In recent years, UAVs have received much attention in both the military and civilian fields for monitoring, emergency relief and searching tasks. UAVs are considered a new technology to obtain data at high altitudes when equipped with sensors. This technology is vital to the success of next-generation monitoring systems, which are expected to be reliable, real-time, efficient and secure. However, due to the bandwidth limitations in UAV-aided networks, the size of the transmitted data is a crucial factor for real-time media data transmission requirements, especially for national defense. To address this issue, in this article, we propose a realtime end-to-end media data transmission mechanism with an unsupervised deep neural network. The proposed mechanism transmutes the media data captured by UAVs into latent codes with a predefined constant size and transmits the codes to the ground console station (GCS) for further reconstruction. We use a real-word dataset containing millions of samples to evaluate the proposed mechanism which achieves a high transmission ratio, low resource usage and good visual quality.

中文翻译：

---

REMT：无人机辅助网络中的实时端到端媒体数据传输机制


近年来，无人机在军事和民用领域都受到广泛关注，用于监控、紧急救援和搜索任务。无人机被认为是一种在配备传感器时在高空获取数据的新技术。这项技术对于下一代监控系统的成功至关重要，预计下一代监控系统将是可靠、实时、高效和安全的。然而，由于无人机辅助网络的带宽限制，传输数据的大小是实时媒体数据传输需求的关键因素，特别是对于国防而言。为了解决这个问题，在本文中，我们提出了一种具有无监督深度神经网络的实时端到端媒体数据传输机制。所提出的机制将无人机捕获的媒体数据转换为具有预定义恒定大小的潜在代码，并将代码传输到地面控制台站（GCS）以进行进一步重建。我们使用包含数百万个样本的真实数据集来评估所提出的机制，该机制实现了高传输率、低资源占用和良好的视觉质量。