A Hybrid Architecture for Video Transmission

Transcription

1 2017 Asia-Pacific Engineering and Technology Conference (APETC 2017) ISBN: A Hybrid Architecture for Video Transmission Qian Huang, Xiaoqi Wang, Xiaodan Du and Feng Ye ABSTRACT With the advance of hardware and network technologies, video transmission becomes increasingly popular, especially through mobile networks. Take different networks and displays into account, a hybrid video transmission architecture is presented. Firstly, captured videos are compressed using a H.264/AVC hard coding module, packed with RTP/RTCP protocols, and sent based on UDP sockets. Secondly, a cloud server is built to forward video streams to end users. To improve the security, Ngrok, an open source reverse proxy project, is included. Finally at the client side, video data are unpacked and decompressed, and then post-processed according to the types of end devices. Experimental results show that no obvious time delays are observed, whereas the transmission bandwidths and packet loss rates are acceptable. INTRODUCTION With the development of hardware and the reduction of cost, an increasing amount of video data are compressed [1] and transmitted over 3G/4G networks. According to the market demand, high video quality and low bandwidth transmission become two key problems of real-time video transmission. Intelligent mobile monitoring terminals, compared with traditional ones, have the advantages of light weight, flexibility, and good user experience. Therefore, it is quite necessary to build a video system on mobile platforms such as Android and IOS. However, in contemporary society, many video systems are PC-oriented. In order to be compatible with the PC platform, we propose a hybrid architecture for video transmission. After capturing and encoding, a cloud server is built to forward video streams through secure channels. Taking different screen sizes and computing capabilities into account, video post-processing is involved when the streams arrive at end devices. The proposed architecture is reliable since the cloud Qian Huang 1,*, Xiaoqi Wang 2, Xiaodan Du 3, and Feng Ye 1 1 College of Computer and Information, Hohai University, Nanjing , P. R. China 2 Nanjing Research Institute of Information Technology, Nanjing , P. R. China 3 The 28th Research Institute of China Electronics Technology Group Corporation, Nanjing , P. R. China * Corresponding author: huangqian@hhu.edu.cn. This work is partly supported by the National Natural Science Foundation of China under Grant No , and , the Fundamental Research Funds of China for the Central Universities under Grant No. 2009B21614 and 2013B

2 server can capture and analyze traffic flows on all the channels. In addition, PC users can directly watch live videos through a web browser. RELATED WORKS J. Dong et al. [2] proposed a solution for live video based on Android system, which was shown to be feasible for H264 soft decoding on Android based on FFmpeg. H.264 encoding and the RTP protocol were discussed in detail, since the paper was focusing on the transmission and the decoding process of H.264 streams. However, the authors talked about only the client side of a whole video system, and neglected the server side implementation. L. Cai [3] presented a scheme of video web surveillance system based on the B/S mode. Therefore, important functions are included at the server side in the video surveillance system, including: Video Capturing and Encoding. Video Packing withe RTP/RTCP. Packet Sending to the Web Server of Boa. This work introduced a relatively complete video transmission system with minor time delay and basically smooth subjective quality. In addition, the system can be tested through a web browser. Unfortunately, the external network transmission quality cannot be guaranteed. THE PROPOSED SYSTEM Taking into account the advantages and disadvantages of the traditional system architecture, a hybrid architecture is presented in this paper, i.e. a B/C/S architecture. The system with a three-layered structure can no only be used for mobile phones, but also for web browsers on PC. With the serious shortage of IPv4, many previous servers cannot be directly used as the cloud server. So we use the proxy server architecture to solve the problem. The proposed architecture can be divided into the following three layers: the first layer is the server; the second layer is the proxy server; and the third layer is the users, which is B/C/S. The designing principle of the system is shown in Fig. 1 [4]. Internet Wifi Computer Wifi camera ARM Processor Wifi Router Cloud Router Phone 3G Phone server cloud server client/browser Figure 1. The Proposed Video Transmission Architecture. 218

3 Firstly, a camera collects video data to the ARM11 microprocessor, which encodes data through a hard encoding module called MFC. Then the compression data are packed with RTP/RTCP, and sent based on UDP Sockets. To forward data to end users, a cloud server is built based on Ngrok, an open source project. When the cloud server receives requests sent from the user layer, it will build a private connection and forward the user layer requests. Finally the user terminal unpacks the data received, calls the decoding module for decoding, and shows the final video information to users. Fig. 2 shows the detailed data flow [5]. QoS control Video Capturing Video Coding Video Display Video Decoding RTCP Analysis RTP Packing RTP Unpacking RTCP Feedback UDP TCP UDP TCP IP(Wifi/3G/4G) Cloud (Ngrok) Figure 2. Detailed Data Flow. Server The server is based on ARM11, which runs an embedded Linux system. The system takes advantages of the universal video capture API interface provided by V4L2 in the Linux kernel, and uses the memory-mapped access to get the video data. Because it is a memory-mapped device to an application s address space, just a pointer to a data buffer is switched and the data itself need not be copied. Thus we manage to reduce the time of acquisition and improve the efficiency of data collection. As shown in Fig. 3, after the video capturing device is turned on, the image format is set and the memory map is initialized. Then it begins to collect video data in cycles and send to the video compression module for processing. When a video stream enters the Video Coding module, which is a hardware encoding module on the ARM processor, it will be coded by H.264. After the H.264 data stream is packaged by RTP, it will be transmitted to a lower layer, and the output rate of the H.264 encoder is adjusted according to the RTCP information. At last, the video stream is sent on UDP. In order to watch live streams by popular browsers like Firefox and Chrome on PC, we also build a web server, like live555 or Boa [6]. 219

4 Open the video device file Get information of devices Set parameters for video capture Apply for a frame buffer for video data Capture video stream start Capture the data from the video output queue Video compression module Repeat capturing data Map the address of the frame buffer to the user space Put the frame buffer into the video capture input queue The frame buffer back into the input queue Stop capturing video data Figure 3. Illustration of Video Capturing. Cloud Server The cloud server is an universal solution for the serious problem with the shortage of IPv4, which builds a private connection to the server and transmits data from servers to clients and browsers. Ngrok is an open source reverse proxy project, which establishes secure channels between servers and the cloud servers. Ngrok can capture and analyze all the traffic on a channel, therefore it is convenient for post analysis and replay. Private connection and proxy connection are set up as in Fig. 4 [7]. Ngrok Ngrokd Users ReqProxy On Control Connection Proxy Connection RegProxy On Proxy Connection StartProxy On Proxy Connection Private Connection DataTransfering On Proxy Connetion Data Transfering on Private Connection Pubilc connection Data transporting Figure 4. Illustration of Cloud Server. Client After mobile phone users connect to the Cloud Server with 3G/4G, the requests will be transmitted to the server, which will build a RTP connection to transmit live 220

5 stream data between mobile phones and the server. Mobile phones receive RTP packets and pack them into NAL slices corresponding to the H.264 standard, then send complete NAL for the Video Decoding Module. To adapt to various sizes of display, we propose to include a post-processing procedure at the client side. For simplicity, a fast de-interlacing algorithm is integrated [7]. Fig. 5 depicts the whole process. register all the file format and codec library open an input stream and read the header open codec get stream info read frames from video stream find the decoder for video stream Video frame? N decode video frame from packet into picture Show and Postprocess Pictures Figure 5. Illustration of Client. Browser As we build a web server for capturing video and use the standard RTP/RTCP protocol to transmit streams, live videos can be played in our PC browsers. At the same time, we can watch live videos in multiple web and Android devices, as the web server uses a multi-threaded processing mechanism. The decoding algorithm of the browser is similar to that of the mobile phone. EXPERIMENTAL RESULTS In the system proposed, a S3C6410 processor is used as the hardware base of the server. And the open source projects JRTPLIB and Live555 are used as the core. The network transmission media are 3G and Wifi based on ARM-Linux, Windows, Centos and Android environments. Via building the Ngrok environment on Centos, we realize the function of the internal network penetration, and solve the problem of remote communication. Also, the proposed client and browser are based on Android and Windows System each other. Different frame rates, network environments and user terminals are employed to test the system delay, transmission rate, and loss packet rate. In Figures 6 and 7, the horizontal axis represents the number of frames in 4G and Wifi, and the vertical axis the information of delay, rate and packet loss. 221

6 Figure 6. The results of the experimental data are tested on the Android device with a 640*480 resolution ratio. Figure 7. The results of the experimental data are tested on the Web with a 640*480 resolution ratio. In those figures, it is easy to find that the relationship between the equipment and the packet loss rate and delay is loose. In generally, 3G and Wifi networks can basically meet the requirements of video transmission; packet loss rate is controlled under 12%; and most of the cases have a packet loss ratio below 6%., In one word, the requirements of real-time video streaming are satisfied. The evaluation environment is as follows: Server: Embedded Linux, ARM S3CC6410, RAM 128MB Cloud Server: Centos, Intel E52650 (16 cores), RAM 8GB Client: Android, Qualcomm MSM8974AC 2.5G, RAM 3GB Browser: Window 10, Intel 2.4G, RAM 8GB, 1024*768 Due to the adoption of H.264 hard coding, the data compression ratio and the network transmission efficiency are guaranteed. Experimental results for surveillance applications show that real-time requirements are satisfied, whereas 222

7 smooth and clear pictures are finally displayed. The proposed architecture results in a low cost and low power consumption solution, with remarkable stability and reliability. CONCLUSION A hybrid video transmission architecture, including B/C/S architecture, has been proposed. It is able to adapt to users' devices and network environments. The proposed system is implemented using Web and JNI. These two techniques completely satisfy the requirements of the self-adaptive software structure; and enable users to utilize a system that is optimized for their environment and networks. Also, the proposed architecture includes monitoring techniques that can deal with network problems by the cloud server Ngrok when the program is running. Furthermore, the transmission time of our system can be reduced when integrated into P2P network architecture mechanisms. REFERENCES [1] O. B. Kwon, The Design and Implementation of Real Time Encoder System, J. Multimedia Ubiquitous Engineering, vol. 9, no. 5, pp (2014). [2] J. Dong, J. Xin and P. Zhuang, A Study on H.264 Live Video Technology with Android System, Commun. Computer Information Science, vol. 525, pp (2015). [3] L. Cai, Research and Implementation of an embedded network video surveillance system based on H.264, Master Thesis, Nanjing University of Posts and Telecommunications (2013). [4] J. Heras, G. Campo, C. Gómez, et al., Multiprotocol Android Application for Smart Control and Monitoring in Buildings, LNCS 7657, pp (2012). [5] T. Takenaka, S. Kato and H. Okamoto, Adaptive Load Balancing Content Address Hashing Routing for Reverse Proxy Servers, in Proc. IEEE Int. Conf. Communications, vol. 3, pp (2004). [6] Q. Ji, H. Yu and H. Chen, A smart Android based Remote Monitoring System, in Proc. 3rd Int. Conf. Technological Advances in Electrical, Electronics and Computer Engineering, pp (2015). [7] Q. Huang and F. Xu, Fast Text Processing for Interlaced Sports or News Videos, Applied Mechanics Materials, vol. 704, pp (2015). 223