In this section, we use simulations to evaluate the performance of the proposed LyDROO algorithm. All the computations are evaluated on a TensorFlow 2.0 platform with an Intel Core i5-4570 3.2GHz CPU and 12 GB of memory. We assume that the average channel gain h ̄i ̄ 􏰆3×108 􏰇de follows a path-loss model hi = Ad 4πfcdi , i = 1,··· ,N, where Ad = 3 denotes the antenna gain, fc = 915 MHz denotes the carrier frequency, de = 3 denotes the path loss exponent, and di in meters denotes the distance between the ith WD and the ES. hi follows an i.i.d. Rician distribution with line-of-sight link gain equal to 0.3h ̄ i . The noise power N0 = W υ0 with noise power spectral density υ0 = −174 dBm/Hz. Unless otherwise stated, we consider N = 10 WDs equally spaced with di = 120+15(i−1), for i = 1,··· ,N. The weight ci = 1.5 if i is an odd number and ci = 1 otherwise. The task data arrivals of all the WDs follow exponential distribution with equal average rate E [Ati ] = λi , i = 1, · · · , N . The values of the other parameters are listed in Table I, which are equal for all the WDs.，列出变量和他们怎么获得的

变量及其获取方式如下：

- h ̄i：平均信道增益。根据论文中所述的路径损耗模型，通过距离di计算得到。
- hi：第i个无线设备（WD）与边缘服务器（ES）之间的信道增益。根据论文中所述的路径损耗模型，通过距离di计算得到。
- N0：噪声功率。根据噪声功率谱密度和带宽W计算得到。
- υ0：噪声功率谱密度。根据论文中所述的参数值得到。
- ci：第i个无线设备（WD）的权重。根据论文中所述的规则计算得到。
- Ati：第i个无线设备（WD）到达的任务数据数量。根据论文中所述的规则，所有无线设备的任务数据到达率均相等。
- λi：第i个无线设备（WD）的任务数据到达率。根据论文中所述的规则，所有无线设备的任务数据到达率均相等。
- Ad：天线增益。根据论文中所述的参数值得到。
- fc：载波频率。根据论文中所述的参数值得到。
- de：路径损耗指数。根据论文中所述的参数值得到。
- di：第i个无线设备（WD）与边缘服务器（ES）之间的距离。根据论文中所述的规则计算得到。
- W：带宽。根据论文中所述的参数值得到。
- vu：信道噪声的方差。根据论文中所述的参数值得到。

这些参数的具体值和计算方式均在论文中给出。