The Technology Study On Video Streaming Over Wireless Networks

With the development of modern wireless communication technology, the wirelessdata transmission rate and reliability increase continually. These technologies are alsorapidly applied to all kinds of wireless communication networks, including convenient,fast and low cost Wireless Local Area Networks (WLANs) and wide coverage cellularwireless access networks. Recently the newest WLANs based on the IEEE802.11nstandard, which adopts the innovative multi-antenna technology (MIMO) andorthogonal frequency division multiplexing technology (OFDM) in Physical layer(PHY), have the transmission data rate of up to600Mbps. Nowadays, the IEEE802.11acand the IEEE802.11ad standard that all are under development have the goal ofachieving data rate up to Gigabit order of magnitude. The increasing of transmissionrate makes it possible to delivery large flow high definition video streams over WLANs;In cellular wireless access networks, due to the advanced MIMO and Code DivisionMultiple Access (CDMA) technology applied, the4G wireless access network that isplanning to be deployed can also achieve500Mbps downlink transmission rate, whichefficiently guarantees the delivery of high definition video streams over wireless accessnetworks. However, video streams are different from general data traffic and have thecharacteristics of error-sensitivity and delay-sensitivity. Error-sensitivity is caused bythe priority structure of video stream data packets. The error or loss of high priority datapackets will result worse consequences than that of low priority data packets.Delay-sensitivity is produced due to the request of real-time, such as video conference,real-time video content sharing, etc. Based on these new characteristics, we need torestudy the transmission control protocol and related parameters in each layer ofnetwork system architecture. The characteristics of video streams also result that theirevaluation criteria are different from that of general data transmissions which usecriteria such as100%reliability and delay unconcerned. To evaluate the users’ watchingexperience offered by video stream transmission service, the concept of decodableframe ratio and then the Quality of Experience (QoE) are presented. So, based on theoff-the-shelf network technology, to provide better video watching experience for users, we fund many aspects worth to research in video delivery over wireless networks.In this dissertation, we focus on the video streams transmission over wirelessnetworks. Based on the knowledge of wireless network transmission and video codec,the key technology of video streams transmission over wireless networks is discussedand studied. The main contributions are:1. Improvement on downlink video streaming contention access protocol inWLANs. The usual application scenario in WLANs is the video streams downlinktransmission which is from Access Point (AP) to the wireless users. For this scenario,we design a new parameters adjustment technology for WLANs contention accessprotocol to guarantee the downlink transmission throughput and also achieve theoptimal system performance. IEEE802.11WLANs applies Distributed CoordinationFunction (DCF) in Media Access Control (MAC) layer to perform fair random accessprotocol. But, by theory analysis, we find this protocol can’t guarantee the throughputand delay of downlink video streams transmission over WLANs. Based on the existingMarkov chain analysis model, the new model is established for our improved accessprotocol and extended to error channel conditions. We proposed the new parameteradjustment technology to guarantee the downlink video streams transmissionthroughput and also achieve the optimal system performance, that is, maximum systemthroughput.2. Video streaming transmission rate control and retry limit setting technology overWLANs. In WLANs, lower transmission rate increases the transmission reliability butdegrades the throughput, which is easier to result the queue overflow. The same thing isthat higher retry limit increases the transmission reliability but degrades the throughput,which is also easier to result the queue overflow. So, using optimal control parametersset in video stream transmission is the guarantees of maximum transmission efficiency.To resolve the transmission rate control and retry limit setting, we proposed twooptimization models. One is based on the Markov decision process (MDP) and the otherone is based on the Markov state transition model. Finally we validate theireffectiveness for performance improvement. At last, combining the transmission rateand retry limit, we study and design the optimal transmission rate control policy underthe condition of retry limit introduced.3. Video streams multicast technology over WLANs. Usually, video streams transmission is unicast, that is, one to one service. In some application scenarios, videostreams need to broadcast or multicast to a group of users. WLANs standard specifiesusing the lowest transmission rate for broadcast or multicast and trying to guaranteeeach members’ reliable packets receive including members which stay in the worstchannel state. This simple rate setting seriously degrades the watching experience ofmembers which stay in the better channel state. On the basis of the priority structure ofvideo stream data packets, the heterogeneous characteristics of group members’wireless channel and the effect of the delay constraint, we use the dynamicprogramming (DP) to resolve the multicast transmission rate control problem. Itsubstantially improves the watching experience of multicast group users.4. Statistic QoS guarantees for video streams downlink transmission over wirelessaccess networks. In downlink transmission of wireless access networks, the schedulingalgorithm in base station affects the QoS performance for multiple users. Consider theclassic and practical queue length based rate maximum scheduling algorithm (QLBRM)and use the effective capacity theory to analyze the relationship between the multi-usersmaximum amounts of input data traffic and their asymptotic stationary queue lengthdistribution bound. Based on the analysis results, given the delay constraint or queuelength threshold, we can estimate the queue overflow probability upper bound of eachuser, that is, the so-called statistical QoS guarantees. Then we can balance theperformance factors between the input traffic, delay constraint and statistical QoSguarantees in system design. In addition, when the queue overflow packets losshappened, on the basis of the priority structure of video stream data packets, we presentthe packets priority drop to achieve the optimal received video frame decoding qualityand provide a satisfying video delivery service for users.