Project Report

 Documents

 4 views
of 8
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Description
1 Streaming MPEG4 video over wired and wireless local area networks Nicole Driscoll Southeast Missouri State University nrdriscoll1s@semo.edu Joshua Liberman University of Central Missouri jsl02020@ucmo.edu Jason Novinger Truman State University jnovinger@truman.edu Robert Williamson University of Missouri - Columbia rdwvy3@mizzou.edu Abstract There is not a clear consensus on how open-standard video streaming technologies perform across wireless computer networks. Wireless networking techn
Share
Tags
Transcript
  1 Streaming MPEG4 video over wired and wirelesslocal area networks Nicole Driscoll Joshua Liberman Jason Novinger Robert Williamson Southeast Missouri State University University of Central Missouri Truman State University University of Missouri - Columbianrdriscoll1s@semo.edu jsl02020@ucmo.edu jnovinger@truman.edu rdwvy3@mizzou.edu Abstract There is not a clear consensus on how open-standard video streaming technologies perform across wireless computernetworks. Wireless networking technologies have become nearly ubiquitous, particularly in residential networks, leadingconsumers to expect that they will retain the performance of wired networks. Fortunately, the advances in videocompression and wireless network bandwidth allow for higher-quality streaming video content over the more limitedwireless network. We seek to evaluate how video, encoded using an implementation of the MPEG4 Part 2 codec,performs when streamed across a simulated residential wired and wireless computer network. In particular, we areinterested in the quality of the received video and how the transmission across the network affects the subjective andobjective appearance on the client computer. Our network test bed comprises nine average desktop computers equippedwith a stream retrieval program, OpenRTSP, and a server running the Darwin Streaming Server from Apple Computer,connected using wired Ethernet connections and 802.11b, 802.11g, and draft 802.11n version 1.0 wireless connections.Each client was monitored while receiving a sample of raw video data encoded at one of a variety of common bit-ratesto note any lost content. In addition, each client saved a copy of the video locally for later comparison with the srcinalusing the PSNR (Peak Signal to Noise Ratio) and SSIM (Structural SIMilarity) metrics. Looking strictly at establishedwireless standards (802.11b and g), we found that they are not capable of streaming multiple High Definition qualityvideo streams across a wireless network link. Wired and draft 802.11n wireless connections did prove more capable of handling multiple High Definition video streams concurrently. Hopefully, our work will lead to a better understanding of the technical issues, performance, and trade-offs in home networking, thus facilitating the rapid deployment of advancedhome networking services and applications. Keywords Video compression, wireless network technologies, video streaming, home networks. I. Introduction Consumers are becoming aware of the capabili-ties that are available to be able to stream videoacross networks. Web browsers srcinally had todownload the entire file before they were able toplay it back. Some of the awareness is because of You-Tube and other popular media-sharing web-sites. There are several advantages and disadvan-tages to streaming video instead of downloading.Aspects crucial to video-streaming are video com-pression, hinting, protocol, physical limitationsand bit-rate.We performed this research to be able to seewhat level of video each of the various wirelesstechnologies could handle. We tested 802.11a,802.11b, 802.11g, and Draft 802.11n Version 1.0.We were streaming video at various bit-rates fromHDTV to VHS quality across the wireless net-works and were comparing the results to our con-trol, wired network. We ran quality analysis to judge the distortion of the videos.Streaming video has several advantages over adownload-and-watch player such as smaller mem-ory use and near instantaneous launching, butthere are also valid reasons to download a videorather then receive a stream. The ability to viewmultiple times without retransmission is one, aswell as capturing the video quality without anypacket loss, watching it as it was intended.In order to stream a video, that video must becompressed, as raw video data is too large to trans-mit over current networks. The basic techniquesof video compression are similar to those of imagecompression, with just a third dimension of local-ity. Each successive frame stores the changes fromthe previous frame. There are two different com-pression schemes used, known as scalable and non-scalable video compression.[2] Non-scalable videocompression uses one compressed stream at one  2 Fig. 1. A diagram of our network testbed, with wire Ethernet and wireless 802.11b, g, and draft n connections. specific bit-rate. Scalable video compression isvery flexible and can change the amount of band-width required to successfully stream dependingon the current network conditions using multiplestreams. We have used non-scalable compressionon our several MPEG4 files.Another thing that is needed to stream thevideo is hinting. Hinting provides the server withthe needed media information, so that it is awareof what tracks are coming. There are two differ-ent types of hinting: content hinting and appli-cation hinting[2]. By using content hinting, weallow the server to know how to form the packetstreams.After hinting a file, the next important topic isthe protocols used to stream multimedia. Real-time Transport Protocol(RTP) is an applicationlayer protocol for transporting audio/video pack-ets in over UDP. Control information is handledout-of-band by a protocol known as Real-timeTransport Control Protocol. RTCP passes Qual-ity of Service(QoS) to the server, including infor-mation like percentage of packet loss, delay, jitter,and out-of-order delivery which allows the serverto adjust to changing network conditions. For thisreason, streaming servers often encode videos at avariety of bit-rates so that they can adapt if net-work congestion is high. Clients can issue VCR-like commands to the server using the Real-timeTransport Streaming Protocol(RTSP).RTSP is an application layer protocol. It usesTCP for the transport of metadata. The basiccommands for RTSP are Describe, Setup, Play,Pause, Record, and Teardown. A Describe com-mand sent to a server includes a URL and the typeof data that the client can handle. The Setup com-mand is used to specify the port that the client willreceive the data on. A Setup command is neededfor each media stream, such as one for video andanother for audio. The Play command will playall streams that have been initialized by the Setupcommand concurrently. The Teardown commandends all media streams and clears all of the clientsdata on the server.If several clients connect and request high-quality media at the same time, the server can eas-ily encounter network congestion. The server mayconnect to each client individually, if the server is  3 using Unicast, even if several clients are request-ing the same file. A more intelligent server willhave implemented Multicast, a technology wherea single signal is sent from the server and is dupli-cated by routers only when required to reach itsdestinations.Along with the technology considerations thereare also a few physical limitations that must betaken into account when streaming media acrossthe various types of networks. The main physi-cal limitations to consider are the bandwidth of the network, the length and resolution of thevideo, and the bit-rate the video was encodedat. Bandwidth is the amount of data that canbe transmitted between two given points, but in-cludes all header information and checksums. Amore accurate description of network transmis-sion capabilities would be throughput. Through-put accounts for all links between the senderand receiver, therefore throughput is the measureof received data without the intermediate head-ers information. Throughput still contains somepacket header information, so the measure of re-ceived data that is application-usable is calledgood throughput or throughput.The size of the video depends on the length andbit-rate. Were a client to request a 30 secondtrailer of a high-definition video, the server wouldsend a total of: Size =30 . 0 s · 15000 . 0 kb/s 8 , 388 . 608 kb/MiB = 53 . 6 MiB (1)The bit-rate of a video is the amount of dataper unit of time, usually a second. Bit-ratesare variable and a higher rate will correspond tohigher quality video assuming the same codec isused. In order for streaming video to be suc-cessful the throughput must be greater than thebit-rate. However throughput will vary dependingupon network usage and bit-rate can vary depend-ing upon the current scene. Therefore at any givenmoment it is possible that the bit-rate will exceedthe throughput. To counter this problem, videoplayers will buffer incoming data. This slight delayin playback ensures that if data transmission is cutoff for a few microseconds or the bit-rate exceedsthroughput that the user will be able to continuewatching the video without interruption.The basic principles of video encoding are de-signed to reduce the bit-rate without sacrificingvideo quality. Before this is done each frame isbroken into YCbCr color space. The Y stands forluminance or brightness. The Cb and Cr are theblue and red chrominance levels, respectively. Thehuman eye is more sensitive to luminance than tochrominance, therefore it is optimal to dedicatemore bits towards brightness than color. The mostcommon ratio used in video compression is 4:2:0.This ratio is luminance to chrominance to the ra-tio of the blue to red chroma. This means thatthe most common ratio has twice as much lumi-nance than chrominance and an equal amount of blue and red chroma. When encoding video, eachof the color spaces is handled individually.To conserve bit rate, most video codec only en-code the differences between frames rather thaneach entire frame. There are different types of frames such as I, P, and B frames. I-frames areintra-frames or key frames. These frames containall the needed data to display the picture withoutneeding knowledge of previous frames. They areoften used at scene changes. These frames takeup the most space and are crucial to display avideo. P-frames are predicted frames which mayrequire a significant amount of knowledge of theframe or frames directly preceding it self. Theyrequire fewer bits than I-frames. B-frames are bi-predicting frames which are similar to P-framesbut may require knowledge of any frames thatcame before it not just the ones directly preced-ing it self. These frames also take up very littlespace since they depend so heavily upon previousframes.Once the video has been encoded, it needs to beplaced into a container. Containers allow multiplestreams to be contained in one file. This is a keystep as the file needs to contain hinting informa-tion that tells the streaming server how to par-cel out chunks of video. MPEG4 containers canbe used to include multiple audio stream (e.g. indiffering languages), hint tracks, and other videostreams.The two lowest layers of networking are thephysical layer and the data link layer. For wiredEthernet, the physical layer is known as the Ether-net Physical Layer and the data link layer is knownas Ethernet. The Ethernet Physical Layer canvary from coaxial cable to fiber optic cable. It can  4 Fig. 2. A illustration of I (Intra) frame and subsequent P (Predicted) frames in an MPEG sequence. also vary in speed, from 3Mbps to 10Gbps. Ether-net, the data link layer, consists of sending smallamounts of data also known as packets. Each Eth-ernet station has a 48 bit MAC address. Ethernetdoes not have a one-to-one connection from senderto receiver. Traffic in current generation switchesis routed to the selected receiver.For wireless networks the physical layer is knowas Wi-Fi and the data link layer is 802.11. Thereare four common protocols for 802.11, they are802.11a, 802.11b, 802.11g, and 802.11n. Currently802.11n can operate in the 2.4GHz and 5GHz fre-quency but is still in the draft stages but productsbased on various draft revisions are available. The802.11b and 802.11g standards operate within the2.4GHz, 802.11b was created first with a speed of 11Mbps and a revision known as 802.11g gave ita speed boost to 54Mbps. Most 802.11g devicesare backwards compatible with 802.11b devicesbut have to fall back to the lower speeds. The802.11a standard was created around the sametime as 802.11b, its difference is that it works inthe less crowded 5GHz frequency which allows forhigher transfer rates of 54Mbps though with re-duced range.Wireless networks can operate in two differenttypes of modes. One type deals with a centralaccess point and the other is ad-hoc, a networkformed from peer-to-peer. Wireless networks arebeing incorporated more and more in the businessworld, individual homes and other venues suchas coffee shops or restaurants, college campuses,and other high-traffic areas. Today’s populationalso has an increasing number of devices that canconnect to wireless networks. These devices in-clude but are not limited to laptops, PDAs/smartphones, portable gaming devices and portable mu-sic players.There are many advantages and disadvantagesof wireless networks. Some advantages includeconnectivity, cost, mobility, and convenience.Connectivity allows people to stay connected tothe internet no matter where they are located de-pending on the location of the access point (AP).Wireless networks can allow multiple clients to beconnected through a single access point. APs maybe a little more expensive than wired hardwaresuch as switches but their initial setup costs onlyrequires a single point and they can scale withoutthe need to buy and run additional wires. Clientsare free to move within the given area of an accesspoint or even from access point to access point.Access points can also allow for quick deploymentor mobility of a network. All these factors com-bined increase the convenience of a client using awireless network over wired networks.Unfortunately wireless networks also suffer froma number of disadvantages such as security, range,reliability and speed. These disadvantages may af-fect the user more depending on the nature of theirwork. While streaming video, the security issuedoes not hinder our success as it would while trans-mitting personal information such as credit cardusage, but an attacker may still attempt a denial-of-service attack by flooding the wireless medium,which would hinder legitimate packet transmis-sions. Range and obstructions also affect the sig-nal strength which in turn will affect the reliabilityand speed. Wireless networks are subject to in-terference from several different types of sources,
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks