Optimal stopping task unloading optimization system and method in energy limited environment

Abstract

The invention discloses an optimal stopping task unloading optimization system and method in an energy limited environment, the maximum allowable observation frequency is calculated through energy consumption constraint, efficient task unloading is realized based on an optimal stopping theory, and the optimal stopping task unloading optimization system and method are suitable for an energy limited car networking scene. The method comprises the following steps: constructing an Internet of Vehicles task unloading model based on an optimal stopping theory, and establishing an objective function for minimizing total unloading cost expectation according to RSU load and observation cost; introducing an energy consumption constraint by establishing an energy consumption inequality; calculating maximum allowable observation times through energy consumption constraint, solving based on an optimal stop theory, and calculating to obtain a series of comparison thresholds; according to the observed current RSU load, the current load is compared with the comparison threshold value obtained through calculation, and the local optimal RSU meeting the condition is found out and unloaded. According to the method, the calculation tasks are optimized and unloaded, so that load balancing of task calculation in the Internet of Vehicles is realized, and the utilization rate of calculation resources is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of vehicle networking, and specifically relates to an optimal stopping task offloading optimization system and method in an energy-constrained environment, which is used to optimize the task offloading decision of vehicles under energy consumption constraints. Background Art

[0002] The computing task offloading strategy in vehicle networking is a popular research direction. Due to limited computing power, vehicles often need to offload computationally intensive tasks to roadside units (RSUs). There is a load value for the RSU server. If the load of the RSU is high, it will affect the task computing time. Therefore, vehicles want to find an RSU with a lower load, so they need to observe the load of the RSU before offloading. In the actual scenario, the energy of vehicles is limited, and frequent observation of RSUs may lead to energy exhaustion and affect the sustainability of the system. In order to help vehicles select an RSU with the smallest possible load for task offloading while meeting the energy consumption constraint, it is necessary to propose a task offloading method based on the optimal stopping theory under the energy consumption constraint to balance load balancing and energy consumption efficiency.

[0003] The following are the differences between this application and the above-mentioned comparative document:

[0004] Technical Comparison with Patent CN119356754A "A Vehicle Networking Task Offloading Method Based on the Optimal Stopping Theory" I. Model Elements and Constraint Conditions

[0005] In patent CN119356754A, only the vehicle speed and the RSU load are used as cost data to establish an objective function for minimizing the expected total offloading cost, and a single reference value is obtained through solving by the optimal stopping theory to determine the offloading moment.

[0006] Based on the above cost model, the present invention additionally introduces an energy consumption constraint, calculates the maximum allowable number of observations in an energy-constrained environment by constructing an energy consumption inequality, and ensures that the offloading decision not only conforms to the cost optimum but also meets the energy consumption upper limit.

[0007] II. Threshold Calculation and Decision-making Mechanism

[0008] Patent CN119356754A only calculates a reference value, observes the size relationship between the current RSU load and this reference value, and judges the optimal stopping moment and performs offloading based on this.

[0009] Based on the optimal stopping theory, the present invention solves a series of increasing comparison thresholds for the maximum allowable number of observations; after observing the current RSU load each time, compares it with the corresponding threshold, and dynamically selects the first locally optimal RSU that meets the conditions for offloading, so as to achieve a finer-grained optimization in terms of energy consumption and decision accuracy. Summary of the Invention

[0010] The main object of the present invention is to overcome the drawbacks and deficiencies of the prior art. The present invention constructs a two-stage detection model, calculates the maximum number of observations by introducing energy consumption constraints, combines the optimal stopping theory to dynamically decide the offloading time, and realizes the load balancing of task calculation in the vehicle network by optimizing the offloading of computing tasks, thereby improving the utilization rate of computing resources.

[0011] To achieve the above object of the invention, the present invention provides an optimal stopping task offloading optimization method in an energy-constrained environment, which is characterized by comprising the following steps:

[0012] S1: Construct a vehicle network task offloading model based on the optimal stopping theory, and establish an objective function for minimizing the expected total offloading cost according to the RSU load and the observation cost; introduce energy consumption constraints by establishing an energy consumption inequality;

[0013] S2: Calculate the maximum allowable number of observations through the energy consumption constraints, and solve based on the optimal stopping theory to calculate a series of comparison thresholds;

[0014] S3: Compare the observed current RSU load with the calculated comparison thresholds, and find a locally optimal RSU that meets the conditions for offloading.

[0015] As a further improvement of the present invention, in S1: construct a vehicle network task offloading model based on the optimal stopping theory, establish an objective function for minimizing the expected total offloading cost according to the RSU load and the observation cost, introduce energy consumption constraints by establishing an energy consumption inequality, and the calculation formula for the energy consumption that vehicle i needs to spend in the whole process is:

[0016]

[0017] Wherein, μ represents an energy consumption coefficient proportional to the square of the moving speed, such as the air resistance coefficient; represents a fixed energy consumption coefficient independent of the vehicle driving speed, the rolling resistance coefficient; v i is the driving speed of vehicle i; ΔL represents the distance between adjacent RSUs; n is the number of the selected RSU; represents the energy consumption required for vehicle i to observe each RSU, is the energy consumption required for vehicle i to offload tasks, is the available energy budget of vehicle i.

[0018] As a further improvement of the present invention, the establishment of the objective function for minimizing the expected total offloading cost in step S1 includes:

[0019] Establish the objective function of vehicle offloading, and the optimization problem is as follows:

[0020]

[0021] Among them, n is the number of the selected RSU, and N is the total number of known RSUs. It represents the cost obtained when vehicle i decides whether to offload the task to RSU n at the current moment, and takes the smaller value between the current offloading cost of the vehicle and the expected offloading cost wherein is the offloading cost generated by stopping the observation and selecting the current RSU at this time, is the expected cost of continuing to observe the subsequent RSUs; It represents the energy consumption required for vehicle i to observe each RSU, is the energy consumption required for vehicle i to offload the task, is the available energy budget of vehicle i.

[0022] As a further improvement of the present invention, the total offloading cost obtained when vehicle i decides to offload at its nth observation moment is:

[0023]

[0024] wherein, x n represents the load of RSU n, that is, the total time delay spent when the task is offloaded to the selected RSU for processing; ΔL represents the distance between adjacent RSUs, n is the number of the current RSU, and v i is the driving speed of vehicle i, and c represents the observation cost required to observe an RSU once.

[0025] As a further improvement of the present invention, S3: According to the observed current RSU load, compare the current load with the calculated comparison threshold, find the locally optimal RSU that meets the conditions for offloading. If vehicle i has not offloaded after observing until the end, it can only choose to offload to the last RSU. The optimal offloading cost of the vehicle's Nth observation is

[0026]

[0027] wherein, is the offloading cost generated by offloading to the last RSU, that is, the Nth RSU.

[0028] The present invention relates to a system using an optimal stopping theory-based task offloading optimization method in an energy-constrained environment. The system includes:

[0029] The model construction module is used to construct a vehicle-to-everything (V2X) task offloading model with energy consumption constraints, establish an objective function for minimizing the expected total offloading cost based on parameters such as vehicle speed, remaining energy, and the coverage range of roadside units (RSUs), and calculate the energy consumption for a single observation. The parameters of the objective function include the load delay of the RSU, the number of observations, and the observation cost.

[0030] The threshold calculation module is used to calculate the maximum allowable number of observations through an energy consumption constraint inequality and solve the comparison thresholds for each observation stage using the backward recursion method based on the optimal stopping theory.

[0031] The offloading execution module is used to obtain the load status of the current RSU and the vehicle energy status in real time, compare the current load with the threshold for the corresponding stage, and perform task offloading when the offloading condition is met; otherwise, continue to observe until forced offloading when reaching the last RSU.

[0032] Beneficial effects:

[0033] By constructing a V2X task offloading model with energy consumption constraints, this invention establishes an objective function for minimizing the expected total offloading cost based on parameters such as vehicle speed, remaining energy, and the coverage range of RSUs. The parameters of the objective function include the load delay of the RSU and the observation cost. By restricting the maximum allowable number of observations, it avoids excessive energy consumption by vehicles during task offloading and saves system energy. The optimal stopping theory is used to find a locally optimal RSU for task offloading, and the observed current RSU load is compared with a calculated comparison threshold to select a low-load RSU and optimize resource utilization.

[0034] The features and advantages of the present invention will become clear by referring to the following drawings and the detailed description of the specific embodiments of the present invention. Description of the Drawings

[0035] Figure 1 is a flowchart of the V2X task offloading method based on the optimal stopping theory and energy consumption constraints;

[0036] Figure 2 is a schematic diagram of the V2X task offloading system based on the optimal stopping theory and energy consumption constraints;

[0037] Figure 3 is a cost comparison graph of the proposed algorithm under different energy consumption constraints;

[0038] Figure 4 is a cost comparison graph of the proposed algorithm and other algorithms. Detailed Embodiments

[0039] Combined with the accompanying drawings in the embodiments of the present invention, the technical solutions in the embodiments of the present invention are clearly and completely described. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0040] Embodiment 1

[0041] Figure 1 is a flowchart of an optimization offloading method for computing tasks in a vehicle-to-everything (V2X) network based on the optimal stopping theory. As Figure 1 shown, this embodiment provides an optimal stopping task offloading optimization system and method in an energy-constrained environment. The method includes the following steps:

[0042] To find the optimal roadside unit (RSU) for task offloading, the optimal stopping theory is used to solve this problem. The currently observed RSU load is compared with a value to determine the optimal offloading moment, achieving load balancing of the RSUs in the V2X network and improving the efficiency of task offloading.

[0043] S1: Construct a V2X task offloading model based on the optimal stopping theory. According to the RSU load and observation cost, establish an objective function to minimize the expected total offloading cost; introduce an energy consumption constraint by establishing an energy consumption inequality;

[0044] S2: Calculate the maximum allowable number of observations through the energy consumption constraint, and solve based on the optimal stopping theory to calculate a series of comparison thresholds;

[0045] S3: Observe the RSUs in real time. According to the currently observed RSU load, compare the current load with the calculated comparison thresholds to find the locally optimal RSU that meets the conditions for offloading.

[0046] The method of this embodiment realizes load balancing of task computing in the V2X network by optimizing the offloading of computing tasks, and improves the utilization rate of computing resources.

[0047] In step S1, the establishment of the objective function to minimize the expected offloading cost includes:

[0048] Establish the objective function for vehicle offloading, and the optimization problem is as follows:

[0049]

[0050] Among them, n is the number of the selected RSU, and N is the total number of known RSUs. represents the cost obtained when vehicle i decides whether to offload the task to RSU n at the current moment, and takes the current offloading cost of the vehicle And the expected offloading cost The smaller value of the two, where Is the offloading cost generated by stopping the observation at this time and selecting the current RSU, Is the expected cost of continuing to observe subsequent RSUs; Represents the energy consumption required for vehicle i to observe each RSU, Is the energy consumption required for vehicle i to offload tasks, Is the available energy budget for vehicle i.

[0051] Preferably, the total offloading cost obtained by vehicle i deciding to offload at its nth observation moment is:

[0052]

[0053] Among them, x n Represents the load of RSUn, that is, the total delay spent when the task is offloaded to the selected RSU for processing; ΔL represents the distance between adjacent RSUs, n is the number of the current RSU, v i Is the driving speed of vehicle i, and c represents the observation cost required to observe an RSU once.

[0054] Preferably, if vehicle i still has not offloaded after observing until the end, it can only choose to offload to the last RSU, and the optimal offloading cost of the vehicle's Nth observation is

[0055]

[0056] Among them, Is the offloading cost generated by offloading to the Nth RSU.

[0057] Preferably, the calculation formula for the energy consumption that vehicle i needs to spend during the whole process is:

[0058]

[0059] Among them, μ represents the energy consumption coefficient proportional to the square of the moving speed, such as the air resistance coefficient; Represents the fixed energy consumption coefficient independent of the vehicle's driving speed, such as the rolling resistance coefficient; v i Is the driving speed of vehicle i; ΔL represents the distance between adjacent RSUs; n is the number of the selected RSU; Represents the energy consumption required for vehicle i to observe each RSU, Is the energy consumption required for vehicle i to offload tasks, Is the available energy budget for vehicle i.

[0060] In step S2, calculate the maximum allowable number of observations through energy consumption constraints And the comparison threshold is solved based on the optimal stopping theory The specific process is as follows:

[0061] In step S2, under the condition of satisfying the energy consumption constraint, the maximum allowable number of observations of vehicle i can be calculated

[0062]

[0063] where represents the energy consumption required for vehicle i to observe each RSU is the energy consumption required for vehicle i to offload tasks[[ID=1^6]] is the available energy budget of vehicle i

[0064] Since there are a total of N RSUs on the roadside during the observation process, the actual maximum allowable number of RSUs that vehicle i can observe is the smaller value between the known total number of RSUs N and the maximum allowable number of observations satisfying the energy consumption constraint, that is;

[0065]

[0066] In step S2, if the vehicle wants to select the optimal value as much as possible, it needs to offload when the current offloading cost is lower than the expected future offloading cost:

[0067]

[0068] where represents the total offloading cost of vehicle i, and x n represents the load of RSUn, that is, the total time delay spent when the task is offloaded to the selected RSU for processing; ΔL represents the distance between adjacent RSUs, n is the number of the current RSU, and v i is the driving speed of vehicle i, c represents the observation cost required to observe an RSU once, represents the cost obtained when vehicle i decides whether to offload the task to RSUn+1 at the current moment

[0069] After arranging the above formula, it can be seen that x n needs to satisfy the following conditions:

[0070]

[0071] For simplicity and convenience in the follow-up, an auxiliary variable [[ID=5^6]]

[0072]

[0073] At this time, only x n needs to be compared with Just compare Substitute for calculation

[0074]

[0075] wherein is the offloading cost generated by stopping the observation at this time and selecting the current RSU is the expected cost of continuing to observe the costs generated by subsequent RSUs

[0076] Here, the random sequence x n is converted into a continuous variable x:

[0077]

[0078] where f(x) represents the probability density function of the variable x, and F(x) represents the cumulative distribution function of the variable x

[0079] In step S3, the reference threshold is calculated through the above process and compared with the RSU load currently observed by the vehicle. If the optimal strategy at this time is to stop the observation and offload the task to the current RSU; if continuing to observe subsequent RSUs is the best practice at this time

[0080] Figure 3 is the situation where the cost changes with the moving speed of the proposed algorithm under different energy constraints. It can be seen from the figure that when the moving speed increases, the cost curve first decreases and then increases. This indicates that there is an optimal moving speed that can minimize the optimal cost. In addition, it is worth noting that the cost curve under more stringent energy constraints is always higher than the cost curve under loose energy constraints or without constraints. This indicates that implementing more stringent energy constraints will result in higher costs

[0081] Figure 4 Compares the costs between the proposed offloading method, random offloading method, best offloading method, and methods 1 and 2 of observing first and then offloading. Methods 1 and 2 maximize the probability of selecting the optimal solution, optimal solution, or sub-optimal solution by optimizing the number of observations. It can be seen that as the driving speed increases, the cost of each method decreases. Compared with other offloading methods, the method proposed in this embodiment can obtain a lower cost. The algorithm proposed in this paper can achieve a cost close to that of the optimal algorithm

[0082] In this embodiment, a vehicle networking task offloading model based on the optimal stopping theory is constructed. According to the RSU load and the observation cost, an objective function for minimizing the expected total offloading cost is established. An energy consumption constraint is introduced by establishing an energy consumption inequality. The maximum allowable number of observations is calculated through the energy consumption constraint, and the solution is carried out based on the optimal stopping theory. A series of comparison thresholds are calculated. During the observation process, according to the observed current RSU load, the current load is compared with the calculated comparison thresholds, and the local optimal RSU that meets the conditions is found for offloading, realizing the load balancing of the RSU in the vehicle networking and improving the utilization rate of resources.

[0083] Embodiment 2

[0084] Figure 2 It is a schematic diagram of a task offloading optimization system based on the optimal stopping theory in an energy-constrained environment. As Figure 2 shown, this embodiment provides a task offloading optimization system based on the optimal stopping theory in an energy-constrained environment. The system includes:

[0085] A model construction module 201, configured to construct a vehicle networking task offloading model with an energy consumption constraint, establish an objective function for minimizing the expected total offloading cost according to parameters such as vehicle speed, remaining energy, and RSU coverage range, and calculate the energy consumption per single observation; the parameters of the objective function include the load delay of the RSU, the number of observations, and the observation cost;

[0086] A threshold calculation module 202, configured to calculate the maximum allowable number of observations through the energy consumption constraint inequality, and solve the comparison thresholds of each observation stage by using the backward recursion method based on the optimal stopping theory;

[0087] An offloading execution module 203, configured to obtain the load status of the current RSU and the vehicle energy status in real time, compare the current load with the threshold of the corresponding stage, and perform task offloading when the offloading condition is met, otherwise continue to observe until forced offloading when reaching the last RSU.

[0088] Preferably, the objective function established by the construction module for minimizing the expected total offloading cost includes:

[0089] Establish the objective function for vehicle offloading, and the optimization problem is as follows:

[0090]

[0091]

[0092] where n is the number of the selected RSU, and N is the total number of known RSUs. represents the cost obtained by the vehicle i's decision at the current moment, taking the current offloading cost of the vehicle and the expected offloading cost The smaller value of the two, where is to stop the observation at this time and select the offloading cost generated by the current RSU, is the expected cost of continuing to observe the costs generated by subsequent RSUs; represents the energy consumption required for vehicle i to observe each RSU, is the energy consumption required for vehicle i to offload the task, is the available energy budget of vehicle i.

[0093] Preferably, the total offloading cost obtained by vehicle i deciding to offload at its nth observation moment is:

[0094]

[0095] where x n represents the load of server n, that is, the total delay spent when the task is offloaded to the selected RSU for processing; ΔL represents the distance between adjacent RSUs, n is the number of the current RSU, v i is the driving speed of vehicle i, and c represents the observation cost required to observe an RSU once.

[0096] Preferably, if vehicle i still has not offloaded after observing until the end, it can only choose to offload to the last RSU, and the optimal offloading cost of the vehicle's Nth observation is

[0097]

[0098] where is the offloading cost generated by offloading to the Nth RSU.

[0099] Preferably, the calculation formula for the energy consumption required for vehicle i throughout the process is:

[0100]

[0101] where μ represents the energy consumption coefficient proportional to the square of the moving speed, such as the air resistance coefficient; represents the fixed energy consumption coefficient independent of the vehicle driving speed, such as the rolling resistance coefficient; v i is the driving speed of vehicle i; ΔL represents the distance between adjacent RSUs; n is the number of the selected RSU; represents the energy consumption required for vehicle i to observe each RSU, is the energy consumption required for vehicle i to offload the task, is the available energy budget of vehicle i.

[0102] The specific implementation process of the functions implemented by each module in this Embodiment 2 is the same as that in Embodiment 1, and will not be elaborated here.

Claims

1. Optimal stopping task offloading optimization method in an energy-constrained environment, characterized in that: It includes the following steps: S1: Construct a vehicle networking task offloading model based on the optimal stopping theory. According to the RSU load and observation cost, establish an objective function to minimize the expected total offloading cost; introduce the energy consumption constraint by establishing an energy consumption inequality; S2: Calculate the maximum allowable number of observations through the energy consumption constraint, and solve it based on the optimal stopping theory to calculate a series of comparison thresholds; S3: According to the observed current RSU load, compare the current load with the calculated comparison thresholds to find the locally optimal RSU that meets the conditions for offloading.

2. The optimal stopping task offloading optimization method in an energy-constrained environment according to claim 1, characterized in that: S1: Construct a vehicle networking task offloading model based on the optimal stopping theory. According to the RSU load and observation cost, establish an objective function to minimize the expected total offloading cost; introduce the energy consumption constraint by establishing an energy consumption inequality. The calculation formula for the energy consumption that vehicle i needs to spend throughout the process is: Among them, μ represents the energy consumption coefficient proportional to the square of the moving speed, such as the air resistance coefficient; represents the fixed energy consumption coefficient independent of the vehicle driving speed, the rolling resistance coefficient; v i is the driving speed of vehicle i; ΔL represents the distance between adjacent RSUs; n is the number of the selected RSU; represents the energy consumption required for vehicle i to observe each RSU, is the energy consumption required for vehicle i to offload tasks, is the available energy budget of vehicle i.

3. The optimal stopping task offloading optimization method in an energy-constrained environment according to claim 2, wherein: The establishment of the objective function for minimizing the expected total offloading cost in step S1 includes: Establish the objective function for vehicle offloading, and the optimization problem is as follows: s.t.C1: C2: n ∈ [1, N] where n is the number of the selected RSU, and N is the total number of known RSUs. denotes the cost obtained when vehicle i decides whether to offload the task to RSU n at the current moment, taking the current offloading cost of the vehicle and the expected offloading cost as the smaller value of the two, where is the offloading cost generated by stopping the observation and selecting the current RSU at this time, is the cost expectation generated by continuing to observe the subsequent RSUs; represents the energy consumption required for vehicle i to observe each RSU, is the energy consumption required for vehicle i to offload the task, is the available energy budget of vehicle i.

4. The optimal stopping task offloading optimization method in an energy-constrained environment according to claim 2, characterized in that: The total offloading cost obtained when vehicle i decides to offload at its nth observation moment is: Among them, x n represents the load of RSUn, that is, the total time delay spent when the task is offloaded to the selected RSU for processing; ΔL represents the distance between adjacent RSUs, n is the number of the current RSU, v i is the driving speed of vehicle i, and c represents the observation cost required to observe an RSU once.

5. The optimal stopping task offloading optimization method in an energy-constrained environment according to claim 1, wherein: S3: According to the observed current RSU load, compare the current load with the calculated comparison thresholds to find the locally optimal RSU that meets the conditions for offloading. If vehicle i still has not offloaded after observing until the end, it can only choose to offload to the last RSU. The optimal offloading cost for the Nth observation of the vehicle is Among them, The offloading cost incurred for offloading to the last RSU, i.e., the Nth RSU.

6. A system for the task offloading optimization method based on the optimal stopping theory in an energy-constrained environment according to any one of claims 1-5, characterized in that The system includes: A model construction module, which is used to construct a vehicle networking task offloading model with energy consumption constraints, establish an objective function to minimize the expected total offloading cost according to parameters such as vehicle speed, remaining energy, and RSU coverage range, and calculate the energy consumption for a single observation; the parameters of the objective function include the load delay, number of observations, and observation cost of the RSU; A threshold calculation module, which is used to calculate the maximum allowable number of observations through the energy consumption constraint inequality, and solve the comparison thresholds for each observation stage based on the optimal stopping theory using the backward recursion method; An offloading execution module, which is used to obtain the load status of the current RSU and the energy status of the vehicle in real time, compare the current load with the thresholds at the corresponding stages, and execute task offloading when the offloading conditions are met, otherwise continue to observe until forced offloading when reaching the last RSU.

Images