based wireless video

8/6/2019 Based Wireless Video

http://slidepdf.com/reader/full/based-wireless-video 1/5

H.263-Based Wireless Video Transmission in Multicode CDMA Systems

Mario R. Hueda and Carlos A. Marqu6sLaboratorio de Comun icaciones Digitales - Universidad Nacional de C6 rdoba

Casilla de Correo 755 - Correo Central - C6rdoba (5000)- ArgentinaEmail: mhueda@ com.uncor.edu

Abstract - We present a simple and accurate semi-analyticmethodology for performance evaluation of wireless videotransmission in multicode DS-CDMA (MC-CDMA). Based onthe novel approach (as well as on entire system simulations), weanalyze performance of H.263 in an IS-95B ystem over slowfading channels at low bit rates. Peak signal-to-n oise ratio(PSNR) values for various video test seq uences demonstrate thatimportant improvements of the video quality can be achieved byusing feedback error control schemes. In ad dition, we inv estigatethe effectiv eness of the co mbin ed use of error correction anderror resilience techniques for reliable video transmission inMC-CDMA systems.

I. INTRODUCTION

As a result of the evolution of wireless communicationstowards transmission of multim edia services, multicode directsequence code division multiple access (MC-CDMA) hasbeen proposed to support variable data rates [l]. In MC-CDMA (such as IS-95B [2]), high speed is provided throughcode aggregation. Up to eight codes may be assigned in IS-

95 B (one fundamental code channel (FCC) and sevensupplemental code channels (SCC)), providing a maximumbit rate of 115.2 kilobits per second (kbps) [2]. However, ithas been shown that multim edia services in MC-CD MA mustoperate at very low b it rates ( c3 2 kbps), in order to provide anadequate capacity for voice users [3].

Video transmission is one of the most importantrequirements for next generation of wireless systems. H.263

is a compression standard developed to transmit video usingthe phone network at data rates less than 64 kbps [4]. Atypical H.263 video stream is composed of inter-codedframes (I-frames) and predictive-coded frames (P-frames).An I-frame is an independ ently coded video fram e that can bedecoded by itself. A P-frame is composed of changes in thecurrent image of the video stream relative to the last I-frameencoded. In general, an H.263 video stream is usually an I -

frame followed by many P-frames, with an I-frame re-introduced to restore image quality in case of transmissionerrors. H.263+ is an extension of H.263 and its aim is toimprove upon the compression efficiency of H.263 andbroaden the range of applications for wireless transmission.How ever, additional protection schem es must be incorporated

in transmission over slow fading channels as a result of thelong block error bursts characteristic of thos e channels.Recently, some papers have analyzed video transmission

over MC -CDM A systems (e.g., [5]). However, none of themhave considered MC-CDMA transmission over slow fadingchannels, as well the effects of important ingredients of theexisting system architecture (e.g., RAKE receivers with

This work was supported in part by SeCyT -UNC - Argentina

maximal ratio combiner (MRC) and convolutional codes(CC) with soft-decision Viterbi decoding (SD-VD)). Theseimportant topics are addressed in this paper. We focus ourstudy on H.263/H.263+ video transmission at low data rates(i.e., c32 kbps) in an IS-95B based system over slow fadingRayleigh chan nels. In order to evaluate performance of videotransmission in MC-CDMA systems, a novel, simple, andaccurate semi-analytic methodology is proposed. Based onthe new approach , as well as on entire sys tem simu lations, wederive a variety of peak signal-to-noise ratio (PSNR) resultsfor various H.263 video test sequences. In addition, we

investigate the effectivenessof retransmission error correctionand error resilience techniques to enhance the robustness ofH.263/H.263+ video transmission over IS-95B systems.

11. BACKGROUND

A . H.263 Based Video Compression

The picture resolution is often Q C F (Quarter CommonIntermediate Format, 176 x144 pixels), which is the mostcommon input format at low bit-rates. At QCIF resolution,each picture is divided into 11x9 macroblocks (MBs), whichcomprise 16x16 luminance samples, and two corresponding8x8 blocks of chrominance samples. A fixed number ofsuccessive MBs is usually grouped into a group of blocks(GOB) and side information that is appropriate for a larger

number of MBs, but not for an entire frame (“$ramet’ efers tovideo), can be comm unicated efficiently on that level.

B .

H.263 provides means to insert synchronization words atthe picture and optionally at the GO B layers. We use the latteroption to allow resynchronization in case of errors. We alsoassume that GOB sync words are inserted at the beginning ofeach macroblock row. This is exploited by the errorconcealm ent technique employed in the decod er, whichdiscards corrupted GOBS and replaces the correspondingimage content with data from the previously decoded frame.This technique works almost perfectly for non-moving partsof the sequ ence, but introd uces severe distortion for mo vingimage regions.

In transmission over mobile channels, transmission errorsdegrade the video quality at the receiver. In order to m itigatethe effects of error transm ission, periodic I-frame refresh hasbeen proposed by several authors. How ever, since the overallimage quality decreases due to the Iower efficiency of I-

frames in comparison to P-frames, and the MC-CDMAsystem has a severe bandwidth limitation, no periodic I-framerefresh is applied in the simulations in order to achieve the

Error Concealment and V ideo Coding Mod e

0-7803-7005-8/01/$10.00 0 2001 IEEE 43 3

Authorized licensed use limited to: UNIVERSIDAD CORDOBA. Downloaded on August 27, 2009 at 14:35 from IEEE Xplore. Restrictions apply.

http://com.uncor.edu/





highest coding perfo rman ce for the error free case, whichserves as a baseline.

C. Loss of Picture Quality

The average peak signal-to-noise ratio (PSNR) has been

widely adopted as a distortion measure. For error-freetransm ission , the PSNR of th e video reproduced at thereceiver (i.e., the baseline) is given by

p s N ~ c o r , e c ,t )= 10lOg10 (255’ 1D s c ( t ) ) . (1)

D sc ( t ) s the video sou rce codin g distortion given by

u= l

where U is the total number of samples of the picture at t ,

while 0, t ) an d r, ( t ) are the u-th amp litudes of the original

and reconstructed pictures, respectively. Owing to

transmission errors, an ad dition al distortion DcH ( t ) occurs atthe video deco der. Thu s, the overall decoded video distortion(i.e.. after the error conc ealm ent stage) results in

Do ( t ) = DS C (t> DC H ( t> (3)

Then, the PSNR of the video sequence at the decoder canbe expressed as

PSNR,,, ( t )=10log lo 255’ / D o t ) ) . (4)

In this paper, the loss of picture quality A P SNR ( t ) ,

defined by

( 5 )ApSNR(t)= PSNRlo,,(0- PSNRcorrec,0)

= lolog10 [Dsc ( t )1Do 011,

is used as a m easure of the video degradation.

In.PERFORMANCEVALUATIONFH.263 IN MC-CDMATRANSMISSIONSINGANALYTICALODELS

A Model for Block Error Process in MC-CDMA.

Our performance studies are based on a novel Markovmodel for the block error process in M C-CDM A transmissionwe introduced in [6]. Unlike previous work, this takes intoaccount the presence of: (a) multicodes, ( b ) MRC RAKEreceivers, (c) CC with SD-VD and (6) on-ideal interleaverperformance (owing to transmission over slow fadingchannels). Consider a certain user that employs M

multicodes in an MC-CDMA system. Let Pi,n be a binaryprocess such that Pi,”=1 if the data block i at the Viterbi

decoder output of the n-th multicode is in error, and 0

otherwise. We define a new process Y as

We showed in [6] that the process Pi is well

approximated by a 2 M state Markov model, with transition

matrix II,, ( x ) = II, (1)’ ,where

(7)

Mwith m , , , ( x ) u , v = 0,1,2,...,2 -1, representing the

probability of B, = v in slot i given that the superblock in

slot i- x was Y -- x = U.The Markov model defined by (6) and

(7) is called the “super-block Markov model” (SBMM). A

method to estimate the elements mu,”n transmission over

slow fading chann els is presented in [6].

B.

Next wc derive an analytical expression for the loss of P-

frames picture quality owing to transmission errors, DcH(t).Towards this end, we assume that an en-or signal is

introduced at t = O (there is no error in the video frames or

D c ~ ( t ) = 0or t <0 ). The resulting error sequence is treated,before the video decoder, by the error concealment techniquedescribed previously. This introduces distortion for movingimage regions, which propagates spatially arid temporally

until an I-fa m e refresh is applied at t =T,, . !$tuhlmiiller et

al. [7] have shown that the variance of the prclpagated errorsignal over a P-frame sequence owing to the m o r introducedat t =0 can be well ap proxim ated at low error riates by

A Model for the Loss o Picture Quality

o f ( t )= f;r(i +e ) - l , O l t < T - . (8)

PB is the block error probability and depends on th etransmission system and channel characteristics (e.g., channelcode, numlxr of RAKE fingers, etc.). Param eter r representsthe sensitivity of the video decoder to an increase in errorrate, and its value depends on several implementation issues,such as packetization, resynchronization, and errorconcealment, as well as the encoded v ideo sequence. It can be

considered as a constant that does not depend on the other

model parameters. Note that P,T is the error variance

introduced in the video seque nce at t =0 [7]. The leakage 0

describ es tlhe efficie ncy of exp licit an d/or imp licit (e.g., due tosub-pel motion com pensat ion, overlapped block transmissio n)loop filtering to remove the introduced error. Its valuedepends on the streng th of the loop filtering as well as on the

shape of the power spectral density of the introduced error.The range of typical values is given by 0<0 :1. From (8)

note that the energy of the error signal decays over time dueto spatial filtering in the pred iction loop.

Unlike 1:7], in this work we are interested not only on thetime avera,ged video distortion ow ing to transmission errors,but also c e th e “temporal progression” of {he distortion

value, DcH ( t ) . This model is very useful because it allows,

434




for example, to evaluate analytically the performance ofnumerous error resilience techniques. Assuming that the errorprocess is stationary ov er suc cessive frames (i.e., the effect oflost blocks is approximately constant for each transmitteddata block) and the error signals are uncorrelated [7], from (8 )

we obtain:

for 05 <T,, ,with

@(t )=8-’ n(1+ et)+ OS(2+et11+et)-’ . (10)

(to derive (10) we approximated the sum in (9) by an integralusing trapezoid based numerical integration).

PE can be easily estimated from the SBMM (this analysis

is not included in this paper). For a given packetization andvideo sequence, we estimate parameters 8 .and r by fitting

(5) (with D, , ( t ) and D, , ( t ) given by (2) and (9),

respectively) to the measurement points obtained fromSBM M based fast simu lation. From experimental results wehave verified that r and 8 are approximately constant overthe range of the video rates considered in this work (i.e., c3 2kbps).

C. A Model fo r the Overall Video Distortion

Using (9) in (3), the total distortion Do ( t ) results in

D, ( t )= D ,, ( t )+ p B r o ( t ) , O I t <T , , . (11)

Then, from (11) we can derive the time averaged totalvideo d istortion:- -

o =D ,, + P B r O ,

where

O - (T,, + I ) - ~ . ~ - ~ [ ( ~ + B T , , ~ ~ ~ ( ~ + B T , , ) - I ) + ~ J

(12)

(13)+ os^,, +e- ’ ln(1+BTm,)]+ 0.5[0(0) O(T,,)]}.

Note that the minimum value, 0 =1, occurs when there is

no error propagation (T, , = O ) . From (12) and (13) we can

conclu de that the minimum distortion (i.e., 0 =1) of a givenvideo sequence achieved by error resilience techniques (suchas error-tracking [9]) in transmission over a channel withblock e rror probability PB , s given by

r n i n E } = D,,+ P B r (14)

Th e distortion rate (DR) model is used to approximate

D,, . Thus, the average video source coding distortion can

be expressed by-where R , is the video so urce rate, while A. , , , and C O are

the parametersof the DR m odel which depend on the encodedsequence [7].

IV.NUMERICALESULTS

To perform our sim ulation experiments, we use theUniversity of British Columbia’s H.263+ Reference codec.Moreover, we utilize the rate control method discussed in

TMN -8. All of the results represent around of 160 pictures oftest sequences having QCIF resolution , coded at 6 frames persecond ( f p s ) . Although numerous simulations have beenperformed, only a few results are discussed in this paper,becaus e of spa ce constraints. Sp ecifically, we analyze the lossof picture quality for the test sequences MotherdDaughterand Foreman. These sequences are selected because of theirdifferent characteristic in motion and spatial detail [7].

We use the parameters, interleaver/deinterleaver, andconv olutio nal code of the downlink of the IS-95B standard[2]. Furth ermo re, we use soft-d ecision decoding of the rate %,

constra int length 9, con volutio nal code (i.e., rate set 2 of IS-

95B). We set the number of RAKE fingers, L , to four

(L = 4 ). We assume that RAKE fingers have equal power.

The block rate is l /T, =50 block/s. A 16-bit cyclicredundancy code (CRC) is used for block error detection.Carrier frequency is 1800 MH z and the maximal Doppler

frequency is f, = 2 Hz. Since the rate variation of the

channel is small, we assume ideal coherent demodulation atth e RAKE receiver. The Rayleigh channel is simulated usingJakes’ model. We set the number of multicodes to two (M=2).FCC and an SCC are used to transmit video. The rate of SCCis 14.4 kbps. The transmission rate of FC C is assumed 7.2kbps (14.4 kbps is its maximal rate). 2.4 kbps are spent foroverhead (i.e., packet headers + CR C + tail block) and thevideo net bit rate results in 19.2 kbps. Three channel states areanalyzed in this paper: (a) “Bad’ , (b) “Good’ and (c) “VeryGood’, which have an SCC average block error of 0.11,0.035

and 0.017, respectively. These values correspondapproximately to 15, 10 and 8 data users per cell (with onecode per user) in a typical multi cell environment.

A. Accuracy of SBMM

We use SBMM to evaluate H.263 video degradation inMC-CDMA transmission over slow fading channels. We alsocompare simulation results derived from both SBMM andentire system sim ulation s in ord er to verify the accuracy ofour channel mod eling. It is impo rtant to realize that, althoughMarkov models have been previously used to evaluateperformance of wireless video transmission, comparisonswith results obtained from simulation of the entire systemhave not been reported so far. These comparisons are

important because it has been demonstrated that the first-order Markov approximation to model slow fading channelscannot be useful in several cases of interest (e.g., applica tionsrequiring a large number of consecutive samples)[8].

Fig. 1 shows the loss of picture quality for various videotest sequences obtaine d from simu lations. Fast simulation-based performance evaluation is achieved by using theSBMM defined by (7). We also present results fromsimulation of the entire system. The coded sequences are

435




transmitted 350 times using different: ( i ) realizations of theblock error process (in SBMM simulations), and ( i i ) startingpoints in the fading simulator (in entire system simulations).The average M S N R ( t ) over all the runs is presented. We canverify the excellent accuracy of the SBMM. Results also

show that video quality is severely degraded in transmissionover MC-CDMA. This is because, in slow fadingenvironments, the probability that all M received data blocksare in error in a deep fading is high [6], herefore long block-error bursts occur at the input of the video decod er. If we tak einto account that P-frames are being transmitted, it is easy toinfer that the error propagation problem in the reproducedvideo is severe, as it can be seen in Fig. 1. Furthermore, wehave verified that several modes incorp orated in H.263+ toimprove performance in wireless channels are not effective(these results are not included in this paper). Fro m the above,we c onclud e that extra protection must be provided in orde r toimpr ove the reliability of H.263 transmission in MC-CDM A.

B.

From Fig. 1 we noted that video quality is severelydegraded in transmission over M C-CDM A. Improve ments ofvideo quality can be achieved by re ducing the packet loss P B

through error correction technique s such as automatic repeatrequest (ARQ) protocols. In this paper, we consider the Non-Selective Variable Bandwidth Retransmission (NSVBR)

scheme, a novel bounded delay modified GBNretransmission-based scheme that uses FCC to transmit, at aproper rate (lower than its maximal), video informationinstead of voice. Because real-time services require a boundeddelay, only one retransmission is allowed. Moreover,depending on the round trip delay and the number of M CCs,only a limited number of packets can be retransmitted. Toretransmit data blocks with low time delay, the proposed

approach increases the FCC da ta rate. Thus, our scheme notonly avoids underutilizing and w asting the scarce bandwidth,but it also significantly reduces the interference and the

complexity of the bandwidth allocation algorithm. Since slowfading and low data rates are assumed, it can be inferred thatNSVBR provides a good tradeoff between efficacy andcomplexity.

Analysis of Error C orrection Techniques

Fnm. Numk r F n m . N um k r

Fig. 1. Results for QCIF Foreman an d Mother&Daughtersequences in IS-95B or different channel states.

Let N , 111 and R,,, deno te the round trip delay in slots, the

number of MCCs assigned to a given user, and the total userbit rate for video, respectively. In "normal conditions" (i.e.,

no errors), the FCC works at a data rate, RFccno, ess than its

maximal RFCcma.The value of RFCCno, will depend on the

link quality (e.g., the system load), and is determined at thebeginning of the comm unication. For ex ample, if the averageerror rate of the link is moderately high, a low value of

RFccn0, s selected (i.e., the video bandwidth (quality) is

reduced), thus a "high" bandwidth (RFccma -RFCc,, ) is

reserved fix retran smiss ions. When the receiv er (e.g., amobile station) detects only one of th e M received blocks inerror, a negative acknowledgm ent (NA K) is sent to thetransmitter (e.g., the base station). This approach simplifiesimplementation at the expense of efficiency. IHowever, wehave verified that the degradation is not imporlant since theblock error processes among MCC s are highly correlated [ 6 ] .

Then, when the transmitter receives an NAK, the data rate ofFCC is increased to RFccmm and the retransmission (only

one) of MC C blocks begins . Unlik e classical GB", not all theNM blocks are sent again since the bandwidth assigned forretransmission is limited. Only a fraction of the information

lost is retr,uwnitted. For example, consider N = 6 and thetransmissice rates defined in Section IV (i.e.,

R,, =RFccno,+ Rscc =21.6 kbps). Then, when an NA K is

received, Ionly 33 % of the NM = 12 blocks can beretransmitted. When the retransmission process finishes, thesystem returns to the "normal" mode.

The ability of NSVB R to improve the reliability of H.263wireless video transmission in MC-CD MA is analyzed in Fig.

2. We adopt M, , ,,, ,RFcc,, and R F C C , , , as given in the

previous example. Moreover, we assume that no feedbackerrors OCCIK. We consider the channel state "Very Good'.Results obtained from entire system simulation andtheoretical values are presented. Residual PBcan,be estimatedby using SBMM (this analysis is not included in this paper).We verify the excellent accuracy of both the distortionmodeling approach (9) and the parameter estimation methodbased on S B M M . Moreover, note that NSVBR achievessignificant gains for both test sequences with small additionalcomplexity. These gains depend on the residud block errorrate and the am ount of m otion present in the sequence. Thebiggest gain can be observed for Foreman [9].

C. Analysis of Error Resilience TechniquesResidual transmission errors cannot be avoided with a

mobile radio channel. Additional schemes to stop the errorpropagatioin have to be considered. In this sense , NSVBR canbe efficiently combined with error resilience techniques tofurther improve the video quality in MC-CDMAtransmission.

436




Fmm. Number Fram. Number

VideoSequence

M& D

Fig. 2. Results for QCIF Foreman and MotherBDaughtersequencesin IS-95B for the channel state “Very Good”.

Average Video Distortion (dB)

N P I-ET I IET+NSVBR I Min.Dist

18.58 17.73 I 16.93 I 16.21

Here we evaluate video performance achieved by NSVBRcombined with a theoretical stop error propagation techniquewe called ideal-error-trucking (I-ET). This approach issimilar to that one presented in [9], that is, it utilizes intra-

coded MBs (I-MBs) refresh to stop temporal errorpropagation. Using a feedback channel, the temporal andspatial Occurrence of an error is reported to the transmitter.Thus, the location and extent of propagated errors isreconstructed at the encoder. However, unlike [9] we assume

that in I-ET the video sou rce coding d istortion D s c ( t ) is not

affected by the I-M Bs refresh, and the error signal at t = O is

successfully canceled at t =Td . It can be verified that the

video performance obtained from I-ET constitutes a theoreticupper bound of the video quality that can be achieved by anyerror resilience technique using the same conditions of

operation (e.g., time delay Td ). The loss of the p icture quality

achieved by I-ET results in

Foreman 23.86

while the time av eraged video distortion is given by

DCH - Im ( t ) pBr@(Td) . (17)

Table I presents results of the average distortion 0, for

NSVBR and I-ET. Channel state “Good’ is used and the tim e

delay between the video coder and decoder is assumed Td =

0.8 s.

21.83 2 1.09 20.30

Several cases are considered: no protection (NF’), -ET(only), and I-ET+NSVB R. The values are analytically derived

from (12 ) with D,, given by (15). 0 is calculated from (13)

for NP, while 0 O(Td) for I-ET and I-ET+NSVBR (see

(17)). We also include the theoretical limit defined by (14),which represents the minimum distortion of the videosequence that can be achieved by NSVBR. Results show thatsignificant gains can be achieved when NSVBR is used incomb ination with I-ET. We also verify that a residual averagedistortion around 0.75 dB is added to the theoretical minim umvalue achievable by NSVBR. This result suggests that newerror resilience schemes could be designed to improve evenmore the performance of NSVBR in video transmission overerror-prone channels.

V. CONCLUDINGEMARKS

This paper has presented a novel semi-analyticalmethodology to estimate performance of wireless videotransmission in MC -CDMA systems. Based on this approach,

we analyzed H.263 video transmission in IS-95B at low bitrates over slow fading channels. Comparisons with valuesderived form entire system simulation have demonstrated theexcellent accuracy of our technique. We found that extraprotection must be provided in order to improve the videoquality of the end user. We also introduced a new reducedcomplexity feedback error control scheme, which has beenshown to significantly improve the reliability of videotransmission. Furthermore, we investigated the effectivenessof error correction and error resilience techniques, and

discussed theoretical limits of the video quality that can beachieved in transmission over error-prone channels. Themethodology we proposed in this paper can be extended toperformance evaluation of several video codecs in MC-

CDM A transmission over generalized fading channels.

REFERENCES

Chih-Lin I, R.D. Gitlin, “Multicode CDMA wireless personalcommunications networks,” IEEE lCC’95, v01.2.TWEIA-95B. “Mobile station-base station compatibilitystandard for dual-mode wideband spread spectrum cellularsystem”. July 1997.D. Ayyagari and A. Ephremides, “Cellular multicode CDMAcapacity for integrated (voice and data) services,” IEEE JSAC,vol. 17, pp. 928-938, May 1999.ITU, el. Stand. Sector ITU, Video coding for narrow tel.channels at< 64 bit/s,” lTU-T Recom. H.263, 1996.B. Deep and W. eng, “Adaptive code allocation in multicode-CDMA for transmitting H.263 video,” Proc of IEEE WCNC,pp. 1003-1007, New Orleans, September 1999.M. R. Hueda, “A Markov model for performance evaluation inmultimedia CDMA wireless transmission” Proc. IEEE-VTC,Boston, Sept. 2000.K. Stuhlmiiller,N. F;irber, M. Link, and B. Girod, “Analysisofvideo transmission over lossy channels,” IEEE JSAC., vol. 18,pp. 1012-1032, June 2000.C. Tan and N. Beaulieu, “First-order Markov modeling for theRayleigh fading channel,” IEEE Globecom98, pp.3669-3674.E. Steinbach, N. F&ber, and B. Girod, “Standard compatibleextension of H.263 for robust video transmission in mobileenvironments,”IEEE T.CSVT., vol.7.p~. 72-881. Dec. 1997.

437

based wireless video

Documents