Resource allocation strategy optimization method and system based on mobile edge computing scene

Abstract

The invention provides a resource allocation strategy optimization method and system based on a mobile edge computing scene, and the method comprises the steps: determining the transmission error probability of each frame in a transmission process based on the mobile edge computing scene; according to the frame-level energy consumption constraint of the terminal equipment, a virtual energy queue model is constructed, the queue length is dynamically updated according to the difference value between the actual energy consumption of each frame and the preset energy budget, and the stability constraint of the virtual energy queue is set; taking minimization of the long-term accumulated average transmission error probability as a target, combining the stability constraint of the virtual energy queue with the minimization target of the long-term accumulated average transmission error probability, and constructing an optimization problem of the current frame; converting into a convex optimization problem and solving; and according to a solving result, allocating a transmission code length and transmitting power to the current frame. According to the method, the long-term average bit error rate can be remarkably reduced while the energy consumption budget is maintained, and the method is suitable for URLLC scenes such as the industrial Internet of Things and augmented reality.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of wireless communication, and particularly relates to a resource allocation strategy optimization method and system based on a mobile edge computing scenario. BACKGROUND

[0002] Mobile edge computing (MEC) as a technology that sinks computing power to the network edge has become an important part of industrial Internet of Things (IIoT) and unmanned aerial vehicle (UAV) communication systems. In typical applications such as industrial manufacturing, warehouse logistics, inspection security, etc., a large number of terminal devices, including industrial sensors, collaborative robots, mechanical arms, automated guided vehicles, and industrial UAVs, need to perform continuous tasks with millisecond-level cycles and complete data acquisition, wireless transmission, edge computing, and control feedback in each frame cycle.

[0003] These tasks generally have strict sequential dependencies, and the processing results of the previous frame often directly affect the decision and action of the next frame. For example, when an industrial UAV is patrolling, it needs to continuously collect images and upload them to the edge node for identification and positioning in real time. Any transmission error of a frame of image will cause deviation in subsequent path planning. Visual inspection in an industrial production line also relies on continuous multi-frame information, and any data error in a frame may cause interruption of the detection link. Robot collaboration or mechanical arm trajectory control requires stable input at the frame level, otherwise it will affect the motion accuracy and even cause safety problems. Therefore, in the industrial and UAV collaborative scenario, communication reliability not only relates to single-frame performance, but also has a cumulative impact on the entire task chain.

[0004] At the same time, industrial devices are often subject to strict energy consumption limits, especially battery-powered terminals such as UAVs and mobile robots. They must complete communication and computing operations within a fixed flight or motion cycle according to the energy budget, and cannot rely on long-term average energy consumption to compensate for energy overruns in a frame. This means that the system must meet the precise energy limit in each frame, and frame-level energy constraints must be strictly followed in actual engineering, which is fundamentally different from the long-term average energy constraints commonly seen in traditional communication systems.

[0005] In order to meet the requirements of low latency and high reliability of industrial and UAV systems, wireless links usually use short packet transmission mechanisms. Since short data packets cannot rely on the infinite code length assumption of the Shannon model, the finite block length (FBL) communication model becomes the only accurate way to describe the error probability. Under the condition of finite block length, the transmission error probability depends on the code length, transmission power, and channel state, and exhibits significantly different behavior characteristics from traditional theory. If the FBL error characteristics are ignored, it will be difficult to analyze and optimize the reliability of industrial or UAV systems.

[0006] Moreover, the availability of edge nodes presents significant dynamics, whether in a factory environment or a UAV network. Industrial UAVs are constantly moving while patrolling, possibly frequently entering or leaving the coverage of different MEC nodes; the wireless links where robots and sensors are located can experience rapid changes in channel gain due to blockage, interference, or device motion; the load of edge servers can also be time-varying due to the number of real-time tasks changing. The dynamics of the wireless environment, computing resources, and the task chain itself make the resource allocation problem in MEC networks even more complex.

[0007] Although there have been attempts to introduce Lyapunov optimization into MEC systems for handling long-term average constraints, they are generally still based on the Shannon model and ignore the key characteristics of finite code length error codes in industrial and UAV scenarios, cannot characterize frame-level energy consumption restrictions, and do not consider the cumulative demand for reliability of sequentially dependent tasks in dynamic network environments. At the same time, thanks to the research of Polyanskiy et al. on finite code length theorems, the decoding error probability during wireless communication transmission can be better modeled, but Polyanskiy et al. only analyzed the decoding error probability of point-to-point wireless communication transmission, and the study of end-to-end reliability is still lacking. In addition, existing methods cannot handle the coupling between dynamic edge node availability, error introduced by short packet communication, and frame-level energy constraints. Therefore, how to make the task chain continuously executed in a dynamic MEC network have a high enough long-term reliability under the condition of limited code length and energy consumption per frame is still a key problem that existing technologies have not effectively solved. SUMMARY

[0008] The present application aims to propose a resource allocation method that can continuously optimize transmission reliability under online, no prior statistical information conditions, in the face of the ubiquitous finite code length short packet transmission, frame-level energy consumption restrictions, and dynamic MEC environment in industrial Internet of Things and UAV cooperative networks. To solve the core problem that existing technologies cannot simultaneously consider "finite code length error probability, per-frame energy consumption constraints, sequentially dependent task chains, and dynamic MEC node availability", the present application constructs a unified framework based on finite code length communication theory, virtual energy queue management, and online optimization, enabling the system to achieve the lowest cumulative average transmission error rate during long-term operation.

[0009] To solve the above technical problems, the present application provides the following technical solutions: A resource allocation strategy optimization method based on a mobile edge computing scenario, comprising the following steps: Based on the mobile edge computing scenario, using finite code length communication theory, determining the transmission error probability of each frame in the transmission process according to the current channel state information; To address the frame-level energy consumption constraints of terminal devices, a virtual energy queue model is constructed, wherein the queue length is dynamically updated based on the difference between the actual energy consumption of each frame and the preset energy budget, and stability constraints are set for the virtual energy queue. With the goal of minimizing the long-term cumulative average transmission error probability calculated from the transmission error probabilities of each frame, the stability constraints of the virtual energy queue are combined with the goal of minimizing the long-term cumulative average transmission error probability to construct the optimization problem for the current frame. The optimization problem of the current frame is transformed into a convex optimization problem with respect to the transmission code length and transmit power, and then solved. Based on the solution results, the transmission code length and transmit power are allocated to the current frame.

[0010] Furthermore, the transmission error probability of each frame during the transmission process is:

[0011] in, This represents the probability of transmission errors for each frame during transmission. Regarding the available code length in each frame Transmission rate The signal-to-noise ratio of the data packets received by the receiver The function; For Shannon capacity, For code length, For channel dispersion, The signal-to-noise ratio of the data packets received by the receiver. is the transmission rate; Q(·) is the Gaussian function.

[0012] Furthermore, the specific steps for updating the length of the virtual energy queue are as follows: If the first... t The queue length of the frame is The allocated code length is m [ t The transmission power is p [ t ], then the first t +1 frame queue length Updated to:

[0013] in This is the threshold for long-term average energy consumption. For the current frame Energy consumed It is a measure of the degree of energy consumption default, and therefore non-negative.

[0014] Furthermore, the stability constraint of the virtual energy queue is:

[0015] wherein, is the current frame and the threshold difference of the energy consumed and the long-term average energy consumption.

[0016] Further, the step of constructing the current frame optimization problem is achieved by minimizing an upper bound of a Lyapunov drift plus penalty function, wherein the penalty term is the transmission error probability of the current frame, and the drift term is related to the virtual energy queue length and the energy consumption of the current frame.

[0017] Further, the upper bound of the Lyapunov drift plus penalty function has the form:

[0018] wherein, B is a constant, V is a non-negative trade-off parameter, is the current frame and the transmission error probability of the current frame.

[0019] Further, the optimization problem of the current frame is:

[0020] wherein, is the transmission error probability of the current frame, and is the minimum available code length and the maximum available code length of each frame, and is the minimum available power and the maximum available power of each frame.

[0021] Further, the step of converting the optimization problem of the current frame into a convex optimization problem about the transmission code length and the transmission power includes: the product term of the code length and the transmission power in the optimization problem of the current frame is replaced by a convex function .

[0022] Further, the convex optimization problem is:

[0023] wherein, .

[0024] In another aspect, the present application provides a resource allocation strategy optimization system based on a mobile edge computing scenario, comprising: ​A transmission error probability calculation module is configured to determine transmission error probabilities of each frame in a transmission process based on a mobile edge computing scenario, using a limited code length communication theory, and according to current channel state information; A queue management module is configured to construct a virtual energy queue model for a frame-level energy consumption constraint of a terminal device, wherein a queue length is dynamically updated according to a difference between actual energy consumption of each frame and a preset energy budget, and a stability constraint of the virtual energy queue is set; An optimization problem construction module is configured to combine the stability constraint of the virtual energy queue with a long-term cumulative average transmission error probability minimization target, and construct an optimization problem of a current frame, with a target of minimizing the long-term cumulative average transmission error probability calculated by transmission error probabilities of each frame; A problem solving module is configured to convert the optimization problem of the current frame into a convex optimization problem about a transmission code length and a transmission power, and solve the problem; A resource allocation module is configured to allocate the transmission code length and the transmission power for the current frame according to a solving result.

[0025] Compared with the prior art, the present application has the following beneficial effects: 1. The present application explicitly introduces an error code probability model under limited code length communication in the resource scheduling process, avoids the reliability distortion problem caused by the traditional approximation based on Shannon capacity, can more accurately depict the transmission reliability in the short packet communication scenario, and thus effectively reduces the long-term average error code probability of the system, and meets the high reliability and low latency application requirements.

[0026] 2. The present application uses the joint convexity characteristics of the limited code length error code probability about the code length and the transmission power, converts the originally non-convex reliability optimization problem into an intra-frame convex optimization problem, makes the resource allocation be efficiently solved by a standard convex optimization method, significantly reduces the calculation complexity, and is suitable for online real-time scheduling in a dynamic MEC environment.

[0027] 3. The present application uses a Lyapunov drift-penalty method to convert the long-term average reliability and the energy consumption constraint into a series of intra-frame deterministic optimization problems, realizes adaptive resource management of a dynamic system without needing to know channel state, task arrival or server availability statistical characteristics in advance, and enhances the robustness and applicability of the method. BRIEF DESCRIPTION OF DRAWINGS

[0028] In order to more clearly illustrate the technical solutions in the present application or the prior art, the following will briefly introduce the drawings needed in the embodiments or prior art description. Obviously, the drawings in the following description are some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor.

[0029] Figure 1 A transmission network model with multiple edge nodes between a source and a destination is provided for the present application; Figure 2 A schematic diagram showing the convergence effect of the optimization method of the present application is provided; Figure 3 A schematic diagram showing the energy consumption of the optimization method of the present application is provided; Figure 4 A schematic diagram showing the sensitivity of the optimization method of the present application to trade-off parameters is provided. DETAILED DESCRIPTION

[0030] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work shall fall within the scope of protection of the present application.

[0031] In specific implementation, the method provided by the technical solutions of the present application can be automatically run by computer software technology, and the system device of the method, such as a computer readable storage medium storing the corresponding computer program of the technical solutions of the present application and a computer device including the running of the corresponding computer program, shall also be within the protection scope of the present application.

[0032] Embodiment 1 The present embodiment provides a resource allocation strategy optimization method based on a mobile edge computing scenario, including the following steps: Step 1. Based on the mobile edge computing scenario, using the limited code length communication theory, the transmission error probability of each frame in the transmission process is determined according to the current channel state information; Step 2. A virtual energy queue model is constructed for the frame-level energy consumption constraint of the terminal device, wherein the queue length is dynamically updated according to the difference between the actual energy consumption of each frame and the preset energy budget, and a stability constraint of the virtual energy queue is set; Step 3. Taking the minimization of the long-term cumulative average transmission error probability calculated from the transmission error probability of each frame as the target, the stability constraint of the virtual energy queue is combined with the long-term cumulative average transmission error probability minimization target to construct an optimization problem of the current frame; Step 4. The optimization problem of the current frame is converted into a convex optimization problem about the transmission code length and the transmission power and is solved; Step 5. According to the solving result, the transmission code length and the transmission power are allocated for the current frame.

[0033] In step 1 of the embodiment, a mobile edge computing scenario is considered, there is a task package to be transmitted, there are N edge nodes, the task is transmitted through the edge nodes, the network knows the transmission link direction, and the current channel state information (CSI) can be obtained, and the task package is transmitted to the terminal through the edge nodes. Wherein the transmission error probability of each frame in the transmission process is characterized by using the finite code length theorem, and the cumulative average error probability is used as an index for measuring performance.

[0034] Figure 1 For the transmission network model between the source and the destination proposed in step 1 of the application, the maximum code length that can be used for each frame is , the maximum transmission power that can be used is , the energy budget is , the noise power is , the transmission error probability threshold is , and the signal-to-noise ratio threshold is . In step 1, the source transmits a data package containing bits to the receiver through a wireless link using a code word with a block length of m. It is assumed that there is current channel state information, and the channel gain (including path loss) is constant over the entire block length . Then the signal-to-noise ratio (SNR) of the data package received by the receiver is , where is the transmit power, denotes the noise spectral density.

[0035] Since the transmission is carried out under the FBL system, i.e. the block length m is no longer large enough to be considered as infinite, the assumption of arbitrary reliable communication under the Shannon limit is no longer accurate. Even if the coding rate is lower than the Shannon capacity, the transmission can still be erroneous. In particular, the maximum achievable transmission rate with a target error probability of is approximately:

[0036] where is the Shannon capacity, is the code length, is the channel dispersion. In particular, in a fixed gain channel there is , and in addition, is the inverse function of , according to the research of Polyanskiy et al. on the finite code length theorem, the error probability under the transmission rate (where is the task package size) can be approximated as follows:

[0037] In step 2 of this embodiment, a virtual queue is introduced, the length of which is... Indicates to frame End the accumulation of energy consumption, that is ,in ,make Queue backlog can be viewed as a measure of how well constraints in previous service frames have been satisfied. If a constraint is temporarily violated, i.e., energy consumption exceeds the energy tolerance of the service frame, the constraint will increase. When stable, the energy constraint will be satisfied.

[0038] The goal of step 3 is to design an optimal resource allocation strategy that minimizes the long-term average error probability under limited resources. Specifically, the error probability for each frame is... ,in For the probability of transmission error, The code length allocated to the transmission task packet is used to calculate the error probability. and transmission power Limited, and the product of code length and transmit power is energy consumption, expressed here as... The long-term average is less than the maximum available energy consumption. Therefore, the optimization problem for the current frame is formulated as follows:

[0039] in, It is the probability of transmission errors. The expectation, m[t] is the current frame The allocated code length, and These are the minimum and maximum available code lengths for each frame, where p[t] is the current frame. Allocated transmission power, and It represents the minimum and maximum available power for each frame.

[0040] This problem is a mixed-integer stochastic optimization problem that requires continuous decision-making to achieve a long-term optimal result. Therefore, this embodiment proposes an online algorithm based on the Lyapunov optimization framework, which decouples the long-term optimization problem into a series of deterministic problems.

[0041] In step 4 of this embodiment, the properties of the virtual energy queue are derived as follows. Based on the queue dynamics, we can obtain:

[0042] in, It is every frame A queue for measuring the degree of energy consumption default. is the energy consumed in the current frame , is a threshold of long-term average energy consumption, the above inequality is summed up to

[0043] Taking expectation and dividing by T, we have

[0044] When the virtual queue mean is stable, i.e., satisfies

[0045] It can be concluded that the system satisfies the long-term average energy constraint:

[0046] where is the expectation of energy consumed in the current frame . Therefore, as long as the virtual queue is stable, the resource allocation strategy of the present application must satisfy the long-term energy consumption limit.

[0047] Let the Lyapunov function and Lyapunov drift be defined as , .

[0048] For any feasible solution of the problem in any service frame, [θ[t]] has an upper bound, i.e. where is a constant, . The proof is as follows: Substitute the Lyapunov function into the queue update formula to obtain:

[0049] Expand the above formula and subtract , and then divide by 2 to obtain the drift upper bound: . Since the energy consumption and the budget both have finite upper bounds, there exists a constant such that

[0050] In order to optimize the error probability under the premise of satisfying the queue stability, we introduce the Lyapunov drift and the error probability ​​​a weighted combination of the Lyapunov drift and the penalty function:

[0051] wherein is a non-negative constant, as a trade-off parameter, adjusting the weighted values of the energy constraint and the error probability, is the transmission error probability. It is obvious that if the above weighted values are directly used for optimization, the information in the next service frame is needed, i.e. , thus leading to the problem that cannot be solved in time. Therefore, the upper bound of the above equation is derived by the present application to solve this problem, i.e.:

[0052] Therefore, the focus of the problem is now to solve the minimum value of the upper bound, which can be convex approximated by , and the problem is converted into the following form:

[0053] wherein , and are the minimum available code length and the maximum available code length of each frame, and are the minimum available power and the maximum available power of each frame.

[0054] The problem in step 4 is jointly convex solved, and the solution of the problem is obtained.

[0055] Figure 2 The convergence effect of the optimization method of the resource allocation strategy based on the mobile edge computing scenario is shown. When T increases, the average error probability gradually decreases, fluctuates, and then tends to be stable. This fluctuation is caused by the Lyapunov queue. In order to maintain a balance between the transmission energy consumption and the error probability of each frame, the energy consumption of this frame is sacrificed to make the error rate as small as possible, which will lead to a larger queue value of this frame. Due to the influence of the queue value, the optimization of the next frame may be biased to ensure that the energy consumption is not too high, so the error rate of the next frame may be larger. At the same time, when the task data increases to a certain extent, the curve does not fluctuate, because when the task data becomes large, the system will focus on optimizing the error rate, resulting in a high error rate, which causes the energy consumption to overflow, so that the queue value is always increasing; Figure 3For the energy consumption of the proposed optimization method of resource allocation strategy in the mobile edge computing scenario, when the data packet size is very small, the demand for long-term cumulative average error rate is not so urgent, the system will try to keep the energy consumption around the threshold, which is why it fluctuates, one frame consumes more energy, in order to maintain the stability of the system, then the energy consumption of the next frame will be reduced. However, when the data volume increases, in order to guarantee the error rate as much as possible, the transmission power and block length of the system remain basically unchanged, and the energy consumption limit is sacrificed, which is the reason why the energy consumption no longer fluctuates.

[0056] Figure 4 For the sensitivity of the proposed optimization method of resource allocation strategy in the mobile edge computing scenario to the trade-off parameter, when the reliability requirement becomes high, the system sacrifices the energy limit to meet the reliability requirement. When V reaches the order of magnitude of , the error rate will change by the order of magnitude of .

[0057] Embodiment 2 The embodiment provides a resource allocation strategy optimization system based on a mobile edge computing scenario, comprising: a transmission error probability calculation module: configured to determine the transmission error probability of each frame in the transmission process based on the mobile edge computing scenario and using the finite code length communication theory according to the current channel state information; a queue management module: configured to construct a virtual energy queue model for the frame-level energy consumption constraint of the terminal device, wherein the queue length is dynamically updated according to the difference between the actual energy consumption of each frame and the preset energy budget, and the stability constraint of the virtual energy queue is set; an optimization problem construction module: configured to combine the stability constraint of the virtual energy queue with the minimum long-term cumulative average transmission error probability obtained by the transmission error probability calculation of each frame, and construct an optimization problem of the current frame with the objective of minimizing the long-term cumulative average transmission error probability; a problem solving module: configured to convert the optimization problem of the current frame into a convex optimization problem about the transmission code length and the transmission power and solve it; a resource allocation module: configured to allocate the transmission code length and the transmission power for the current frame according to the solving result.

[0058] It should be understood that parts not elaborated in the specification are all prior art.

[0059] It should be understood that the above description is merely a detailed example of the preferred embodiment and is not to be taken in a limiting sense. There can be many variations to the embodiments described herein without departing from the spirit of the application. The scope of the application should be determined by a fair reading of the appended claims, along with the full text of the specification.

Claims

1. A method for optimizing resource allocation strategies in mobile edge computing scenarios, characterized in that, Includes the following steps: Based on the mobile edge computing scenario, the transmission error probability of each frame during transmission is determined by using the finite code length communication theory and the current channel state information. To address the frame-level energy consumption constraints of terminal devices, a virtual energy queue model is constructed, wherein the queue length is dynamically updated based on the difference between the actual energy consumption of each frame and the preset energy budget, and stability constraints are set for the virtual energy queue. With the goal of minimizing the long-term cumulative average transmission error probability calculated from the transmission error probabilities of each frame, the stability constraints of the virtual energy queue are combined with the goal of minimizing the long-term cumulative average transmission error probability to construct the optimization problem for the current frame. The optimization problem of the current frame is transformed into a convex optimization problem with respect to the transmission code length and transmit power, and then solved. Based on the solution results, the transmission code length and transmit power are allocated to the current frame.

2. The resource allocation strategy optimization method based on mobile edge computing scenarios according to claim 1, characterized in that, The transmission error probability of each frame during the transmission process is: in, This represents the probability of transmission errors for each frame during transmission. Regarding the available code length in each frame Transmission rate The signal-to-noise ratio of the data packets received by the receiver The function; For Shannon capacity, For code length, For channel dispersion, The signal-to-noise ratio of the data packets received by the receiver. is the transmission rate; Q(·) is the Gaussian function.

3. The resource allocation strategy optimization method based on mobile edge computing scenarios according to claim 1, characterized in that, The specific steps for updating the length of the virtual energy queue are as follows: If the length of the first virtual energy queue is... t The queue length of the frame is The allocated code length is m [ t The transmission power is p [ t ], then the first t +1 frame queue length Updated to: in This is the threshold for long-term average energy consumption. For the current frame Energy consumed It is a measure of the degree of energy consumption default, and therefore non-negative.

4. The resource allocation strategy optimization method based on mobile edge computing scenarios according to claim 3, characterized in that, The stability constraint of the virtual energy queue is: in, For the current frame The expected value of the difference between the energy consumed and the threshold of the long-term average energy consumption.

5. The resource allocation strategy optimization method based on mobile edge computing scenarios according to claim 4, characterized in that, The steps for constructing the current frame optimization problem are: to minimize the upper bound of the Lyapunov drift plus penalty function, where the penalty term is the transmission error probability of the current frame, and the drift term is related to the virtual energy queue length and the energy consumption of the current frame.

6. The resource allocation strategy optimization method based on mobile edge computing scenarios according to claim 5, characterized in that, The upper bound of the Lyapunov drift plus penalty function is as follows: Where B is a constant and V is a nonnegative tradeoff parameter. For the current frame The probability of transmission errors.

7. The resource allocation strategy optimization method based on mobile edge computing scenarios according to claim 6, characterized in that, The optimization problem for the current frame is as follows: in, It is the probability of transmission errors. Expectations and These are the minimum and maximum available code lengths for each frame. and It represents the minimum and maximum available power for each frame.

8. The resource allocation strategy optimization method based on mobile edge computing scenarios according to claim 7, characterized in that, The steps to transform the optimization problem of the current frame into a convex optimization problem with respect to the transmission code length and transmit power include: In the optimization problem of the current frame, the code length With transmission power product term Using convex functions To make a substitution.

9. The resource allocation strategy optimization method based on mobile edge computing scenarios according to claim 8, characterized in that, The convex optimization problem is as follows: in, .

10. A resource allocation strategy optimization system based on mobile edge computing scenarios, characterized in that, include: Transmission error probability calculation module: It is used in mobile edge computing scenarios to determine the transmission error probability of each frame during transmission based on the current channel state information using the finite code length communication theory. Queue Management Module: It is used to construct a virtual energy queue model for frame-level energy consumption constraints of terminal devices. The queue length is dynamically updated according to the difference between the actual energy consumption of each frame and the preset energy budget, and the stability constraints of the virtual energy queue are set. The optimization problem construction module is used to construct the optimization problem of the current frame by combining the stability constraints of the virtual energy queue with the goal of minimizing the long-term cumulative average transmission error probability calculated from the transmission error probabilities of each frame. Problem-solving module: It is used to transform the optimization problem of the current frame into a convex optimization problem with respect to the transmission code length and the transmit power, and then solve it. Resource allocation module: It is used to allocate transmission code length and transmit power to the current frame based on the solution results; The resource allocation strategy optimization system based on mobile edge computing scenarios is used to execute the steps in the resource allocation strategy optimization method based on mobile edge computing scenarios according to any one of claims 1-9.

Images