基于混合粒子群算法的计算卸载成本优化

为了满足用户日益增长的计算密集型和时延敏感型服务需求，同时最小化计算任务的处理成本，在时延约束下，该文针对超密集异构边缘计算网络，构建了有关任务卸载、无线资源管理、计算资源块分配的联合优化问题。考虑到所规划的问题具有非线性和混合整数的形式，且为满足约束条件及提升算法收敛速率，通过改进分层自适应搜索(HAS)算法设计了混合粒子群优化 (HPSO)算法来求解所提出的问题。仿真结果表明，HPSO算法明显优于现有算法，能有效降低任务处理成本。

Computation Offloading Cost Optimization Based on Hybrid Particle Swarm Optimization Algorithm

In order to meet the ever-increasing computation-intensive and delay-sensitive service requirements of users, as well as minimizing the processing cost of computation tasks, an optimization problem of joint task offloading, wireless resource management, and computation resource block allocation are formulated for ultra-dense heterogeneous edge computing networks under users’ delay constraints. Such a formulated problem is in a nonlinear and mixed-integer form. In order to meet the constraints and improve the convergence speed of algorithm, a Hybrid Particle Swarm Optimization (HPSO) algorithm is developed by improving Hierarchical Adaptive Search (HAS) algorithm. The simulation results show that HPSO algorithm is superior to other benchmark algorithms under users’ delay constraints, and can reduce the task processing cost effectively.