无线供能移动边缘计算中联合定价和资源分配的双层博弈算法

A bi-level game algorithm for joint pricing and resource allocation in wireless powered mobile edge computing

关键词:  双层优化  多目标进化优化  博弈论  移动边缘计算  优化算法

Key words:  Bi-level optimization  Multi-objective evolutionary optimization  Game theory  Mobile edge computing  Optimization algorithm

摘要:

针对多服务器多用户的无线供能移动边缘计算系统计算复杂以及能量传输等特性,为了提高边缘服务器供应商与用户收益,同时对任务卸载与资源分配进行合理规划,提出了一个双层优化问题模型。上层由边缘服务器供应商负责计算资源和能源价格的优化,而下层则关注用户基于这些价格调整任务卸载、任务传输功率和资源分配方案。通过简化和转化下层博弈优化问题,提出了一种双层梯度优化算法(BIGDA)。该算法利用差分进化算法在上层优化资源价格,并在下层运用梯度法优化用户的任务卸载、传输功率及资源分配策略。将该算法与进行微调后的两种对比算法BIDE与BIPSO比较,仿真实验证明,该算法在解决无线供能移动边缘计算中的资源与能量定价问题上展现出高效性和实用性,有效提升了系统性能并优化了资源配置。

Abstract:

In response to the computational complexity and energy transmission characteristics of wireless powered mobile edge computing systems with multiple servers and users, a bilevel optimization problem model is proposed to enhance the profits of edge server providers and users, while ensuring rational planning of task offloading and resource allocation. The upper level involves optimization of computing resource and energy prices by edge server providers, while the lower level focuses on users adjusting task offloading, task transmission power, and resource allocation based on these prices. By simplifying and transforming the lower-level game optimization problem, a bilevel gradient optimization algorithm (BIGDA) is proposed. This algorithm utilizes differential evolution algorithm for upper-level optimization of resource prices, and employs gradient descent for lower-level optimization of users' task offloading, transmission power, and resource allocation strategies. Comparisons with two contrastive algorithms, BIDE and BIPSO, after fine-tuning, are conducted through simulation experiments, demonstrating that the proposed algorithm exhibits efficiency and practicality in addressing resource and energy pricing issues in wireless powered mobile edge computing. It effectively enhances system performance and optimizes resource allocation.