基于强化学习的多小区协同干扰协调仿真数据集

本数据集含有五个数据文件，数据量2.54G。其中（1）QuaDRiGa生成基于D-DDQN在cell-free场景下的热点干扰避免仿真系统数据集，本数据集为仿真数据，包括用户信道数据，维度分别是（用户数+测点数，基站数，天线数，信道时隙）和干扰避免性能分析数据，例如用户速率分析；（2）用户密集型无蜂窝网络中公平调度与波束成形的分布式联合设计仿真系统数据集为仿真数据，包括用户信道数据，维度分别是（时隙数，用户数，天线数）和调度及预编码性能分析数据，例如用户速率分析；（3）QuaDRiGa生成基于低维离散软演员评论家在cell-free场景下的混合预编码仿真系统数据集，本数据集为仿真数据，包括用户信道数据，维度分别是（时隙数，用户数，天线数）和混合预编码性能分析数据，例如用户速率分析;（4）基于集中-分布式深度强化学习框架的无蜂窝网络用户关联、波束成形与功率分配联合优化算法的仿真训练数据集，本数据集为所提算法及所用基线算法在仿真训练过程中产生的数据，包括用户关联动作、波束成形动作、功率分配动作、高级函数奖励、低级函数奖励、用户速率数据、用户服务质量数据和网络能效数据；（5）仿真平台小区间协同干扰消除数据集，本数据集为小区间干扰协调仿真系统对应的仿真数据，包括仿真中使用的用户信道数据、各用户关联的基站索引号以及对应的长时调度结果，即不同方案长时调度各用户平均速率性能分析数据。This dataset contains 5 data files with a total size of 2.54 GB. The specific contents are as follows: (1) Dataset for hotspot interference avoidance simulation system in cell-free scenarios based on D-DDQN generated by QuaDRiGa: This is simulation data including user channel data with dimensions (number of users + number of measurement points, number of base stations, number of antennas, channel time slots), and interference avoidance performance analysis data such as user throughput analysis. (2) Dataset for distributed joint design of fair scheduling and beamforming in user-dense cell-free networks: This is simulation data including user channel data with dimensions (number of time slots, number of users, number of antennas), and scheduling and precoding performance analysis data such as user throughput analysis. (3) Dataset for hybrid precoding simulation system in cell-free scenarios based on low-dimensional discrete Soft Actor-Critic generated by QuaDRiGa: This is simulation data including user channel data with dimensions (number of time slots, number of users, number of antennas), and hybrid precoding performance analysis data such as user throughput analysis. (4) Simulation training dataset for joint optimization of user association, beamforming and power allocation in cell-free networks based on a centralized-distributed deep reinforcement learning framework: This dataset contains data generated during the simulation training process of the proposed algorithm and its employed baseline algorithms, including user association actions, beamforming actions, power allocation actions, high-level function rewards, low-level function rewards, user throughput data, user quality of service data and network energy efficiency data. (5) Dataset for inter-cell cooperative interference cancellation simulation platform: This dataset is simulation data corresponding to the inter-cell interference coordination simulation system, including user channel data used in the simulation, base station index numbers associated with each user, and corresponding long-term scheduling results, i.e., performance analysis data of average user throughput for each user under long-term scheduling across different schemes.

搜集汇总

背景与挑战

背景概述

该数据集包含五个仿真数据文件，总数据量为2.54GB，涉及基于强化学习的多小区协同干扰协调技术。数据内容涵盖用户信道、干扰避免、公平调度、波束成形、混合预编码以及用户关联与功率分配等性能分析，主要用于无蜂窝网络场景下的算法仿真与优化。