FUTURE VIDEO TECHNOLOGY
Video over IEEE802.11 Wireless LAN: A Brief Survey Chang Wen CHEN1, FENG Zhengyong2, 3 1
Department of Computer Science and Engineering, University at Buffalo, State University of New York, New York 14260, USA School of Physics and Electronic Information, China West Normal University, Nanchong 637002, China 3 School of Communication and Information Engineering, University of Electronic Science and Technology of China, Chengdu 611731, China 2
Abstract: With relatively high transmission capacity and usually unconstrained connections, IEEE802.11 WLANs provide the ideal infrastructure for pervasive video content sharing and communications. However, the delivery of high-performance video streams over 802.11 WLANs remains a challenging task due to the inherent characteristics of compressed video and dynamic channels. In this paper, we present a brief survey of various recent innovations that have been developed to enhance the Quality of Service (QoS) performance for video over WLANs. Based on the application scenarios, the solutions have focused mainly on three network layers, that is, Application layer (APP), Media Access Control layer (MAC), and Physical layer (PHY). After reviewing the video compression technology, we first examine various single-layer solutions for video over WLANs. We then discuss several cross-layer solutions that take advantage of mutual interactions between different network layers. Finally, several technical issues beyond QoS performance, including energy and security, are also addressed. We conclude that the application of video over WLANs will continue to increase in future. Key words: IEEE802.11 WLANs; video delivery; QoS; cross-layer solutions
I. INTRODUCTION In recent years, the family of IEEE802.11 [1] Wireless Local Area Network (WLAN) stanChina Communications • May 2013
dards has become the de facto deployment specification for various wireless infrastructure providers to offer pervasive unconstrained mobile Internet connections. More and more public and enterprise locations, including office buildings, airports, campuses, train stations, hot spots, and home environments, have been deployed with IEEE802.11 WLANs. Numerous new applications are booming to enable emerging multimedia delivery services over WLAN, from conventional VoIP and video streaming to mobile gaming and interactive video conferencing. In particular, the new wireless smart home gateways, equipped with more powerful processing capability than the traditional wireless access point, are expected to provide desired Quality of Services (QoS) for delivering high definition video stream to High Definition Television (HDTV) terminals in home wireless environment. Furthermore, the popularity of smartphones equipped with Wi-Fi access functionality has led to an explosive increase in image and video centric applications, enabling real-time camera video content sharing under the mobile social media settings. According to Cisco, by the year 2016, the mobile video will account for 66% of global mobile data traffic [2]. Among these future applications, we expect that substantial percentage of mobile video services shall be delivered over the IEEE802.11 WLANs. However, video delivery over WLANs continues to be a challenging task, especially when these services need to guarantee desired QoS
Received: 2013-03-06 Revised: 2013-04-07 Editor: Ming-Ting SUN
1
A brief survey is presented that summarizes the recent inovations for the QoS performance enhancement of video over WLANs. These inovations are mainly related to three layers of network: APP layer, MAC layer and PHY layer. According to which layers these inovations are based on, they are reviewed by dividing into two categories: single-layer solutions
and
layer solutions.
2
cross-
for their consumers. First of all, the wireless channel is dynamic and error-prone. Video data packet transmission over wireless channel is very difficult since the compressed video content is both error sensitive and time critical. The video compression algorithms attempting to achieve bandwidth reduction would create complex dependencies among video blocks and frames. During the wireless transmissions, video data packet errors or losses often affect not only the current video frame but also the following video frames. In addition, the delay in video data packet transmission needs to be kept within certain level of latency threshold. This threshold is often decided by the frame decoding time at the receiver, demanding the on-time transmission and reception of packets from next frame when current video frame is being displayed. It is especially important for real-time video streaming in which the video packets that exceed the delay threshold will become useless, even though they may be reliably transmitted to receivers after the deadline. These two conflicting demands in video over WLAN, error reduction and delay constraint, have been the source of inspiration for developing new class of wireless transmission technologies to guarantee reliable and on-time delivery of video contents over WLANs. Over the past few years, significant advances in video over WLAN have been achieved. The IEEE802.11 standards have been continuously enhanced to facilitate new applications. The first WLAN standard was published in 1997 with its data rate of only 2 Mb/s. However, the most recent standard, IEEE802.11n [3] amendment, was published in 2009 and can achieve high data rate of up to 600 Mb/s owing to the new transmission technology used in the physical layer. Two future amendment standards IEEE 802.11ac and 802.11ad [4] are currently under development and will provide higher throughput in the 5 GHz and 60 GHz band respectively. In particular, the goal of IEEE802.11ad is to achieve a theoretical maximum throughput of up to 7 Gb/s which will be used for transmitting high definition video streams in home wireless environment. The
continuously increase of throughput in the Physical layer (PHY) promises enough bandwidth to support such high capacity transmission of video streams. Correspondingly, we will also need to carefully design innovative schemes for Media Access Control layer (MAC) error recovery and channel access mechanism to ensure seamless video delivery over WLANs. The two amendment standards IEEE802.11e [5] and IEEE802.11aa [6] have been dedicated to improve the efficiency of video data transmission in the MAC layer. Explicitly, the IEEE802.11e specifies a set of higher priority channel access parameters for video stream category to reduce the transmission delay while the IEEE802.11aa specifies a group of new error recovery mechanisms for video Multicast/Broadcast over WLANs. Besides these PHY and MAC layer mechanisms specified in the standard, there are also other non-standard mechanisms that have been developed to improve the performance of video transmission over WLANs, including admission control, application layer rate control, cross layer optimization, and so on. Among all these, the cross layer technology is the most attractive strategy to resolve the inherent problems of video transmission over wireless channels. A large number of researchers have been working in various cross layer solutions and have made significant advances in recent years. In this survey paper, we introduce and summarize several relevant technologies in video over WLANs. These technologies are classified based on their implementation layers within the wireless network. Due to the page limit, only main technologies are discussed in this paper. For the convenience of the readers, the abbreviations used in this paper are listed in Table I. In Section II, a brief review of video compression and streaming techniques is introduced. In Section III, single-layer solutions for video delivery over WLANs, including APP layer, MAC layer and PHY layer, are presented. In Section IV, various cross-layer solutions for video delivery over WLANs are also presented. We discuss several important issues in video over WLANs that cannot be classified by implementation layer in Section V. China Communications • May 2013
Table I Abbreviation list PHY
Physical layer
MAC
Media Access Control layer
APP
Application layer
QoS
Quality of Service
GOP
Group of Picture
MCP
Motion Compensated Prediction
DCT
Discrete Cosine Transform
SVC
Scalable Video Coding
AP FMO NALU
Access Point Flexible Macroblock Ordering Network Abstraction Layer Unit
RDO
Rate Distortion Optimization
JSCC
Joint Source and Channel Coding
UEP
Unequal Error Protection
FEC
Forward Error Correction
MIMO
Multiple Input and Multiple Output
RTT
Round Trip Time
ARF
Auto Rate Fallback
AMC
Adaptive Modulation and Coding
DCF
Distributed Coordination Function
PCF
Point Coordination Function
CSMA/CA
Carrier Sense Multiple Access with Collision Avoidance
CAP
Controlled Access Period
CFP
Contention-Free Period
CP
Contention Period
EDCA
Enhanced Distributed Channel Access
HCCA
HCF-Controlled Channel Access
AIFS
Arbitration Inter Frame Space
TXOP
Transmission Opportunity
CBR
Constant Bit Rate
VBR
Variable Bit Rate
RED
Random Early Detection
ACK
Acknowledgement
AC
Access Category
QoE
Quality of Experience
MOS
Mean Opinion Score
ARQ
Auto Repeat reQuest
RTS
Request to Send
CTS
Clear to Send
Finally, the key open problems related to video over WLANs are discussed in Section VI. Section VII concludes this paper with a summary.
II. VIDEO COMPRESSIONS AND STREAMING The aim of video compression is to more effiChina Communications • May 2013
ciently transmit the video data information over wired and wireless channels or more compactly store the video data information in the digital media. An increasing number of video content sharing services and a growing popularity of high definition TV in recent years are creating greater demands for higher coding efficiency in video compression. Moreover, the wireless transmission channels such as UMTS or WLANs usually offer much lower data rates than wired links such as cable broadcast channels. Therefore, an enhanced coding efficiency shall enable the transmission of more video channels or higher-quality video representations within existing transmission channel capacities. The video compression coding has evolved over 20 years since the first widespread standard ITU-T H.262 [7] (also known as MPEG2) was established. Throughout this evolution, continued efforts have been made to maximize the coding efficiency. Over these years, the most successful coding standard is considered to be the ITU-T H.264 [8] (also known as MPEG4 part 10) which was formally approved in 2003. The H.264 video coding standard adopts the MCP and DCT to obtain extremely high coding efficiency while keeping an acceptable perceptual video quality. This efficient coding scheme results in the priority structure of video frame sequence based on how each video frame is encoded. Figure 1 denotes one GOP for a video sequence. The successive concatenation of GOPs forms the complete video stream. For simplicity, we use picture and frame interchangeably. From Figure 1, we can see that the I frame is the coding reference frame for the subsequent P and B frames and has the first level of priority, while a P frame is encoded with reference to the preceding I(P) frame but will be referenced by the B frames and has the second level of priority. A B frame belongs to the lowest level of priority since it is encoded with reference to preceding I(P) frame and succeeding (I)P frame but is not referenced by any other frames. This inherently unequal priority characteristic embedded in the video coding strategy distinguishes the
3
Fig.1 Typical H.264 GOP pattern [9]
Fig.2 Examples for SVC [9]
compressed video streams from other data traffic and can be exploited to design some unique technology solutions for video stream transmission over wireless channels. To address the scalability of video transmission over networks, the extension of H.264 called SVC [8] was developed and finalized in 2007. In SVC standard, the video is encoded hierarchically into a base layer and one or
4
more enhancement layers. Successful decoding of the base layer provides a basic video quality, while the decoding of the base layer together with one or more enhancement layers provides an enhanced video quality. SVC specifies the following scalability modes: data partitioning, temporal, spatial, and Signal-to-Noise Ratio (SNR). To explain the structure of SVC streams, temporal and spatial modes are illustrated in Figure 2. For more information, interested readers can refer to Ref. [9] for more details. We can see that the SVC streams introduce the layered priority for a single video stream. The base layer has the first level priority, while the first enhancement layer has the second level priority, and the second enhancement layer has the third level priority, and so on. From the transmission point of view, the layered priority of the SVC multilayer video streams can be utilized when we design the video transmission technology over wireless channels. In addition to the layered priority structure of H.264 SVC as we have just described above, there is another important characteristic of a video stream which is its stringent delay constraint in the process of transmission. The delay constraint is predominantly decided by the frame decoding deadlines at the receiver. The two distinct characteristics of video coding and streaming in layered priority and delay constraints pose significant challenges as well as offer prime opportunities in developing robust and highly effective schemes for video transmission over WLANs. The priority structure of the video streams dictates that the effect of packet loss would be different one packet from another and will most likely to be China Communications • May 2013
content-dependent. For example, the error or loss of those frames that contain higher priority video data will degrade more the overall decoding quality because subsequent frames need to refer to this frame for reconstruction. The delay constraint of the video streams will restrict applicability of certain channel access control and/or error recovery mechanisms, especially for the real-time applications of video transmission over the dynamic and error-prone wireless channels. How to reliably transmit video data over error-prone channels while meeting the stringent delay constraint is indeed a daunting task for wireless video applications. We can see that almost every solution to be discussed in this survey paper attempts to simultaneously resolve the problems originated from unequal error sensitivity and stringent delay constraint. In WLANs, video applications can be classified into two different scenarios: real-time video transmission and video streaming. As shown in Figure 3, in real-time video transmission scenario, multiple users transmit their video contents to each other within one WLAN cell or via the WLAN AP to connect them to remote users. This real-time video transmission scenario may include contemporary applications such as mobile video conferencing, smartphone video content sharing, and other interactive video services that need immediate video content play-out at the receivers. This scenario usually demands onboard video coding and stringent delay constraint. For video streaming scenarios, the video server located in the wired network side delivers the video content to multiple users that connect to a WLAN AP. It can often be assumed that the wired network provides sufficient bandwidth and causes negligible effects on video stream delivery. This video streaming scenario may include video-on-demand, video download and play, and home HDTV sharing. At the receiving end, such scenario usually would buffer video frames for a while before the play-out starts. At the server side, offline video encoding is usually adopted and hence can tolerate more relaxed delay constraints. In China Communications • May 2013
Fig.3 Video applications scenarios
each of the scenarios outlined above, the video streams can be transported via either unicast or multicast. In this paper, both real-time video transmission and video streaming scenarios will be covered when discussing all the solutions developed for video delivery over WLANs. From transmission point of view, video delivery over WLANs is mainly associated with three communication layers: APP layer, MAC layer and PHY layer. The pre-coded video data in storage or online encoding of video data are involved with APP layer which can conveniently provide video stream characteristic information such as video coding parameters, video frame priority structure, and so on. The
5
channel access and error recovery mechanism are involved with MAC layer which will directly impact the transmission performance of video streams. The time-varying and errorprone channel information can be dealt with in PHY layer which can be appropriately utilized to improve the overall transmission efficiency. To clearly present the solutions developed for video delivery over WLANs, we consider various video delivery solutions based on their associated layers. In particular, we will begin our discussion on those single-layer solutions and then follow up with more recent crosslayer solutions.
III. SINGLE-LAYER SOLUTIONS 3.1 APP layer solutions The solutions developed and implemented at APP layer are mostly associated with the video coding techniques employed to generate the video bitstreams. These solutions are dependent on particular video codec algorithms. This is particularly true for real-time video transmission in which the onboard coding algorithm in the application layer is a crucial component for guaranteeing the QoS performance upon deployment. In fact, such single APP layer solutions for video transmission are not restricted to WLANs applications. These APP layer solutions can be deployed in any wired or wireless network. Usually these solutions are only able to adapt to a few channel parameters, such as transmission rate or error probability. In the following, we shall focus on the APP layer solutions based on H.264/AVC video coding standard. Other video coding solutions are based on similar technology principles. 3.1.1 Error resilience tools The H.264/AVC standard has a profound impact on video transmission due to its error resilience tools in terms of supporting video applications in various types of networks. The most useful error resilience tools of H.264/ AVC required in a wireless transmission environment are Slice structuring, FMO, Data par-
6
titioning and Error concealment. The objective of slice structuring is to avoid error propagation while the aim of FMO is to avoid error accumulation in a frame. Data partitioning is designed to encapsulate the syntax elements with different importance levels into separate NALU. It enables the unequal error protection according to the importance of syntax elements. These error resilience functionalities can also be fully exploited at the receiving end when error concealment can be designed to make use of error resilience embedded in the video bitstreams. With error concealment, the decoding frame quality degradation due to transmission errors or losses can be mitigated and the quality of the received video can be substantially improved. Many researchers have verified these error resilience tools’ performance in wireless environment [10-11]. 3.1.2 Rate distortion optimization The H.264/AVC coding standard has another profound impact on video transmission due to its rate distortion optimization function. The rate-distortion function can be formulated as
o* = arg min( D (o ) + λ R ( o )) o∈ X
(1)
Here, o is a block coding mode (inter/intra mode and block size) selected from the set of coding modes X. The D(o) is the distortion introduced by encoding with mode o; R(o) is the corresponding coding rate; and λ is the Lagrange parameter for appropriate weighting of rate and distortion. This RDO mode can be used for rate control in APP layer to adapt the video traffic to the network bandwidth dynamic [12]. Among the RDO solutions, Philip A. Chou introduced the most powerful RDO framework based on packets granularity called RaDiO [13]. The benefits of introducing packets granularity is the priority structure of video streams that can be easily handled. RaDiO improves the packetized video streaming performance greatly. 3.1.3 Joint Source and Channel Coding (JSCC) JSCC schemes attempt to develop APP layer solutions that consider both source coding rate as well as channel coding rate simultaneously. China Communications • May 2013
Channel coding is introduced to combat the transmission errors to improve the quality of received video. However, since the channel coding is competing for the limited channel bandwidth, an appropriate allocation of bit budget between video source coding and channel coding needs to be carefully executed to achieve optimal solution. This can also be considered as an efficient APP layer error control for real-time video transmission over packet lossy networks [14]. For a given channel loss characteristic, JSCC can efficiently utilize the UEP algorithm to increase the transmission robustness of the video bitstreams and therefore can be considered as application layer FEC. Interested reader should refer to Ref. [15] for more detailed treatment of JSCC schemes for various lossy channels.
3.2 PHY layer solutions The most useful characteristic of IEEE802.11 physical layers is the multiple transmission rates it can provide based on different modulation and coding schemes. This is often called AMC. For example, the transmission rate modes based on different AMC in 802.11a are shown in Table II. The original 802.11 standard operates at 1 and 2 Mb/s. Three high speed versions were added to the original version. The 802.11b supports four physical rates up to 11 Mb/s [3]. The 802.11a/g [3] provides eight and twelve physical rates up to 54 Mb/s, respectively. The most recent standard 802.11n offers possibly sixty-two different physical rates up to 600 Mb/s due to its MIMO and dual band technologies. Currently, new technologies in physical layer are still under development and the new amendments, 802.11ac and 802.11ad, are being formed aiming at higher throughput in the range of Gigabit per second. As a result, mobile devices are able to select the appropriate link rate depending on the required QoS and the instantaneous channel conditions in order to enhance the overall system performance. In addition, the MIMO technology introduced into the physical layer leads to more complicated signal processing techniques that can be used to achieve higher China Communications • May 2013
Table II Rate modes for 802.11a Index Modulation
Bit rate/(Mb·s−1)
Coding rate
0
BPSK
6
1/2
1
BPSK
9
3/4
2
QPSK
12
1/2
3
QPSK
18
3/4
4
QAM-16
24
1/2
5
QAM-16
36
3/4
6
QAM-64
48
2/3
7
QAM-64
54
3/4
throughput or better reliability. 3.2.1 Rate adaptation With the technology evolutions in the physical layer of the IEEE802.11 WLAN, more and more transmission rate modes can be made available for selection by the mobile receivers. For example, with 802.11n, there are sixty-two transmission rate modes that can be selected by a receiver depending on its current channel status and QoS requirement. How to efficiently design the rate adaptation scheme from so many choices of transmission rate modes in order to achieve higher overall transmission throughput is really a difficult task. Ref. [16] is among the first to introduce a simple algorithm in which the deployed rate adaptation is based on the received signal strength measured in the current state. In Ref. [17], Rebai, et al. proposed a new algorithm, called Modified Adaptive Auto Rate Fallback (MAARF), based on RTT-decision to provide rate adaptation procedure. The intention of this RTT-based approach is to pro-actively detect channel condition variation in its incipient stages. Since this solution can be easily implemented, a practical algorithm of rate adaptation has been recently developed [18]. 3.2.2 Operable MIMO The recently finalized IEEE802.11n standard has introduced the MIMO technology into the physical layer. The main objective of introducing MIMO technology is to increase the throughput of physical layer. However, such technology also provides the flexibility of trade-off between the throughput and reliability because the MIMO can operate in two
7
modes, spatial multiplexing to increase throughput and diversity to enhance reliability. In Ref. [19], Niu and Ngo proposed an efficient and simple algorithm that switches between these modes based on the instantaneous channel state information. In Ref. [20], Zhang, et al. presented an antenna selection algorithm in MIMO available 802.11n physical layer to support high throughput multimedia transmission. From the discussion above, we can see that the single physical layer solutions mainly concentrate on the increasing throughput efficiency which is not limited to aim at video transmission. However, when PHY layer technology such as pre-coding is designed, taking video characteristics into consideration [21], additional performance gains in effective throughput will result in QoS improvement for video over MIMO wireless channels.
3.3 MAC layer solutions The IEEE802.11 MAC layer aims to provide access control functions to the wireless channel such as access coordination, frame retransmission and check sequence generation. There are several ongoing activities to extend the IEEE802.11 MAC layer protocols. One of them is the well-known IEEE802.11e standard which has been developed in order to enhance the QoS performance. Two medium access coordination functions are defined in the original 802.11 MAC: a mandatory DCF and an optional PCF. The basic DCF uses a CSMA/CA mechanism to regulate access to the shared wireless channel. The most important parameter of DCF is the size of the Contention Window (CW). PCF was introduced to support multimedia transmissions and can only be used as a centralized control unit implemented at an AP in a WLAN. When a WLAN system is set up with PCF enabled, the channel access time is divided into periodic intervals called beacon intervals. See Figure 4 (a). A beacon interval is composed of a CFP and a CP in which PCF and DCF are applied in each period, respectively. During a CFP, the AP maintains a list of registered users and polls them according to the list.
8
In DCF, only the best effort service is provided. The point is that with DCF, all the users compete for the channel with the same priority. There is no differentiation mechanism to provide better service for real-time video traffic than for data applications. While PCF was designed to support time-critical multimedia applications, this mode has three major problems that lead to poor QoS performance. PCF defines only a single class round-robin scheduling algorithm which cannot handle the various QoS requirements. The AP has to contend to seize the channel to start the PCF process and the transmission time of each polled user in PCF cannot be strictly controlled. Therefore, PCF cannot provide guaranteed delay-constraint for real-time multimedia transmission. To improve the MAC layer QoS performance, IEEE802.11e was developed. It replaces DCF and PCF with EDCA and HCCA correspondingly. It also introduced CAP to generate periodic contention free access intervals. See Figure 4 (b). EDCA is designed to provide prioritized QoS by enhancing the contention-based DCF. HCCA provides the same contention free access as PCF but it controls the time delay more strictly than PCF. Another common QoS problem for both DCF/EDCA and PCF/HCCA is that no admission control mechanism is specified in the 802.11 legacy MAC. When traffic load is very high, the performance of both functions can be degraded. The access mechanism is the most important part of the MAC layer solutions for performance enhancement of video transmissions. Furthermore, the retry limit adaptation and MAC-FEC are the other two important technologies in the MAC layer and will be discussed in details in the following. 3.3.1 Contention access mechanism For MAC layer solutions based on contention access mechanism, the schemes mainly focus on the access parameter adjustment. In fact, EDCA itself is already an improvement over original DCF for contention access mechanism. However, using different access parameter sets, including CW, AIFS and TXOP, China Communications • May 2013
Fig.4 IEEE802.11(e) MAC layer media access control time sequence
can provide priority access for different categories of data flows. During the early attempts, some CW adjustment algorithms were introduced in DCF. In Ref. [22], an adaptive p-persistent CW adjustment algorithm was introduced. This algorithm adaptively assigns differentiated permission probabilities to transmission stations which are in different access categories and with various waiting delays. In EDCA, another parameter Transmission Opportunity (TXOP) can also be adaptively regulated. Ref. [23] presented a method that takes into account the estimated incoming video frame size and the current transmit queue length to tune the EDCA TXOP after winning a contention. More recently, a comprehensive and accurate EDCA access parameter adjustment algorithm was developed to achieve optimal performance for video transmission over WLANs [24]. China Communications • May 2013
3.3.2 Contention free access mechanism The contention free access mechanism has been developed to provide QoS guarantees for real-time multimedia transmission over WLANs. Many researchers introduced the improved algorithm either in PCF or in HCCA for multimedia transmission. Ref. [25] introduced the priority characteristics to PCF and proposed a Unified PCF which permits that the high-priority users always join the polling list earlier and get on the polling list in bounded time through the fast reservation scheme. In addition, UPCF can employ dynamic channel time allocation to provide bandwidth assurance. We can see that the improvement on PCF mainly comes from overcoming its shortcomings that were discussed above. In a sense, the improvement on PCF facilitates the emergence of HCCA even though HCCA is still not suffi-
9
cient to support the QoS requirement for video transmission over WLANs. One major disadvantage is that the 802.11e proposal for HCCA is merely a simple scheduler providing a CBR service, which is not well suited for VBR video media flows. In Ref. [26], Gao, et al. exploited the statistical multiplexing gain among multiple VBR traffic flows. They applied the existing statistical multiplexing framework to study the admission control problem, with all features of the HCCA being taken into consideration. A similar approach has recently been developed [27] to adapt the HCCA to the VBR traffic. Compared to the original HCCA, it provides higher channel utilization and adapts better to the characteristics of the VBR traffic. In fact, the VBR video traffic is dynamic and time-varying and therefore it is difficult to be standardized into the HCCA. An interesting strategy in designing the MAC layer access mechanism is well illustrated in a recently published approach toward Virtual Contention Free (VCF) environment [28]. This scheme is able to offer burst transmission of many video packets within one successful contention to ensure the timely transmission of video data. However, this VCF scheme, although being able to improve video QoS, creates an unfair environment for nonvideo data traffics. Mitigation of such unfairness can be designed when the deadline for each video packet is considered and the VCF can be made deadline aware [29]. Once the deadline for a given video packet is estimated, it serves its dual purpose: one is to guarantee the transmission of any given video packet before it expires while the other one is to allow other types of data traffic to transmit when there is no pressing need to transmit any video packet. It has been shown that this interesting MAC layer access mechanism can achieve virtually no loss for video data packets and noticeable improvement in transporting non-video traffics. 3.3.3 Retry limit adaptation and MAC-FEC In 802.11 MAC layer, the retransmission of data frame is a very efficient error recovery
10
scheme. The default retransmission limit in 802.11 is set to 4 and 7 for short and long data frames, respectively. Ref. [30] provided a retry limit adaptation analysis model and presented a nice design of a real-time retry limit adaptation scheme for video transmission. Following this work, Ref. [31] also carried out further analysis and improved the performance of retry limit adaptation. In MAC layer, FEC is often applied when the multicast mode is implemented. This is because, in this mode, multiple users in the WLANs will receive data frame at the same time. If each of the receivers sends back ACK frames, there is potential collision occurring. To increase the multicast reliability, the MAClayer FEC is introduced. In Ref. [32], a REDFEC scheme has been developed. The proposed algorithm tunes the number of redundant FEC packets in accordance with the access point queue length in such a way that the redundant packets injected for loss recovery purposes will not result in network congestion. From the implementation point of view, the MAC-FEC would have the same structure as the APP-FEC. However, the APP-FEC has been much more widely used in practice and is not limited to the applications in WLANs. Moreover, researchers have attempted to introduce a multiple-user ACK scheme in multicast instead of applying the MAC-FEC in the case of multicasting over WLANs. In particular, the IEEE802.11aa proposes many enhanced ACK schemes for multimedia multicast transmission.
IV. CROSS-LAYER SOLUTIONS Single-layer approaches are generally limited due to the lack of interaction between the network layers. Although each network layer may be well-defined, they usually exhibit intimate relations because some layers are heavily dependent on corporations from other layers to achieve the maximum potential. Cross-layer architecture has become the trend of networking research for the past decade. Brand new cross-layer mathematical models have been China Communications • May 2013
proposed, which may lead to a revolutionary network architecture design [33-34]. Numerous cross-layer approaches have been developed in recent years. In this paper, we shall focus on such cross-layer solutions that are based on the off-the-shelf technologies in WLANs, especially based on these single-layer technologies introduced in Section III. How to efficiently implement cross-layer solutions to increase the QoS performance of video transmission over WLANs is still an open question. Until now, there is still no unified strategy when designing the interaction between different network layers for cross-layer solutions. Very different considerations have been attempted to enhance the overall QoS performance in video over WLANs. Based on the video transmission scenarios in WLAN as outlined in Section II, such transmission is mainly associated with three layers of the network, which are, the PHY layer, the MAC layer and the APP layer. For the scenario of real-time video transmission as shown in Figure 3 (a), it is obviously that the video transmission performance is mostly affected by these three layers mentioned above. For the scenario of video streaming as shown in Figure 3 (b), some researchers have also considered the wired network part. This will result in two additional layers, the network layer or the transport layer, to be considered. More researchers focusing on video over WLANs usually ignore the impact of wired network because the bottleneck in such a hybrid networks lies in the wireless counterpart. As a result, the cross layer solutions we shall discuss in this paper will be limited to the three most intimately coupled layers: the PHY layer, the MAC layer and the APP layer. In Section III, we have already discussed the main characteristics of each of these three layers. The objective of cross-layer solutions is to design appropriate interaction algorithms between these layers based on the characteristics of each layer and their mutual dependence in order to maximize the overall QoS performance of the video transmission. For the APP layer, it is very important to understand China Communications • May 2013
how a particular video data format generated by the video coding algorithm can be utilized by other layers. For example, whether or not the video content is encoded at the APP layer by H.264/AVC or H.264/SVC and whether or not the video coding is executed online or offline will have significant implications for the other two layers to adopt the proper strategy. For MAC layer, how to utilize the characteristics of video coding algorithms will determine which access mechanism, either contention-based or contention-free, should be adopted. Some common transmission parameters in the MAC layer can also be adopted in cross-layer design, including the retry limit, the MAC-FEC, and the contention window size. For the PHY layer, the two frequently adopted strategies for consideration in the cross-layer design are rate adaptation and MIMO technology. Based on the network layers associated with the cross-layer design, we shall discuss the cross-layer solutions in the following four categories: the MAC-APP, the MAC-PHY, the PHY-APP and the PHY-MACAPP.
4.1 MAC-APP cross-layer solutions The MAC-APP cross-layer design constitutes the majority among all categories of crosslayer solutions. The design principle for the MAC-APP category of cross-layer design can be best illustrated by exploiting the intimate interactions between the inherent features of the video streams and the characteristics of WLAN MAC layer mechanism. For contention-based channel access mechanism, this category of cross-layer solutions often focus on how to map the video priority frames into 802.11e EDCA priority queues. Ksentini, et al. may be the first to summarize such mapping structure in Ref. [35]. They use H.264 data partitioning technology to generate different priorities of video data packets and map them into the priority queue of the EDCA. In Ref. [36], Li, et al. reported their investigation on how to map the SVC video streams into EDCA priority queues in order to achieve performance enhancement. In Ref. [37], San-
11
tos, et al. proposed a novel QoE-aware multicast scheme. In their scheme, the base layer and the enhancement layer of the SVC video streams will be injected into two new access categories of the EDCA, primary and alternate queues, respectively, as defined in 802.11aa, which is one of the newly developed WLAN standards. For contention free channel access mechanism, the cross-layer solutions are usually focused on how to utilize the limited transmission bandwidth to schedule as many important video stream units as possible. In Ref. [38], Ven der Schaar, et al. proposed an optimized and scalable HCCA-based admission control for delay-constrained video streaming applications. This cross-layer strategy combining adaptation, scheduling, and error protection is facilitated by the fine-grain layering structure provided by the H.264 SVC bitstreams. In Ref. [39], Fallah, et al. reported their study on how the priority data of video streams with H.264 data partitioning can be mapped into MAC layer AC queues of 802.11e EDCA and HCCA. The possible mapping situations can be single EDCA AC, multiple EDCA ACs, single HCCA AC, multiple HCCA ACs (the authors introduced two priority ACs in HCCA) and combination ACs of EDCA and HCCA. In the MAC layer of WLANs, another important strategy in addition to the access mechanism is the implementation of Auto Repeat reQuest (ARQ) and MAC-FEC. In Section 3.3.3, we have introduced the retry limit adaptation algorithm. Notice that this retry limit adaptation algorithm can also be combined with video streams priority information to design MAC-APP cross-layer solutions. In Ref. [40], an adaptive algorithm has been proposed for the retransmission of video packets according to their different priorities, that is, timestamp urgent level and error propagation level. In Ref. [41], based on the same idea, an improved retry adaptation scheme with better video content analysis and packet retry scheduling was proposed. In Section 3.3.3, we have also introduced the MAC-FEC scheme. Having the same goal to resolve the
12
video multicast optimization over WLANs, an optimal solution based on Multiple Description Coding with FEC (MDFEC) was proposed in Ref. [42]. In addition, authors in Ref. [42] presented a Hybrid ARQ scheme for video unicast, that is, a combination of ARQ and FEC schemes. Another cross-layer solution associated with the combination of ARQ and FEC is an adaptive MAC layer ARQ scheme, combined with APP layer FEC, to achieve an improved video transmission over WLAN [43]. A more complicated MAC-APP cross-layer solution was presented in Ref. [44] in which a Hybrid-ARQ is implemented in the MAC layer in combination with conditional video frame skipping and reference frame selection in the APP layer. These MAC-APP cross-layer solutions implement ARQ or FEC or a combination of them according to some specific characteristics of video data generated in the APP layer.
4.2 PHY-APP cross layer solutions As indicated early in Section 3.2, two main characteristics of the WLAN PHY layer are AMC and MIMO technology. With proper design, MIMO technology can also be viewed as some type of extended adaptive modulation and coding scheme. Therefore, this category of PHY-APP cross-layer solutions usually aims at mapping the PHY layer transmission rate adaptation (AMC) with APP-layer video data characteristics. This strategy of mapping the application layer video rate variation to the physical layer rate adaptation has been illustrated very well in two recently developed schemes [45-46]. Ref. [45] reported a scheme of PHY-APP cross-layer solution that assigns different layers of SVC streams to different PHY rate modes. Ref. [46] reported another PHY-APP cross-layer scheme that makes use of the video streaming rate waveform from the APP layer to guide the adaptation of the wireless PHY layer rate selection. Furthermore, it has been shown in Ref. [47] that the error resilient video coding implemented in application layer can also be combined with PHY layer rate adaptation to achieve enhanced perChina Communications • May 2013
formance in video over WLANs. More specifically, the strategy of JSCC and error concealment has been integrated with rate adaptation in PHY layer for an end-to-end optimal solution. For the application of MIMO technology, an elegant and powerful cross-layer solution has been developed in Ref. [48] to match different layers of H.264/SVC encoded video bitstreams with different virtual MIMO channels under the spatial multiplexing MIMO settings. A very effective and low-complexity solution has been implemented by an adaptive channel selection based on partial channel information.
4.3 PHY-MAC cross-layer solutions In general, this category of cross-layer solutions needs not to take the video data characteristics into account in the design process. Like the single-layer solutions in either PHY or MAC layers, the main objective of PHYMAC cross-layer solutions is to achieve network maximized throughput. Among them, a joint adaptation of MIMO configuration at the PHY layer and contention window size for different access categories’ traffic at the MAC layer has been developed in Ref. [49]. In Ref. [50], two cross-layer access mechanisms were presented for contention access and contention-free access, respectively. Specifically, one such mechanism is implemented through the physical layer assisted link differentiation-distributed queuing MAC protocol and another one is implemented through the physical layer assisted link differentiation-multiple polling MAC protocol. These cross layer protocols facilitate an improved throughput for all types of data traffics in WLAN, including the video traffic which usually demands high throughput due to volumetric nature of the video data.
4.4 PHY-MAC-APP cross-layer solutions In theory, more than two network layers can be combined to further improve the performance of video over WLAN applications. The combination of PHY, MAC, and APP layers has been attempted in recent years [51-54]. China Communications • May 2013
However, the combination of more than two network layers usually results in very sophisticated interactions among them. The key to the success of such PHY-MAC-APP crosslayer solutions is how to efficiently mitigate the risk in terms of exponential increase in the state parameters for cross-layer interactions. A neural network based approach was proposed in Ref. [51] by adjusting the multidimensional MAC layer back-off parameters online in accordance with the APP layer QoS requirements as well as the PHY layer channel conditions. In Ref. [52], an effective QoS model defined by the H.264 group-of-pictures level estimator of decodable slice rate has been proposed. This APP layer QoS metric is integrated with an adaptive cross-layer feedback mechanism based on channel status estimation from PHY layer and maximum retry limit value as well as RTS/CTS function from the MAC layer to achieve significant performance gain over schemes without channel state feedback. Another triple cross-layer solution was developed in Ref. [53] for multiuser video streaming over WLANs. A general-sum switching control dynamic Markovian game was formulated by modeling the system state information as a finite state Markov chain. In this scheme, the system state information includes the PHY layer fading channel qualities, the MAC layer queue length and the queue state to reflect the SVC video priority structure which can be seen as information from APP layer. More recently, a cross-layer solution for video multicast over WLANs was also developed [54]. The cross-layer optimization architecture has been devised through a threelayer scheme in which the priority structure based on H.264 SVC in the APP layer, selective retransmission for erroneous multicast frame in the MAC layer, and the rate modes adaptation for SVC layers in the PHY layer have been seamlessly integrated to achieve much enhanced performance for video over WLANs. In summary, we have discussed four categories of cross-layer solutions for video transmission over WLANs. The classification of
13
video over WLAN approaches based on the associated network layers is able to clearly reveal the factors that need to be considered in the design of cross-layer solutions. In particular, the unique priority structure of video streams can be effectively utilized by multiple relevant network layers, MAC and PHY, and contribute to the success of several cross-layer solutions. By exploiting the interaction among different network layers, we show that the cross-layer solutions can achieve substantial performance improvement for video transmission over WLANs, compared with those singlelayer solutions we discussed in Section III.
V. BEYOND QOS PERFORMANCE Up to now, we have discussed mainly the QoS performance related solutions for video transmission over WLANs. Both single-layer and cross-layer solutions share the same goal in achieving the QoS guaranteed service for timecritical and error-sensitive video streaming over WLANs. However, in practical applications, several non-QoS measures will need to be considered to develop a globally optimal video over WLAN system. Among these non-QoS measures, the most frequently addressed issue besides QoS is the energy efficiency of the video over WLAN system simply because the mobile devices are often used as the receiving terminals for WLAN. For any mobile device, low energy consumption is an important aspect that needs to be carefully considered in the overall system design. The devices in WLANs are usually not so energy demanding as the mobile devices in cellular wireless network or multi-hop wireless network since devices in these two networks need to execute more connection or management functions, including handover, probing, routing, and so on. In WLANs, the energy consumption is often associated with the issue of QoS guarantees. In general, providing better QoS performance would consume more energy. An appropriate trade-off between the energy consumption and QoS guarantee needs to be carefully carried out to
14
develop globally optimal video over WLAN systems. In Ref. [55], Lagkas et al. proposed a scheme of low energy priority oriented adaptive control with QoS guarantee to support low energy consumption while guaranteeing the QoS for all types of multimedia networking applications. In Ref. [56], Gan et al. developed a cross-layer optimization scheme for multi-user video streaming over the uplink of IEEE 802.11e HCCA wireless network. The objective of this scheme is to minimize the energy consumption for all users, including both video encoder and wireless transmission, while delivering desired video quality for each user. Another important property to consider is the secure and privacy issue in the wireless video systems. This is particularly true with tremendous increase in social media applications via WLANs. For social media, the security and authentication requirements will become more and more critical. In Ref. [57], Sun et al. reviewed existing end-to-end packetloss-tolerant media authentication schemes including both stream-based and content-based methods and then described how to design authentication schemes for multimedia streaming that are tolerant to packet loss and capable of exploiting the unequal importance of different packets. In Ref. [58], Mokhtarian and Hefeeda developed an efficient authentication scheme for SVC video streams that can account for the full scalability of video streams and enables the verification of all possible substreams that can be extracted and decoded from the original stream. In Ref. [59], Yi et al. proposed a quality-optimized authentication of scalable media streams with flexible transcoding over wireless networks. For security and authentication issue, the research has been focused on the twisted problem of video stream’s quality performance and authentication performance over the error-prone wireless channels. This is because the security measures would always consume additional resource in wireless transmission which will lead to un-avoided degradation in video QoS. An optimal resource allocation is often the design China Communications • May 2013
target when security measures need to be integrated to the systems for video over WLANs.
VI. KEY OPEN PROBLEMS In this survey paper, we have briefly discussed QoS related problems in video delivery over WLANs. Towards the goal of QoS performance enhancement for video delivery over WLANs, single-layer and cross-layer solutions have been introduced. Single-layer solutions are aimed at designing improvement based on the single control parameter or algorithms of video codec and network layers. Single-layer design of video codec may not be just for WLAN applications and single-layer control optimization of WLANs is not limited to video delivery application, either. However, the cross-layer solutions which take advantage of both video and WLAN characteristics in their design demonstrate the highlight in combining multiple solutions to further enhance the desired QoS. These cross-layer solutions we have discussed represent some emerging trends in video over 802.11 WLAN. In spite of these recent advances, there are several key open problems associated with cross-layer solutions. 1) The optimization criterion As we move into a new era in user experience based evaluation for video transmission performance, the existing optimization criterion based on QoS, such as packets loss ratio, delay and delay jitter is gradually replaced by the criterion based on QoE, a more subjective evaluation criterion. QoE adopts the MOS, which has been originally proposed for voice applications [60] as a common metric for user-perceived quality. In Section 4.1, some approaches we reviewed have used QoE as their optimization criterion in the MAC-APP cross-layer solutions. It has been widely believed that QoE is a more efficient criterion to evaluate the performance of video transmission. Recently, more QoE-driven or QoE-aware cross-layer optimization schemes for wireless multimedia systems have been developed [61]. They attempt to optimize the video transmisChina Communications • May 2013
sion performance over wireless networks towards the goal of maximizing the QoE. However, many direct performance evaluation criterions of video transmission over networks are still based on QoS. How to properly transform from QoS criteria to QoE criteria still remains an open problem. 2) The cross-layer architecture In Section IV, we summarized a variety of cross-layer solutions for video transmission over WLANs. They can be classified into four categories, MAC-APP, PHY-APP, PHY-MAC and PHY-MAC-APP. These four categories are differentiated by which layers their cross layer architecture is based on. The category of cross-layer solutions based on PHY and MAC layers can be seen as a design for regular network performance enhancement. They are not limited to video delivery application. From the implementation point of view, this category of solutions can be easily realized and bear relatively low implementation cost. For the other three categories of solutions, their implementations are all related to the APP layer. They utilize the video stream characteristics which come from the APP layer and often have higher implementation cost. In particular, the category of PHY-MAC-APP cross-layer solutions that work across three layers needs more complicated cross-layer architecture and higher cost of implementation. Which category of the cross-layer solutions to adopt can be a difficult decision to make in practice? How to evaluate the advantage and disadvantage of each of these cross-layer solutions from even only the implementation point of view remains an open problem.
VII. CONCLUSION We have presented in this paper a brief survey of major techniques developed for video over 802.11 WLAN in the past decade. We focused on the review of approaches designed to achieve QoS performance gains based on both single-layer strategies and cross-layer strategies. When reviewing these approaches, we highlighted the unique characteristic of video
15
streams in their priority structure inherent in the popular video coding standard H.264/AVC and H.264/SVC. This characteristic can be efficiently utilized by various network layers for improved transmission. In the discussion of application scenarios for video over WLANs, we concluded that most schemes designed for video transmission over WLANs are associated with three network layers: the APP layer, the MAC layer and the PHY layer. Under the layer architecture, we summarized the solutions to improve QoS performance for video transmission over WLAN in two main categories: single-layer solutions and cross-layer solutions. Single-layer solutions are essentially following the technology evolution in video coding and WLANs themselves. However, crosslayer solutions still remain as open problems due to their complexity but clearly they can achieve greater performance improvement than the single-layer solutions. Finally, we also discussed two important issues associated with video transmissions over WLANs that are beyond the QoS performance. These issues are energy efficiency and media security which will become even more critical when the mobile social media is penetrating the global population. The low cost in WLAN deployment and the relatively high throughput of WLAN systems have made the WLAN a popular choice for mobile applications. With the advances in roaming support and integration with other wireless networks, such as LTE or WiMAX, we believe video over WLANs applications will continue to flourish for years to come. WLAN is also the most attractive network for a new generation of interactive video applications such as mobile video conferencing and mobile video gaming. With video consumers being the ultimate judge in the performance of any video application, we envision that the QoS driven technology innovation will soon be replaced by a new class of innovations aiming at improving QoE for the user, instead of the QoS performance of the networks. Holistic solutions are expected for next generation video over WLAN applications.
16
References [1]
IEEE. IEEE Standard 802.11 Part 1, Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) Specifications[S], 1997.
[2]
Cisco White Paper. Cisco Visual Networking Index: Global Mobile Data Traffic Forecast Update 2011-2016[EB/OL]. http://www.cisco.com/ en/US/solutions/collateral/ns341/ns525/ns53 7/ns705/ns827/white paper c11-520862.html.
[3]
IEEE. IEEE Standard 802.11a/b/g/n Part 11, Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) Specifications: High(er) Data Rate Enhancements[S], 1999/2000/2003/ 2009.
[4]
IEEE. IEEE Standard 802.11ac/ad Part 11, Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) Specifications (Under Development)[S], 2012.
[5]
IEEE. IEEE Standard 802.11e Part 11, Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) Specifications Amendment 8: Media Access Control (MAC) Quality of Service Enhancements [S], 2009.
[6]
IEEE. IEEE Standard 802.11aa Part 11, Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) Specifications Amendment 3: MAC Enhancements for Robust Audio Video Streaming[S], 2011.
[7]
ITU-T and ISO/IEC JTC 1. Generic Coding of Moving Pictures and Associated Audio Information — Part 2: Video, ITU-T Rec. H.262 and ISO/IEC 13818-2 (MPEG-2 Video)[S], 1994.
[8]
ITU-T and ISO/IEC JTC 1. Advanced Video Coding for Generic Audiovisual Services, ITU-T Rec. H.264 and ISO/IEC 14496-10 (MPEG-4 AVC) [S]. Version 1, 2003; Version 8 (including SVC extension), 2007.
[9]
SEELING P, REISSLEIN M, KULAPALA B. Network Performance Evaluation Using Frame Size and Quality Traces of Single-Layer and Two-Layer Video: A Tutorial[J]. IEEE Communications Surveys and Tutorials, 2004, 6(3): 5878.
[10]
KUMAR S, XU Liyang, MANDAL M K, et al. Error Resiliency Schemes in H.264/AVC Standard[J]. Journal of Visual Communication and Image Representation, 2006, 17(2): 425-450.
[11]
WAJAHAT J, AAMIR K, KHURRAM S, et al. Analysis of Video Coding and Error Resilience Tools of H.264 in Wireless Environment[J]. International Journal of Computer Applications, 2012, 50(1): 1-7.
[12]
HE Zhihai, CAI Jianfei, CHEN C W. Joint Source Channel Rate-Distortion Analysis for Adaptive Mode Selection and Rate Control in Wireless
China Communications • May 2013
Video Coding[J]. IEEE Transactions on Circuits
WLAN[J]. IEEE Communications Letters, 2006,
and Systems for Video Technology, 2002, 12(6): 511-523. [13]
[14]
10(11): 737-739. [23]
gorithm Using Video Prediction for Multime-
Optimized Streaming of Packetized Media[J].
dia IEEE802.11e WLAN[C]// Proceedings of
IEEE Transactions on Multimedia, 2006, 8(2):
International Conference on Wireless and
390-404.
Mobile Communications (ICWMC’06): July
ZHAI Fan, EISENBERG Y, PAPPAS T N, et al. An
29-31, 2006. Bucharest, USA. IEEE Press, 2006:
Integrated Joint Source and Channel Coding Framework for Video Transmission over Pack-
[15]
10-16. [24]
Theoretic Scheme for Efficient Video Trans-
national Conference on Image Processing
mission over IEEE802.11e EDCA WLANs[J].
(ICIP’04): October 24-27, 2004. Singapore. IEEE
ACM Transactions on Multimedia Computing,
Press, 2004: 2531-2534.
Communications and Applications, 2012, 8(3):
FRESIA M. Joint Source and Channel Coding[J].
32-56. [25]
Point Coordination Function with QoS and
PAVON J P, SUNGHYUN C. Link Adaptation
Power Management for Multimedia over Wire-
Strategy for IEEE802.11 WLAN via Received
less LANs[J]. IEEE Transactions on Networking,
ings of IEEE International Conference on Com-
Control Based on Rate-Variance Envelop for VBR Traffic Over IEEE 802.11e HCCA WLANs[J].
1113.
IEEE Transactions on Vehicular Technology,
REBAI A R, FLISS M, JARBOUI S, et al. A New
2008, (57)3: 1778-1788. [27]
Scheduling Algorithm for Multimedia Services
gies, Mobility and Security (NTMS’08): No-
in IEEE 802.11e WLAN[C]// Proceedings of
vember 5-7, 2008. Tangier, Morocco. IEEE Press,
IEEE Vehicular Technology Conference (VTC’11):
2008: 1-5.
September 5-8, 2011. San Francisco, CA, USA.
MAKHLOUF A B, HAMDI M. Practical Rate
IEEE Press, 2011: 1-6. [28]
Wenbin, et al. Downlink Channel Dynamic
cations, 2013, 12(2): 1-9.
Contention Free Access Mechanism Based on
NIU Huaning, NGO C. Diversity and Multi-
IEEE 802.11e EDCA[J]. Application Research of Computers, 2010, 27(8): 3083-3085. [29]
LIU Qian, ZOU Zixuan, CHEN C W. A Dead-
ference on Signals, Systems and Computers
line-Aware Virtual Contention Free EDCA
(ACSSC’06): October 29-November 1, 2006.
Scheme for H.264 Video over IEEE 802.11e
Pacific Grove, CA, USA. IEEE Press, 2006: 1644-
Wireless Networks[C]// Proceedings of IEEE
1648.
International Symposium on Circuits and
ZHANG Hongyuan, MOLISCH A F, ZHANG Jin.
Systems (ISCAS’11): May 15-18, 2011. Rio De
Applying Antenna Selection in WLANs for Achieving Broadband Multimedia Communi-
Janeiro, Brazil. IEEE Press, 2011: 625-628. [30]
LI Qiong, VAN DER SCHAAR M. Providing
cations[J]. IEEE Transactions on Broadcasting,
Adaptive QoS to Layered Video over Wireless
2006, 52(4): 475-482.
Local Area Networks through Real-Time Retry
LIU Qian, LIU Shujie, CHEN C W. A Novel Pri-
Limit Adaptation[J]. IEEE Transactions on Mul-
oritized Spatial Multiplexing for MIMO Wireless System with Application to H.264 SVC
[22]
FENG Zhengyong, WEN Guangjun, ZHANG
[J]. IEEE Transactions on Wireless Communi-
[C]// Proceedings of Fortieth Asilomar Con-
[21]
LEE K Y, CHO K S, RYU W. Efficient QoS
WLANs[C]// Proceedings of New Technolo-
plexing Switching in 802.11n MIMO Systems
[20]
GAO Deyun, CAI Jianfei, CHEN C W. Admission
chorage, Alaska, USA. IEEE Press, 2003: 1108-
Adaptation for Very High Throughput WLANs
[19]
2006, 14(4): 807-820. [26]
munications (ICC’03): May 11-15, 2003. An-
Link Adaptation Scheme for IEEE 802.11
[18]
CHOU Z T, HSU C C, HSU S N. UPCF a New
104-113.
Signal Strength Measurement[C]// Proceed-
[17]
PATRAS P, BANCHS A, SERRANO P. A Control
ets Loss Networks[C]// Proceedings of Inter-
IEEE Signal Processing Magazine, 2010, 27(6): [16]
LIU Huabing, ZHAO Yun. Adaptive EDCA Al-
CHOU P A, MIAO Zhourong. Rate-Distortion
timedia, 2004, 6(4): 278-290. [31]
BOBARSHAD H, VAN DER SCHAAR M,
video[C]// Proceedings of IEEE International
SHIKH-BAHAEI M R. A Low-Complexity Ana-
Conference on Multimedia and Expo (ICME’10):
lytical Modeling for Cross-Layer Adaptive Er-
July 19-23, 2010. Singapore. IEEE Press, 2010:
ror Protection in Video over WLAN[J] IEEE
968-973.
Transactions on Multimedia, 2010, 12(5):
YEN C M, CHANG C J, CHEN Y S. An Adaptive P-Persistent MAC Scheme for Multimedia
China Communications • May 2013
427-438. [32]
LIN C H, CE-KOEN S, CHILAMKURTI N K, et al.
17
[33]
A RED-FEC Mechanism for Video Transmission
tortion
over WLANs[J]. IEEE Transactions on Broad-
ror-Control[C]// Proceedings of the 10th In-
casting, 2008, 54(3): 517-524.
ternational Conference on Advanced Com-
MUNG C, LOW S H, CALDERBANK A R, et al.
munication Technology (ICACT 2008): Febru-
Layering as Optimization Decomposition: A
ary 17-20, 2008. Gangwon-Do, Korea. IEEE
Mathematical Theory of Network Architectures[J]. IEEE, 2007, 95(1): 255-312. [34]
[35]
Reference Frame Selecting Scheme for Real-
Layer Optimization[J]. IEEE Transactions on
Time Video Transport over Wireless LAN[J].
Vehicular Technology, 2009, 58(4): 1887-1903.
IEEE Transactions on Consumer Electronics,
KSENTINI A, NAIMI M, GUEROUI A. Toward an
2004, 50(1): 158-167. [45]
Link Adaptation Scheme for Efficient Trans-
chitecture[J]. IEEE Communications Magazine,
mission of H.264 Scalable Video over Multi-
2006, 44(1): 107-114.
rate WLANs[J]. IEEE Transactions on Circuits
LI Maodong, CHEN Zhenzhong, TAN Y P.
and Systems for Video Technology, 2008, 18(7): 875-887. [46]
Aware Rate Adaptation for MIMO WLANs[C]//
and Image Representation (JVCIR), 2011, 22(3):
Proceedings of the 19th IEEE International
284-296.
Conference on Network Protocols (ICNP): Oc-
SANTOS M A, VILLALON J, OROZCO-BARBOSA
tober 17-20, 2011. Vancouver, BC, Canada. IEEE Press, 2011: 321-330. [47]
and Transmission in Multirate Wireless LANs
Communications, 2012, 30(7): 1205-1214.
[J]. IEEE Transactions on Multimedia, 2008, 10(5): 691-700.
VAN DER SCHAAR M, ANDREOPOULOS Y, HU [48]
Video Transmission Over MIMO Wireless Sys-
Under Delay Constraints[J]. IEEE Transactions
tems With Adaptive Channel Selection Based
on Mobile Computing, 2006, (5)6: 755-768.
on Partial Channel Information[J]. IEEE Trans-
FALLAH Y P, NASIOPOULOS P, ALNUWEIRI H.
actions on Circuits and Systems for Video Technology, 2007, 17(9): 1218-1226. [49]
LIN Yuxia, WONG V W S. Adaptive Tuning of
Journal on Wireless Communications and
MIMO-Enabled 802.11e WLANs with Network
Networking, 2008: 1-14.
Utility Maximization[C]// Proceedings of IEEE
LU M H, STEENKISTE P, CHEN T. Video Strea-
Wireless Communications and Networking Con-
ming Over 802.11 WLAN with Content-Aware
ference (WCNC’08): March 31-April 3, 2008.
Adaptive Retry[C]// Proceedings of IEEE In-
Las Vegas, NS, USA. IEEE Press, 2008: 17511756.
ternational Conference on Multimedia and Expo (ICME’05): July 6-9, 2005. Amsterdam,
[50]
SUN Fanglei, LI V O K, DIAO Zhifeng, et al.
USA. IEEE Press, 2005: 723-726.
Medium Access Control with Physical-Layer
CHEN C M, LIN C W, CHEN Y C. Cross-Layer
-Assisted Link Differentiation[J]. IEEE Transac-
Packet Retry Limit Adaptation for Video Tran-
tions on Vehicular Technology, 2008, 57(2): 1255-1271.
sport over Wireless LANs[J]. IEEE Transactions on Circuits and Systems for Video Technology,
[51]
WANG C, LIN T, CHEN J L. A Cross-Layer
2010, 20(11): 1448-1461.
Adaptive Algorithm for Multimedia QoS Fair-
MAJUMDAR A, SACHS D G, KOZINTSEV I V, et
ness in WLAN Environments Using Neural
al. Multicast and Unicast Real-Time Video
Networks[J]. IET Communications, 2007, 1(5): 858-865.
Streaming over Wireless LANs[J]. IEEE Transactions on Circuits and Systems for Video
18
SONG D, CHEN C W. Scalable H.264/AVC
over IEEE 802.11a/e HCCA Wireless Networks
Multirate IEEE 802.11e WLANs[J]. EURASIP
[43]
ARGYRIOU A. Error-Resilient Video Encoding
WLANs[J]. IEEE Journal of Selected Areas in
Efficient Transmission of H.264 Video over
[42]
CHAN An, LUNDGREN H, SALONIDIS T. Video-
works[J]. Journal of Visual Communication
Zhiping. Optimized Scalable Video Streaming
[41]
FALLAH Y P, MANSOUR HM, KHAN S, et al. A
over IEEE802.11e Through a Cross-Layer Ar-
for Video Communications over IEEE802.11
[40]
CHEN Min, WEI Gang. Multi-Stages Hybrid ARQ with Conditional Frame Skipping and
L. A Novel QOE Aware Multicast Mechanism
[39]
Er-
Systematic Framework for Autonomous Cross-
livery over the IEEE 802.11e Wireless Net-
[38]
MAC-Level
Press, 2008: 1232-1237. [44]
Cross-Layer Optimization for SVC Video De-
[37]
and
FU Fangwen, VAN DER SCHAAR M. A New
Improvement of H.264 Video Transmission
[36]
Measure
[52]
OH B J, HUA Guogang, CHEN C W. Seamless
Technology, 2002, 12(6): 524-534.
Video Transmission over Wireless LANs Based
CHOI J Y, YOO J J, SHIN J. Cross-Layer Trans-
on an Effective QoS Model and Channel State
mission Scheme for Wireless H.264 Using Dis-
Estimation[C]// Proceedings of 17th IEEE In-
China Communications • May 2013
[53]
[54]
[55]
[56]
[57]
ternational Conference on Computer Com-
puter Science and Engineering at the University at
munications and Networks (ICCCN'08): Au-
Buffalo, State University of New York, since 2008.
gust 3-7, 2008. St. Thomas, US Virgin Islands,
Previously, he was Allen S. Henry Distinguished Pro-
USA. IEEE Press, 2008: 1-6.
fessor in the Department of Electrical and Computer
HUANG J W, MANSOUR H, KRISHNAMURTHY
Engineering, Florida Institute of Technology, from
V. A Dynamical Games Approach to Transmis-
2003 to 2007. He was on the faculty of Electrical and
sion-Rate Adaptation in Multimedia WLAN[J].
Computer
IEEE Transactions on Signal Processing, 2010,
Missouri — Columbia from 1996 to 2003 and at the
58(7): 3635-3646.
University of Rochester, Rochester, NY, from 1992 to
LIM W S, KIM D W, SUH Y J. Design of Efficient
1996. From 2000 to 2002, he served as the Head of
Multicast Protocol for IEEE 802.11n WLANs
the Interactive Media Group at the David Sarnoff
and Cross Layer Optimization for Scalable
Research Laboratories, Princeton, NJ. He has also
Video Streaming[J]. IEEE Transactions on Mo-
consulted with Kodak Research Labs, Microsoft Re-
bile Computing, 2012, 11(5): 780-792.
search, Mitsubishi Electric Research Labs, NASA
LAGKAS T D, PAPADIMITRIOU G I, NICOPO-
Goddard Space Flight Center, and Air Force Rome
LITIDIS P, et al. A New Approach to the Design
Laboratories.
[59]
at
the
University
of
of MAC Protocols for Wireless LANs Combing
Dr. Chen has been the Editor-in-Chief for IEEE Tran-
QoS Guarantee with Power Saving[J]. IEEE Com-
sactions on Circuits and Systems for Video Technol-
munications Letters, 2006, 10(7): 537-539.
ogy (CSVT) since January 2006. He has served as an
GAN Tong, DEJONGHE A, LENOIR G, et al.
Editor for Proceedings of IEEE, IEEE Transactions on
Cross-Layer Optimization for Multi-User Video
Multimedia, IEEE Journal of Selected Areas in Com-
Streaming Over IEEE802.11e HCCA Wireless
munications, IEEE Multimedia Magazine, Journal of
Networks[C]// Proceedings of IEEE Interna-
Wireless Communication and Mobile Computing,
tional Conference on Multimedia and Expo
EUROSIP Journal of Signal Processing: Image Com-
(ICME’08): June 23-26, 2008. Hannover, USA.
munications, and Journal of Visual Communication
IEEE Press, 2008: 505-508.
and Image Representation. He has also chaired and
SUN Q, APOSTOLOPOULOS J, CHEN C W, et al.
served in Numerous Technical Program Committees
Quality-Optimized and Secure End to End
for IEEE and other international conferences.
Authentication for Media Delivery[J]. IEEE, [58]
Engineering
He received his B.S. degree from University of Sci-
2008, 96(1): 97-111.
ence and Technology of China in 1983, M.S.EE degree
MOKHTARIAN K, HEFEEDA M. Authentication
from University of Southern California in 1986, and
of Scalable Video Streams with Low Commu-
Ph.D. degree from University of Illinois at Urbana-
nications Overhead[J]. IEEE Transactions on
Champaign in 1992. He was elected an IEEE Fellow for
Multimedia, 2010, 12(7): 730-742.
his contributions in digital image and video process-
YI Xiaowei, LI Mingyu, ZHENG Gang, et al.
ing, analysis, and communications and an SPIE Fellow
Quality-Optimized Authentication of Scalable
for his contributions in electronic imaging and visual
Media Streams with Flexible Transcoding over
communications. Email:
[email protected]
Wireless Networks[C]// Proceedings of 3rd
[60]
[61]
FTRA International Conference on Mobile, Ubi-
FENG Zhengyong, was born in Sichuan, China in
quitous and Intelligent Computing (MUSIC):
1978. He received his B.S. degree in physics from
June 26-28, 2012. Vancouver, BC, Canada. IEEE
China West Normal University, Nanchong, China in
press, 2012: 148-153.
2001 and the M.E. degree in communication engi-
International Telecommunication Union. Met-
neering from Institute of Electronics, Chinese Acad-
hod for Subjective Determination of Trans-
emy of Sciences, Beijing, China in 2004. Since 2004 to
mission Quality[S]. ITU-T Recommendation
2007, he was a Research Associate at China West
P.800, 1996.
Normal University. From 2007, he has been a Ph.D.
MARTINI M G, CHEN C W, CHEN Zhibo, et al.
candidate at School of Communication and Informa-
QoE-Aware Wireless Multimedia Systems[J].
tion Engineering, University of Electronic Science and
IEEE Journal on Selected Areas in Communi-
Technology of China, Chengdu, China. His research
cations, 2012, 30(7): 1153-1281.
interests include wireless networks, QoS of multim-
Biographies Chang Wen CHEN, has been a Professor of Com-
China Communications • May 2013
edia streaming over wireless networks and cross-layer optimization of wireless networks. Email:
[email protected]
19
