is stipulated that the amount of data dequeued cannot be greater than the queue length, that is,

[0033] The edge computing server maintains two queues, namely edge computing queue Q ec and downstream sending queue Q down , ec in the symbol represents edge computing (EC). In frame t, the edge computing queue Q ec The amount of queued data is the amount of data received by the base station from the terminal uplink Amount of data dequeued The computing resources allocated to the queue by the edge computing server The amount of data completed can be calculated using formula (6);

[0034]

[0035] Also given the edge computing queue Q ec The amount of backlog data The dynamic expression of is shown in formula (7):

[0036]

[0037] In frame t, the edge computing queue Finished The amount of data generated The result of the data volume is the downlink sending queue Qdown The amount of data entering the queue is based on the length of the downlink time slot. And the communication rate The calculation is as shown in formula (8):

[0038]

[0039] Similarly, given the downstream sending queue Q down The amount of backlog data The dynamic expression of is shown in formula (9):

[0040]

[0041] (2) Problem model for maximizing the average processing rate of all frames

[0042] There are four queues in the system: 1) Unload send queue Q up ; 2) Local computing queue Q lc ; 3) Edge computing queue Q ec ; 4) Edge computing results are returned to queue Q down To simplify the notation, we use Represents the set of the above 4 queues; use It represents the set of backlog data of the above four queues at frame t;

[0043] Establish the following optimization problem, the optimization goal is to maximize the two queues maintained locally by the terminal, the local calculation queue Q lc and unload the send queue Q up , the average processing rate on all frames is the average throughput of the terminal, while satisfying constraints such as power limit. The specific optimization problem P1 is shown in formula (10):

[0044]

[0045] Constraints All of them are established. Constraints C1 to C4 indicate that the queues need to satisfy causality. The amount of data dequeued from all queues at each moment cannot be greater than the amount of data currently in the queues. Constraint C5 indicates that the uplink transmit power p of the terminal up and local computing power p lc The time average must satisfy the average power constraint P ave , is the maximum power of the base station, is the maximum power of the terminal, is the artificial noise power in the uplink phase, is the artificial noise power in the downlink phase.

[0046] Furthermore, according to the queue model of task offloading in the optoelectronic fusion system and the problem model of maximizing the average processing rate of all frames, the steps of solving the above optimization problem by using the minimization drift plus penalty algorithm based on Lyapunov optimization theory for resource allocation are as follows:

[0047] (1) Introduce a virtual power queue Y and give the backlog data volume Y of queue Y t The dynamic expression of is shown in formula (11):

[0048]

[0049] Joint Control Data Queue and virtual power queue Y, defining a queue set Given Lyapunov equation and Lyapunov drift equation As shown in formula (12):

[0050]

[0051] (2) Using the MDPP algorithm, minimize the upper bound of drift plus penalty at each moment and give the drift plus penalty The expression of , V is the penalty term weight;

[0052]

[0053] (3) Give all L(Q t+1 )-L(Q t ), As shown in formula (14):

[0054]

[0055] According to formula (14), After the expansion, all the terms are sorted out to obtain formula (15), where is the upper limit of the mathematical expectation of some terms in formula (14);

[0056]

[0057] Using inequalities An upper limit can be given The specific expression of is as shown in formula (16):

[0058]

[0059] Where R up is the maximum secure channel capacity that the terminal can obtain, F lc is the maximum computing rate of the terminal, Fec is the maximum computing rate of the edge computing server;

[0060] Let the term on the right side of the inequality sign be The value is not affected by The variables are affected, so we need to minimize is to minimize Other items except

[0061] (4) Uplink communication power Solve it and get The optimal value of Then calculate the power locally Solve it and get The optimal value of

[0062] (5) Computing resources for MEC servers Solve and get its optimal value Finally, the downlink power Solve and get its optimal value

[0063] (6) Confirm After taking the best value in various situations, we can then determine The complete solution to the problem can be obtained, thus completing the allocation of time slot resources;

[0064] (7) Update the virtual power queue, Get the maximum throughput of the terminal.

[0065] The present invention has the following beneficial effects:

[0066] (1) The processing method of transmitting information to the edge computing system of the present invention improves the information processing capability.

[0067] (2) The present invention introduces a message queue and establishes a model, uses Lyapunov optimization theory to solve the optimization problem, and maximizes the average processing rate of the system.

[0068] (3) The present invention adopts precoding and uses optical and electrical dual channels for communication, which greatly guarantees the security of user information. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] Figure 1 A system diagram that integrates the transmitting user, the receiving base station, the edge computing system, and takes into account the illegal eavesdropping communication.

[0070] Figure 2 It is a diagram of the queue model and the optoelectronic dual-channel model of the present invention. DETAILED DESCRIPTION

[0071] The present invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0072] The present invention designs a resource allocation method for a single-user optoelectronic fusion edge computing system oriented to physical layer security. Figure 1 , which includes the transmitting end user, the receiving end base station, the edge computing system, and the system that takes into account the illegal eavesdropping end communication. In this communication process, wireless communication and wireless optical communication are used for information transmission, such as Figure 2 , establish a queue model in the form of task offloading of message queues to exchange information.

[0073] (1) First, the terminal device sends information. The terminal device's memory stores electrical queues and optical queues. When transmitting electrical queues, multiple antennas are used for transmission. When transmitting optical queues, wireless optical communication uses LEDs or lasers as transmitters, uses optoelectronic devices to receive signals and convert them into electrical signals, and completes data decoding at the receiving end.

[0074] (2) Consider the choice of channel. When exchanging information, it is necessary to consider the transmission rate and security. The transmission rate of wireless optical communication is greatly affected by weather conditions. For example, under severe weather conditions such as haze, rain and snow, the transmission rate will be greatly attenuated. Therefore, when the weather is bad, choose the telecommunication channel for transmission. At the same time, the system is equipped with an eavesdropping detection device, which adopts a strategy of security over rate. When eavesdropping is detected on one channel, it will immediately switch to another channel for transmission.

[0075] (3) Queue modeling. Two queue models are established, namely, optical queue and electrical queue. The queue model has an unloaded sending queue Q up , local computing queue Q lc , edge computing queue Q ec , the edge computing results are returned to queue Q down Etc., set power limits and build an optimization model with the maximum average throughput as the goal.

[0076] (4) Solve the problem with the help of Lyapunov optimization theory. Use the Lyapunov optimization theory algorithm to minimize the upper bound of drift plus penalty at each moment, and give an expression for drift plus penalty. Solve for the optimal uplink communication power, MEC server computing resources, downlink power, time slot resources, etc., and obtain a stable resource allocation method with maximum throughput.

[0077] It is further explained that the resource allocation method for a single-user optoelectronic fusion edge computing system for physical layer security provided by the present invention includes establishing a queue model for each step of task unloading in the optoelectronic fusion system, deriving a dynamic expression of the queue length, and modeling the problem with the goal of maximizing the average processing rate on all frames; a physical layer security resource transmission allocation strategy based on an optoelectronic dual-channel channel is proposed. While considering the physical layer security, the optimization problem is established using the Lyapunov optimization theory with the goal of maximizing the system throughput.

[0078] (I) Constructing a queue model for task offloading in an optoelectronic fusion system and a problem model for maximizing the average processing rate of all frames

[0079] The communication system involved in the present invention includes a transmitting end user, a receiving end base station, an edge computing system, and considers an illegal eavesdropping end, see Figure 1 shown.

[0080] The first is the definition of the application scenario. Here, the terminal information is mainly unloaded to the edge computing system for calculation and processing. Therefore, the present invention believes that the queue model needs to be considered in the terminal and the edge computing server.

[0081] The terminal maintains two queues locally, namely the local computing queue Q lc and unload queue Q up , where the unload queue Q up It is an optical queue or an electrical queue. When a task is generated at the terminal, it is necessary to decide whether to calculate the task locally or offload the calculation. If it is decided to calculate locally, then put the task into the calculation queue; if it is decided to offload the calculation, then put the task into the sending queue. First, establish the model of the local calculation queue and the sending queue. Whether the sending queue is an optical queue or an electrical queue is determined by the subsequent channel selection.

[0082] After receiving the uplink data from the terminal, the base station immediately sends it to the edge computing server. The edge computing server maintains two queues, namely edge computing queue Q ec And the result is returned to queue Q down , where the results are returned to queue Q down The same is true for optical or electrical queues. The task first enters the edge computing queue Q ec The edge computing server provides a certain amount of computing power to process the edge computing queue Q ec The tasks in the queue and the results generated enter the result return queue Q down , the result is returned to queue Q down The choice of optical queue or electrical queue is also determined by the subsequent channel selection. Finally, the execution result of the task is sent to the terminal through the base station, completing the entire offloading process.

[0083] Select the time division duplex communication mode, divide the duplex system time into frames of duration Δ, and then divide the uplink time slot and downlink time slot in each frame. Configure the proportion of uplink time slot and downlink time slot in each frame in a dynamic time slot manner. The proportion of uplink time slot in each frame is τ up , τ up ∈[0,1], the downlink time slot ratio is τ down , and similarly τ down ∈[0,1]. The length of the uplink time slot in frame t is The downlink time slot duration is The data volume of a single task is λ, in bits or bytes. The computational density of a task, that is, the amount of computation required per unit of data, is ρ, in units of CPU computation cycles per bit or Byte; the ratio of the amount of data returned by a task to the amount of data in the task itself is υ. That is, the amount of computation required to execute each task is ρ·λ, and the amount of data generated by the execution is υ·λ. This system model assumes that tasks cannot be divided at the same time, that is, all data of a task must either enter the local computation queue Q or enter the local computation queue Q. lc , or all enter the unloading queue Q up , but it can perform downlink reception and local calculation at the same time. As long as the current time slot is a downlink time slot, the terminal can perform downlink reception. When it is an uplink time slot, the terminal must suspend the processing of the local calculation queue.

[0084] According to the above scenario, the system model is established as follows:

[0085] (1) Task offloading queue model:

[0086] In frame t, the amount of data generated by the terminal is d t Modeled as a random variable that follows a 0-1 distribution, the probability of generating a task is denoted by P task , if a task is generated, then d t =λ, otherwise d t =0, recorded as Therefore, in frame t, the local computation queue Q lc The amount of data entered into the queue is x t ·d t , unload queue Q up The amount of data to be entered is (1-x t )·d t , x t ∈{0,1}.

[0087] With the help of the power-frequency scheduling algorithm, the local computing power p of the terminal is given lc and computing power f lc The relationship is shown in formula (1). The lc in the symbol represents local computing (LC).

[0088]

[0089] Among them, f lc represents the computing frequency of the chip, that is, the number of computing cycles that can be executed per unit time. This value reflects the computing power of the terminal. If the duration of local computing in frame t is Then in this frame, the local calculation queue Q lc The amount of data dequeued As shown in formula (2).

[0090]

[0091] in is the time proportion of local calculation in frame t. Since the terminal cannot perform local calculation and uplink transmission at the same time, τ lc ≤1-τ up .

[0092] The terminal's local computing queue Q lc The amount of backlog data (i.e. queue length) The dynamic expression of is shown in formula (3).

[0093]

[0094] x t ∈{0,1} represents the execution strategy of the task, x t =1 means putting the task into the local execution queue, x t =0 means putting the task into the unloaded send queue. The formula shows the length of the queue in the next frame. is the current length Add column data amount x t ·d t , reduce the amount of column data And it is stipulated that the amount of data dequeued cannot be greater than the queue length, that is,

[0095] Similar to the local calculation queue model, in frame t, according to the uplink timeslot length of the frame And the communication rate You can get the unloaded send queue Q up The amount of data dequeued As shown in formula (4).

[0096]

[0097] Given the terminal's offload send queue Q up The amount of backlog data The dynamic expression of is shown in formula (5).

[0098]

[0099] That is, the length of the queue in the next frame is the current length Add column data amount (1-x t )d t , reduce the amount of column data And it is stipulated that the amount of data dequeued cannot be greater than the queue length, that is,

[0100] Due to the close binding between the base station and the edge computing server, the edge computing server maintains two queues, namely edge computing queue Q ec and downstream sending queue Q down , ec in the symbol represents edge computing (EC). In frame t, the edge computing queue Q ec The amount of queued data is the amount of data received by the base station from the terminal uplink Amount of data dequeued The computing resources allocated to the queue by the edge computing server The amount of data completed can be calculated using formula (6).

[0101]

[0102] Also given the edge computing queue Q ec The amount of backlog data The dynamic expression of is shown in formula (7).

[0103]

[0104] In frame t, the edge computing queue Finished The amount of data generated The result of the data volume is the downlink sending queue Q down The amount of data that is dequeued from the downlink transmission queue is based on the length of the downlink time slot. And the communication rate The calculation is shown in formula (8).

[0105]

[0106] Similarly, given the downstream sending queue Q down The amount of backlog data The dynamic expression of is shown in formula (9).

[0107]

[0108] (2) Problem model for maximizing the average processing rate of all frames:

[0109] There are four queues in the system: 1) Unload send queue Q up ; 2) Local computing queue Q lc ; 3) Edge computing queue Q ec ; 4) Edge computing results are returned to queue Q down To simplify the notation, we use Represents the set of the above 4 queues; use Represents the set of backlog data amounts of the above four queues at frame t.

[0110] Establish the following optimization problem, the optimization goal is to maximize the two queues maintained locally by the terminal, the local calculation queue Q lc and unload the send queue Q up , the average processing rate over all frames is the average throughput of the terminal, while satisfying constraints such as power limit. The specific optimization problem P1 is shown in formula (10).

[0111]

[0112] Constraints Constraints C1 to C4 indicate that the queues must satisfy causality, and the amount of data dequeued from all queues at each moment cannot be greater than the amount of data currently in the queues. Constraint C5 indicates that the uplink transmit power p of the terminal up and local computing power p lc The time average must satisfy the average power constraint P ave . is the maximum power of the base station, is the maximum power of the terminal, is the artificial noise power in the uplink phase, is the artificial noise power in the downlink stage. Solving this optimization problem can obtain the maximum average processing rate on all frames.

[0113] 2. Physical layer security resource transmission allocation strategy based on optical-electrical dual-channel

[0114] The system uses optical and electrical dual channels for information transmission. While using precoding to ensure a certain physical layer security, if eavesdropping is found in the telecommunication channel, it will immediately switch to the optical channel for transmission, and if eavesdropping is found in the optical channel, it will immediately switch to the telecommunication channel for transmission, further ensuring the security of information transmission.

[0115] To obtain the maximum average processing rate of all frames in the optimization problem in (a), we need to maximize the average throughput of the terminal. We need to use Lyapunov optimization theory to decouple the problem into a deterministic planning problem at each moment. Using the drift minimization plus penalty algorithm, the solution to the resource allocation problem can maximize the terminal throughput while maintaining the stability of the queue.

[0116] According to the task offloading model and optimization problem described in claim 1, the specific steps of using the minimization drift plus penalty algorithm based on Lyapunov optimization theory to perform resource allocation and solve the above optimization problem are as follows:

[0117] (1) Introduce a virtual power queue Y. Given the backlog of queue Y, t The dynamic expression of is shown in formula (11).

[0118]

[0119] Joint Control Data Queue and virtual power queue Y, defining a queue set Given Lyapunov equation and Lyapunov drift equation As shown in formula (12).

[0120]

[0121] (2) Using the MDPP algorithm, minimize the upper bound of drift plus penalty at each moment and give the drift plus penalty The expression of , V is the penalty term weight.

[0122]

[0123] (3) Give all L(Q t+1 )-L(Q t ), As shown in formula (14).

[0124]

[0125] According to formula (14), After the expansion, all the terms are sorted out to obtain formula (15), where It is the upper limit of the mathematical expectation of some terms in formula (14).

[0126]

[0127] Using inequalities An upper limit can be given The specific expression of is shown in formula (16).

[0128]

[0129] Where R up is the maximum secure channel capacity that the terminal can obtain, F lc is the maximum computing rate of the terminal, F ec is the maximum computing rate of the edge computing server.

[0130] Let the term on the right side of the inequality sign be The value is not affected by The variables are affected. So we need to minimize is to minimize The remaining items outside.

[0131] (4) Uplink communication power Solve it and get The optimal value of Then calculate the power locally Solve it and get The optimal value of

[0132] (5) Computing resources for MEC servers Solve and get its optimal value Finally, the downlink power Solve and get its optimal value The allocation of power and computing resources is completed.

[0133] (6) Confirm After taking the best value in various situations, we can then determine The complete solution to the problem can be obtained, thereby completing the allocation of time slot resources.

[0134] (7) Update the virtual power queue, Get the maximum throughput of the terminal.

[0135] Compared with the prior art, the advantages of the technology of the present invention are not only profoundly reflected in the significant improvement of information processing capabilities, but also in the significant progress in systematic optimization and security assurance brought about by its technological innovation, which is specifically reflected in the following aspects:

[0136] 1. Dramatic improvement in information processing capabilities

[0137] The present invention greatly improves the system's ability to process information by innovatively integrating the information transmission process into the edge computing system. In traditional network architectures, information processing mainly relies on cloud computing centers, with large processing delays and more network bandwidth occupancy. However, with the increase in the number of terminal devices and the explosive growth in data volume, traditional centralized processing methods are difficult to meet the application requirements of low latency and high bandwidth. The present invention introduces edge computing to migrate computing and storage resources from the cloud to the edge of the network. Terminal devices do not need to transmit large amounts of data to remote servers, but can quickly process them at local or nearby edge nodes. This approach not only reduces the delay in information transmission, but also improves the response speed and processing efficiency of the overall network, and can effectively meet the needs of real-time application scenarios, such as smart transportation, the Internet of Things, and video streaming media, which require fast data processing.

[0138] 2. The exquisite combination of message queue and Lyapunov optimization theory

[0139] Combination of message queue model and Lyapunov optimization theory. The present invention systematically manages complex task offloading and resource scheduling problems by introducing a message queue model. The message queue is used to coordinate the processing order of tasks to ensure that task data enters and leaves the edge computing system in an orderly manner, thereby avoiding waste of computing resources and network congestion. At the same time, the present invention innovatively uses Lyapunov optimization theory to solve the resource allocation problem in the system. Lyapunov optimization theory ensures that the system remains stable during long-term operation and maximizes the average processing rate of the entire system by controlling the dynamic changes in queue length. This means that the system can effectively schedule computing resources and communication bandwidth within different time frames, improve the overall throughput, and thus achieve higher efficiency when processing tasks. This method can not only adapt to changes in task load, but also maintain stable performance when processing high-concurrency tasks.

[0140] 3. Combination of precoding and optical-electrical dual-channel communication ensures information security

[0141] The present invention effectively improves the security of communication by introducing a combination of precoding technology and optoelectronic dual-channel communication. Precoding technology can encrypt and encode the transmitted data through intelligent processing of signals, making it difficult to be eavesdropped or cracked during transmission. In addition, the present invention innovatively adopts an optoelectronic dual-channel communication method. When the risk of eavesdropping is detected in the radio channel, the system can quickly switch to the optical communication channel to continue transmission, and vice versa. When the optical communication channel is attacked, the system will automatically switch back to the radio channel. This dynamic switching mechanism greatly improves the confidentiality of information transmission, ensures that the user's data is always in a safe environment during the transmission process, and greatly reduces the risk of information being eavesdropped or leaked. This has extremely high application value in scenarios involving sensitive data, such as financial transactions, medical data transmission, and other fields.

[0142] Today, as information transmission security is increasingly valued, this invention adopts an optoelectronic dual-channel communication solution and combines it with advanced precoding technology to provide double protection for the secure transmission of user information. The optoelectronic dual-channel effectively resists the risks of external attacks and eavesdropping through the integration of physical isolation and encryption technology. At the same time, advanced precoding technology further enhances the signal's anti-interference ability and concealment by preprocessing and optimizing the transmission signal, ensuring the integrity and confidentiality of the information during transmission. This comprehensive security strategy not only improves the security of the system, but also provides users with a more reliable and reassuring user experience. It is an important innovation in the field of information security.

[0143] In summary, the advantages of the present invention are not only its significant improvement in information processing capabilities, but also its significant progress in system optimization and security assurance through technological innovation. These advantages not only bring revolutionary changes to the existing technology, but also point out the direction for the development of future technology.

[0144] The above embodiments are only for illustrating the technical concept and features of the present invention, and their purpose is to enable people familiar with the technology to understand the content of the present invention and implement it accordingly, and they cannot be used to limit the protection scope of the present invention. Any equivalent changes or modifications made according to the spirit of the present invention should be included in the protection scope of the present invention.

Claims

1. A resource allocation method for a single-user optoelectronic fusion edge computing system for physical layer security, characterized in that: This includes establishing a queue model for each step of task unloading in an optoelectronic fusion system, deriving a dynamic expression for queue length, and modeling the problem with the goal of maximizing the average processing rate over all frames. A physical layer security resource transmission allocation strategy based on optical-electrical dual-channel is proposed. Taking physical layer security into consideration, the Lyapunov optimization theory is used to establish an optimization problem solution with the goal of maximizing the system throughput.

2. The resource allocation method for a single-user optoelectronic fusion edge computing system for physical layer security according to claim 1, characterized in that: The modeling of the queue model includes establishing two queue models, namely, an optical queue and an electrical queue. The queue model includes an unloading sending queue Q up , local computing queue Q lc , edge computing queue Q ec , edge computing results return queue Q down .

3. The resource allocation method for a single-user optoelectronic fusion edge computing system for physical layer security according to claim 2, characterized in that: The queue model sets power limits and establishes an optimization model with maximum average throughput as the goal.

4. The resource allocation method for a single-user optoelectronic fusion edge computing system for physical layer security according to claim 1, characterized in that: The Lyapunov optimization theory is used to establish the optimization problem solution. The Lyapunov optimization theory algorithm is used to minimize the upper bound of drift plus penalty at each moment, give the expression of drift plus penalty, solve the optimal uplink communication power, MEC server computing resources, downlink power, and time slot resources, and obtain a stable resource allocation method with maximum throughput.

5. The resource allocation method for a single-user optoelectronic fusion edge computing system for physical layer security according to claim 1, characterized in that: The optoelectronic fusion system integrates wireless communication and optical communication, including: a transmitting end user, a receiving end base station, an edge computing system, and takes into account an illegal eavesdropping end; wherein the transmitting end includes a wireless queue transmitting end and an optical queue transmitting end.

6. The resource allocation method for a single-user optoelectronic fusion edge computing system for physical layer security according to claim 5, characterized in that: The terminal information is unloaded to the edge computing system for computing and processing. The queue model is considered between the terminal and the edge computing server: The terminal maintains two queues locally, namely the local computing queue Q lc and unload queue Q up , where the unload queue Q up It is an optical queue or an electrical queue. When a task is generated at the terminal, it is necessary to decide whether to calculate the task locally or offload the calculation. If it is decided to calculate locally, then the task is placed in the calculation queue; if it is decided to offload the calculation, then the task is placed in the sending queue. First, establish the model of the local calculation queue and the sending queue. Whether the sending queue is selected as an optical queue or an electrical queue is determined by the subsequent channel selection.

7. The resource allocation method for a single-user optoelectronic fusion edge computing system for physical layer security according to claim 6, characterized in that: After receiving the uplink data from the terminal, the base station immediately sends it to the edge computing server. The edge computing server maintains two queues, namely edge computing queue Q ec And the result is returned to queue Q down , where the results are returned to queue Q down In the same optical or electrical queue, the task first enters the edge computing queue Q ec , the edge computing server provides a certain amount of computing power to process the edge computing queue Q ec The tasks in the queue and the results generated enter the result return queue Q down , the result is returned to queue Q down The choice of optical queue or electrical queue is also determined by the subsequent channel selection. Finally, the execution result of the task is sent to the terminal through the base station to complete the entire offloading process.

8. The physical layer security-oriented single-user optoelectronic fusion edge computing system resource allocation method according to claim 7, characterized in that: Select the time division duplex communication mode, divide the time of the duplex system into frames of duration Δ, and then divide the uplink time slot and the downlink time slot in each frame, and configure the proportion of the uplink time slot and the downlink time slot in each frame in a dynamic time slot manner, where the proportion of the uplink time slot in each frame is τ up , τ up ∈[0,1], the downlink time slot ratio is τ down , and similarly τ down ∈[0,1], the uplink time slot length in frame t is The downlink time slot duration is The data volume of a single task is λ, in bits or bytes. The computational density of the task, that is, the amount of computation required per unit data volume is ρ, and the unit is CPU computation cycle per bit or Byte; the ratio of the amount of data returned by the task to the amount of data of the task itself is v, that is, the amount of computation required to execute each task is ρ·λ, and the amount of data generated by the execution is v·λ. The system model assumes that tasks cannot be divided at the same time, that is, all the data of the task either enters the local computation queue Q lc , or all enter the unloading queue Q up However, downlink reception and local calculation can be performed at the same time. As long as the current time slot is a downlink time slot, the terminal can perform downlink reception. When it is an uplink time slot, the terminal must suspend the processing of the local calculation queue.

9. The resource allocation method for a single-user optoelectronic fusion edge computing system for physical layer security according to claim 5, characterized in that: The system model is established as follows: (1) Task offloading queue model In frame t, the amount of data generated by the terminal is d t Modeled as a random variable that follows a 0-1 distribution, the probability of generating a task is denoted by P task , if a task is generated, then d t =λ, otherwise d t =0, recorded as In frame t, the local computation queue Q lc The amount of data entered is x t ·d t , unload queue Q up The amount of data to be entered is (1-x t )·d t ,x t ∈{0,1}; With the help of the power-frequency scheduling algorithm, the local computing power p of the terminal is given lc and computing power f lc The relationship is shown in formula (1): Among them, f lc represents the computing frequency of the chip, that is, the number of computing cycles that can be executed per unit time. If the duration of local computing in frame t is Then in this frame the local calculation queue Q lc The amount of data dequeued As shown in formula (2): in is the time proportion of local calculation in frame t. Since the terminal cannot perform local calculation and uplink transmission at the same time, τ lc ≤1-τ up ; The terminal's local computing queue Q lc The amount of backlog data (i.e. queue length) The dynamic expression of is shown in formula (3): x t ∈{0,1} represents the execution strategy of the task, x t =1 means putting the task into the local execution queue, x t =0 means putting the task into the unloaded send queue. The formula shows the length of the queue in the next frame. The current length Add column data amount x t ·d t , reduce the amount of column data And it is stipulated that the amount of data dequeued cannot be greater than the queue length, that is, Similar to the local calculation queue model, in frame t, according to the uplink timeslot length of the frame And the communication rate You can get the unloaded send queue Q up The amount of data dequeued As shown in formula (4): Given the terminal's offload send queue Q up The amount of backlog data The dynamic expression of is shown in formula (5): That is, the length of the queue in the next frame The current length Add column data amount (1-x t )d t , reduce the amount of column data And it is stipulated that the amount of data dequeued cannot be greater than the queue length, that is, The edge computing server maintains two queues, namely edge computing queue Q ec and downstream sending queue Q down , ec in the symbol represents edge computing (EC). In frame t, the edge computing queue Q ec The amount of queued data is the amount of data received by the base station from the terminal uplink Amount of data dequeued The computing resources allocated to the queue by the edge computing server The amount of data completed can be calculated using formula (6); Also given the edge computing queue Q ec The amount of backlog data The dynamic expression of is shown in formula (7): In frame t, the edge computing queue Finished The amount of data generated The result of the data volume is the downlink sending queue Q down The amount of data entering the queue is based on the length of the downlink time slot. And the communication rate The calculation is as shown in formula (8): Similarly, given the downstream sending queue Q down The amount of backlog data The dynamic expression of is shown in formula (9): (2) Problem model for maximizing the average processing rate of all frames There are four queues in the system: 1) Unload send queue Q up ; 2) Local computing queue Q lc ; 3) Edge computing queue Q ec ; 4) Edge computing results are returned to queue Q down To simplify the notation, we use Represents the set of the above 4 queues; use It represents the set of backlog data of the above four queues at frame t; Establish the following optimization problem, the optimization goal is to maximize the two queues maintained locally by the terminal, the local calculation queue Q lc and unload the send queue Q up , the average processing rate on all frames is the average throughput of the terminal, while satisfying constraints such as power limit. The specific optimization problem P1 is shown in formula (10): C12:t up +t lc ≤1 Constraints All of them are established. Constraints C1 to C4 indicate that the queues need to satisfy causality. The amount of data dequeued from all queues at each moment cannot be greater than the amount of data currently in the queues. Constraint C5 indicates that the uplink transmit power p of the terminal up and local computing power p lc The time average must satisfy the average power constraint P ave , is the maximum power of the base station, is the maximum power of the terminal, is the artificial noise power in the uplink phase, is the artificial noise power in the downlink phase.

10. The resource allocation method for a single-user optoelectronic fusion edge computing system for physical layer security according to claim 1, characterized in that: According to the queue model of task offloading in the optoelectronic fusion system and the problem model of maximizing the average processing rate of all frames, the steps of solving the above optimization problem by using the minimization drift plus penalty algorithm based on Lyapunov optimization theory for resource allocation are as follows: (1) Introduce a virtual power queue Y and give the backlog data volume Y of queue Y t The dynamic expression of is shown in formula (11): Joint Control Data Queue and virtual power queue Y, define the queue set Given Lyapunov equation and Lyapunov drift equation As shown in formula (12): (2) Using the MDPP algorithm, minimize the upper bound of drift plus penalty at each moment and give the drift plus penalty The expression of , V is the penalty term weight; (3) Give all As shown in formula (14): According to formula (14), After the expansion, all the terms are sorted out to obtain formula (15), where is the upper limit of the mathematical expectation of some terms in formula (14); Using inequalities An upper limit can be given The specific expression of is as shown in formula (16): Where R up is the maximum secure channel capacity that the terminal can obtain, F lc is the maximum computing rate of the terminal, F ec is the maximum computing rate of the edge computing server; Let the term on the right side of the inequality sign be The value is not affected by The variables affect the is to minimize Other items except (4) Uplink communication power Solve it and get The optimal value of Then calculate the power locally Solve it and get The optimal value of (5) Computing resources for MEC servers Solve and get its optimal value Finally, the downlink power Solve and get its optimal value (6) Confirmation After taking the best value in various situations, we can then determine The complete solution to the problem can be obtained, thus completing the allocation of time slot resources; (7) Update the virtual power queue, Get the maximum throughput of the terminal.

Images