THE DEMAND for wireless packet-data applications

Transcription

1 1218 IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY, VOL. 17, NO. 9, SEPTEMBER 2007 Scalable H.264/AVC Video Transmission Over MIMO Wireless Systems With Adaptive Channel Selection Based on Partial Channel Information Daewon Song, Student Member, IEEE, and Chang Wen Chen, Fellow, IEEE (Invited Paper) Abstract In this paper, we present a novel joint application physical-layer design (JAPLD) strategy to cost-effectively transmit scalable H.264/AVC video over multi-input multi-output (MIMO) wireless systems. With this approach, the application layer cooperates with the physical layer to maximize the system performance. First, in physical layer, we propose a new layered video transmission scheme over MIMO: adaptive channel selection (ACS). ACS-MIMO is fundamentally different from parallel transmission MIMO (PT-MIMO). While each bit stream is continuously transmitted through a fixed antenna in PT-MIMO, ACS-MIMO is able to periodically switch each bit stream among multiple antennas. In application layer, Scalable Video Coding (SVC) generates layered bit streams that need prioritized delivery. Then, we obtain the ordering of each subchannel s SNR strength as partial channel information (CI) at the receiver. The partial CI is acquired via the estimated channel state information based on training sequences. The JAPLD strategy we developed in this research shall switch the bit stream automatically to match the ordering of SNR strength for the subchannels. Essentially, we will launch higher priority layer bit stream into higher SNR strength subchannel by the proposed JAPLD algorithm. In this fashion, we can implicitly achieve automatic unequal error protection (UEP) for layered SVC transmission over MIMO system without power control at the transmitter. Experimental results show that the proposed ACS-MIMO system is able to achieve UEP with the obtained partial CI and the reconstructed video peak signal-to-noise ratio demonstrate the performance improvement of the proposed system as compared with open loop PT-MIMO system. Index Terms Adaptive channel selection (ACS), joint application physical-layer design (JAPLD), multi-input multi-output (MIMO), partial channel information (CI), Scalable Video Coding (SVC), unequal error protection (UEP). I. INTRODUCTION THE DEMAND for wireless packet-data applications (wireless web browsing, real-time mobile multimedia streaming, interactive applications, etc.) beyond conventional voice communication is increasing at an explosive rate. However, the inherently limited channel bandwidth and the Manuscript received May 2, 2007; revised July 15, This research was supported in part by the Florida Institute of Technology Allen S. Henry Endowment Fund. This paper was recommended by Guest Editor T. Wiegand. The authors are with the Department of Electrical and Computer Engineering, Florida Institute of Technology, Melbourne, FL USA ( dsong@fit. edul cchen@fit.edu). Color versions of one or more of the figures in this paper are available online at Digital Object Identifier /TCSVT unpredictability of the propagation channel becomes significant obstacle for wireless communication providers to offer high quality multimedia services. To overcome such obstacles, multi-input multi-output (MIMO) system has recently emerged as one of the most prominent techniques [1] [4]. In the past, MIMO systems were normally used to increase receiver or transmit (spatial) diversity so as to reduce the high bit-error rate (BER) of mobile wireless channel [1], [2]. Spatial multiplexing techniques [3], [4] have been investigated to simultaneously transmit independent data in order to achieve high data rate wireless multimedia communications. If perfect channel state information (CSI) at the receiver is available at the transmitter [5] through feedback, closed loop (CL) MIMO systems can maximize the channel capacity through well-known water-filling (WF) solution based on singular value decomposition (SVD). Assuming that the CSI is only available at the receiver side, open loop (OL) MIMO systems such as V-BLAST [4] attempt to decompose high-rate bit stream into independent sequences and transmit simultaneously. A. Existing Approaches in Multimedia Over MIMO One niche area of application in MIMO wireless systems is the transmission of multimedia (image/video) data. There have been several research works to report multimedia transmission over MIMO systems. In order to achieve robust image/video transmission over mobile wireless channels, space-time block codes (STBC) that is capable of alleviating the deep fades due to the time-varying nature of the channel was employed so as to utilize the spatial diversity [26] [28]. Those systems demonstrated that STBC-MIMO is able to achieve PSNR gain as compared with SISO system. In this work, we focus on video application over MIMO using spatial multiplexing since the high data rate is imperative for high quality video reconstruction. Scalable video codec was normally adopted to generate multi-layered video bit streams so as to launch independent layer bit stream to each antenna over parallel transmission [15] MIMO (PT-MIMO) as shown in Fig. 1. Video delivery over OL PT-MIMO could be found in [14]. This scheme adopted the hybrid of spatial multiplexing and spatial diversity to realize the robust video transmission over wireless channels. Video sequences were layered coded with different priority and each layer bit stream was transmitted over two antennas to provide a diversity gain. An unequal error protection (UEP) scheme based on the product Reed Solomon /$ IEEE

2 SONG AND CHEN: SCALABLE H.264/AVC VIDEO TRANSMISSION OVER MIMO WIRELESS SYSTEMS 1219 Fig. 1. PT-MIMO. (RS) codes was employed to provide the different priority protection. More research works on video transmission schemes over CL PT-MIMO are informed to maximize the advantage of MIMO capacity with assumption: perfect CSI is available at the transmitter. In [15], power allocation was performed to optimize the error performance according to the importance of the source layer so as to achieve UEP. Layered scalable video transmission schemes over SVD-based MIMO systems could be found in [16] [19]. By SVD, MIMO system is transformed to parallel SISO subchannels. The optimal power allocation scheme for minimizing the total system distortion based on joint source channel coding (JSCC) principles was proposed in [16] and video broadcast scenario was considered in [17]. In order both to maximize MIMO system throughput and to guarantee quality of service (QoS), we proposed optimal power allocation scheme with adaptive modulation for 2 2 MIMO system in [18] and extended for general MIMO systems in [19]. B. Limitations of Existing Approaches In MIMO wireless systems, independent symbols are transmitted across multiple antennas by sharing the available frequency and therefore, to detect the transmitted symbols, the CSI has to be estimated at the receiver side. However, in OL PT-MIMO, this estimated valuable information is useless at the transmitter since the estimated CSI is not feedback to the transmitter. In addition, equal power as optimal solution has to be allocated across multiple antennas and hence OL PT-MIMO systems are not appropriate for transmitting compressed video data that need prioritized transmission. In order to successfully transmit scalable video, this approach has to assign UEP channel codes according to the importance as described in [14]. As shown in Section V-B, this would result in losing the peak signal-to-noise ratio (PSNR) gain in high BER situation (or low SNR) because it allocates high redundant channel codes to strongly protect the base layer (loss of the source coding rate). On contrast, for scalable video transmission over CL PT-MIMO, all existing schemes [15] [19] assume that the perfect CSI is available at the both transmitter and receiver. From practical point of view, the perfect CSI is not attainable and the delay of feedback CSI is inevitable. The performance of the existing layered video transmission schemes over CL PT-MIMO systems is clearly dependent on the accuracy of the estimated CSI and feedback delay. Therefore, a more practical scalable video transmission scheme over MIMO in which the estimated CSI at the receiver is utilized to improve the system performance has become a challenging research topic. C. Main Contributions of the Research In this research, in order to overcome the challenge issue of scalable video delivery over OL PT-MIMO and CL PT-MIMO, we present a novel joint application physical-layer design (JAPLD) strategy to efficiently transmit scalable video over MIMO wireless systems. In physical layer, we propose a new layered video transmission scheme: adaptive channel selection (ACS). In ACS-MIMO system, each bit stream can be periodically switched among multiple antennas. In application layer, layered bit streams that need prioritized delivery are created by scalable video codec. Then, in order to make the application layer cooperated with the physical layer, we obtain the partial channel information (CI), the ordering of each subchannel s SNR strength, at the receiver and feedback to the transmitter. The partial CI is acquired via the estimated CSI based on training sequences. With the obtained partial CI, the JAPLD algorithm will launch higher priority layer bit stream into higher SNR strength subchannel. Therefore, we can implicitly achieve automatic UEP for layered Scalable Video Coding (SVC) transmission over MIMO system without power control at the transmitter. Following are the main contributions of the proposed JAPLD algorithm. 1) The estimated valuable CI is utilized to obtain the partial CI which is able to enhance the system performance as compared with OL PT-MIMO. 2) The obtained partial CI is small enough so that overcome the limited feedback bandwidth and delay. 3) The proposed system is very robust to the error of the obtained partial CI. In other words, with the worst partial CI, the long-term average BER performance of the proposed system converges to the one of OL PT-MIMO system. II. PRINCIPLE OF BITSTREAM SWITCHING In this section, we describe how the ACS scheme is different from parallel transmission [15]. The main purpose of the proposed transmission scheme is that each bit stream can be transmitted over a proper subchannel among multiple antennas. Thus, the proposed bit stream switching technique in physical layer can be collaborated with the application layer. Two transmission schemes at the transmitter are illustrated in Figs. 1 and 2, respectively. Video sequences are firstly compressed and partitioned into -layers according to the importance of the compressed data. In OL PT-MIMO, independent each bit stream is continuously transmitted through a fixed antenna as shown in Fig. 1. Then, the long-term average BER of each subchannel is almost same because of no knowledge on the CSI at the transmitter and allocating equal power across multiple antennas. Thus, in order to achieve the prioritized delivery of compressed bit streams over OL PT-MIMO, this scheme has to adjust the channel coding rate according to the importance of each bit stream. On contrast, in ACS-MIMO, each bit stream can be periodically switched among multiple antennas as shown in Fig. 2. Although the transmitter allocates equal power (no power control) across multiple antennas, UEP would be automatically achieved by not the channel coding but the proposed JAPLD technique based on the partial CI. The partial CI can be obtained through the simple channel estimation (CE) and calculation as described

3 1220 IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY, VOL. 17, NO. 9, SEPTEMBER 2007 Fig. 2. ACS-MIMO. Fig. 3. Proposed video delivery over ACS-MIMO system. in Section IV. Based on this information, the transmitter should allocate each bit stream to a proper antenna every CE (or bursty) period. III. SYSTEM DESCRIPTION The proposed layered scalable video transmission over ACS-MIMO wireless system is shown in Fig. 3. An example of 4 4 MIMO system is described. At the encoder side, 4-layer scalable video bit streams are generated by scalable H.264/AVC [8], channel codes are added, fed to the proper subchannel by JAPLD algorithm, modulated, and launched via multi transmitter antennas. At the receiver side, the CSI is estimated using pilot symbols. The transmitted signals are detected by linear zero-forcing (ZF) [or minimum mean-squared error (MMSE)] receiver [30] and decided by demodulation. Then, decided symbols are unloaded to the proper subchannel buffer by JAPLD algorithm, bit errors are corrected by channel coding, and finally transmitted video sequences are reconstructed. A. Scalable H.264/AVC With the proposed video transmission scheme over MIMO systems, multilayered video bit streams are essential and thus created by SVC [8] in this research. In this subsection, we briefly review this video codec which is an extension of the H.264/AVC [12], [13] video coding standard. Traditionally, there are two different ways for scalable video codec: either by using a technique that is intrinsically scalable (such as bit plane arithmetic coding) or by using a layered approach. SVC supports a combination of the two approaches so that a full spatio-temporal and quality scalable codec is achieved. A coded SVC video sequence consists of a series of Network Abstraction Layer (NAL) units, each containing the layer information. A 4-byte SVC NAL unit header, an extension of H.264/AVC NAL, indicates the decoding dependency relationship of spatial, temporal, and quality scalability. The temporal scalability of SVC is typically given based on the principle of hierarchical B-pictures. In a video sequence, the first picture (key picture) is intra-coded as IDR picture and next key picture will be either intra-coded or inter-coded using previous key picture. All pictures between two key pictures are hierarchically predicted and encoded. Therefore, a group of pictures (GOP) is created by a key picture and all pictures that are temporally located between a key picture and the previous key picture. Spatial scalability [11] is also supported based on

4 SONG AND CHEN: SCALABLE H.264/AVC VIDEO TRANSMISSION OVER MIMO WIRELESS SYSTEMS 1221 Fig. 4. Block diagram of a PAT transceiver. existing multilayered coding approach. For spatial scalability, inter-layer prediction tools are employed not only improve coding efficiency compared to simulcast scenarios but also reduce decoding complexity with single loop decoding. For quality (or SNR) scalability, SVC supports two types: coarse grain scalability (CGS) using various inter-layer prediction techniques and fine grain scalability (FGS) known as progressive refinement. In this work, we adopt FGS for SNR scalability in order to both satisfy each layer s target bit rate and increase the error robustness. Within each spatial resolution FGS is achieved by encoding successive refinements of the transform coefficients. Therefore, a picture is represented by base representation and FGS refinement representations by repeatedly decreasing the quantization step size. The NAL units of FGS refinement layers can be truncated at any arbitrary point at the encoder and thus error robustness is increased by the decoder capable of arbitrarily discarding corrupted NAL unit streams. For a more detailed explanation on SVC, see [8] and [10]. B. Wireless MIMO Channel Model In this work, the primary MIMO channel model under consideration is a quasi-static, frequency nonselective, and Rayleigh fading channel model. For a single user flat-fading channel over MIMO system with transmitter antennas and receiver antennas, the system equation is where is the quasi-static and complex channel matrix, is the received signal vector, is the transmitted signal vector, and is the noise vector from i.i.d. Gaussian collection with zero mean, independent real and imaginary parts, with variance. When the CSI is only available at the receiver, the optimal power allocation at the transmitter is with global power. IV. JOINT APPLICATION PHYSICAL-LAYER DESIGN In this section, we present the JAPLD in detail. The partial CI, the ordering of each subchannel s SNR strength, is critical for the proposed ACS-MIMO system so that the application layer is able to cooperate with the physical layer to increase the system performance. Before obtaining the partial CI, the CE method will be given. (1) A. Channel Estimation Normally, the CSI can be obtained by two different methods. One is called blind CE [20], [21], which uses the statistical property of the channel and properties of the transmitted signals. The other is called training-based CE [22], which is based on the training sequences which are known at the receiver. Though blind CE does not cause any increase in overhead, it requires long data record. In other words, it is very sensitive to the CSI feedback delay and is only applicable to slowly time varying channels. Therefore, in this work, the CSI is estimated by employing pilot symbols [22]. The pilot symbols are traditionally time multiplexed and the block diagram of a simplified pilot-assisted transmission (PAT) is illustrated in Fig. 4. The training sequence of th transmit antenna is represented as where and is the length of the training sequence. Then, the training sequence matrix is generated as follows: The training sequence received signals can be represented as the matrix where and are matrices. Then, the maximum-likelihood (ML) estimation of the channel matrix is given by where represents pseudo-inverse. Here the estimated MIMO CI will be used for the partial CI. B. Partial Channel Information Through the estimated MIMO CI in the previous subsection, now we are able to obtain the partial CI. By employing linear ZF detection algorithm [30], the received signal can be modified as where. With assumption, the (5) can be written as (2) (3) (4) (5) (6)

5 1222 IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY, VOL. 17, NO. 9, SEPTEMBER 2007 Then, th subchannel s SNR is given by (7) where is th row of is norm, and is Hermitian conjugate. Therefore, based on the estimated MIMO CI, the ordering of each subchannel s SNR strength is calculated and will be used as the partial CI. C. Summary of JAPLD Algorithm In this subsection, the proposed JAPLD algorithm that automatically switches the bit stream so as to match the ordering of SNR strength for the subchannels can be summarized as following five steps. 1) Training sequences are transmitted for MIMO CE at the transmitter every CE (or bursty) period. 2) The MIMO channel is estimated by known training sequences using (4) at the receiver side. 3) Based on the estimated CSI, each subchannel s SNR strength is calculated using (7) and the ordering of SNR strength as the partial CI is feedback to the transmitter. 4) At the transmitter side, each layer bit stream is loaded to the proper channel according to the obtained partial CI and transmitted during bursty period. As an example, if the ordering of SNR strength is 3, 1, 2, 4, the bit loading of each layer bit stream is therefore described in Fig. 3. 5) The bit unloading of the estimated transmission symbols is inversely processed at the receiver side as described in Fig. 3. We assume in this work that the estimated partial CI can be feedback over reliable channel and synchronized at the both transmitter and receiver through keeping fixed feedback delay. D. Detection and Video Reconstruction The transmitted signals are detected by linear ZF or MMSE receiver. To estimate the transmitted signals through ZF, the received signals are calculated using (5). The received signals via MMSE detector [30] is given as where is identity matrix. Then, the adopted demodulation decides the transmitted signals based on the received signals and the decided symbols are unloaded to the proper subchannel buffer by JAPLD algrithm. Corrupted NAL unit streams might be corrected by channel coding or discarded and finally transmitted video sequences are reconstructed. V. EXPERIMENTAL RESULTS We provide numerical examples to show how the proposed ACS-MIMO system is able to achieve the promising goals as described in Section I-C through the JAPLD algorithm. In this section, we will use a 4 4 MIMO system under independent Rayleigh fading. The elements of the MIMO channel matrix are obtained from Clarke and Jakes model [23], [24] with (Doppler frequency) 10 Hz. Noise vector is from i.i.d. Gaussian collection with zero mean, independent real and (8) Fig. 5. Long-term average BER of OL PT-MIMO with ZF. Fig. 6. Long-term average BER of OL PT-MIMO with MMSE. imaginary parts, with variance. Equal power is allocated to each subchannel and QPSK is used for modulation. The data rate of each subchannel is considered with 256 kbps. The known training sequences with is transmitted every s for MIMO CE. The partial CI, the ordering of subchannel s SNR strength, is obtained at the receiver and feedback to the transmitter with time delay. At the receiver side, for the detection of transmitted signals, we adopt liner ZF or MMSE receiver [30]. A. BER Performance of ACS-MIMO versus OL PT-MIMO In traditional OL PT-MIMO systems, each subchannel s long-term average BER will be converged since the transmitter assigns equal power and each independent path across multiple antennas has the same statistical characteristics. Figs. 5 and 6 show the long-term average BER of a 4 4 OL PT-MIMO system with ZF and MMSE receiver over various SNR (subchannel transmit power to noise variance) in case of CSI (perfect ) and CE, respectively. We assume that the base layer bit stream is launched to first subchannel, first enhancement bit stream to second subchannel, and so on. These figures show that MMSE detector shows better performance than ZF detector since the noise variance is considered for detecting

6 SONG AND CHEN: SCALABLE H.264/AVC VIDEO TRANSMISSION OVER MIMO WIRELESS SYSTEMS 1223 Fig. 7. Short-term average subchannel BER of OL PT-MIMO with MMSE receiver. Fig. 8. Short-term average subchannel BER of RCS-MIMO with MMSE receiver. transmitted signals. Therefore, we will consider only MMSE detector from now. The short-term average BER of each subchannel with SNR 20 db against CE error is illustrated in Fig. 7. This figure clearly demonstrates that the BER of each subchannel is timely fluctuated from 10 to around 10 and thus we have to assign different channel coding rates according to the importance of each layer [14]. Specially, in order to protect base-layer bit stream during bad status short-term period, allocating strong channel codes (high redundancy) is indispensable and, therefore, it would induce the loss of source coding rate as compared with the proposed ACS-MIMO system. Before comparing the proposed system with OL PT-MIMO, we conduct an experiment to show the short-term average BER performance of random channel selection (RCS)-MIMO system in which each bit stream is randomly assinged among multiple antennas. As shown in Fig. 8, each subchannel short-term BER is converged cause of random allocation and this approach still needs UEP channel codes to meet the need of prioritized delivery in the application layer. If the partial CI of the proposed system is totally wrong, the proposed ACS-MIMO system will be RCS-MIMO system. Then, in order to demonstrate the BER performance of the proposed JAPLD algorithm, we conduct experiments in terms of CE error and feedback time delay. First, the short-term average BER performance of ACS-MIMO with SNR 20 db against CE error is illustrated in Fig. 9. This figure obviously manifests that an unequal BER according to the importance of bit stream is accomplished in ACS-MIMO during short-term period compared to OL PT-MIMO. It is worth to note that the BER of base-layer bit stream is from 10 to around 10. In order to protect the base layer bit stream, we need to employ FEC codes for target BER 10 instead of 10 in OL PT-MIMO. Less redundancy in channel codes (increased source coding rate) results in better base layer representation and eventually improves the system performance of the reconstructed PSNR. Fig. 10 shows the long-term average BER performance with CSI (perfect ) versus CE while CSI Fig. 9. Short-term average subchannel BER of ACS-MIMO with MMSE receiver. (perfect ) versus feedback time delay in Fig. 11. One important observation from these figures is that, if the MIMO channel is slowly time-varying such as indoor wireless system, the performance of the proposed system more depends on the CE accuracy. The BER performance against both CE error and feedback time delay is given in Fig. 12. This figure also illustrates that the proposed ACS-MIMO system can achieve automatic UEP for layered video coding over MIMO system if the feedback delay time is reasonable. Note that the obtained automatic UEP in the proposed system is from not the power control but the feedback of the partial CI: the ordering of each subchannel s SNR strength. As described in Section I-B, in both OL PT-MIMO and ACS-MIMO systems, the CE is essential in order to detect the transmitted signals at the receiver. Compared to OL PT-MIMO, the overhead of the proposed system is the simple calculation of each subchannel SNR using the estimated CSI and feedback to the transmitter over a reliable feedback channel. It is worth to note that the

7 1224 IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY, VOL. 17, NO. 9, SEPTEMBER 2007 Fig. 10. error. Long-term average BER of ACS-MIMO system against CE Fig. 12. Long-term average BER of ACS-MIMO system against both CE error and feedback time delay. Fig. 13. Reconstructed average PSNR of Mobile. Fig. 11. The long-term average BER of ACS-MIMO system against feedback time delay. feedback data (partial CI) is small enough so that overcome the limited feedback bandwidth and delay. Then, the transmitter should allocate each bit stream to a proper subchannel via the JAPLD strategy. In this fashion, the implicitly obtained automatic UEP will contribute the reconstructed PSNR improvement as described below. In addition, the robustness of the proposed system against incorrect partial CI is also shown in Fig. 12. It shows that the longterm average BER performance of the proposed ACS-MIMO system is converged to OL PT-MIMO system in worst case (long feedback delay). This robustness is from that the longterm system average BER of OL PT-MIMO, RCS-MIMO, and ACS-MIMO are all same whether the obtained partial CI is correct or not. B. Reconstructed Video PSNR Performance of ACS-MIMO versus OL PT-MIMO In this subsection, we conduct experiments to show the reconstructed average PSNR of the decoded video sequences over various SNR. Three video sequences, Mobile, Carphone, and Foreman, are tested. All test sequences are 256 frames with CIF 30[Hz] and encoded by the JSVM reference encoder [29] to generate 4-layer scalable video bit streams with GOP 8 and intra-period 16. RS codes [25] are adopted to protect the transmitted bit streams since it maintains maximum erasure protection while produces a minimum of redundancy. We allocate optimal UEP channel codes to both systems. The average reconstructed PSNR (luminance component) for Mobile, Foreman, and Carphone with and data rate 256 kbps per subchannel are shown in Figs , respectively. These figures undoubtedly show that the PSNR improvement is achieved by the proposed JAPLD algorithm in ACS-MIMO system against OL PT-MIMO system. The performance improvement is from the automatically obtained UEP of ACS-MIMO system. Particularly, the gap of PSNR between OL PT-MIMO and ACS-MIMO in the low SNR (or high BER) is outstanding since the high redundant channel codes (loss of source coding rate) have to be assigned in OL PT-MIMO to protect bit streams under high BER. In this section, experimental results proved that the proposed system is able to achieve automatic UEP and the reconstructed

8 SONG AND CHEN: SCALABLE H.264/AVC VIDEO TRANSMISSION OVER MIMO WIRELESS SYSTEMS 1225 This proposed JAPLD algorithm in ACS-MIMO system enables us to both overcome the challenge of the perfect CSI that was assumed in the existing approaches and utilize the estimated CSI from the receiver so as to achieve UEP for layered SVC transmission over MIMO system. In addition, the automatically obtained UEP in this scheme does not require any power control at the transmitter. It was shown by various simulation results that the proposed ACS-MIMO system provides a better PSNR performance against OL PT-MIMO and shows the robustness against incorrect partial CI in terms of CE error and feedback delay time. Fig. 14. Reconstructed average PSNR of Foreman. ACKNOWLEDGMENT The authors would like to thank the anonymous reviewers who provided very valuable feedback and constructive comments on an early version of this paper. Fig. 15. Reconstructed average PSNR of Carphone. video PSNR demonstrates the performance improvement as compared with OL PT-MIMO. The proposed JAPLD algorithm in ACS-MIMO enables us to not only overcome the challenge of the perfect CSI that was assumed in the existing approaches but also utilize the estimated valuable CI to enhance the system performance through the feedback of the partial CI. VI. CONCLUSION We described in this paper a novel JAPLD strategy that successfully transmit scalable video over MIMO wireless systems. This strategy makes proper use of the feedback of the obtained partial CI in such a way that the application layer cooperates with the physical layer. In physical layer, we proposed the ACS- MIMO system in which each bit stream can be periodically switched among multiple antennas. SVC generates layered bit streams that need prioritized delivery in application layer. Then, we obtain the ordering of each subchannel s SNR strength as partial CI at the receiver through simple CE and calculation. Finally, via the JAPLD algorithm developed in this research, we are able to launch higher priority layer bit stream into higher SNR strength subchannel based on the partial CI. Therefore, we can implicitly achieve automatic UEP for layered scalable video coding transmission over MIMO system without power control at the transmitter. REFERENCES [1] S. M. Alamouti, A simple transmit diversity scheme for wireless communications, IEEE J. Sel. Areas Commun., vol. 16, no. 10, pp , Oct [2] V. Tarokh, H. Jafarkhani, and A. R. Calderbank, Space-time block coding for wireless communications: Performance results, IEEE J. Sel. Areas Commun., vol. 17, no. 3, pp , Mar [3] G. J. Foschini, Layered space-time architecture for wireless communication in a fading environment when using multi-element antennas, Bell Labs Tech. J., pp , [4] P. W. Wolniansky, G. J. Foschini, G. D. Golden, and R. A. Valenzuela, V-BLAST; An architecture for realizing very high data rates over the rich-scattering wireless channel, in Proc. IEEE ISSSE, Sep. 1998, pp [5] I. E. Telatar, Capacity of multi-antenna Gaussian channels, Eur. Trans. Commun., vol. 10, pp , [6] H. M. Radha, M. Schaar, and Y. Chen, MPEG-4 fine-grained scalable video coding method for multimedia streaming over IP, IEEE Trans. Multimedia, vol. 3, no. 1, pp , Mar [7] W. Li, Overview of fine granularity scalability in MPEG-4 video standard, IEEE Trans. Circuits Syst. Video Technol., vol. 11, no. 3, pp , Mar [8] T. Wiegand, G. Sullivan, J. Reichel, H. Schwarz, and M. Wien, Joint Draft of SVC Amendment, Joint Video Team (JVT), Doc. JVT-W201, Apr [9] J. Reichel, H. Schwarz, and M. Wien, JSVM-7 Software, Joint Video Team (JVT), Doc. JVT-T202, Jul [10] H. Schwarz, D. Marpe, and T. Wiegand, Overview of the scalable video coding extension of the H.264/AVC standard, IEEE Trans. Circuits Syst. Video Technol., vol. 17, no. 9, pp , Sep [11] A. Segall and G. Sullivan, Spatial scalability within the H.264/AVC scalable video coding extension, IEEE Trans. Circuits Syst. Video Technol., vol. 17, no. 9, Sep [12] Advanced Video Coding for Generic Audiovisual Services, v3, ITU-T Recommendation H.264 & ISO/IEC AVC, [13] T. Wiegand, G. J. Sullivan, G. Bjontegaard, and A. Luthra, Overview of the H.264/AVC video coding standard, IEEE Trans. Circuits Syst. Video Technol., vol. 13, no. 7, pp , Jul [14] C. Kuo, C. Kim, and C. C. J. Kuo, Robust video transmission over wideband wireless channel using space-time coded OFDM systems, in Proc. WCNC, Mar. 2002, vol. 2, pp [15] H. Zheng and K. J. R. Liu, Space-time diversity for multimedia delivery over wireless channels, in Proc. IEEE ISCAS, May 2000, pp [16] Z. Ji, Q. Zhang, W. Zhu, Z. Guo, and J. Lu, Power efficient MPEG-4 FGS video transmission over MIMO-OFDM systems, in Proc. IEEE ICC, 2003, vol. 5, pp [17] Z. Ji, Q. Zhang, W. Zhu, J. Lu, and Y. Zhang, Video broadcasting over MIMO-OFDM systems, in Proc. IEEE ISCAS, May 2003, vol. 2, pp [18] D. Song and C. W. Chen, QoS-guaranteed SVC-based video transmission over MIMO wireless systems with channel state information, in Proc. IEEE ICIP, Oct. 2006, pp

9 1226 IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY, VOL. 17, NO. 9, SEPTEMBER 2007 [19] D. Song and C. W. Chen, QoS-guaranteed scalable video transmission over time-varying MIMO channel capacity, in Proc. IEEE ICME, Jul. 2007, pp [20] H. Bolcskei, D. Gesbert, and A. J. Paulraj, On the capacity of OFDM based spatial multiplexing systems, IEEE Trans. Commun., vol. 50, no. 2, pp , Feb [21] Z. Liu, G. Giannaksis, S. Barbarosa, and A. Scaglione, Transmit-antenna space-time block coding for generalized OFDM in the presence of unknown multipath, IEEE J. Sel. Areas Commun., vol. 19, no. 7, pp , Jul [22] K. Lee and J. Chun, On the interference nulling operation of the V-BLAST under channel estimation errors, Proc. IEEE VT, pp , [23] W. C. Jakes, Microwave Mobile Communications. New York: Wiley, [24] R. H. Clarke, A statistical theory of mobile-radio reception, Bell Syst. Tech. J., vol. 47, pp , [25] S. Lin and D. J. Costello, Error Control Coding-Fundamentals and Applications. Englewood Cliffs, NJ: Prentice-Hall, [26] M. Farshchian and W. A. Pearlman, Real-time video transmission over MIMO OFDM channels using space-time block codes, in Proc. 40th Annual Conf. Inf. Sci. Syst., Mar. 2006, pp [27] J. Song and K. J. R. Liu, Robust progressive image transmission over OFDM systems using space-time block code, IEEE Trans. Multimedia, vol. 4, no. 9, pp , Sep [28] D. Song and C. W. Chen, Robust image transmission over MIMO space-time coded wireless systems, presented at the SPIE, Commun. Netw. Technol. Syst., May 2006, vol [29] J. Vieron, M. Wien, and H. Schwarz, JSVM7 Software Joint Video Team (JVT), Doc. JVT-T203, Jul [30] A. Paulraj, R. Nabar, and D. Gore, Introduction to Space-Time Wireless Communications. Cambridge, U.K.: Cambridge Univ. Press, Daewon Song (S 04) was born in Jinhae, Korea. He received the B.S. and M.S. degrees in electrical engineering from Changwon National University, Changwon, Korea, in 1995 and 1997, respectively, and is currently working toward the Ph.D. degree in electrical and computer engineering at the Florida Institute of Technology, Melbourne, FL. He worked as a Research Engineer for Network and Application Software Team, Daewoo Heavy Industries from 1997 to During Summer 2007, he conducted research at Home & Networks Mobility, Motorola, San Diego, CA. His interests lie in the areas of wireless visual communication, scalable video coding, and multimedia communications over MIMO systems. Chang Wen Chen (S 86 M 90 SM 97) received the B.S. degree from University of Science and Technology of China, Beijing, China, in 1983, the M.S.E.E. degree from University of Southern California at Los Angeles in 1986, and the Ph.D. degree from University of Illinois at Urbana-Champaign in He has been Allen S. Henry Distinguished Professor in the Department of Electrical and Computer Engineering, Florida Institute of Technology, Melbourne, since July Previously, he was on the Faculty of Electrical and Computer Engineering at the University of Missouri Columbia from 1996 to 2003 and at the University of Rochester, Rochester, NY, from 1992 to From September 2000 to October 2002, he served as the Head of the Interactive Media Group at the David Sarnoff Research Laboratories, Princeton, NJ. He has also consulted with Kodak Research Labs, Microsoft Research, Mitsubishi Electric Research Labs, NASA Goddard Space Flight Center, and Air Force Rome Laboratories. Dr. Chen is the Editor-in-Chief for IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY (TCSVT) since January He was an Associate Editor for IEEE TRANSACTIONS ON MULTIMEDIA from 2002 to 2005 and for TCSVT from 1997 to He was also on the Editorial Board of IEEE Multimedia Magazine from 2003 to 2006 and was an Editor for the Journal of Visual Communication and Image Representation from 2000 to He has been a Guest Editor for the PROCEEDINGS OF THE IEEE (Special Issue on Distributed Multimedia Communications), a Guest Editor for IEEE JOURNAL OF SELECTED AREAS IN COMMUNICATIONS (Special Issue on Error-Resilient Image and Video Transmission), a Guest Editor for IEEE TSVT (Special Issue on Wireless Video), a Guest Editor for the Journal of Wireless Communication and Mobile Computing (Special Issue on Multimedia over Mobile IP), a Guest Editor for Signal Processing: Image Communications (Special Issue on Recent Advances in Wireless Video), and a Guest Editor for the Journal of Visual Communication and Image Representation (Special Issue on Visual Communication in the Ubiquitous Era). He has also served in numerous Technical Program Committees for IEEE and other international conferences. He was the Chair of the Technical Program Committee for ICME2006 held in Toronto, Canada. He was elected an IEEE Fellow for his contributions in digital image and video processing, analysis, and communications and an SPIE Fellow for his contributions in electronic imaging and visual communications. He has received research awards from NSF, NASA, Air Force, Army, DARPA, and the Whitaker Foundation. He also received the Sigma Xi Excellence in Graduate Research Mentoring Award from the University of Missouri Columbia in Two of his Ph.D. students have received Best Paper Awards in visual communication and medical imaging, respectively.