Trajectory-Aware Rate Adaptation for Aerial Networks Simulation Results

<strong>Introduction</strong> Even though the concept of ubiquitous wireless connectivity is becoming a reality, there are scenarios where wireless communications coverage is insufficient or does not exist. Considering natural and man-made disaster scenarios, communications infrastructures may be damaged and become unavailable. In temporary crowded events, the existing infrastructure may not have been designed to cope with the additional traffic demand, resulting in overload. In maritime scenarios, environmental monitoring activities using autonomous vehicles will take place in offshore zones, typically not in range of existing onshore communications infrastructures. Flying networks, composed of Unmanned Aerial Vehicles (UAV), are emerging as a flexible and cost-effective solution to provide on-demand wireless connectivity in such scenarios. UAVs have the possibility to operate virtually everywhere, and the growing payload capacity makes them suitable platforms to carry wireless communications hardware, playing the role of mobile base stations, access points or relay nodes. A flying network may typically be composed of a fleet of UAVs, organized in a multi-tier topology with so-called Flying Edge Nodes (FENs) and Flying Gateways (FGWs) [1]. FENs can play the role of Flying Access Points that provide the access network to the users on the ground, or the role of Flying Sensor Nodes that can perform video surveillance missions. The FENs forward the traffic to the FGWs, that act as relay nodes and are responsible for forwarding the traffic to/from the backhaul (BKH) network and ultimately to/from the Internet. The flying network concept brings up new challenges. The flying nodes need to be properly positioned and their wireless link configuration dynamically adjusted in order to ensure the Quality of Service (QoS) expected by the end users. In addition, these scenarios are typically highly unpredictable due to the varying locations as well as the concentration/dispersion of end-users and their movements regarding direction and velocity - e.g., vehicles or pedestrians. Therefore, a static wireless link configuration and UAV positioning are not adequate. State of the art work has been mainly focused on the optimal positioning of the flying nodes, having most of the wireless link parameters statically configured with default values. The Rate Adaptation challenge is well-known in fixed or low mobility IEEE 802.11 networks, and Minstrel High Throughput (HT) [2] is the default Wi-Fi rate adaptation algorithm used in the Linux kernel since the IEEE 802.11n version. However, few works propose solutions designed to consider the characteristics of other communications environments, such as flying and vehicular networks [3]. To the best of our knowledge, solutions that use the node trajectory information to predict the wireless channel conditions and perform rate adaptation are yet to be developed. The main contribution of this paper is the Trajectory-Aware Rate Adaptation (TARA) algorithm. TARA takes advantage of knowing the trajectory of all nodes in the flying network to estimate future changes in the wireless link quality and perform rate adaptation accordingly. The network performance improvement achieved with TARA was evaluated using ns-3 [4]. The simulation results presented in this dataset show significant throughput gains when compared with conventional rate adaptation algorithms. <strong>Folder Organization</strong> The following dataset presents the results of the TARA Paper, organized in different folders for each Rate Adaptation Algorithm, as well as the random seeds that were used to obtain such results: <strong>Naming Convention:</strong> Rate Adaptation Algorithm<strong> </strong> <strong>tara </strong>– Trajectory-Aware Rate Adaptation <strong>min </strong>– MinstrelHTWifiManager <strong>id </strong>– IdealWifiManager <strong>Folder Content: </strong> <em>distances.csv - </em><strong>Distances between nodes</strong> Column 1 – <strong>Simulation Time </strong>(seconds) Column 2 – <strong>Distance between BKH and FGW</strong> (meters) Column 3 – <strong>Distance between FEN and FGW </strong>(meters) <em>positions.csv</em> <em>- </em><strong>Current 3D position of nodes</strong> Column 1 – <strong>Simulation Time </strong>(seconds) Column 2 – <strong>BKH x </strong>(meters) Column 3 – <strong>BKH y </strong>(meters) Column 4 – <strong>BKH z </strong>(meters) Column 5 – <strong>FEN x </strong>(meters) Column 6 – <strong>FEN y </strong>(meters) Column 7 – <strong>FEN z </strong>(meters) Column 8 – <strong>FGW x </strong>(meters) Column 9 – <strong>FGW y </strong>(meters) Column 10 – <strong>FGW z </strong>(meters) <em>throughput.csv</em> - <strong>Link Specific Throughput, at MAC layer level</strong> Column 1 – <strong>Simulation Time </strong>(seconds) Column 2 – <strong>Relay Link (BKH - FGW), Throughput measured in BKH </strong>(Mbit/second) Column 3 – <strong>Access Link (FEN - FGW), Throughput measured in FEN </strong>(Mbit/second) Column 4 – <strong>Relay Link (BKH - FGW), Throughput measured in FGW </strong>(Mbit/second) Column 5 – <strong>Access Link (FEN - FGW), Throughput measured in FGW </strong>(Mbit/second)引言 尽管泛在无线互联的概念正逐步成为现实，但仍存在无线通信覆盖不足或完全无覆盖的场景。以自然与人为灾害场景为例，通信基础设施可能遭到损毁而无法使用；在临时大型集会活动中，现有通信基础设施往往未针对激增的流量需求进行设计，进而引发网络过载；在海事场景中，使用自主无人平台开展的环境监测活动通常位于近海区域，超出了现有陆上通信基础设施的覆盖范围。由无人机（Unmanned Aerial Vehicle, UAV）组成的飞行网络正逐渐成为一种灵活且高性价比的解决方案，可在上述场景中按需提供无线互联服务。无人机几乎可在任意区域部署，且有效载荷能力的不断提升使其成为搭载无线通信硬件的理想平台，可充当移动基站、接入点或中继节点。飞行网络通常由多架无人机组成，采用多层拓扑架构，包含所谓的飞行边缘节点（Flying Edge Nodes, FENs）与飞行网关（Flying Gateways, FGWs）[1]。飞行边缘节点可作为飞行接入点，为地面用户提供接入网络；亦可作为飞行传感节点，执行视频监控任务。飞行边缘节点将流量转发至飞行网关，后者作为中继节点，负责将流量往返传输至回传（Backhaul, BKH）网络，最终接入互联网。 飞行网络的概念也带来了全新的挑战：飞行节点需要进行合理部署，并动态调整无线链路配置，以确保终端用户所需的服务质量（Quality of Service, QoS）。此外，由于终端用户的位置不断变化、聚集/分散状态随时间波动，且其移动方向与速度存在差异（例如车辆或行人），这类场景通常具有高度的不可预测性。因此，固定的无线链路配置与无人机部署方案并不适用。 现有研究主要聚焦于飞行节点的最优部署，且多数无线链路参数采用默认静态配置。速率适配问题在固定或低移动性的IEEE 802.11网络中已是公认的挑战，自IEEE 802.11n标准以来，Minstrel高吞吐量（Minstrel High Throughput, HT）[2]便是Linux内核中默认采用的Wi-Fi速率适配算法。然而，鲜有研究提出针对飞行网络、车载网络等特殊通信环境的适配方案[3]。据我们所知，目前尚未有研究提出利用节点轨迹信息预测无线信道状态并据此开展速率适配的解决方案。 本文的核心贡献为轨迹感知速率适配（Trajectory-Aware Rate Adaptation, TARA）算法。TARA通过获取飞行网络中所有节点的轨迹信息，预测无线链路质量的未来变化，并据此执行速率适配。本文采用ns-3[4]模拟器对TARA带来的网络性能提升进行了评估。本数据集所呈现的仿真结果表明，相较于传统速率适配算法，TARA可显著提升系统吞吐量。 文件夹组织 本数据集收录了TARA相关论文的仿真结果，按照不同速率适配算法分为多个文件夹，同时包含了生成上述结果所使用的随机种子： 命名规则： - tara：轨迹感知速率适配（Trajectory-Aware Rate Adaptation） - min：MinstrelHTWi-Fi管理器 - id：理想Wi-Fi管理器（IdealWifiManager） 文件夹内容： - distances.csv：节点间距离 第1列：仿真时长（单位：秒） 第2列：回传节点与飞行网关间的距离（单位：米） 第3列：飞行边缘节点与飞行网关间的距离（单位：米） - positions.csv：节点当前三维位置 第1列：仿真时长（单位：秒） 第2列：回传节点x轴坐标（单位：米） 第3列：回传节点y轴坐标（单位：米） 第4列：回传节点z轴坐标（单位：米） 第5列：飞行边缘节点x轴坐标（单位：米） 第6列：飞行边缘节点y轴坐标（单位：米） 第7列：飞行边缘节点z轴坐标（单位：米） 第8列：飞行网关x轴坐标（单位：米） 第9列：飞行网关y轴坐标（单位：米） 第10列：飞行网关z轴坐标（单位：米） - throughput.csv：MAC层链路专属吞吐量 第1列：仿真时长（单位：秒） 第2列：回传链路（回传节点-飞行网关）：回传节点侧测得的吞吐量（单位：Mbit/秒） 第3列：接入链路（飞行边缘节点-飞行网关）：飞行边缘节点侧测得的吞吐量（单位：Mbit/秒） 第4列：回传链路（回传节点-飞行网关）：飞行网关侧测得的吞吐量（单位：Mbit/秒） 第5列：接入链路（飞行边缘节点-飞行网关）：飞行网关侧测得的吞吐量（单位：Mbit/秒）