Wireless Video Streaming for Ultra-low-power Cameras
Wireless Video Streaming for Ultra-low-power Cameras
复制标题
适用于超低功耗相机的无线视频流
DOI：
--
复制
发表时间：
2018
期刊：
影响因子：
--
通讯作者：
Joshua R. Smith
中科院分区：
文献类型：
--
作者：
Mehrdad Hessar;Saman Naderiparizi;Ye Wang;Ali Saffari;Shyamnath Gollakota;Joshua R. Smith
传统上，无线视频流被认为是非常耗电的操作。现有方法分别优化相机和通信模块以最小化它们的功耗。然而，设计视频流传输设备需要功耗硬件组件和视频编解码算法，这使得无电池视频流传输目前不可行。现有的射频供电的无线摄像头原型需要数十分钟的时间来进行大量的工作循环，以捕获、处理和传输单个帧。自供电摄像头可以每隔几秒钟捕捉一次图像，但不具备无线传输视频的能力。为了理解这种情况，让我们看看视频流传输设备1（a）中的不同组件：光学透镜、视频压缩和通信。光学透镜是连接到放大器和ADC的光电二极管阵列，用于将模拟像素转换为数字值。然后，视频编解码器执行帧压缩以压缩视频，然后在无线介质上发送视频。现有的方法分别优化相机和通信模块，以最大限度地减少它们的功耗。然而，设计视频流传输设备需要功耗硬件组件和视频编解码器算法，这些硬件组件和视频编解码器算法将相机和通信模块连接起来。我们提出超低功率视频流传输设备1（B）的设计。我们创建了不使用放大器、ADC和AGC的"模拟"视频反向散射系统。在高水平上，我们将模拟像素从光电二极管直接馈送到反向散射硬件。我们通过将天线连接到光电二极管阵列来实现这一点，光电二极管阵列的输出电压/阻抗随像素值而变化;因此，消除了功耗高的硬件组件，包括放大器、AGC和ADC。这样的方法将具有附加的益处，即视频质量随着变化的无线信道而平滑地缩放，而不需要显式的速率适配。我们在[1，2]中提供了我们的视频流架构的更多细节。我们使用CentEye的112 × 112灰度随机像素访问相机在超低功耗FPGA平台上实现了反向散射设计的原型，该相机提供对单个模拟像素的读出访问。我们的视频流媒体设备原型的功耗低至2.36 mW，同时以13 fps的速度流媒体直播视频。我们的接入点（AP）由两个组件组成。采用RTL2832U软件无线电接收来自标签的反馈信号，Semtech SX 1232和SE2435L-EK5功放产生辅助信号。我们使用与SDR接口的Python脚本演示了视频帧的实时显示。我们评估我们的原型在不同的条件下，研究其性能在不同的房间照明条件和不同的距离接入点。我们可以以每秒7 - 13帧的速度在距离AP最远27英尺的地方进行流媒体传输。我们在下面的链接中展示了我们的实时演示视频。
Wireless video streaming has traditionally been considered an extremely power-hungry operation. Existing approaches optimize the camera and communication modules individually to minimize their power consumption. However, designing a video streaming device requires power-consuming hardware components and video CODEC algorithms which makes battery-free video streaming currently infeasible. Existing RF-powered wireless camera prototypes require extensive duty-cycling on the order of tens of minutes, to capture, process and communicate a single frame. Self-powered cameras can capture an image once every few seconds, but do not have the capability to stream video wirelessly. To understand this case, let us look at the different components in a video-streaming device 1(a): optical lens, video compression and communication. Optical lens is an array of photo-diodes connected to amplifiers and an ADC to translate the analog pixels into digital values. A video CODEC then performs frame compression to compress video, which is then transmitted on the wireless medium. Existing approaches optimize the camera and communication modules separately to minimize their power consumption. However, designing a video streaming device requires power consuming hardware components and video CODEC algorithms that interface the camera and the communication modules. We present the design of an ultra-low-power video streaming device 1(b). We create "analog" video backscatter system that does not use amplifiers, ADCs and AGCs. At a high level, we feed analog pixels from the photo-diodes directly to the backscatter hardware. We achieve this by connecting the antenna to an array of photo-diodes whose output voltage/impedance varies as a function of the pixel value; thus, eliminate power-hungry hardware components including amplifiers, AGCs and ADCs. Such an approach would have the added benefit that the video quality scales smoothly with a varying wireless channel, without the need for explicit rate adaptation. We present our video streaming architecture with more details in [1, 2]. We implement a prototype of our backscatter design on an ultra-low power FPGA platform using a 112 × 112 gray-scale random pixel access camera from CentEye, which provides readout access to the individual analog pixels. The prototype of our video streaming device burns as low as 2.36 mW while streaming live video at 13 fps. Our access point (AP) consist of two components. We use RTL2832U SDR to receive backsactter signal from the tag and Semtech SX1232 and SE2435L-EK5 power amplifier to generate helper signal. We demonstrate real-time display of video frames using Python scripts which interfaces with the SDR. We evaluate our prototype under different conditions to study its performance under different room lighting conditions and different distances from the access point. We can stream at 7-13 frames per second at distances of up to 27 feet from the AP. We show a video of our real-time demonstration in following link.