混合线性依赖和独立子任务的卸载与服务缓存计算资源分配

文章在设备到设备(device-to-device, D2D)辅助的单边缘服务器多用户移动设备场景下, 研究混合线性依赖和独立子任务的卸载与服务缓存计算能力分配问题, 目标为最小化能耗与时延的加权和。首先将该问题模型化为非凸问题, 接着提出混合任务两层优化(mixed task two-layer optimization, MTTLO)算法。该算法第1层通过固定各个设备或边缘服务器上计算能力的分配, 将混合线性依赖和独立子任务的非凸问题使用部分固定最优成本卸载算法得到优先级, 并求出任务卸载策略; 第2层使用卡罗需-库恩-塔克(Karush-Kuhn-Tucker, KKT)条件求出固定任务卸载策略下设备或边缘服务器服务缓存的计算能力的封闭解。实验结果表明, MTTLO算法优于其他基准算法, 能够有效减少系统的能耗与时延加权和, 验证了算法的有效性。

In device-to-device (D2D) assisted single-edge server multi-user mobile device scenario, this paper studies the offloading and service cache computing power allocation for mixed linearly dependent and independent subtasks, aiming to minimize the weight sum of energy consumption and delay. Firstly, the problem is modeled as a non-convex problem, and then the mixed task two-layer optimization (MTTLO) algorithm is proposed. In the first layer of MTTLO algorithm, the computing power allocation on each device or edge server is fixed, the non-convex problem of mixed linearly dependent and independent subtasks is given priority by using the partial fixed optimal cost offloading algorithm, and the task offloading strategy is obtained. The second layer uses the Karush-Kuhn-Tucker (KKT) conditions to find the closed solution of computing power of device or edge server service cache under fixed task offloading strategy. Simulation results show that the proposed MTTLO algorithm is superior to other benchmark algorithms, and can effectively reduce the weight sum of system energy consumption and delay, which verifies the effectiveness of the algorithm.