Game Frame Time data files for Multi Service Provider scenario

To cater to many players in AR/VR applications in the MEC network (MEN), MEN is divided into regions by each Service Provider (SP), where each region consists of multiple MEC hosts. Every region has one server, called the Region Coordinator (RC), responsible for hosting the services to process the action data for players within that region. This architecture is decentralized as the processing of action data does not occur centrally, in the cloud. In such a multiplayer decentralized architecture, the response time for a user is the sum of a) Stage 1: the time taken by the task, whether offloaded or not, b) Stage 2: time to send the output of the task from the MEC host to the RC, c) Stage 3: the time to process the output of the tasks at the RC and the time to communicate the result to the player. The maximum value of the sum of the duration for stages 1, 2, and 3 constitutes a Game Frame Time (GFT). GFT represents the maximum response time experienced by any player. An MEC Orchestrator/Controller decides which MEC hosts are RCs, which MEC hosts a UE needs to be associated with, and whether to offload UE tasks while reducing GFT.

We formulated the Optimal Profit Problem (OPP) problem to maximize the minimum profit of all SPs with a constraint on the upper limit of the GFT. We also proposed the heuristic Decentralized multi-SP Profit maximization and Game Frame Time reduction (DmPmGFTr). Text file "Files_description_rGFT.txt" contains the description about files available in folder "DmPmGFTr". These data files are used to run the heuristic experiments. Text file "Files_description_rGFT-OPT.txt" contains the description about files available in folder "DmPmGFTr-OPP" . These files are used to run the experiments when comparing the DmPmGFTr and OPP. Note if any parameter values is missing in this raw data, it may be referred from Research paper.

为适配多接入边缘计算网络（Multi-Access Edge Computing Network, MEN）中AR/VR应用的众多玩家，各服务提供商（Service Provider, SP）将MEN划分为多个区域，每个区域包含多台多接入边缘计算（Multi-Access Edge Computing, MEC）主机。每个区域设有一台名为区域协调器（Region Coordinator, RC）的服务器，负责部署服务以处理该区域内玩家的动作数据。该架构采用去中心化设计，动作数据的处理无需在云端集中完成。在此类多人去中心化架构中，用户的响应时间由三部分构成：a）阶段1：任务（无论是否被卸载）的处理耗时；b）阶段2：将任务输出从MEC主机传输至区域协调器的耗时；c）阶段3：在区域协调器处处理任务输出，并将结果回传给玩家的耗时。阶段1、2、3的总时长的最大值即为游戏帧时间（Game Frame Time, GFT），其代表任意一名玩家所经历的最长响应时间。多接入边缘计算编排器/控制器负责决策哪些MEC主机将作为区域协调器、用户设备（User Equipment, UE）需关联哪些MEC主机，以及是否需要卸载用户设备任务以缩减游戏帧时间。我们构建了最优收益问题（Optimal Profit Problem, OPP），旨在实现所有服务提供商的收益最小值最大化，同时以游戏帧时间上限作为约束条件。此外，我们还提出了启发式去中心化多服务提供商收益最大化与游戏帧时间缩减算法（Decentralized multi-SP Profit maximization and Game Frame Time reduction, DmPmGFTr）。文本文件"Files_description_rGFT.txt"对"DmPmGFTr"文件夹中的可用文件进行了说明，这些数据文件用于运行启发式实验。文本文件"Files_description_rGFT-OPT.txt"对"DmPmGFTr-OPP"文件夹中的可用文件进行了说明，这些文件用于开展对比DmPmGFTr与OPP的实验。请注意，若本原始数据中存在参数值缺失的情况，可参考相关研究论文获取对应参数。