Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Resource Optimization Technology Using Genetic Algorithm in UAV-Assisted Edge Computing Environment

Resource Optimization Technology Using Genetic Algorithm in UAV-Assisted Edge Computing Environment Hindawi Journal of Robotics Volume 2022, Article ID 3664663, 8 pages https://doi.org/10.1155/2022/3664663 Research Article Resource Optimization Technology Using Genetic Algorithm in UAV-Assisted Edge Computing Environment Huijuan Sun and Hongqi Xi College of Computer and Information Technology, Henan Finance University, Zhengzhou, Henan 450046, China Correspondence should be addressed to Huijuan Sun; sunhuijuan@hafu.edu.cn Received 16 February 2022; Revised 18 March 2022; Accepted 19 March 2022; Published 6 April 2022 Academic Editor: Shan Zhong Copyright © 2022 Huijuan Sun and Hongqi Xi. ,is is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. As fixed edge computing systems can hardly meet the demand of mobile users for massive data processing, a computational resource allocation strategy using the genetic algorithm in UAV-assisted edge computing environment is proposed. First, a UAV-assisted mobile edge computing (MEC) system is designed to help users execute computation tasks through the UAV or relaying to the ground base station. ,en, a communication model and a computation model are constructed to minimize the total system energy consumption by jointly optimizing the UAV offloading ratio, user scheduling variables, and UAV trajectory. Finally, the mini- mization of total system energy consumption is modeled as a nonconvex optimization problem and solved by introducing an improved genetic algorithm, so as to achieve a rational allocation of computational resources. Based on the experimental platform, the simulation of the proposed method is carried out. ,e results show that the total energy consumption is 650 J when the execution time is 110 s and the execution time is 17.5 s when the number of users is 50, which are both better than other comparison methods. high maneuverability and flexibility, UAV plays an in- 1. Introduction creasingly prominent role in the communication system and is widely used in aerial surveillance, imaging, and traffic [5]. In recent years, as cloud computing are gradually moving to To maintain the safe operation of UAVs through real-time the edge of the network, mobile edge computing (MEC) has command/control and to enable new applications with ar- emerged and developed rapidly, which mainly utilizes the tificial intelligence, it is becoming more important to en- spare computing capacity and storage of the edge network to hance the communication and computing capabilities of execute computation-intensive and delay-sensitive tasks UAVs [6, 7]. [1, 2]. Currently, scholars are actively trying to integrate However, UAVs are limited by size, load capacity, and MEC-related technologies with a wireless communication power consumption, which make the communication, theory to better address long propagation delays, and the computing capacity, and endurance time all suffer from a deployment of mobile base stations on unmanned aerial large impact. In order to settle this limitation, it is of great vehicle (UAV) is a possible solution to extend the coverage practical significance to design resource allocation strategies of wireless systems to the areas with limited available in- and optimal control methods for UAV trajectories to make frastructure for wireless access points [3]. the best use of limited energy of UAV, so as to better provide UAV is a non-manned vehicle operated by remote high-capacity, high-speed, and low-delay services for control, with advantages of good maneuverability and small wireless communication edge computing systems [8]. size, which is easy to carry and transport. Meanwhile, with At present, there have been some studies at home and great stability during low altitude flight, UAV can better abroad for computational resource allocation strategies in acquire images obscured by clouds and other objects, which edge computing environment. According to [9], the mixed cannot be captured by the satellite light-sensing remote integer and nonconvex optimization problem was decoupled controller and other aerospace photography [4]. Due to its 2 Journal of Robotics into two sub-problems of task offloading decision and re- source optimization, which were jointly optimized by using UAV V a designed iterative algorithm. ,e experimental results showed that the optimization performance is comparable to Wireless mm Wave that of the exhaustive search. However, this algorithm takes Link backhaul a longer time for optimization and the efficiency needs to be improved. ,e authors of [10] proposed a dynamic multi- win game model based on incomplete information to solve User the task offloading and edge resource allocation problems for multiple terminal users. Further optimization is needed for MEC Server multiobjective task offloading and resource allocation in complex network situations. ,e authors of [11] proposed a computing resource allocation scheme for MEC scenarios based on DRL networks. ,e cost of task offloading still Figure 1: System model. needs to be optimized. In order to investigate the rela- tionship between the cost of computation tasks and time and Both UAV and ground users have communication energy consumption, a computation offloading strategy was circuits and computing processors and are powered by proposed, which reduces the execution cost by collabora- batteries with a certain capacity. It is worth noting that tively allocating computational resources between mobile different users have different computing capacitates. ,e devices and edge servers [12]. ground base station is deployed with multiple built-in MEC With the continuous development of machine learning servers with high computing capacity, which can be regarded and UAV communication technologies, deep learning and as an access point [16, 17]. In this scenario, the user has UAV assistance have been gradually applied in computa- delay-sensitive computation tasks that require UAV to as- tional resource allocation. Authors in Ref. [13] proposed a sists in computation. ,e UAV is deployed with MEC multiobjective optimization algorithm based on the servers to help users to execute the task on itself or relay to MOACO algorithm combined with reinforcement learning, the ground base station for execution. which effectively improves the quality of service in edge computing by establishing an effective cognitive agent model to evaluate the resource allocation. However, with more 2.2. Communication Model. In the system model, users are iteration, the time efficiency needs to be improved. In the fixed on the ground and the position can be denoted as q, study [14], a multiagent reinforcement learning framework q � [x, y], where x and y are the horizontal coordinates of based on independent learners was proposed to achieve the user. UAV keeps flying at the same altitude H(H > 0) computation offloading and resource allocation for multiple and its position at i th timeslot can be expressed as private users in IoT edge computing networks. Authors in q � [x [i], y [i]], 0 ≤ i ≤ I. u u u Ref. [15] introduced agents into task offloading and pro- It is assumed that the wireless channel between the UAV posed a new framework for agent-enabled task offloading in and users is a Line of Sight (LOS) channel. ,erefore, the UAV-aided MEC to help users, UAV, and edge clouds channel power gain between the UAV and users can be perform computation task offloading. UAV can improve expressed as follows: wireless communication coverage to a certain extent, but the computation offloading strategy still needs further − 2 0 h[i] � δ d [i] � � � , i ∈ I, 0 (1) � �2 optimization. � � H + �q [i] − q� Based on the above analysis, a computational resource allocation strategy using the genetic algorithm in the UAV- where δ is the channel power gain within unit distance − 2 assisted edge computing environment is proposed to address (d � 1m) and d [n] is the horizontal distance between the the problem that the computing capacity of existing mobile UAV and the user at i th timeslot,i � 1, 2, . . . , I. ‖ · ‖ is the devices (computers, cell phones) cannot fully meet the users’ Euclidean norm. demand for communication quality. Based on the UAV- assisted edge computing system, a minimization model of total system energy consumption with jointly optimized 2.3. Computation Model UAV offloading ratio, user scheduling variables, and UAV 2.3.1. Local Execution. When users execute the task locally, trajectory is constructed. ,e minimization problem is the execution time can be calculated as follows: solved using an improved genetic algorithm to achieve a reasonable allocation of computational resources. 1 − η [i]ϑ [i]􏼁C D [i] loc n n n n (2) T [i] � , 2. System Model and Optimization Objective where η [i] is the scheduling variable of user n. ϑ [i] is the n n 2.1. System Model. As shown in Figure 1, (a) UAV-assisted offloading ratio of user n at i th time slot. D [i] denotes the MEC system consists of a cellular-connected UAV and bit size of the computation task of user n at ith timeslot. C multiple ground-requesting users. denotes the CPU frequency required when user n executes a Journal of Robotics 3 task of 1 bit. f denotes the local computing capacity of user η [i]ϑ [i]D [i]C UAV n n n n T [i] � , (6) n. UAV ,en, the energy consumption for local execution can be UAV written as follows: where f is the computing capacity of the UAV when serving user n. Hence, the energy consumption for the UAV- loc 2 (3) E [i] � c1 − η [i]ϑ [i]􏼁D [i]C f , n n n n n n assisted computation can be formulated as follows: UAV UAV2 where c is a coefficient that depends on the chip architecture. ′ (7) E [i] � c η [i]ϑ [i]D [i]C f , n n n n n n where c is a coefficient related to the chip architecture of 2.3.2. UAV-Assisted Computation. For UAV-assisted UAV. computation, it can be divided into two parts: the trans- mission process and the assisted computation. ,us, the UAV transmission time delay can be calculated as follows: 2.4. Optimization Objective. According to the aforemen- η [i]ϑ [i]D [i] tr n n n tioned UAV-assisted computation model, the correspond- T [i] � , (4) v [i] ing optimization problem is designed. It is assumed that user local execution and UAV-assisted computation are per- where v [i] is the transmission rate between user n and the formed simultaneously, and the UAV needs to perform the UAV at i th time slot. ,en, the transmission energy con- computation and return the results at the same time slot. sumption can be written as follows: Because the size of the returned results is much smaller than η [i]ϑ [i]D [i] that of the UAV-assisted computation task, the energy tr n n n E [i] � P , (5) n 0 consumption for returning results can be neglected [18, 19]. v [i] ,e objective is to minimize the total system energy con- where P is the rated transmission power of the UAV. sumption by jointly optimizing the UAV offloading ratio, For the UAV-assisted computation, in each timeslot, the the user scheduling variables and the UAV trajectory. ,us, UAV serves multiple ground users, the execution time of the optimization problem can be formulated as follows: UAV serving user n at ith timeslot for the assisted com- putation can be calculated as follows: I I N fly UAV tr loc Γ � min 􏽘 E [i] + 􏽘 􏽘 E [i] + E [i] + E [i] , (8) 􏼐 􏼑 n n n η [i],ϑ [i],q[i] { } n n i�1 i�1 n�1 s.t. q[1] � q[I], (9) ‖q[i + 1] − q[i]‖ ≤ V , i ∈ I, max 0≤ ϑ [i] ≤ 1, ∀n, i, (10) 􏽘 η [i] ≤ 1, ∀i (11) n�1 η [i] ∈ {0, 1}, ∀n, i, v [i] ≥ η [i] · ], ∀n, i, (12) n n I N loc 􏽘 􏽘 T [i]η [i] ≤ T, ∀n, i, (13) n n i�1 n�1 I N tr UAV 􏽘 􏽘􏼐T [i] + T [i]􏼑η [i] ≤ T, ∀n, i, (14) n n i�1 n�1 fly where E [i] is the energy consumption of the UAV during Equation (9) shows the constraint of the UAV trajectory. its flight. V is the maximum flight speed of the UAV. τ is Equation (10) defines the constraint of the offloading ratio. max the length of each timeslot. ] is the flight speed of the UAV. Equation (11) represents the user scheduling variables. 4 Journal of Robotics Equation (12) ensures the communication quality during Start offloading. Equation (13) represents the constraint of user local execution delay. Equation (14) represents the con- straint of user transmission delay and UAV-assisted com- Coding and initial putation delay. population generation 3. Computational Resource Allocation Strategy Calculate individual Mutation operation fitness Using Improved Genetic Algorithm Crossover operation Genetic algorithm is a relatively mature algorithm that searches for the global optimal solution by simulating the Meet the termination conditions? Selection operation process of biological evolution in nature. Its parallelism is suitable for distributed edge computing. Moreover, the genetic algorithm has good global search capability, so that End an appropriate solution set for resource allocation can be obtained by using the genetic algorithm with multiobjective Figure 2: Flow chart of basic genetic algorithm. optimization. ,erefore, this paper introduces an improved genetic algorithm, which can not only improve the search efficiency mutation operations, the value of genes converted to real but also obtain the further optimization objective by numbers may exceed the range of edge node numbers, so a establishing a dynamic adjustment model of crossover rate remainder approach is used [25, 26]. For example, the total and mutation rate, while reflecting the advantages of dis- number of edge nodes is defined to be 7 and thus the binary tributed execution of tasks. In the following sections, the code is {1101}, which will be converted to a decimal number implementation of resource scheduling of the proposed of 13. ,e remainder of 13 divided by 7 is 6, then the model is addressed based on the genetic algorithm. corresponding edge node number of this binary code is 6. Set the population size as S, the task sequence as S , the length as a, and the number of edge nodes in the random 3.1. Basic Genetic Algorithm. ,e basic genetic algorithm network topology graph as m. Each chromosome is com- was proposed by J. H. Holland of the University of Michigan, posed of a random combination of binary numbers whose USA, which is the basis for other genetic algorithms. length is 4 a, and every four binary numbers will be con- Moreover, other improved genetic algorithms are developed verted into a decimal number, which represents the number by adding new mechanisms. ,e flow chart of basic genetic of edge nodes. algorithm is shown in Figure 2. ,e basic genetic algorithm generates optimal individ- uals by simulating the selection, crossover, and mutation 3.2.2. Fitness Function. ,e fitness function is the key to operations in biological genetic and evolutionary processes evaluate the direction of population evolution in task and finds the optimal solution through an adaptive search scheduling of terminal users in edge computing. In order to process [20, 21]. ensure that the fitness function can remain meaningful when the value of optimization objective Γ is equal to 0, it is 3.2. Improved Genetic Algorithm defined as follows: − Γ f(Γ) � e . (15) 3.2.1. Encoding and Population Initialization. ,e proposed method adopts binary encoding where a chromosome corresponds to a solution of the optimization problem. ,e defined chromosomes are encoded using real numbers. 3.2.3. Genetic Operations. ,e selection operation is per- ,erefore, it is necessary to convert the genes in the formed by a roulette wheel. Assuming that the size of the chromosome to be represented in binary. Moreover, the population is S and the fitness value of an individual s is f , genes in the chromosome are all real numbers, which can be and p is the probability of an individual s being selected, the converted into a four-bit binary number and then be probability can be calculated as follows: replaced in the corresponding genes. Moreover, this is the encoding method used in the improved genetic algorithm p � . (16) [22, 23]. For example, a chromosome sequence is (3, 5, 8) j�1 j when encoded in real numbers, then the chromosome se- quence is converted into (0011, 0101, 1000) after binary ,e crossover operator is used to generate new indi- encoding. viduals by swapping the positions of chromosome genes of However, the objective function value is calculated the parent individuals. ,e traditional single-point crossover mainly based on real numbers, which requires reconverting method is used, in which a crossover point is set between any the binary-coded chromosome sequence into a real-coded two adjacent genes in each chromosome sequence S , then a − 1 chromosome sequence [24]. Moreover, after crossover and crossover point is arbitrarily selected in the first w genes, Journal of Robotics 5 and all genes after that crossover point are replaced with is 0.75, the mutation rate is 0.01, and the initial population each other [27]. size is 25. In the mutation operation, each chromosome sequence S uses a basic position mutation, where one gene is ar- 4.1. UAV Trajectory. ,e UAV trajectory optimized by the bitrarily selected in w number of genes and replaced by a improved genetic algorithm during different task execution random number between 1 to w. ,us, a new chromosome time T is shown in Figure 3, where the red circles indicate the is generated and the diversity of the population can be positions of ground users. ensured. From Figure 3, it can be seen that when T � 30 s, the UAV flies along the closed-loop trajectory and the flight range is small, the UAV is far from users when offloading 3.3. Steps of Task Scheduling. ,e steps of the improved genetic algorithm are as follows. data, and the communication quality is poor. With the increase of T, and when T � 60 s and T �120 s, the distance (1) Population initialization. ,e user submits a ter- between the UAV and users decreases and the flight range minal request and randomly sets the initial pop- expands, thus the communication quality improves. When T ulation g(0) based on the request. Set the iteration is large enough, the UAV continues to fly along the closed- counter t � 0, and g(t) indicates the tth generation loop trajectory and serves one user in several timeslots. In of population. addition, it can be found that the UAV has a larger flight (2) Calculate the fitness value. ,e fitness value of an range and is closer to users when T is larger, and the individual is calculated according to the fitness communication quality is better. function in the improved genetic algorithm. (3) Judge the termination condition. When the number 4.2. Convergence Performance of Algorithms under Different of evolutionary generations reaches the specified Computation Tasks. In order to demonstrate the conver- number of iterations Θ, the result should be given gence of the improved genetic algorithm, it is compared with and an edge computation task scheduling strategy is the traditional genetic algorithm, and the results are illus- found; otherwise, go to step (4). trated in Figure 4. (4) Selection operation. ,e selection operation of the It is depicted from Figure 4 that the proposed improved populations is performed using the roulette wheel. genetic algorithm executes more tasks within the same (5) Crossover operation. According to the crossover number of iterations, and when the number of iterations method in the designed improved genetic algorithm, exceeds 70, the amount of computation tasks tends to two individuals are randomly selected among the converge to 3450 M bit. Because it is optimized in encoding individuals produced by the selection operation and and genetic operations, the amount of computation in- are performed based on the crossover rate, so as to creases. However, the process is slightly more complicated, generate new individuals. resulting in longer processing time, and the number of it- erations that tend to be stable is about 10 times higher than (6) Mutation operation. According to the mutation that of the original genetic algorithm. On the whole, the method in the improved genetic algorithm, new proposed algorithm can achieve effective convergence individuals are generated by crossover operation and performance. the selected individuals among the new individuals are operated according to the mutation rate to generate new individuals. 4.3. Relationship between Total System Energy Consumption (7) Compare the fitness value. If the fitness value of the with Task Execution Time. To demonstrate the performance mutated individuals is smaller than that of parent of the total energy consumption of the proposed method, it individuals, they will be replaced by the parent is compared with those of the [9, 13, 15], and the results are generation. If it is greater than or equal to the fitness shown in Figure 5. value of the parent generation, the parent generation As can be seen in Figure 5, with the increase of task will be replaced. ,en, a new generation of indi- execution time, the total energy consumption of the system viduals g(i + 1) is generated. also increases. Meanwhile, the proposed UAV-assisted computational resource allocation method requires lower (8) Update the iteration counter. Set the iteration counter as t � t + 1, then go to step (2). energy consumption and has better performance than the other three methods. Moreover, the total energy con- sumption required by the proposed method is always the 4. Experiment and Analysis least while the task execution time is increasing. ,e total ,e UAV trajectory and the performance of the proposed energy consumption is 650 J when the execution time is improved genetic algorithm are demonstrated by numerical 110 s. As the proposed method executes the computation simulations. ,e system simulation parameters are set as task with the assistance of UAV and optimizes the UAV shown in Table 1. trajectory using the improved genetic algorithm, it can dynamically and timely adjust the computation tasks so as to ,e parameters of the improved genetic algorithm are set as follows. ,e number of iterations is 00, the crossover rate reduce the energy consumption. Reference [9] introduces 6 Journal of Robotics Table 1: System simulation parameters. System parameters Value Number of users N 10 Flight altitude H 150 m Channel power gain δ −75 dB CPU frequency C 1200 cycles/bit −27 c � c 10 UAV UAV computing capacity f 1400 Hz User local computing capacity f [200, 500] Hz User fixed transmission capacity P 50 dB·m 600 Maximum flight speed of UAV V 60 m/s max 10 30 50 70 90 110 Task execution time (s) Ref. [9] Ref. [15] Ref. [13] Proposed method Figure 5: Total energy consumption of the system under different methods. mixed integer and nonconvex optimization to obtain task offloading decision and resource optimization. However, the optimization method is traditional, the obtained resource 100 allocation strategy is not reasonable enough, and the total energy consumption of the system is higher than 1600 J. Reference [13] combines MOACO algorithm and rein- forcement learning for multiobjective resource allocation, 0 100 200 300 400 500 x (m) which can effectively improve the quality of users’ com- puting services, but the computational process is complex T=120s and the energy consumption is high. Similarly, [15] intro- T=60s duces UAV agents to assist in the offloading of computation T=30s tasks. However, it lacks an effective optimization algorithm Figure 3: UAV trajectory. to schedule its trajectory, so the energy consumption during UAV flight is high and its total system energy consumption is 62.5% higher than the proposed method. 4.4. Performance Comparison with Other Methods. In ad- dition, to further validate the advantage of the proposed method in term of execution time, it is compared with [9, 13, 15], and the results are depicted in Figure 6. It can be seen from Figure 6 that the task execution time increases continuously with more users, but the proposed method has the slowest increase and an obvious advantage in terms of execution time. ,e execution time is 17.5 s when the number of users is 50. ,e proposed method uses an improved genetic algorithm to optimize the UAV trajectory, it reduces energy consumption by lessening the blind flight time of the UAV and it offloads the tasks nearby to further reduce the transmission time. Compared with the proposed 1000 method, [15] lacks an effective algorithm to optimize the 0 50 100 150 200 250 UAV trajectory, resulting in a longer flight time and a higher Number of iterations overall execution time, which is similar to that of [13]. ,us, Improved genetic algorithm the maximum execution time is more than 25 s. ,e opti- Genetic algorithm mization method in [9] is simple and easy to implement, and the execution time is shorter when the number of users is Figure 4: Convergence performance of genetic algorithms before and after improvement. small. However, when the number of users increases, it is Calculate task quantity (Mbit) y (m) Total energy consumption of the system (J) Journal of Robotics 7 30 ,e UAV-assisted MEC system only considers the scenarios where UAV is connected with users or the ground. When the UAV only has limited computing capacity, the case that UAV collects users’ data and offloads tasks to the ground base station for assisted computing is also worth studying. 15 Data Availability ,e data used to support the findings of this study are in- cluded within the article. Conflicts of Interest ,e authors declare that they have no conflicts of interest regarding the publication of this paper. 10 20 30 40 50 Number of users References Ref. [9] Ref. [15] Ref. [13] Proposed method [1] X. Liu, J. Yu, J. Wang, and Y. Gao, “Resource allocation with Figure 6: Execution time of different methods. edge computing in IoT networks via machine learning,” IEEE Internet of <ings Journal, vol. 7, no. 4, pp. 3415–3426, 2020. [2] H. C. Ke, H. Wang, H. W. Zhao, and W. J. Sun, “Deep re- unable to handle the huge amount of data, resulting in the inforcement learning-based computation offloading and re- significant rise of the execution time. source allocation in security-aware mobile edge computing,” Wireless Networks, vol. 27, no. 5, pp. 3357–3373, 2021. [3] J. Feng, L. Liu, Q. Pei, F. Hou, T. Yang, and J. Wu, “Service 5. Conclusion characteristics-oriented joint optimization of radio and computing resource allocation in mobile-edge computing,” With the explosive growth of IoT devices, cloud computing IEEE Internet of <ings Journal, vol. 8, no. 11, pp. 9407–9421, can no longer meet the demand of some time-sensitive computation tasks. For this reason, a computational resource [4] L. Deng, P. Yang, W. Liu, L. Wang, S. Wang, and X. Zhang, allocation strategy using the genetic algorithm in the UAV- “NAAM-MOEA/D-Based multitarget firepower resource al- assisted edge computing environment is proposed. With the location optimization in edge computing,” Wireless Com- advantages of high mobility, low cost, and easy carrying, the munications and Mobile Computing, vol. 2021, Article ID UAV is introduced in the edge computing environment to 5579857, 14 pages, 2021. assist in task offloading. ,e flexible scheduling of the UAV [5] Z. Song, Y. Liu, and X. Sun, “Joint task offloading and resource enables efficient allocation of user tasks and reduces execution allocation for NOMA-enabled multi-access mobile edge computing,” IEEE Transactions on Communications, vol. 69, time. In order to optimize the flight trajectory of the UAV, the no. 3, pp. 1548–1564, 2021. problem of minimizing the total energy consumption of the [6] B. Baek, J. Lee, Y. Peng, and S. Park, “,ree dynamic pricing system is solved by applying an improved genetic algorithm. schemes for resource allocation of edge computing for IoT ,en, the optimal flight scheme can be obtained, that is, the environment,” IEEE Internet of <ings Journal, vol. 7, no. 5, optimal resource allocation strategy, which greatly reduces pp. 4292–4303, 2020. the system energy consumption and improves the trans- [7] Z. Wang, S. Zheng, Q. Ge, and K. Li, “Online offloading mission efficiency. ,e results based on the UAV simulation scheduling and resource allocation algorithms for vehicular platform show as follows: edge computing system,” IEEE Access, vol. 8, no. 9, pp. 52428–52442, 2020. (1) By real number encoding, selection, crossover, and [8] W. Zhan, C. Luo, G. Min, C. Wang, Q. Zhu, and H. Duan, mutation operations, the performance of the genetic “Mobility-Aware multi-user offloading optimization for algorithm in finding the optimal solution is effec- mobile edge computing,” IEEE Transactions on Vehicular tively improved, and the data size of computational Technology, vol. 69, no. 99, pp. 3341–3356, 2020. tasks can reach 3450 M bit when the number of it- [9] Y. Fu, X. Yang, P. Yang et al., “Energy-efficient offloading and erations exceeds 70. resource allocation for mobile edge computing enabled mission-critical internet-of-things systems,” EURASIP Jour- (2) ,e introduction of the UAV allows dynamic re- nal on Wireless Communications and Networking, vol. 2, no. 1, sponse to users’ computational demands and im- pp. 1–16, 2021. proves the efficiency of the computation offloading. [10] B. Wang and M. Li, “Resource allocation scheduling algo- When the number of users is 50, the execution time rithm based on incomplete information dynamic game for is 17.5 s, and at the same time, when the execution edge computing,” International Journal of Web Services Re- time is 110 s, the total energy consumption is 650 J, search, vol. 18, no. 2, pp. 1–24, 2021. which effectively reduces the system energy [11] Y. Zhang, M. Zhang, C. Fan, F. Li, and B. Li, “Computing consumption. resource allocation scheme of IOV using deep reinforcement Execution time (s) 8 Journal of Robotics learning in edge computing environment,” EURASIP Journal edge computing,” Science China (Information Sciences), vol. 63, no. 04, pp. 154–169, 2020. on Applied Signal Processing, vol. 21, no. 1, pp. 1–19, 2021. [12] S. Chen, Q. Li, M. Zhou, and A. Abusorrah, “Recent advances [27] M. Zeng and V. Fodor, “Energy minimization for delay constrained mobile edge computing with orthogonal and in collaborative scheduling of computing tasks in an edge computing paradigm,” Sensors, vol. 21, no. 3, pp. 779–787, non-orthogonal multiple access,” Ad Hoc Networks, vol. 98, no. 3, pp. 1–13, Article ID 102060, 2020. [13] V. Shanmuganthan, M. Khari, N. Dey, R. G. Crespo, and Y. H. Robinson, “Enhanced resource allocation in mobile edge computing using reinforcement learning based MOACO algorithm for IIOT,” Computer Communications, vol. 151, no. 4, pp. 355–364, 2020. [14] X. Liu, J. Yu, Z. Feng, and Y. Gao, “Multi-agent reinforcement learning for resource allocation in IoT networks with edge computing,” China Communications, vol. 17, no. 9, pp. 220– 236, 2020. [15] W. A. Rui, C. A. Yong, B. An, T. A. Alamoudi, and R. Nour, “Agent-enabled task offloading in UAV-aided mobile edge computing,” Computer Communications, vol. 149, no. 2, pp. 324–331, 2020. [16] S. Li, S. Lin, L. Cai, W. Li, and G. Zhu, “Joint resource al- location and computation offloading with time-varying fading channel in vehicular edge computing,” IEEE Transactions on Vehicular Technology, vol. 69, no. 3, pp. 3384–3398, 2020. [17] J. Yan, S. Bi, and Y. Zhang, “Offloading and resource allo- cation with general task graph in mobile edge computing: a deep reinforcement learning approach,” IEEE Transactions on Wireless Communications, vol. 19, no. 8, pp. 5404–5419, 2020. [18] Z. Yang, C. Pan, J. Hou, and M. Shikh-Bahaei, “Efficient resource allocation for mobile-edge computing networks with NOMA: completion time and energy minimization,” IEEE Transactions on Communications, vol. 67, no. 11, pp. 7771– 7784, 2019. [19] J. Zhao, Q. Li, Y. Gong, and Z. Ke, “Computation offloading and resource allocation for cloud assisted mobile edge computing in vehicular networks,” IEEE Transactions on Vehicular Technology, vol. 68, no. 8, pp. 7944–7956, 2019. [20] L. I. Shulei, D. Zhai, D. U. Pengfei, and T. Han, “Energy- efficient task offloading, load balancing, and resource allo- cation in mobile edge computing enabled IoT networks,” Science China Information Sciences, vol. 62, no. 2, pp. 1–3, [21] L. Yang, G. Xu, J. Ge, P. Liu, and X. Fu, “Energy-efficient resource allocation for application including dependent tasks in mobile edge computing,” KSII Transactions on Internet and Information Systems, vol. 14, no. 6, pp. 2422–2443, 2020. [22] Y. Guo, F. Liu, N. Xiao, and Z. Chen, “Task-based resource allocation bid in edge computing micro datacenter,” Com- puters, Materials and Continua, vol. 58, no. 2, pp. 777–792, [23] H. T. Chien, Y. D. Lin, C. L. Lai, and W. Chien-Ting, “End-to- End slicing with optimized communication and computing resource allocation in multi-tenant 5G systems,” IEEE Transactions on Vehicular Technology, vol. 69, no. 2, pp. 2079–2091, 2020. [24] Y. Chen, Z. Li, B. Yang, K. Nai, and K. Li, “A Stackelberg game approach to multiple resources allocation and pricing in mobile edge computing,” Future Generation Computer Sys- tems, vol. 108, no. 3, pp. 273–287, 2020. [25] M. Dai, Z. Su, Q. Xu, and N. Zhang, “Vehicle assisted computing offloading for unmanned aerial vehicles in smart city,” IEEE Transactions on Intelligent Transportation Systems, vol. 22, no. 3, pp. 1932–1944, 2021. [26] K. Wang, Z. Xiong, C. Lin, Z. Pan, and H. Shin, “Joint time delay and energy optimization with intelligent overclocking in http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal of Robotics Hindawi Publishing Corporation

Resource Optimization Technology Using Genetic Algorithm in UAV-Assisted Edge Computing Environment

Sun, Huijuan; Xi, Hongqi
Journal of Robotics , Volume 2022 – Apr 6, 2022
Download PDF
8 pages

Loading next page...
 
/lp/hindawi-publishing-corporation/resource-optimization-technology-using-genetic-algorithm-in-uav-Z3YpSj4kvx

References (27)

Publisher
Hindawi Publishing Corporation
Copyright
Copyright © 2022 Huijuan Sun and Hongqi Xi. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
ISSN
1687-9600
eISSN
1687-9619
DOI
10.1155/2022/3664663
Publisher site
See Article on Publisher Site

Abstract

Hindawi Journal of Robotics Volume 2022, Article ID 3664663, 8 pages https://doi.org/10.1155/2022/3664663 Research Article Resource Optimization Technology Using Genetic Algorithm in UAV-Assisted Edge Computing Environment Huijuan Sun and Hongqi Xi College of Computer and Information Technology, Henan Finance University, Zhengzhou, Henan 450046, China Correspondence should be addressed to Huijuan Sun; sunhuijuan@hafu.edu.cn Received 16 February 2022; Revised 18 March 2022; Accepted 19 March 2022; Published 6 April 2022 Academic Editor: Shan Zhong Copyright © 2022 Huijuan Sun and Hongqi Xi. ,is is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. As fixed edge computing systems can hardly meet the demand of mobile users for massive data processing, a computational resource allocation strategy using the genetic algorithm in UAV-assisted edge computing environment is proposed. First, a UAV-assisted mobile edge computing (MEC) system is designed to help users execute computation tasks through the UAV or relaying to the ground base station. ,en, a communication model and a computation model are constructed to minimize the total system energy consumption by jointly optimizing the UAV offloading ratio, user scheduling variables, and UAV trajectory. Finally, the mini- mization of total system energy consumption is modeled as a nonconvex optimization problem and solved by introducing an improved genetic algorithm, so as to achieve a rational allocation of computational resources. Based on the experimental platform, the simulation of the proposed method is carried out. ,e results show that the total energy consumption is 650 J when the execution time is 110 s and the execution time is 17.5 s when the number of users is 50, which are both better than other comparison methods. high maneuverability and flexibility, UAV plays an in- 1. Introduction creasingly prominent role in the communication system and is widely used in aerial surveillance, imaging, and traffic [5]. In recent years, as cloud computing are gradually moving to To maintain the safe operation of UAVs through real-time the edge of the network, mobile edge computing (MEC) has command/control and to enable new applications with ar- emerged and developed rapidly, which mainly utilizes the tificial intelligence, it is becoming more important to en- spare computing capacity and storage of the edge network to hance the communication and computing capabilities of execute computation-intensive and delay-sensitive tasks UAVs [6, 7]. [1, 2]. Currently, scholars are actively trying to integrate However, UAVs are limited by size, load capacity, and MEC-related technologies with a wireless communication power consumption, which make the communication, theory to better address long propagation delays, and the computing capacity, and endurance time all suffer from a deployment of mobile base stations on unmanned aerial large impact. In order to settle this limitation, it is of great vehicle (UAV) is a possible solution to extend the coverage practical significance to design resource allocation strategies of wireless systems to the areas with limited available in- and optimal control methods for UAV trajectories to make frastructure for wireless access points [3]. the best use of limited energy of UAV, so as to better provide UAV is a non-manned vehicle operated by remote high-capacity, high-speed, and low-delay services for control, with advantages of good maneuverability and small wireless communication edge computing systems [8]. size, which is easy to carry and transport. Meanwhile, with At present, there have been some studies at home and great stability during low altitude flight, UAV can better abroad for computational resource allocation strategies in acquire images obscured by clouds and other objects, which edge computing environment. According to [9], the mixed cannot be captured by the satellite light-sensing remote integer and nonconvex optimization problem was decoupled controller and other aerospace photography [4]. Due to its 2 Journal of Robotics into two sub-problems of task offloading decision and re- source optimization, which were jointly optimized by using UAV V a designed iterative algorithm. ,e experimental results showed that the optimization performance is comparable to Wireless mm Wave that of the exhaustive search. However, this algorithm takes Link backhaul a longer time for optimization and the efficiency needs to be improved. ,e authors of [10] proposed a dynamic multi- win game model based on incomplete information to solve User the task offloading and edge resource allocation problems for multiple terminal users. Further optimization is needed for MEC Server multiobjective task offloading and resource allocation in complex network situations. ,e authors of [11] proposed a computing resource allocation scheme for MEC scenarios based on DRL networks. ,e cost of task offloading still Figure 1: System model. needs to be optimized. In order to investigate the rela- tionship between the cost of computation tasks and time and Both UAV and ground users have communication energy consumption, a computation offloading strategy was circuits and computing processors and are powered by proposed, which reduces the execution cost by collabora- batteries with a certain capacity. It is worth noting that tively allocating computational resources between mobile different users have different computing capacitates. ,e devices and edge servers [12]. ground base station is deployed with multiple built-in MEC With the continuous development of machine learning servers with high computing capacity, which can be regarded and UAV communication technologies, deep learning and as an access point [16, 17]. In this scenario, the user has UAV assistance have been gradually applied in computa- delay-sensitive computation tasks that require UAV to as- tional resource allocation. Authors in Ref. [13] proposed a sists in computation. ,e UAV is deployed with MEC multiobjective optimization algorithm based on the servers to help users to execute the task on itself or relay to MOACO algorithm combined with reinforcement learning, the ground base station for execution. which effectively improves the quality of service in edge computing by establishing an effective cognitive agent model to evaluate the resource allocation. However, with more 2.2. Communication Model. In the system model, users are iteration, the time efficiency needs to be improved. In the fixed on the ground and the position can be denoted as q, study [14], a multiagent reinforcement learning framework q � [x, y], where x and y are the horizontal coordinates of based on independent learners was proposed to achieve the user. UAV keeps flying at the same altitude H(H > 0) computation offloading and resource allocation for multiple and its position at i th timeslot can be expressed as private users in IoT edge computing networks. Authors in q � [x [i], y [i]], 0 ≤ i ≤ I. u u u Ref. [15] introduced agents into task offloading and pro- It is assumed that the wireless channel between the UAV posed a new framework for agent-enabled task offloading in and users is a Line of Sight (LOS) channel. ,erefore, the UAV-aided MEC to help users, UAV, and edge clouds channel power gain between the UAV and users can be perform computation task offloading. UAV can improve expressed as follows: wireless communication coverage to a certain extent, but the computation offloading strategy still needs further − 2 0 h[i] � δ d [i] � � � , i ∈ I, 0 (1) � �2 optimization. � � H + �q [i] − q� Based on the above analysis, a computational resource allocation strategy using the genetic algorithm in the UAV- where δ is the channel power gain within unit distance − 2 assisted edge computing environment is proposed to address (d � 1m) and d [n] is the horizontal distance between the the problem that the computing capacity of existing mobile UAV and the user at i th timeslot,i � 1, 2, . . . , I. ‖ · ‖ is the devices (computers, cell phones) cannot fully meet the users’ Euclidean norm. demand for communication quality. Based on the UAV- assisted edge computing system, a minimization model of total system energy consumption with jointly optimized 2.3. Computation Model UAV offloading ratio, user scheduling variables, and UAV 2.3.1. Local Execution. When users execute the task locally, trajectory is constructed. ,e minimization problem is the execution time can be calculated as follows: solved using an improved genetic algorithm to achieve a reasonable allocation of computational resources. 1 − η [i]ϑ [i]􏼁C D [i] loc n n n n (2) T [i] � , 2. System Model and Optimization Objective where η [i] is the scheduling variable of user n. ϑ [i] is the n n 2.1. System Model. As shown in Figure 1, (a) UAV-assisted offloading ratio of user n at i th time slot. D [i] denotes the MEC system consists of a cellular-connected UAV and bit size of the computation task of user n at ith timeslot. C multiple ground-requesting users. denotes the CPU frequency required when user n executes a Journal of Robotics 3 task of 1 bit. f denotes the local computing capacity of user η [i]ϑ [i]D [i]C UAV n n n n T [i] � , (6) n. UAV ,en, the energy consumption for local execution can be UAV written as follows: where f is the computing capacity of the UAV when serving user n. Hence, the energy consumption for the UAV- loc 2 (3) E [i] � c1 − η [i]ϑ [i]􏼁D [i]C f , n n n n n n assisted computation can be formulated as follows: UAV UAV2 where c is a coefficient that depends on the chip architecture. ′ (7) E [i] � c η [i]ϑ [i]D [i]C f , n n n n n n where c is a coefficient related to the chip architecture of 2.3.2. UAV-Assisted Computation. For UAV-assisted UAV. computation, it can be divided into two parts: the trans- mission process and the assisted computation. ,us, the UAV transmission time delay can be calculated as follows: 2.4. Optimization Objective. According to the aforemen- η [i]ϑ [i]D [i] tr n n n tioned UAV-assisted computation model, the correspond- T [i] � , (4) v [i] ing optimization problem is designed. It is assumed that user local execution and UAV-assisted computation are per- where v [i] is the transmission rate between user n and the formed simultaneously, and the UAV needs to perform the UAV at i th time slot. ,en, the transmission energy con- computation and return the results at the same time slot. sumption can be written as follows: Because the size of the returned results is much smaller than η [i]ϑ [i]D [i] that of the UAV-assisted computation task, the energy tr n n n E [i] � P , (5) n 0 consumption for returning results can be neglected [18, 19]. v [i] ,e objective is to minimize the total system energy con- where P is the rated transmission power of the UAV. sumption by jointly optimizing the UAV offloading ratio, For the UAV-assisted computation, in each timeslot, the the user scheduling variables and the UAV trajectory. ,us, UAV serves multiple ground users, the execution time of the optimization problem can be formulated as follows: UAV serving user n at ith timeslot for the assisted com- putation can be calculated as follows: I I N fly UAV tr loc Γ � min 􏽘 E [i] + 􏽘 􏽘 E [i] + E [i] + E [i] , (8) 􏼐 􏼑 n n n η [i],ϑ [i],q[i] { } n n i�1 i�1 n�1 s.t. q[1] � q[I], (9) ‖q[i + 1] − q[i]‖ ≤ V , i ∈ I, max 0≤ ϑ [i] ≤ 1, ∀n, i, (10) 􏽘 η [i] ≤ 1, ∀i (11) n�1 η [i] ∈ {0, 1}, ∀n, i, v [i] ≥ η [i] · ], ∀n, i, (12) n n I N loc 􏽘 􏽘 T [i]η [i] ≤ T, ∀n, i, (13) n n i�1 n�1 I N tr UAV 􏽘 􏽘􏼐T [i] + T [i]􏼑η [i] ≤ T, ∀n, i, (14) n n i�1 n�1 fly where E [i] is the energy consumption of the UAV during Equation (9) shows the constraint of the UAV trajectory. its flight. V is the maximum flight speed of the UAV. τ is Equation (10) defines the constraint of the offloading ratio. max the length of each timeslot. ] is the flight speed of the UAV. Equation (11) represents the user scheduling variables. 4 Journal of Robotics Equation (12) ensures the communication quality during Start offloading. Equation (13) represents the constraint of user local execution delay. Equation (14) represents the con- straint of user transmission delay and UAV-assisted com- Coding and initial putation delay. population generation 3. Computational Resource Allocation Strategy Calculate individual Mutation operation fitness Using Improved Genetic Algorithm Crossover operation Genetic algorithm is a relatively mature algorithm that searches for the global optimal solution by simulating the Meet the termination conditions? Selection operation process of biological evolution in nature. Its parallelism is suitable for distributed edge computing. Moreover, the genetic algorithm has good global search capability, so that End an appropriate solution set for resource allocation can be obtained by using the genetic algorithm with multiobjective Figure 2: Flow chart of basic genetic algorithm. optimization. ,erefore, this paper introduces an improved genetic algorithm, which can not only improve the search efficiency mutation operations, the value of genes converted to real but also obtain the further optimization objective by numbers may exceed the range of edge node numbers, so a establishing a dynamic adjustment model of crossover rate remainder approach is used [25, 26]. For example, the total and mutation rate, while reflecting the advantages of dis- number of edge nodes is defined to be 7 and thus the binary tributed execution of tasks. In the following sections, the code is {1101}, which will be converted to a decimal number implementation of resource scheduling of the proposed of 13. ,e remainder of 13 divided by 7 is 6, then the model is addressed based on the genetic algorithm. corresponding edge node number of this binary code is 6. Set the population size as S, the task sequence as S , the length as a, and the number of edge nodes in the random 3.1. Basic Genetic Algorithm. ,e basic genetic algorithm network topology graph as m. Each chromosome is com- was proposed by J. H. Holland of the University of Michigan, posed of a random combination of binary numbers whose USA, which is the basis for other genetic algorithms. length is 4 a, and every four binary numbers will be con- Moreover, other improved genetic algorithms are developed verted into a decimal number, which represents the number by adding new mechanisms. ,e flow chart of basic genetic of edge nodes. algorithm is shown in Figure 2. ,e basic genetic algorithm generates optimal individ- uals by simulating the selection, crossover, and mutation 3.2.2. Fitness Function. ,e fitness function is the key to operations in biological genetic and evolutionary processes evaluate the direction of population evolution in task and finds the optimal solution through an adaptive search scheduling of terminal users in edge computing. In order to process [20, 21]. ensure that the fitness function can remain meaningful when the value of optimization objective Γ is equal to 0, it is 3.2. Improved Genetic Algorithm defined as follows: − Γ f(Γ) � e . (15) 3.2.1. Encoding and Population Initialization. ,e proposed method adopts binary encoding where a chromosome corresponds to a solution of the optimization problem. ,e defined chromosomes are encoded using real numbers. 3.2.3. Genetic Operations. ,e selection operation is per- ,erefore, it is necessary to convert the genes in the formed by a roulette wheel. Assuming that the size of the chromosome to be represented in binary. Moreover, the population is S and the fitness value of an individual s is f , genes in the chromosome are all real numbers, which can be and p is the probability of an individual s being selected, the converted into a four-bit binary number and then be probability can be calculated as follows: replaced in the corresponding genes. Moreover, this is the encoding method used in the improved genetic algorithm p � . (16) [22, 23]. For example, a chromosome sequence is (3, 5, 8) j�1 j when encoded in real numbers, then the chromosome se- quence is converted into (0011, 0101, 1000) after binary ,e crossover operator is used to generate new indi- encoding. viduals by swapping the positions of chromosome genes of However, the objective function value is calculated the parent individuals. ,e traditional single-point crossover mainly based on real numbers, which requires reconverting method is used, in which a crossover point is set between any the binary-coded chromosome sequence into a real-coded two adjacent genes in each chromosome sequence S , then a − 1 chromosome sequence [24]. Moreover, after crossover and crossover point is arbitrarily selected in the first w genes, Journal of Robotics 5 and all genes after that crossover point are replaced with is 0.75, the mutation rate is 0.01, and the initial population each other [27]. size is 25. In the mutation operation, each chromosome sequence S uses a basic position mutation, where one gene is ar- 4.1. UAV Trajectory. ,e UAV trajectory optimized by the bitrarily selected in w number of genes and replaced by a improved genetic algorithm during different task execution random number between 1 to w. ,us, a new chromosome time T is shown in Figure 3, where the red circles indicate the is generated and the diversity of the population can be positions of ground users. ensured. From Figure 3, it can be seen that when T � 30 s, the UAV flies along the closed-loop trajectory and the flight range is small, the UAV is far from users when offloading 3.3. Steps of Task Scheduling. ,e steps of the improved genetic algorithm are as follows. data, and the communication quality is poor. With the increase of T, and when T � 60 s and T �120 s, the distance (1) Population initialization. ,e user submits a ter- between the UAV and users decreases and the flight range minal request and randomly sets the initial pop- expands, thus the communication quality improves. When T ulation g(0) based on the request. Set the iteration is large enough, the UAV continues to fly along the closed- counter t � 0, and g(t) indicates the tth generation loop trajectory and serves one user in several timeslots. In of population. addition, it can be found that the UAV has a larger flight (2) Calculate the fitness value. ,e fitness value of an range and is closer to users when T is larger, and the individual is calculated according to the fitness communication quality is better. function in the improved genetic algorithm. (3) Judge the termination condition. When the number 4.2. Convergence Performance of Algorithms under Different of evolutionary generations reaches the specified Computation Tasks. In order to demonstrate the conver- number of iterations Θ, the result should be given gence of the improved genetic algorithm, it is compared with and an edge computation task scheduling strategy is the traditional genetic algorithm, and the results are illus- found; otherwise, go to step (4). trated in Figure 4. (4) Selection operation. ,e selection operation of the It is depicted from Figure 4 that the proposed improved populations is performed using the roulette wheel. genetic algorithm executes more tasks within the same (5) Crossover operation. According to the crossover number of iterations, and when the number of iterations method in the designed improved genetic algorithm, exceeds 70, the amount of computation tasks tends to two individuals are randomly selected among the converge to 3450 M bit. Because it is optimized in encoding individuals produced by the selection operation and and genetic operations, the amount of computation in- are performed based on the crossover rate, so as to creases. However, the process is slightly more complicated, generate new individuals. resulting in longer processing time, and the number of it- erations that tend to be stable is about 10 times higher than (6) Mutation operation. According to the mutation that of the original genetic algorithm. On the whole, the method in the improved genetic algorithm, new proposed algorithm can achieve effective convergence individuals are generated by crossover operation and performance. the selected individuals among the new individuals are operated according to the mutation rate to generate new individuals. 4.3. Relationship between Total System Energy Consumption (7) Compare the fitness value. If the fitness value of the with Task Execution Time. To demonstrate the performance mutated individuals is smaller than that of parent of the total energy consumption of the proposed method, it individuals, they will be replaced by the parent is compared with those of the [9, 13, 15], and the results are generation. If it is greater than or equal to the fitness shown in Figure 5. value of the parent generation, the parent generation As can be seen in Figure 5, with the increase of task will be replaced. ,en, a new generation of indi- execution time, the total energy consumption of the system viduals g(i + 1) is generated. also increases. Meanwhile, the proposed UAV-assisted computational resource allocation method requires lower (8) Update the iteration counter. Set the iteration counter as t � t + 1, then go to step (2). energy consumption and has better performance than the other three methods. Moreover, the total energy con- sumption required by the proposed method is always the 4. Experiment and Analysis least while the task execution time is increasing. ,e total ,e UAV trajectory and the performance of the proposed energy consumption is 650 J when the execution time is improved genetic algorithm are demonstrated by numerical 110 s. As the proposed method executes the computation simulations. ,e system simulation parameters are set as task with the assistance of UAV and optimizes the UAV shown in Table 1. trajectory using the improved genetic algorithm, it can dynamically and timely adjust the computation tasks so as to ,e parameters of the improved genetic algorithm are set as follows. ,e number of iterations is 00, the crossover rate reduce the energy consumption. Reference [9] introduces 6 Journal of Robotics Table 1: System simulation parameters. System parameters Value Number of users N 10 Flight altitude H 150 m Channel power gain δ −75 dB CPU frequency C 1200 cycles/bit −27 c � c 10 UAV UAV computing capacity f 1400 Hz User local computing capacity f [200, 500] Hz User fixed transmission capacity P 50 dB·m 600 Maximum flight speed of UAV V 60 m/s max 10 30 50 70 90 110 Task execution time (s) Ref. [9] Ref. [15] Ref. [13] Proposed method Figure 5: Total energy consumption of the system under different methods. mixed integer and nonconvex optimization to obtain task offloading decision and resource optimization. However, the optimization method is traditional, the obtained resource 100 allocation strategy is not reasonable enough, and the total energy consumption of the system is higher than 1600 J. Reference [13] combines MOACO algorithm and rein- forcement learning for multiobjective resource allocation, 0 100 200 300 400 500 x (m) which can effectively improve the quality of users’ com- puting services, but the computational process is complex T=120s and the energy consumption is high. Similarly, [15] intro- T=60s duces UAV agents to assist in the offloading of computation T=30s tasks. However, it lacks an effective optimization algorithm Figure 3: UAV trajectory. to schedule its trajectory, so the energy consumption during UAV flight is high and its total system energy consumption is 62.5% higher than the proposed method. 4.4. Performance Comparison with Other Methods. In ad- dition, to further validate the advantage of the proposed method in term of execution time, it is compared with [9, 13, 15], and the results are depicted in Figure 6. It can be seen from Figure 6 that the task execution time increases continuously with more users, but the proposed method has the slowest increase and an obvious advantage in terms of execution time. ,e execution time is 17.5 s when the number of users is 50. ,e proposed method uses an improved genetic algorithm to optimize the UAV trajectory, it reduces energy consumption by lessening the blind flight time of the UAV and it offloads the tasks nearby to further reduce the transmission time. Compared with the proposed 1000 method, [15] lacks an effective algorithm to optimize the 0 50 100 150 200 250 UAV trajectory, resulting in a longer flight time and a higher Number of iterations overall execution time, which is similar to that of [13]. ,us, Improved genetic algorithm the maximum execution time is more than 25 s. ,e opti- Genetic algorithm mization method in [9] is simple and easy to implement, and the execution time is shorter when the number of users is Figure 4: Convergence performance of genetic algorithms before and after improvement. small. However, when the number of users increases, it is Calculate task quantity (Mbit) y (m) Total energy consumption of the system (J) Journal of Robotics 7 30 ,e UAV-assisted MEC system only considers the scenarios where UAV is connected with users or the ground. When the UAV only has limited computing capacity, the case that UAV collects users’ data and offloads tasks to the ground base station for assisted computing is also worth studying. 15 Data Availability ,e data used to support the findings of this study are in- cluded within the article. Conflicts of Interest ,e authors declare that they have no conflicts of interest regarding the publication of this paper. 10 20 30 40 50 Number of users References Ref. [9] Ref. [15] Ref. [13] Proposed method [1] X. Liu, J. Yu, J. Wang, and Y. Gao, “Resource allocation with Figure 6: Execution time of different methods. edge computing in IoT networks via machine learning,” IEEE Internet of <ings Journal, vol. 7, no. 4, pp. 3415–3426, 2020. [2] H. C. Ke, H. Wang, H. W. Zhao, and W. J. Sun, “Deep re- unable to handle the huge amount of data, resulting in the inforcement learning-based computation offloading and re- significant rise of the execution time. source allocation in security-aware mobile edge computing,” Wireless Networks, vol. 27, no. 5, pp. 3357–3373, 2021. [3] J. Feng, L. Liu, Q. Pei, F. Hou, T. Yang, and J. Wu, “Service 5. Conclusion characteristics-oriented joint optimization of radio and computing resource allocation in mobile-edge computing,” With the explosive growth of IoT devices, cloud computing IEEE Internet of <ings Journal, vol. 8, no. 11, pp. 9407–9421, can no longer meet the demand of some time-sensitive computation tasks. For this reason, a computational resource [4] L. Deng, P. Yang, W. Liu, L. Wang, S. Wang, and X. Zhang, allocation strategy using the genetic algorithm in the UAV- “NAAM-MOEA/D-Based multitarget firepower resource al- assisted edge computing environment is proposed. With the location optimization in edge computing,” Wireless Com- advantages of high mobility, low cost, and easy carrying, the munications and Mobile Computing, vol. 2021, Article ID UAV is introduced in the edge computing environment to 5579857, 14 pages, 2021. assist in task offloading. ,e flexible scheduling of the UAV [5] Z. Song, Y. Liu, and X. Sun, “Joint task offloading and resource enables efficient allocation of user tasks and reduces execution allocation for NOMA-enabled multi-access mobile edge computing,” IEEE Transactions on Communications, vol. 69, time. In order to optimize the flight trajectory of the UAV, the no. 3, pp. 1548–1564, 2021. problem of minimizing the total energy consumption of the [6] B. Baek, J. Lee, Y. Peng, and S. Park, “,ree dynamic pricing system is solved by applying an improved genetic algorithm. schemes for resource allocation of edge computing for IoT ,en, the optimal flight scheme can be obtained, that is, the environment,” IEEE Internet of <ings Journal, vol. 7, no. 5, optimal resource allocation strategy, which greatly reduces pp. 4292–4303, 2020. the system energy consumption and improves the trans- [7] Z. Wang, S. Zheng, Q. Ge, and K. Li, “Online offloading mission efficiency. ,e results based on the UAV simulation scheduling and resource allocation algorithms for vehicular platform show as follows: edge computing system,” IEEE Access, vol. 8, no. 9, pp. 52428–52442, 2020. (1) By real number encoding, selection, crossover, and [8] W. Zhan, C. Luo, G. Min, C. Wang, Q. Zhu, and H. Duan, mutation operations, the performance of the genetic “Mobility-Aware multi-user offloading optimization for algorithm in finding the optimal solution is effec- mobile edge computing,” IEEE Transactions on Vehicular tively improved, and the data size of computational Technology, vol. 69, no. 99, pp. 3341–3356, 2020. tasks can reach 3450 M bit when the number of it- [9] Y. Fu, X. Yang, P. Yang et al., “Energy-efficient offloading and erations exceeds 70. resource allocation for mobile edge computing enabled mission-critical internet-of-things systems,” EURASIP Jour- (2) ,e introduction of the UAV allows dynamic re- nal on Wireless Communications and Networking, vol. 2, no. 1, sponse to users’ computational demands and im- pp. 1–16, 2021. proves the efficiency of the computation offloading. [10] B. Wang and M. Li, “Resource allocation scheduling algo- When the number of users is 50, the execution time rithm based on incomplete information dynamic game for is 17.5 s, and at the same time, when the execution edge computing,” International Journal of Web Services Re- time is 110 s, the total energy consumption is 650 J, search, vol. 18, no. 2, pp. 1–24, 2021. which effectively reduces the system energy [11] Y. Zhang, M. Zhang, C. Fan, F. Li, and B. Li, “Computing consumption. resource allocation scheme of IOV using deep reinforcement Execution time (s) 8 Journal of Robotics learning in edge computing environment,” EURASIP Journal edge computing,” Science China (Information Sciences), vol. 63, no. 04, pp. 154–169, 2020. on Applied Signal Processing, vol. 21, no. 1, pp. 1–19, 2021. [12] S. Chen, Q. Li, M. Zhou, and A. Abusorrah, “Recent advances [27] M. Zeng and V. Fodor, “Energy minimization for delay constrained mobile edge computing with orthogonal and in collaborative scheduling of computing tasks in an edge computing paradigm,” Sensors, vol. 21, no. 3, pp. 779–787, non-orthogonal multiple access,” Ad Hoc Networks, vol. 98, no. 3, pp. 1–13, Article ID 102060, 2020. [13] V. Shanmuganthan, M. Khari, N. Dey, R. G. Crespo, and Y. H. Robinson, “Enhanced resource allocation in mobile edge computing using reinforcement learning based MOACO algorithm for IIOT,” Computer Communications, vol. 151, no. 4, pp. 355–364, 2020. [14] X. Liu, J. Yu, Z. Feng, and Y. Gao, “Multi-agent reinforcement learning for resource allocation in IoT networks with edge computing,” China Communications, vol. 17, no. 9, pp. 220– 236, 2020. [15] W. A. Rui, C. A. Yong, B. An, T. A. Alamoudi, and R. Nour, “Agent-enabled task offloading in UAV-aided mobile edge computing,” Computer Communications, vol. 149, no. 2, pp. 324–331, 2020. [16] S. Li, S. Lin, L. Cai, W. Li, and G. Zhu, “Joint resource al- location and computation offloading with time-varying fading channel in vehicular edge computing,” IEEE Transactions on Vehicular Technology, vol. 69, no. 3, pp. 3384–3398, 2020. [17] J. Yan, S. Bi, and Y. Zhang, “Offloading and resource allo- cation with general task graph in mobile edge computing: a deep reinforcement learning approach,” IEEE Transactions on Wireless Communications, vol. 19, no. 8, pp. 5404–5419, 2020. [18] Z. Yang, C. Pan, J. Hou, and M. Shikh-Bahaei, “Efficient resource allocation for mobile-edge computing networks with NOMA: completion time and energy minimization,” IEEE Transactions on Communications, vol. 67, no. 11, pp. 7771– 7784, 2019. [19] J. Zhao, Q. Li, Y. Gong, and Z. Ke, “Computation offloading and resource allocation for cloud assisted mobile edge computing in vehicular networks,” IEEE Transactions on Vehicular Technology, vol. 68, no. 8, pp. 7944–7956, 2019. [20] L. I. Shulei, D. Zhai, D. U. Pengfei, and T. Han, “Energy- efficient task offloading, load balancing, and resource allo- cation in mobile edge computing enabled IoT networks,” Science China Information Sciences, vol. 62, no. 2, pp. 1–3, [21] L. Yang, G. Xu, J. Ge, P. Liu, and X. Fu, “Energy-efficient resource allocation for application including dependent tasks in mobile edge computing,” KSII Transactions on Internet and Information Systems, vol. 14, no. 6, pp. 2422–2443, 2020. [22] Y. Guo, F. Liu, N. Xiao, and Z. Chen, “Task-based resource allocation bid in edge computing micro datacenter,” Com- puters, Materials and Continua, vol. 58, no. 2, pp. 777–792, [23] H. T. Chien, Y. D. Lin, C. L. Lai, and W. Chien-Ting, “End-to- End slicing with optimized communication and computing resource allocation in multi-tenant 5G systems,” IEEE Transactions on Vehicular Technology, vol. 69, no. 2, pp. 2079–2091, 2020. [24] Y. Chen, Z. Li, B. Yang, K. Nai, and K. Li, “A Stackelberg game approach to multiple resources allocation and pricing in mobile edge computing,” Future Generation Computer Sys- tems, vol. 108, no. 3, pp. 273–287, 2020. [25] M. Dai, Z. Su, Q. Xu, and N. Zhang, “Vehicle assisted computing offloading for unmanned aerial vehicles in smart city,” IEEE Transactions on Intelligent Transportation Systems, vol. 22, no. 3, pp. 1932–1944, 2021. [26] K. Wang, Z. Xiong, C. Lin, Z. Pan, and H. Shin, “Joint time delay and energy optimization with intelligent overclocking in

Journal

Journal of Robotics – Hindawi Publishing Corporation

Published: Apr 6, 2022

There are no references for this article.