Block chain enabled and MA-UAV assisted Internet of Vehicles multi-target computing unloading method

Abstract

The invention discloses a block chain enabled and mobile antenna (MA)-unmanned aerial vehicle (UAV) assisted Internet of Vehicles multi-target calculation unloading method. Combination of the Internet of Vehicles and a mobile edge computing server (MEC) is the current development trend of the Internet of Vehicles, and the system can greatly improve the processing capacity on the premise of ensuring low time delay by unloading a computing task to an adjacent MEC. However, sensitive data is frequently exchanged between the vehicle and the MEC, and centralized vulnerabilities and data tampering risks are faced. Therefore, the block chain technology is introduced, a non-tampering record is generated for each task unloading process, and the transparency and credibility of the data are guaranteed through a consensus mechanism of the block chain. Meanwhile, in order to prevent eavesdropping, leakage of signals in the direction of an eavesdropper is reduced by adjusting the position of the MA and beamforming. Specifically, a vehicle in the system generates a calculation task, and the calculation task is relayed and forwarded through a UAV equipped with an MA and finally transmitted to a base station. And the base station allocates computing resources according to the cost provided by the vehicle and the task size. According to the scheme, the time delay, the energy consumption and the cost generated in the task unloading and calculation process are balanced with the communication secrecy rate through the MA position and beam forming optimization, and the weighted total utility of the system is maximized.

Description

Technical Field

[0001] This solution designs a blockchain - empowered and mobile antenna (MA) - unmanned aerial vehicle (UAV) - assisted multi - objective computing offloading method for vehicle - to - everything (V2X) networks, belonging to the field of wireless communication. Background Art

[0002] In recent years, with the continuous evolution of vehicle - to - everything (V2X) technology, mobile vehicle users (VUs) undertake increasingly complex sensing, decision - making, and control tasks during driving. These tasks not only require extremely high computing performance but also extremely low processing latency and high reliability. However, limited by the resource capabilities of in - vehicle computing units, local processing by VUs often fails to meet the requirements of high - real - time applications. Therefore, the task offloading mechanism based on vehicle - edge computing has gradually become one of the key technologies. By offloading computing tasks to nearby mobile edge computing servers (MECs), the system can significantly improve processing capabilities while ensuring low latency. However, data transmission and remote computing during the task offloading process also introduce potential security risks, such as eavesdropping on transmission information by eavesdroppers (Eves), which seriously threatens the stability and security of V2X networks. How to improve the security of task offloading while ensuring computing performance has become an important challenge that urgently needs to be solved in the V2X field.

[0003] To address the above challenges, this solution uses physical - layer security (PLS) technology. PLS realizes the security protection of data by exploiting the characteristics of the wireless channel itself. In particular, multi - antenna transmission technology is considered one of the key means to enhance physical - layer security, which can effectively improve the quality of legitimate communication links while weakening the channel performance of potential eavesdroppers.

[0004] Due to excessive distance or terrain blockage, a direct communication link may not be established between the VU and the base station (BS). At this time, the UAV can act as a wireless relay to effectively improve system connectivity. With its high - altitude advantage, the UAV can establish good line - of - sight (LoS) communication with ground devices, reducing signal blockage and shadow effects, thereby expanding the network coverage and enhancing communication quality. Therefore, in the communication process of task offloading for VUs in this solution, UAV - assisted communication equipped with MAs is adopted. During the task offloading process, sensitive data needs to be frequently exchanged between nodes, facing risks of centralization vulnerabilities and data tampering. To address this challenge, this solution uses blockchain technology to generate an immutable record for each task offloading process and ensures the transparency and credibility of data through the consensus mechanism of the blockchain. These records can be used as evidence for task execution, providing a complete audit trail for the system and preventing malicious actors from tampering with task data or forging offloading records. Summary of the Invention

[0005] Technical problem to be solved: In view of the deficiencies of the existing solutions, the present invention provides a blockchain-enabled MA-UAV-assisted multi-objective computing offloading solution, which makes full use of the spatial degrees of freedom of the MA to improve the secrecy rate in the communication process. At the same time, the energy consumption, delay, task cost, and fairness of computing resource allocation are comprehensively considered to maximize the total system utility.

[0006] Technical solution:

[0007] The present invention introduces a blockchain-enabled MA-UAV-assisted multi-objective computing offloading solution in an Internet of Vehicles (IOV). The task offloading solution is characterized by the following steps:

[0008] Step 1, establish a scenario for MA-UAV-assisted VU task offloading in the blockchain-oriented IOV. The ground facilities include K I single-antenna VUs, K M multi-access edge computing (MEC) servers, a hovering UAV equipped with a multiple-antenna (MA), where the UAV receives and forwards the VU transmission signals. The MA is divided into a receiving MA with a total of N r antennas and a transmitting MA with a total of N t antennas. There is also a multi-antenna base station with K B antennas. At the same time, there are multiple single-antenna eavesdroppers (Eves);

[0009] Step 2, establish communication models, energy consumption models, and cost models for the MA position and beamforming respectively, and construct an optimization model based on these three models;

[0010] Step 3, according to the current initial values, use an improved particle swarm optimization algorithm to solve for the position of the MA;

[0011] Step 4, according to the current optimal MA position, use successive convex approximation (SCA) to solve for beamforming;

[0012] Step 5, use an alternating optimization algorithm, and loop through Step 4 and Step 5 until the system utility value of the latter iteration is less than or equal to that of the previous iteration between two adjacent iterations, or when the maximum preset number of iterations is reached, the iteration ends;

[0013] Step 6, based on the current state of the entire IOV, optimize the MA position and beamforming, and allocate computing resources, and perform task offloading and computing for the VUs in the IOV with the aim of maximizing the system weighted total utility value. Description of the drawings

[0014] Figure 1 is the scenario diagram of this solution.

[0015] Figure 2is the MA position solving algorithm proposed in this solution.

[0016] Figure 3 is the beamforming solving algorithm proposed in this solution.

[0017] Figure 4 is the algorithm for alternating optimization of MA position and beamforming.

[0018] Figure 5 is the comparison experimental graph between the standard particle swarm optimization (PSO) and the improved PSO in this solution.

[0019] Figure 6 is the comparison graph between this solution and a blockchain-enabled and IRS-UAV-assisted multi-objective computing offloading method for vehicle-to-everything networks [1] of. Specific implementation manners

[0020] The following further describes this solution in conjunction with the accompanying drawings and specific implementation manners.

[0021] As Figure 1 shown, a MA-UAV-assisted multi-objective computing offloading solution for a blockchain-enabled vehicle-to-everything network, the task offloading solution includes the following steps:

[0022] Step 1, establish a blockchain-enabled secure full-duplex (FD) multi-user scenario. The blockchain is used to record transaction information to achieve the non-tamperability and data integrity of transactions. The VUs in the vehicle-to-everything network need to offload computing tasks to the MEC. Due to the existence of environmental obstacles, they cannot be directly offloaded to the MEC. Therefore, a UAV carrying a MA is used for relaying. There are multiple single-antenna Eves in the environment. To improve the communication secrecy rate during the offloading process, the position of the MA and beamforming are designed to ensure secure communication. For the MA, there are N t receiving MAs and N r transmitting MAs. For the n t -th transmitting MA and the n r -th receiving MA, their positions are represented in the Cartesian coordinate system, that is, C t and C r are the specified transmission area and reception area respectively.

[0023] The sets of transmitting and receiving MAs are respectively defined as: and The self-interference (SI) channel of the UAV, the channel from the VU to the UAV, the channel from the UAV to the antenna of the BS, the channel from the UAV to The channels are respectively represented as: k I The vehicle arrives at k E Eve's channel is

[0024] When the UAV sends data to the BS, beamforming is added to prevent Eve from eavesdropping. It represents the transmit beamforming of all users. Therefore, the data information transmitted by the UAV can be expressed as:

[0025]

[0026] Then the data information of all VUs can be expressed as:

[0027]

[0028] The received signals at the UAV, BS base station, and Eve are respectively:

[0029]

[0030]

[0031] Among them, v U represents the artificial noise (AN), where V ≥ 0. Note that tr(v U ) will reflect the total transmit power invested in the AN. In the following, the optimization process involves the participation of the covariance of v U , that is, V. respectively represent the additive white Gaussian noise (AWGN) at the UAV, k I VU, and Eve k E . This scheme establishes a channel model based on the field response channel model, considers the quasi-static block fading channel, and focuses on a specific fading block, where the multipath channel components are fixed at any position in the given area. The channel will be introduced by node below.

[0032] (1) First is the UAV node SI channel: Suppose there are transmission paths and receiving paths. The position of the j-th transmission path MA and the signal propagation distance difference between the origin of the transmission area (O t ) is (related to AOA, AOD), where, θ t , and θ r , The AoDs and AoAs of the transmitting and receiving MAs, respectively.

[0033] The transmitting field response vector (FRV) is:

[0034]

[0035] where λ is the carrier wavelength, is the phase difference.

[0036] The receiving field response vector is:

[0037]

[0038] is the position of the MA on the i-th receiving path The difference in the signal propagation distance between the origin of the receiving area.

[0039] The SI channel matrix is expressed as:

[0040]

[0041] where, is the field response matrix (FRM) of the N r receiving MAs, is the path response matrix, and the i-th row and j-th column represent the channel response between the i-th receiving path from O t to O r and the j-th transmitting path. is the field response matrix of the N t transmitting MAs.

[0042] (2) Next is the channel between the VU node and the BS node: Since the VU, BS, and eavesdropper are all single-antenna, for the channel from the VU to the UAV, the field response matrix (FRM) only exists at the receiving end; for the channel from the UAV to the BS, the field response matrix (FRM) only exists at the transmitting end. Define as the number of receiving paths from the k I VU to the UAV, the number of transmitting paths from the UAV to the base station, and the number of eavesdropping paths from the UAV to Eve, respectively.

[0043]

[0044] where, is the receiving field response matrix (FRM) of the k I VU channel. and For BS k respectively B and Eve k E The corresponding transmitted field response matrix (FRM). and are path response vectors, representing respectively the channels from k I VU to O r , from O t to BS k B , and from O t to Eve k E channel responses. and have a structure similar to that of and and can be calculated by replacing the corresponding values (number of paths and AoA and AoD). and

[0045]

[0046] We assume that the CSI of the legitimate channels is known and that Eves have perfect CSI of the VU and UAV channels. From a secrecy perspective, this is the worst-case scenario.

[0047] To protect communication security, the UAV will continuously broadcast AN signals to interfere with Eves. Additionally, for ease of understanding, the transmission process from k I to the UAV and the transmission process from the UAV to the BS are denoted as Stage I and Stage II respectively.

[0048] We introduce the task transmission in three cases. First is the initial stage of the system, where only Stage I sends data packets and there is no data packet transmission in Stage II. In this case, the communication secrecy rate of Stage I is

[0049] Normal stage of the system: The UAV has received the task data packets and performs decoding and forwarding. In this case, there are task data packet transmissions in both Stage I and Stage II. At this time, the communication secrecy rates of Stage I and Stage II are respectively and

[0050] Final stage of the system: If all VU tasks have been sent and there are still data packets not transmitted at the UAV at this time, then the communication secrecy rate of Stage II is

[0051] The purpose of this solution is to perform secure offloading while considering system latency, energy consumption, and resource allocation. Therefore, this solution uses the secrecy rate in the worst case of the two-stage communication in Stage I and Stage II to determine the security of the system. As long as the worst secrecy rate is greater than the threshold, the entire system is secure.

[0052] The calculation of the communication secrecy rate in the system is as follows:

[0053]

[0054] Among them, for the communication in Stage I: Eavesdropping: For the communication in Stage II: Eavesdropping:

[0055] Before the calculation, we need to know the SINR between each node. First, we define k I The SINR between the VU and the UAV is:

[0056]

[0057] Where is the receive beamformer, is k I is the transmit power of the VU, and ρ is the SI loss coefficient, representing the path loss and the SI cancellation at the analog and digital thresholds. It can also be rewritten as:

[0058]

[0059] The SINR between the UAV and the BS is:

[0060]

[0061] In the formula, is the transmit beamforming. It can also be written as:

[0062]

[0063] When calculating the eavesdropping rate of the eavesdropper, for the SINR calculation of Eves, we consider two cases. One is that when the eavesdropper is eavesdropping on the target k I VU and UAV, all other signals are regarded as noise.

[0064] The other is to consider the most unfavorable case for secure transmission, that is, the eavesdropping end can perform multi-user interference cancellation. When obtaining k IWhen receiving the information sent by the VU or UAV, it can eliminate the signals sent by other users. This method considers this most adverse situation, and at this time, the SINRs of Stage I and Stage Eves are respectively:

[0065]

[0066] It can also be written as:

[0067]

[0068] It can also be written as:

[0069]

[0070] Step2: Establish the optimization problem of maximizing the weighted system total utility of the above system. By jointly optimizing the MA position, beamforming, and resource allocation, the system utility is maximized.

[0071] Specifically:

[0072] After determining the secrecy rate of the task transmission process, we need to consider the time required for the entire process of task offloading, including the transmission delay and the computing delay. The transmission delay is divided into two parts, namely the delay from the k I VU to the UAV and the transmission delay from the UAV to the BS. They are respectively defined as

[0073]

[0074] Among them is the size of the k I VU task, is the size of the data packet after the UAV decodes and forwards.

[0075] So the total transmission delay of the k I task is:

[0076]

[0077] The computing delay at the MEC is defined as:

[0078]

[0079] is the number of CPU cycles required for the MEC to calculate each bit, is the size of the available computing resources of the MEC.

[0080] According to the task offloading process, there are multiple users in this system who need to offload their computing tasks to the BS, and the BS allocates the MEC to perform task calculations. Therefore, the latency is divided into two parts: offloading latency and computing latency. The latency of the system is the sum of the maximum values of these two latencies.

[0081]

[0082] The energy consumption of the entire system mainly includes computing energy consumption, transmission energy consumption, and block generation energy consumption. Next, we will introduce the energy consumption situation from the perspective of different nodes.

[0083] First is the VU node. The energy consumption of the VU is the energy consumption for offloading data tasks, which we use to represent.

[0084]

[0085] Secondly is the UAV node. The UAV hovers at a fixed altitude H and forwards data packets, and adjusts the position of the MA. The UAV hovers at a fixed altitude H throughout the entire process of task offloading in the system. Therefore, the energy consumption of the UAV includes hovering energy consumption, transmission energy consumption, energy consumption generated by the MA, and block generation energy consumption. The hovering energy consumption of the UAV is represented by to represent,

[0086]

[0087] where is the hovering power, and the hovering power depends on the mass of the UAV, the radius of the propeller, gravity, and air density. Theoretically, these parameters are invariant at a fixed altitude and stable environmental conditions. The UAV forwards tasks from k I and adds AN to prevent eavesdropping. The forwarding energy consumption of the UAV is expressed as

[0088]

[0089] In the formula, is the transmission power of the UAV, is the size of the data packet forwarded by the UAV. The total energy consumption of the UAV is represented by to represent,

[0090]

[0091] Finally, there is the MEC. The MEC node needs to perform task calculations, which will generate corresponding energy consumption. In addition, after receiving the data, the BS will allocate a suitable MEC to complete the task according to the MEC computing power and quotation it has. After the MEC completes the calculation, it will generate a block and record the calculation result and the identity information of the MEC that executed the calculation in the block to achieve traceability of the result source. Therefore, their energy consumption includes the computing energy consumption of the MEC and the energy consumption for generating the block. The energy consumption of the MEC computing task is represented by indicating that

[0092]

[0093] where κ is the capacitance coefficient, is the amount of computing resources available to the MEC server assigned to the k I VU task, is the number of CPU cycles required for the MEC to calculate 1 bit, and the energy consumption for generating the block is represented by indicating that

[0094]

[0095] In the formula, P blc is the power of the MEC to generate the block, and Γ blc is the time for generating the block.

[0096] Therefore, the total energy consumption required for the MEC to complete the k I task is:

[0097]

[0098] The total energy consumption required to complete the k I task is the sum of the VU energy consumption, UAV energy consumption, and MEC server energy consumption, expressed as:

[0099]

[0100] The total system energy consumption is:

[0101]

[0102] There is also the cost part. To ensure the authenticity and accuracy of transactions, a lightweight private blockchain is proposed, integrating the nodes in the system into the blockchain network. The MEC aggregates the transaction records and then encrypts and digitally signs these records. The MEC receives a reward in the form of Gas fees. The service fees and Gas fees are divided into three tiers according to the computing performance, which are and where and After the allocation is completed, the cost is settled according to the task size. That is, for k I in terms of, the cost to be paid for completing the task calculation is:

[0103]

[0104] The total cost generated by the system is:

[0105]

[0106] According to the above model, an optimization problem is established

[0107] The system utility function is defined as:

[0108] μ = αR min -β1Γ sum -β2ε sum -β3c sum . (38)

[0109] The optimization problem can be expressed as:

[0110]

[0111] Among them, (39.1) restricts the moving range of the MA, (39.2)(39.3) are the minimum distance constraints between the moving antennas, and (39.4)(39.5) respectively represent k I and the transmission power constraints of the UAV, and (39.6) normalizes the receiving beamformer.

[0112] Since the original optimization problem is non-convex, and there are many variables and they are coupled with each other. Therefore, we decompose the original problem P1 into two sub-problems, namely, the MA position optimization and the beamforming optimization.

[0113] Step3: As Figure 2 shown, an improved PSO algorithm is proposed, that is, MVPSO is used to solve the MA position.

[0114] Specifically:

[0115] First, since the MA position will directly affect the communication secrecy rate, a secrecy rate constraint is added, and the sub-problem of MA position optimization is written as

[0116]

[0117] Among them,

[0118] Since the traditional PSO iterates while fixing other MAs when moving one MA, however, the positions of other given MAs shrink the optimization space of the current MA to a very small region, which may converge to a local optimal solution. To address this issue, the traditional PSO is improved by iteratively replacing the single velocity of each particle with multiple candidate velocities while optimizing the positions of MAs.

[0119] The first thing to do is to randomly initialize the positions and velocities of N particles, denoted as: and The position of each particle represents an antenna position vector of a possible solution, that is,

[0120] The moving area is a square of C×C (C = C t = C r ), and each element in U (0) follows a uniform distribution in the real number interval This ensures that the initial positions of MAs do not exceed the corresponding area for movement, that is, constraint (39.1). Then, the personal best position u pbest,n of the n-th particle is initially Then, the global best position u gbest .

[0121] (1) Define the fitness function: Considering the constraint conditions (39.1)(39.2)(39.3), first define the penalty function as:

[0122]

[0123] where is the position of the n-th particle in the q-th iteration (1 ≤ q ≤ maxq), and maxq is the maximum number of iterations. is an indicator function that equals 1 when the condition in the parentheses is true, and vice versa, equals 0. and are positive penalty factors used to adjust the degree of penalty. Assuming that the fitness value of the best position is the largest, based on the given penalty function, to maximize R sum , define the fitness function as:

[0124]

[0125] and always maintain the value of Therefore, the penalty function can drive the particles to satisfy the minimum distance between MAs. In other words, as the iteration progresses, will converge to 0.

[0126] (2) U Update Position and Velocity: The candidate position of the n-th particle in the q-th iteration is Candidate Velocity Update

[0127]

[0128] where is the candidate velocity, is a function that projects to the nearest boundary when each term of u exceeds the feasible region to satisfy constraint (1), i.e.:

[0129]

[0130] Regarding the inertia weight factor, the inertia weight of particles in the standard PSO decreases as the number of iterations increases. The decrease in inertia weight results in a relatively small update amplitude of the particle velocity, which easily causes the particle swarm to gather in a local area and thus fall into a local optimal solution. We improve the traditional PSO inertia weight:

[0131]

[0132] where k is a constant coefficient, ω max and ω min are the maximum inertia weight and the minimum inertia weight respectively, q and maxq are the current iteration number and the maximum iteration number respectively.

[0133] The standard PSO lacks exploitation in the early stage and exploration in the later stage. To improve it, each particle can select the optimal velocity from multiple candidate velocities. The candidate velocity is generated by the weighted sum combination of velocity components, i.e.:

[0134]

[0135] where is the constant vector of the n-th particle, representing the combined weight for generating the candidate velocity.

[0136] is the set of velocity components of the n-th particle, is the velocity component. It is calculated as:

[0137]

[0138] Among them, c1 and c2 are the personal and global acceleration coefficients respectively, and r1 and r2 are two random vectors randomly distributed between [0, 1]. is the inertia weight of the velocity component.

[0139] Finally, select the position of the n-th particle in the q-th iteration from the candidate solutions to maximize the value of its fitness function, that is, satisfy:

[0140]

[0141] The corresponding velocity update is:

[0142]

[0143] (3) Update the personal and global optimal positions.

[0144] After obtaining the particle positions, if the fitness values of the current positions exceed the personal and global best fitness values respectively, update them to the personal and global best positions.

[0145] The MA position optimization by the improved PSO improves the signal strength of legal communication, thus enhancing communication security.

[0146] Step4, given the MA position, use the algorithm shown in Figure 3 to solve the beamforming.

[0147] Specifically:

[0148] The beamforming sub-problem is written as:

[0149]

[0150] It can be known from P3 that this problem is non-convex and not easy to solve. We use the SCA method to transform it into a convex problem and then solve it with CVX. Due to the non-convexity of the secrecy rate constraint, we introduce slack variables to formulate the problem P3 as follows:

[0151]

[0152]

[0153] and is k I the minimum acceptable SINR between the UAV and the UAV and between the UAV and k B BS, similarly, and is the maximum acceptable SINR for Eve to eavesdrop on two communication links. To illustrate the equivalence between P3 and P3.1, define K I ·E 个 Positive real number

[0154]

[0155] where and Under the third constraint of P3, using equation k I to the definition of the worst-case secrecy rate of the UAV, we obtain

[0156]

[0157] The upper bound of the third constraint in P3.1 can be used through (51) in (52) and the lower bound of the fifth constraint to satisfy the third constraint in P3. Similarly, the fourth and sixth constraints of P3.1 are argued similarly.

[0158] First, optimize all the received beamforming A for all points. The goal is to maximize the SINR of all k I to the UAV, that is

[0159]

[0160] After rewriting Regarding Form a generalized Rayleigh quotient, in the form of:

[0161]

[0162] where

[0163]

[0164] but Since it is of rank one, the optimal beamforming vector that maximizes the signal-to-noise ratio at k I VU can be expressed as

[0165]

[0166] Regarding as an irrelevant parameter, introduce the above expression into P3.1 to obtain a problem independent of A:

[0167]

[0168] When the beamforming for k I VU is performed, the SINR seen at the UAV is

[0169]

[0170] where \(C = B\) H \(B + V\). The first constraint in P3.2 is non - convex. To correct this non - convexity, we adopt the SCA technique to relax the non - convex constraint. Thus, through sequential convex programming in an iterative manner, we can provide a locally optimal solution.

[0171] First, write the first constraint as:

[0172]

[0173] where

[0174]

[0175] Then, perform a first - order Taylor expansion on it to obtain an approximate estimate of Let This gives the estimate for the SCA method in the m - th iteration.

[0176]

[0177] For all the lower - bound constraints on SINR can be written at each iteration as: where

[0178] The first - order derivative of

[0179]

[0180] Since the expression is twice - differentiable, from the above equation, all second - order derivatives are zero, except for

[0181]

[0182] This shows that the Hessian matrix is an all - zero matrix, except for at the \(k\) - th position on the diagonal I - th. Thus, if then This is clearly true because is restricted to be positive, and is a given positive number. Therefore, we conclude that is convex with respect to \(B\) and \(V\).

[0183] The second constraint in P3.2 is expressed in terms of the trace as:

[0184]

[0185] Among them, This represents a convex set.

[0186] The third constraint in P3.2 can be expressed as follows:

[0187]

[0188] The fourth constraint in P3.2 is expressed as follows:

[0189]

[0190] The optimization problem of P3.2 can be written as:

[0191]

[0192] In this way, problem P3.3 is a convex problem and can be solved by the CVX toolbox in MATLAB.

[0193] Through beamforming, signals can be transmitted more effectively in a directional manner, enhancing the received signals of legitimate users and improving the security during the communication process.

[0194] As Figure 5 shown, a comparative experiment is conducted between the improved algorithm of this scheme and the standard PSO. The abscissa is the number of iterations, and the ordinate is the average fitness value after running 100 times.

[0195] As Figure 6 shown, a comparative experiment is conducted between this scheme and the single - antenna scheme using IRS. The abscissa is the number of iterations, and the ordinate is the system utility value. Figure 6 The reason for the large gap between this scheme and the IRS scheme in is that in the single - antenna scheme using IRS, tasks can only be transmitted one by one, which increases the system delay. At the same time, due to the increase in delay, the system energy consumption also increases. While the multi - antenna full - duplex nodes considered in this scheme greatly reduce the task transmission delay. Since the system energy consumption is proportional to the delay, the reduction in delay leads to a reduction in energy consumption. In addition, this scheme uses MA and beamforming, further enhancing the system communication security. Generally speaking, the performance of this scheme is significantly better than the IRS scheme.

[0196] The specific implementation manners of this scheme are described in detail above in conjunction with the accompanying drawings. However, this scheme is not limited to the above - mentioned implementation manners. Within the scope of knowledge possessed by those of ordinary skill in the art, various changes can be made without departing from the purpose of this scheme.

[0197] [1]Qufu Normal University. A Multi-objective Computation Offloading Method for Vehicular Networks Empowered by Blockchain and Assisted by IRS-UAV: 202411059802.X[P]. 2025-01-07.

Claims

1. A blockchain-enabled and MA-UAV-assisted multi-objective computing offloading method for the Internet of Vehicles The features are as follows: In the vehicle networking, the local computing resources of vehicles are limited and cannot meet the requirements of high performance and high real-time. By offloading computing tasks to neighboring mobile edge computing servers (MECs), the system can significantly improve the processing capacity while ensuring low latency. However, data transmission and remote computing during the task offloading process also introduce potential security risks, such as eavesdropping on the transmitted information by eavesdroppers (Eves), which seriously threatens the stability and security of the vehicle networking. This solution uses an unmanned aerial vehicle (UAV) equipped with a movable antenna (MA) to relay and forward task data, and this problem can be effectively solved by adjusting the position of the MA and beamforming. To address the centralized vulnerability problem during the vehicle task data offloading process, this solution uses blockchain technology to generate an immutable record for each task offloading process, providing a complete audit record for the system; Specifically: Step 1, establish a scenario where MA-UAV in the blockchain-oriented Internet of Vehicles assists mobile vehicles in task offloading. The ground facilities include single-antenna mobile vehicle users (VUs), MECs, a hovering UAV equipped with MA. The UAV receives and forwards the VU transmission signal. The MA is divided into a receiving MA with a total of antennas and a transmitting MA with a total of antennas. There is a multi-antenna base station (BS) with antennas, and multiple single-antenna Eves exist simultaneously; Step 2, respectively establish the communication model, energy consumption model, and cost model of the MA position and beamforming, and construct an optimization model based on these three models; Step 3, improve the traditional particle swarm optimization (PSO) according to the current initial value. Specifically, the inertia weight of the particles in the standard PSO decreases as the number of iterations increases. The decrease in the inertia weight leads to a relatively small update amplitude of the particle velocity, which easily causes the particle swarm to aggregate in a certain local area and thus fall into a local optimal solution. This solution improves the inertia weight of the traditional PSO; the standard PSO lacks exploration in the early stage and exploitation in the later stage. This solution improves it so that each particle can select the optimal velocity from multiple candidate velocities, and uses the improved PSO to solve the position of the MA; Step 4, according to the current optimal MA position, use successive convex approximation (SCA) to solve the beamforming; Step 5, use the alternating optimization algorithm, and loop to execute Step 4 and Step 5 until the system utility value of the latter iteration is less than or equal to that of the previous iteration between two adjacent iterations, or when the maximum preset number of iterations is reached, the iteration ends; Step 6, based on the current state of the entire vehicle networking, optimize the MA position and beamforming, allocate computing resources, and offload and calculate tasks for the vehicles in the vehicle networking with the aim of maximizing the system weighted total utility value.

Images