Natl. Acad. Sci. Lett. (March–April 2015) 38(2):113–119 DOI 10.1007/s40009-014-0297-5
RESEARCH ARTICLE
MPEG-4 Video Transmission Over IEEE 802.11e Wireless Mesh Networks Using Dynamic-Cross-Layer Approach Rowayda Sadek • Aliaa Youssif • Amal Elaraby
Received: 8 October 2013 / Revised: 28 January 2014 / Accepted: 14 February 2014 / Published online: 15 February 2015 Ó The National Academy of Sciences, India 2015
Abstract This paper proposes a dynamic cross-layer approach to enhance quality of MPEG-4 video stream transmission over an IEEE 802.11e wireless mesh networks. IEEE 802.11e standard either classifies video data to a specific access category (AC) which causes high packet loss at heavy load conditions or uses static mapping to map video data to lower priority AC which causes unnecessary transmission delays and packet losses in case of light load network. Some developed techniques dynamically map video frames to the other priority queues causing highly degradation in video quality. The paper work proposes a new approach that combines a new mapping cross-layer between the medium access control and the application layer for dynamically mapping video frames based on frame priority and queue load congestion with low priority queue parameters adaptation such as minimum and max contention window to reduce queuing delay. Proposed approach is compared with already developed schemes. Simulation results give high performance under light and heavy loads network traffic. All implementations are simulated using NS2 Simulator tool with required configuration to examine all possible types of traffics. Keywords IEEE 802.11e MPEG-4 video compression QoS Cross layer optimization Access category queue
Introduction Wireless multimedia transmission across wireless mesh networks (WMNs) [1] has been gaining attention in the R. Sadek (&) A. Youssif A. Elaraby Arab Academy for Science Technology and Maritime Transport (AASTMT), Cairo, Egypt e-mail:
[email protected]
recent years. The breakthrough in wireless networking has extended its domain from simple file transfer to real time multimedia services [2]. Providing multimedia services over WMN is always challenging due to the dynamic QoS requirement of applications and time varying nature of the wireless link. In order to cope with the above requirements, the parameter setting and tuning of protocol at design time can be done, but with poor performance and inefficient resource utilization. The optimal allocation of resources is possible only through dynamic adaptation of parameters in run time. It is generally possible to exchange parameter information in between layers in a timely manner during the period of network operation through cross-layer design architecture [3] ensuring interaction between layers to provide optimization for QoS guarantee. This new approach allows knowledge sharing between different layers to obtain high possible Adaptability and helps to increase network efficiency and better QoS support [4]. The design of efficient cross-layer framework is dependent on the selection of appropriate cross-layer architecture which entirely depends on the requirements of an application. The various crosslayer architectures involve creation of interfaces, merging adjacent layers, design coupling and vertical calibration [4]. Most existing link layer and network layer protocols are not adequate to support QoS for multimedia applications. In the MAC layer, it has been shown that distributed QoS mechanisms are difficult for IEEE 802.11 [5]. Many existing buffer management schemes do not consider the QoS requirements of multimedia applications. Traditional routing protocols strive for the shortest path but do not explicitly support QoS. A prominent feature of WMNs is that there are generally multiple routes between a single source–destination pair, which benefits QoS routing schemes to find better routes with less interference.
123
114
The concept of compressed video such as MPEG-4 has been proposed to cope with limited bandwidth constraint that has made MPEG-4 video codec popular in wireless environment and in internet streaming applications in recent years [6]. MPEG-4 video defines three types of video frames namely I (Intra-coded) frame, P (Predictive-coded) frame, and B (Bi-Directionally predictive-coded) frame. Each of these frames has different compression rates and frame dependency. Therefore, each frame needs discriminating QoS support according to frame dependency where I frame is highest priority and P frame is lowest priority video frame [7]. It has an advantage in the implementation of multimedia services because MPEG-4 improves compatibility compared to the existing MPEG-1 and MPEG-2, and also highly efficient compression is available. IEEE 802.11e is defined by task group E to provide QoS supportability through differentiated class of service. The key function of EDCA is to provide differentiated services to different traffic classes via Access category (AC) queues. EDCA maintains four AC queues named as AC_ VO, AC_ VI, AC_BE and AC_BK in order to transmit voice, video, best effort and background traffic, respectively [8]. Instead of classifying video data to a specific AC in an 802.11e network which causes high packet loss at heavy load conditions.And instead of using static mapping where if the network load is light, the video data which is mapped to lower priority AC will result in unnecessary transmission delays and packet losses. This paper has proposed the dynamic cross layer approach named AMLPQA. In which the decision of the proposed mapping algorithm is entirely based on the priority of video frames and traffic load of AC queue. In addition, low priority queue parameters have to be adjusted to receive also video frames in efficient way. Simulation has shown that AMLPQA provides better video quality at receiver maintaining the better average end-to-end delay. The rest of the paper is organized as follows. Section II describes the relating works. Section III is an overview on IEEE 802.11e. Section IV covers the proposed approach. Section V covers simulation model configuration. Section VI covers simulation results and discussions and concludes the paper in Section VII.
Related Work Cross-layer dynamic mapping technique [9] maps the arriving video packets into the four ACs to optimize video traffic and packet loss ratio. When the video queue is almost full the packets will get mapped to best effort or background traffic AC with equal probability. The crosslayer QoS-optimized EDCA adaptation algorithms in [10]
123
R. Sadek et al.
takes into account the unequal error protection characteristics of video streaming, the IEEE 802.11e EDCA parameters and the lossy wireless nature. It makes use of two models, video distortion model and channel throughput estimation model to predict the video quality. A cross-layer optimization for video streaming over wireless multimedia sensor networks is attempted in [11]. In 802.11 s mesh networks, packets are differentiated and higher priorities are given to forward packets. When queue length of AC2 fills up, forward packets are remapped to lower access category AC1. The Interaction between network abstraction layer (NAL) and IEEE 802.ll e MAC layer with cross layer approach conveys the priority information of scalable video coding (SVC) slices. Later the SVC slices are mapped into the appropriate ACs statically providing differentiated services for base layer and enhancement layers. This improved the video quality compared to EDCA [12]. The dynamic mapping of H.264 video packets in proper ACs based on priority information and also in presence of network congestion using random early detection (RED) concept was proposed [13]. This improved the video quality as compared to static mapping. Nevertheless the approach was only for single hop. Recently, [14] has also proposed the dynamic mapping algorithm for MPEG-4 video frames (I/P/B) in multihop MANET using AODV as routing protocol. This has improved the video quality but with expense of end-to-end delay.
Overview on IEEE 802.11e IEEE 802.11e EDCA is designed to enhance the 802.11 distributed coordination function (DCF) mechanisms by providing a distributed access method that can support service differentiation among different classes of traffic. EDCA classifies traffic into four different AC (as illustrated in Fig. 1 [15]). The four access categories include AC_VO (for voice traffic), AC_VI (for video traffic), AC_BE (for best effort). The EDCA parameter set includes minimum Contention Window size (CWmin), maximum Contention Window size (CWmax), arbitration inter frame space (AIFS), and transmission opportunity limit (TXOPlimit). The preferred values of each mechanism parameters that the standard recommends are shown in Table 1 [16]. Figure 2 demonstrates the operations in 802.11e EDCA. To achieve differentiation, instead of using fixed DIFS (Distributed Inter frame Space) (as in 802.11 DCF), EDCA assigns higher priority ACs with smaller CWmin, CWmax, and AIFS to influence the successful transmission probability (statistically) in favor of high-priority ACs. The AC with the smallest AIFS has the highest priority, and a station needs
MPEG-4 Video Transmission over IEEE 802.11e
115
Fig. 2 IEEE 802.11e EDCA operations
Fig. 1 Four access categories in IEEE 802.11e
to defer for its corresponding AIFS interval. The smaller the parameter values (such as AIFS, CWmin and CWmax) the greater the probability of gaining access to the medium. Each AC within a station behaves like an individual virtual station: it contends for access to the medium and independently starts its back off procedure after detecting the channel being idle for at least an AIFS period. The back off procedure of each AC is the same as that of DCF. When a collision occurs among different ACs within the same station, the higher priority AC is granted the opportunity to transmit, while the lower priority AC suffers from a virtual collision, similar to a real collision outside the station. IEEE 802.11e EDCA defines a TXOPlimit as the time interval during which a particular station can initiate transmissions. During this period, defined by a starting time and a maximum duration, stations are allowed to transmit multiple data frames from the same AC continuously within the time limit defined by TXOPlimit. In 802.11e EDCA the higher priority ACs have a longer TXOPlimit, while lower priority ACs have a shorter TXOPlimit. Priority differentiation used by EDCA ensures better service to high priority class while offering a minimum service for low priority traffic. Although this mechanism improves the quality of service of real-time traffic, the performance obtained is not optimal since Table 1 Access categories and relevant parameters Access category
AC_VO
AC_VI
AIFS
2
2
3
7
CWmin
7
15
31
31
1,023
1,023
0
0
Cwmax
15
31
TXOPlimit
0.0031
0.006016
AC_BE
AC_BK
EDCA parameters cannot be adapted according to the network conditions. In current IEEE802.11e [14], there are two approaches to map MPEG-4 video frames (l/P/B) into ACs of IEEE 802.11e MAC: EDCA approach and static approach, as shown in Fig. 3. The EDCA approach allocates all video frames into a single AC for Video (AC_ VI) queue, while the static approach allocates I, P and B video frames into three separate ACs namely AC_ VI, AC_BE (Access Category for Best Effort) and AC_BK (Access Category for Background), respectively. These video frames mapping techniques cannot distinguish the frames’ importance, which lead towards degradation of video quality at the receiver. There is no cross-layer framework to allow dynamic mapping of video frames into ACs of MAC.
Adaptive Mapping Cross-Layer with Low Priority Queues Parameters Adjustment (AMLPQA) Algorithm In this framework, the Application layer passes its video frames (I/P/B) information to MAC layer through the Network layer see Fig. 4. The video frame information is passed into differentiated service code point (DSCP) field of Network layer and then to the MAC layer. Based on the frame received, MAC layer will perform dynamic mapping to allocate video frames (l/P/B) into different AC queues. The decision of the proposed mapping algorithm is entirely based on the priority of video frames and traffic load of AC queue. When a video data frame arrives, the algorithm checks the queue length against the threshold_low. If the condition satisfies the algorithm keeps the packet in AC [2] queue. If not, the algorithm checks AC [2] queue whether it is less than threshold_high, and if the condition satisfies then it calculates the probability using formula (1) of the packet. Next, a uniformly distributed random number is generated in the range [0.0–1.0]. If this random number is greater than the probability then, the packet is kept in AC [2]. Otherwise the packet is
123
116
R. Sadek et al.
Fig. 3 EDCA and static mapping of video frames
placed in AC [1]. If the queue length of AC [2] is greater than the threshold_high, and the random number generated is greater than Prob_TYPE, then the packet is kept in AC [1], otherwise AC [0] is used for this packet. ProbNew ¼ Probtype
qlenðAC½2Þ threshlow threshhigh threshlow
ð1Þ Fig. 4 Proposed adaptive cross-layer (AMLPQA) approach
The following section describes the algorithm in details: Simulation Model Configuration • • • •
Define thresh_low ……… Define thresh_High …….. Define Random function ( R ) range [0:1] Define Prob_type: • Set Prob_I = 0…..…(Probability of dropping I-Frame) • Set Prob_P = 0.6…...(Probability of dropping P-Frame) • Set Prob_B = 0.9…...(Probabilityof dropping B-Frame) • Define qlen (AC [2]) ………………getAvgQueue.AC [2] 1- Adaptive Mapping Cross-Layer Section - when frame arrived - Check Dropping probability of video frame into Prob_type - Check average queue length of AC [2] - If qlen < thresh_low Map Packet to AC[2] else if (thresh_low < qlen < thresh_High) { - Calculate Porb_New as per the formula in (1) - If (R> Porb_New) Map packet to AC[2] else Map packet to AC [1] ) else if (qlen(AC[2] > thresh_high) { - If (R > Prob_new) { Map packet to AC[1] else Map packet to AC [0] } 2- Adjusting low priority queues parameters CWnew [AC] = (( CWcurrent [AC] + 1)*2) – 1)
Video streaming performance is examined in a grid WMN where the network topology and traffic are well defined and can be easily controlled. We add 16 nodes in a 200 mx200 m square area with equal 50m horizontal and vertical spacing as shown in Fig. 5. This grid network has the key features of a WMN: all nodes are connected to each other via multihop links and multiple paths exist between the source and the destination. Table 2 lists the characteristics of the traffic we inject into the network. A congested network scenario common to the bandwidth-consuming video applications is created. For the video streams, we use the standard QCIF (176 9 144) ‘‘foreman’’ clip (400 frames in raw YUV format) see
(2)
Use formula (2) to adapt the following queues parameters : • Set (CWmin, CWmax) For AC[1]= (31-63) • Set (CWmin, CWmax) For AC [0] = (63-127) • Set AIFS= 2 for all AC’s
Fig. 5 A grid wireless mesh network
123
MPEG-4 Video Transmission over IEEE 802.11e Table 2 Network Traffic Flows Parameters
Table 3 Parameters of encoded video sources
117
Flow parameters
Voice
Video
BE
BK
Packet size Data rate
160 bytes
1,500 bytes
200 bytes
1,000 bytes
64 Kb
0.2 Mb
125 Kb
512 Kb
No. of packets
2,250
659
1,172
960
Duration
(5–50) s
(10–50) s
(20–35) s
(15–30) s
Video source
Forman
Video format
QCIF
Frame number Packet number
Table 3. It is compressed using FFMPEG [17] into a MPEG-4 stream (30 frames per second, 9 frames in a Group of Pictures). We used different number of flows for each traffic type one flow, two flows, three flows, five flows and eight flows represented by case1, case2, case3, case4, and case5, respectively. Simulation parameters can be seen in Table 4.All simulations are done with the NS2 simulator [18].
Simulation Results and Discussions Configured scenario contains five different loading cases, including different loads of voice, video, best effort, and background. Traffic flows were randomly generated and transmitted over the entire simulation. Table 4 Simulation parameters
I
P
B
Total
45
89
266
400
I
P
B
Total
237
149
273
659
In this scenario, the received video quality is analyzed to evaluate the efficacy of the proposed scheme under various network loading conditions using Average Peak Signal to Noise Ratio (PSNR) as a measure for video quality, Packet Delivery Ratio (PDR) as a packet loss measure, video frame loss rate and average end-to-end delay. Figures 6, 7 and Table 5 show that PDR and PSNR under heavy load are less for EDCA 802.11e, the static mapping because static mapping places the frames into lower priority queues and EDCA 802.11e puts all video frames in one queue which causes overflow and packet loss in case of heavy loads. Proposed system had overcome these problems by mapping the video frames to access categories (AC’s) based on frame significance and network
Parameter
Value
Simulation environment
NS2 integrated with evalvid tool
Area size
200 9 200 m2
Data rate
1 Mbps
Dropping probability of I,P & B frames (Prob I, Prob P, Prop B)
0,0.6 and 0.9 respectively
Maximum queue length
50 packets
Minimum, maximum queue threshold (minith , maxth )
10 and 40 packets respectively
Video Source
Foreman YUV QCIF(176 9 I44) (foreman)_qcif.yuv)
Number of video frames
400 Frames
Number of video packets
659 Packets
Group of picture (GOP)
9
Frame Sequence in GOP
IBB PBB PBB
Simulation time
50 s
Packet Size
1,500
Access category for voice (AC_VO),For video (AC_VI) For Best Effort (AC_BE) and For Background (AC_BK)
AC [3], AC [2], AC [1], AC [0] respectively
123
118
R. Sadek et al.
Fig. 6 Video packet delivery ratio
Fig. 8 Voice over IP (VOIP) packets received
Fig. 7 Video average peak signal to noise ratio
Fig. 9 Constant bit rate (CBR) packets received
Table 5 Average peak signal to noise ratio IEEE 802.11e
Static
Proposed AMLPQA
Case 1
34.887196
33.69253
34.887196
Case 2
33.752071
33.00534
34.175568
Case 3
29.023055
28.52341
32.831722
Case 4
28.556275
27.55426
31.075495
Case 5
27.261033
26.38219
29.381371
traffic load and adapting for low priority queues to be ready and fast to serve highly significant video Frames. From Fig. 8 it is observed that VoIP received packets have been affected a little due to queues parameters adaptation but we consider this still under acceptable percentage and we will try to enhance this in our future work. We notice from Fig. 9 that proposed system achieved high packets received percentage for CBR streams case and this due to adapting low priority queue parameters to become more efficient for video frames mapping. The proposed
123
Fig. 10 Video frames loss rate
System achieved superior results in case of video frame loss in compared with EDCA, static mapping approach see Fig. 10. Figure 11 shows that proposed AMLPQA approach reduced average End-to-End delay and this due to adapting queue parameters for lower priority queues. Hence, it can be concluded that proposed AMLPQA performs better than
MPEG-4 Video Transmission over IEEE 802.11e
Fig. 11 Average packet end-to-end delay
EDCA and static approaches and exploiting such a priority scheme, queue length management strategy and adapting low priority queues parameters the transmissions are prioritized and the drop rate of video is minimized, along with efficient utilization of network resource.
Conclusion In this paper, a new adaptive approach is proposed to improve the video delivery quality of MPEG-4 video transmission over IEEE 802.11e WMNs. The proposed approach combines two steps; the first step is dynamically allocating video packets to the most appropriate access categories according to network traffic loading conditions. The second step is carried out by adapting queuing parameters for lower priority queues. Simulations results ensure that proposed approach successfully enhances the video transmission quality. The results show that, under both light and heavy network loading conditions. The proposed scheme performs better than 802.11e EDCA and the static mapping scheme in terms of PDR with increasing up to 62 %. Average increase in PSNR has reached up to 29.38 %. Video frame loss rate has decreased to less than 60 % as well as highly limitation on average end-to-end delay under heavy load network conditions.
119 2. Fallah YP, Koskinen D, Shahabi A, Karim F, Nasiopoulos P (2007) A cross layer optimization mechanism to improve H.264 video transmission over WLANs. consumer communications and networking conference, Las Vegas, pp 875–879 3. Khan S, Peng Y, Steinbach E, Sgroi M, Kellerer W (2006) Application-driven cross-layer optimization for video streaming over wireless networks. IEEE Commun Mag 44(1):122–130 4. Srivastava V, Motani M (2005) Cross-layer design: a survey and the road ahead. IEEE Commun Mag 43(12):112–119 5. Pattara-atikom W, Krishnamurthy P, Banerjee S (2003) Distributed mechanisms for quality of service in wireless LANs. IEEE Wirel Commun Mag 10(3):26–34 6. Strachan D (1996) Video compression, society of motion pictures and television engineers. J SMPTE 105:68–73 7. Lin C, Shieh C, Ke C, Chilamkurti N, Zeadally S (2009) An adaptive cross-layer mapping algorithm for MPEG-4 video transmission over IEEE S02.lle WLAN, telecommunication systems. Springer, Netherlands, pp 223–234 8. Ke C-H, Lin C-H (2009) Evaluation of video stream quality over IEEE 802.lle EDCF. 2009 [Online]. Available: http://csie. nqu.edu.tw/smallko/ns2/ns2.htm 9. Zhu X, Girod B (2010) Distributed media-aware rate allocation for wireless video streaming. IEEE Trans Circuits Syst Video Technol 20(11):1462–1474 10. Saesue W, Chou CT, Zhang J (2010) Cross-layer QOS optimized EDCA adaptation for wireless video streaming. In: Proceedings of 2010 IEEE 17th international conference on image processing 11. Zhang J, Ding J (2010) Cross layer optimization for video streaming over wireless multimedia sensor networks. In: 2010 international conference on computer application and system modeling (ICCASM 2010) 12. Chen HL, Lee PC, Hu SH (2008) Improving scalable video transmission over IEEE S02.11e through a cross-layer architecture. In: proceeding on 2008 IEEE fourth international conference on wireless and mobile communication (lCWMC’OS), pp 241–246 13. Soni R, Chilamkurti N, Giambene G, Zeadally S (2008) A crosslayer design for H.264 video stream over wireless local area networks. In: Proceeding on 2008 IEEE international symposium on computer science and its applications, pp 3S7–392, 13–15 Oct 2008 14. Sanguankotchakorn T, Pudashine K (2011) A cross-layer framework for efficient MEPG-4 video streaming over IEEE S02.11e in MANETs. In: Proceeding on IEEE Sth international conference of ECT!-CON 2011, Khon Kaen, 17–19 May 2011 15. Kong Z-N, Tsang DHK, Bensaou B, Gao D (2004) Performance analysis of IEEE 802.11e contention-based channel access. IEEE J Sel Areas Commun 22(10):2095–2106 16. Mangold S, Choi S, Hiertz GR, Klein O, Walke B (2003) Analysis of IEEE 802.11e for QoS support in wireless LANs. IEEE Wirel Commun Mag 10:40–50 17. FFMPEG video codec, http://ffmpeg.org/. Accessed March 2013 18. The Network Simulator ns2 can be found at http://www. isi.edu/nsnam/ns/. Accessed March 2013
References 1. Akyildiz IF, Wang X, Wang W (2005) Wireless mesh networks: a survey. Comput Netw J 47(3):445–487
123
