In this work, we focus our attention on real-time multimedia flows from Unmanned Aerial Vehicless (UAV) to the ground, presenting and analysing the data collected in field trials during a real testbed. The objective is assessing whether a video feed of reasonable quality can be provided to the pilot of an UAV to enable Beyond Visual Line of Sight (BVLoS) operations, by exploiting the multiple cellular operators available in the area. Three cellular networks have been jointly used in a multihoming/multipath setup, leveraging the variable coverage offered in both urban and suburban environments. Taking into account both Quality of Service (QoS) and Quality of Experience (QoE) metrics, the target parameters measured in this testbed are: latency, packet error rate, and video quality, which accounts for frames integrity, continuity, and fluidity. Data collected on the field allow to evaluate both QoS and QoE in the presence of a multipath architecture, showing how the latter, in the presence of network diversity, offer the possibility to improve the QoE at the receiver. We also design a framework to characterize the error model and to map it into a QoE model, therefore providing an analytical characterisation of a multipath channel.

在这项工作中，我们将注意力集中在从无人驾驶飞行器 (UAV) 到地面的实时多媒体流上，展示和分析在真实测试台期间现场试验中收集的数据。目标是通过利用该地区可用的多个蜂窝运营商来评估是否可以向无人机的飞行员提供合理质量的视频馈送，以实现超视距 (BVLoS) 操作。三个蜂窝网络已在多宿主/多路径设置中联合使用，利用城市和郊区环境中提供的可变覆盖范围。考虑到服务质量 (QoS) 和体验质量 (QoE) 指标，该测试台中测量的目标参数是：延迟、数据包错误率和视频质量，这些参数考虑了帧的完整性、连续性和流动性。现场收集的数据允许在存在多径架构的情况下评估 QoS 和 QoE，显示后者如何在存在网络多样性的情况下提供改善接收器 QoE 的可能性。我们还设计了一个框架来表征误差模型并将其映射到 QoE 模型中，从而提供多径信道的分析表征。