20_Perceptually Unequal Packet Loss Protection by Weighting Saliency and Error Propagation_Wuweiming

Perceptually Unequal Packet Loss Protection by Weighting Saliency and Error Propagation

Hojin Ha,Jincheol Park,Sanghoon Lee,Member,IEEE,and Alan Conrad Bovik,Fellow,IEEE

Abstract—We describe a method for achieving perceptually minimal video distortion over packet-erasure networks using perceptually unequal loss protection(PULP).There are two main ingredients in the algorithm.First,a perceptual weighting scheme is employed wherein the compressed video is weighted as a function of the nonuniform distribution of retinal photoreceptors. Secondly,packets are assigned temporal importance within each group of pictures(GOP),recognizing that the severity of error propagation increases with elapsed time within a https://www.sodocs.net/doc/ee13142500.html,ing both frame-level perceptual importance and GOP-level hierar-chical importance,the PULP algorithm seeks ef?cient forward error correction assignment that balances ef?ciency and fairness by controlling the size of identi?ed salient region(s)relative to the channel state.PULP demonstrates robust performance and signi?cantly improved subjective and objective visual quality in the face of burst packet losses.

Index Terms—Forward error correction,human visual system, internet video,perceptual coding,unequal loss protection(ULP).

I.Introduction

W ITH THE EXPLOSIVE growth of multimedia envi-ronments,the robust transmission of video data has become an important requirement to enable smooth and seam-less interaction with multimedia content[1],[2].In error-prone environments,signi?cant spatio-temporal dependencies in the video data may be lost owing to congestion,jitter, or delays over packet-erasure networks.This leads to sub-stantial deterioration of received video quality from error propagation.To minimize visual quality degradation from packet losses,it is necessary to simultaneously consider the question of perceptual video quality[3]–[11]while accounting for error propagation effects arising from the video coding structure[13]–[21].

Manuscript received March21,2009;revised September25,2009;accepted January21,2010.Date of publication May27,2010;date of current version September9,2010.This work was supported by the Agency for Defense Development under Contract UD1000221D,Korea.This paper was recommended by Associate Editor H.Sun.

H.Ha is with the Digital Media and Communications Research and Development Center,Samsung Electronics,Yeongtong-gu,Suwon-si443-373, Korea(e-mail:hojiniha@yonsei.ac.kr).

J.Park and S.Lee are with the Department of Electrical and Electron-ics Engineering,Yonsei University,Seoul120-749,Korea(e-mail:dewof-dawn@yonsei.ac.kr;slee@yonsei.ac.kr).

A.C.Bovik is with the Laboratory for Image and Video Engineering,Center for Perceptual Systems,University of Texas,Austin,TX78712-1084USA (e-mail:bovik@https://www.sodocs.net/doc/ee13142500.html,).

Color versions of one or more of the?gures in this paper are available online at https://www.sodocs.net/doc/ee13142500.html,.

Digital Object Identi?er10.1109/TCSVT.2010.2051368

Here,we present a packet loss resilience scheme that is based on an unequal loss protection(ULP)method that seeks to minimize perceptual distortions in the compressed video bit stream.This is accomplished by assigning unequal importance to different levels in the video coding structure using models of human visual sensitivity.We begin by quantifying the relative importance of video frames within each group-of-pictures (GOP)[13]–[15].The motion compensation that is computed from the frames in each GOP causes the picture quality of a current reconstructed frame to be strongly dependent on the reconstructed version of its preceding frames.Generally,when packet losses occur earlier in a GOP,the reconstructed quality of following frames will be more severely compromised owing to the longer error propagation.In this sense,the perceptual importance of each frame descends from the?rst frame to the last frame in each GOP.We then de?ne a procedure for incorporating perceptual weights into an ULP scheme[22], [23].Similar approaches have been used to improve visual quality in other resource allocation schemes,by allocating more resources to perceptually important bit information via the use of visual saliency weights[3],[4].In[6]–[9],a nonuni-form spatial?ltering law,called foveation,was employed to de?ne spatial perceptual weights on coding macroblocks (MBs).Larger weights were applied near presumed visual ?xation points,which were represented at high resolution, while lower weights were assigned to peripheral points.This process of foveal weighting attempts to match the nonuniform density of photoreceptors over the retina to achieve better vi-sual quality.In that approach,a nonuniform foveation?ltering method causes the local spatial bandwidth(LSB)to rapidly decrease with distance from the presumed?xation point(s). Fig.1depicts the dependence of packet loss induced per-ceptual quality degradation on the error resilience scheme. Speci?cally,the degree of visual quality degradation that oc-curs in the45th frame of the Foreman sequence due to a packet loss in the30th frame.Fig.1(a)and(b)shows the30th and 40th reconstructed frames when a perceptually salient region is protected from packet loss in the30th frame.It is assumed in this example that the point of?xation is on the face in the spatial center of the30th frame,although this need not be the case.By contrast,Fig.1(c)and(d)shows the same frames suffering from the same degradation,but without the per-ceptual weighting mechanism.In previous paper,it has been observed that higher perceptual video quality can be obtained by protecting those portions of the bitstream corresponding to salient regions from packet loss[14],[22],[23].However,the

1051-8215/$26.00c 2010IEEE

question remains as to how to achieve the minimum degree of visual quality degradation from packet loss for a given video coding algorithm,using limited channel resources.The general approach we take is that,for a given number of channel coding bits and a given video coding structure,formulate an optimiza-tion procedure that enables forward error correction(FEC) based on an appropriate perceptual weighting mechanism. More speci?cally,we propose a performance metric based on both foveal weighting and on the temporal error propaga-tion effect.The metric consists of two factors.One is the LSB obtained using a foveation?lter model[7]–[11].The other is called the perceptual weight on error propagation(PWEP).Us-ing this metric,we develop an optimal FEC assignment algo-rithm which perceptually allocates channel coding resources. There have been related studies of error propagation modeling [24],[25].However,these are computationally formidable when applied on a video server providing multiple concurrent video streams.Therefore,we developed a simple,alternative temporal error propagation model that requires much less com-putation.Of course,any type of error propagation modeling could be applied to our proposed scheme without dif?culty.We chose to adopt a performance metric derived from the packet loss rate,and concentrating on the perceptual application of FEC in terms of fairness and ef?ciency.

This optimal allocation is de?ned in terms of ef?ciency and fairness as a function of the spatio-temporal weight carried in each packet.At high-packet loss rates,ef?ciency is given greater emphasis by allocating increased protection to localized salient regions.In this way,if degradation from packet loss occurs in less salient regions,higher quality can be still attained in more salient region(s).On the other hand, at low-packet loss rates,the size of the salient region(s)can be expanded.Fairness among data packets is ful?lled by allocating more bits to region(s)of low saliency.Thus,a tradeoff between ef?ciency and fairness is mediated based on the number of available channel coding bits and the channel status.In the simulations,it is shown that de?nite performance gains are achieved in terms of visual quality,ef?ciency and fairness,relative to conventional algorithms.

II.Related Work

Retransmission-based error control techniques such as au-tomatic retransmission request have been shown to enhance the reliability of video transmission[26].Nevertheless,simple techniques of this sort present limitations in real-time situa-tions,owing due to delays arising from retransmitted packets. As an alternative,FEC deployed at the application layer yields a greater degree of ef?ciency.FEC can be adapted to variable bandwidths with reduced delay in wireless networks as well as in best-effort Internet networks.A number of researchers have proposed unequal FEC assignments to improve the quality of videos corrupted by packet loss[13]–[36].For single layer videos,unequal protection can be conducted as a function of the coding type of each frame along the temporal axis. In[16]and[17],FEC codes were unequally assigned to I-and P-frames in each GOP according to the channel status. Unequal importance can also be assigned at the packet

https://www.sodocs.net/doc/ee13142500.html,parison of perceptual quality.(a)30th frame using a perceptual ULP when a packet loss occurs in the frame.(b)45th frame after error propagation from the30th frame in(a).(c)30th frame using a conventional ULP when a packet loss occurs in the frame.(d)45th frame after error propagation from the30th frame in(c).

For example,the packet header,motion information,and text information can be adapted to improve video quality in packet erasure networks[15].In multilayered coding schemes,such as set partitioning in hierarchical trees,different degrees of im-portance can be assigned to the base and enhancement layers. By assigning unequal importance to the packets in different layers,unequal FEC schemes have been ef?ciently applied to multilayered coding[19],[20].In[21],unequal error protection(UEP)is applied to MPEG-4?ne granular scalable (FGS)compressed video data using rate-distortion information for each layer.UEP schemes for multiple description coding (MDC)and for hybrid space-time coding have also been de-veloped to achieve more robust video transmission[30],[31]. In recent years,source and channel rate allocation schemes have been deeply investigated for video communications[33]–[36].A rate allocation scheme with a delay constraint was presented in[33].The number of FEC codes is determined based on the network delay and the packet generation interval. The number of redundant packets is then allocated to attain a required packet loss ratio.In[34],a lower bound on the total transmission rate was computed by exploiting both source coding bits to attain minimum quality and channel coding bits to achieve the required packet loss ratio.

III.Perceptually Unequal Loss Protection(PULP)

A.Motivation

Fig.2shows the mechanics of PULP as compared to a conventional approach.If the available resources for video coding or transmission are plentiful,then we do not expect a performance improvement of the proposed scheme relative to conventional ones.However,when the resources are insuf?cient,then noticeably better performance can be attained by protecting perceptually important regions.The new approach balances a tradeoff between fairness and ef?ciency from the perspective of perceptual improvement.Fairness and ef?ciency mediate the visual quality by controlling the

HA et al.:PERCEPTUALLY UNEQUAL PACKET LOSS PROTECTION BY WEIGHTING SALIENCY AND ERROR PROPAGATION1189

size(s)of identi?ed salient region(s).As the fairness level is increased,salient regions are increased in size,leading to improved visual quality of the reconstructed video.The conventional approach,shown in Fig.2(b)employs equal perceptual weighting across the fairness levels.No spatial assignment of visual importance or salience is used in de?ning the fairness levels.Nevertheless,there are opportunities for incorporating perceptual relevance.For example,if regions that attract visual attention can be identi?ed,then resources can be allocated to them,while also taking into account human contrast sensitivity when selecting the quantization level or the prediction block size.By comparison,Fig.2(c)shows the proposed PULP framework,which adaptively con?gures each fairness level to the channel behavior.FEC assignments are based on perceptual weights,which are dynamically selected as a function of the channel state.The size of the salient region(s)is adaptively adjusted as a function of the fairness level and of the channel state.The larger the size of the salient region,the higher the fairness level is.For example, the fairness of level1is larger than that of level0in Fig.2(c). When the packet loss rate increases,the size of the salient region is reduced,to improve ef?ciency by setting a low-fairness level.When the packet loss rate decreases,the size of the salient region is expanded by setting a high-fairness level.

B.Overview of the PULP Algorithm

Fig.3diagrams various essential aspects of PULP.Fig.3(a) shows the?ow of PULP.Raw video frames are?rst fed into the video encoding module.During the encoding process,the degree of degradation due to packet loss is estimated used a quality metric called PWEP.The packet loss rate is estimated using the Markov model in[46].It can then be reported by the underlying protocol,such as the real time control proto-col(RTCP)[32].Furthermore,by deploying cross-layer co-operation,the channel signal-to-interference-plus-noise-ratio (SINR)can be measured using the pilot channel.From the SINR,the bit error rate and the packet error rate(PER) can then be estimated.If the PER information is fed back periodically to the end-user via RTCP packets,the QoS may be more reliably controlled.PWEP values are obtained using a foveal weighting model and a GOP-level hierarchical weight-ing model.The foveal weighting model calculates the LSB for each video packet.The LSB is decreased exponentially from the centers of each salient region,which are called foveation points.The exponential drop-off is such that,when a visual?xation falls on the salient region,the projection of the distribution of LSBs onto the retina will approximately match the nonuniform distribution of retinal photoreceptors[3]–[9]. Fig.3(b)shows the reconstructed35th frame of the video test clip"Silent"after applying perceptual weighting,where three salient regions were identi?ed.The?gure also depicts the foveal weighting model.Assume that the face and the left hand,both of which are in motion,are selected as a region of heightened visual interests.Picture-level perceptual weighting is allocated as a function of the spatial placement within the indicated iso-contours of the foveation-induced LSBs.In this example,video packets in region A are located in a highly salient region and are thus well protected.Video packets

in Fig.2.Proposed PULP framework compared to conventional ULP.

(a)Channel status.(b)Conventional ULP.(c)PULP.

region B are located in a low-saliency region and are less well protected.The spatial weighting is obtained for each frame in the GOP.In addition,the GOP-level hierarchical weighting model is used to identify regions that have unequal importance in the compressed video packets.Speci?cally,in each GOP the pictures have importance that descends with time relative to the?rst reference frame(I-frame),owing to the increasing severity of error propagation with elapsed time within the https://www.sodocs.net/doc/ee13142500.html,ing both of frame-level perceptual importance and GOP-level hierarchical importance,the PULP FEC assigner seeks to balance and optimize ef?ciency and fairness in order to achieve improved visual quality.The right portion of Fig.3(c)depicts the architecture of the model-based FEC assignment algorithm.Video packets from the video encoder are assembled into blocks of packets(BOP) by the BOP assembler for each GOP.Since channel error propagation is terminated within each GOP,this assembling leads to improved FEC capacity.The bit stream is sequentially packetized without considering regions of interest.Reed–Solomon(RS)codes are used across packets for FEC in the face of packet loss in packet erasure networks[15],[39].The (N,K)RS code has a code rate of K/N,where N packets are transmitted over the channel for K video packets.These N packets build a BOP,and this code rate can be adjusted for each BOP as a function of the unequal importance of the visual quality degradation and the channel state.

1)Problem Formulation:Let F j denote the number of FEC packets assigned into BOP j,Then,the FEC assignment vector for the current GOP is F=[F1,F2,...,F J].The optimal FEC assignment vector F?can be obtained by minimizing an

appropriate performance metric D( F)which is de?ned in the next section.For a given F j in BOP j,we denoteγ(F j)to be the packet loss rate after recovering with RS(N j,K j)codes.

A two-state Markov model is used to model the packet loss rate[46].If Pr(m,N)is the probability of losing m packets among N packets,then the original data can be recovered if the number of lost packets is less than the number of protection packets.γ(F j)can be formulated as

γ(F j)=

N j

m=N j?K j+1

Pr(m,N j)(1)

1190IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY,VOL.20,NO.9,SEPTEMBER

2010

Fig.3.Overview of the proposed VULP scheme.(a)Block diagram of the proposed VULP scheme.(b)Depiction of the foveation-based perceptual weighting model.(c)Architecture of the model-based FEC assignment algorithm.

An optimal FEC assignment vector F?can be obtained

by minimizing the spatio–temporal performance metric D( F),

which is the measure of quality degradation due to packet loss

from the previous section.The problem of?nding the optimal

FEC assignment vector F?is then

min F

D( F)

(2)

subject to

γ(F j)≤γ(F?j)if j≤?j(3) where j and?j are BOP indices.The constraint(3)is referred as“descending priority”from the sorted performance metric in the GOP

J

j=1

H j·F j≤B ch(4)

where J is the total number of BOPs for a given K.In constraint(4),B ch is the available number of channel bits for constructing FEC packets.H j is the length of FEC packets

in BOP j as determined by H j=max

k=1,2,...,K j {h k,j}where h k,j

is the length of the k th video packet in BOP j.For smaller video packets,?ller bytes are used to equalize the length of the video packets before FEC encoding.

IV.Perceptual Weight of Error Propagation

(PWEP)

Here,we describe a model-based performance metric for measuring the amount of visual quality degradation that occurs due to packet loss in a GOP.

A.Salient Point Selection

Selection of salient points is designed under the following assumption:The HVS often directs more attention to moving objects than to a stationary background[10],[11],[42],[44]. The HVS deals unequally with incoming visual information according to selective focal attention.This selectivity implies that if the human gaze is directed toward any speci?c loca-tions in a video,then that observer is less likely to notice defects in other areas of the video[43].In PULP,available MB saliency information,such as velocity magnitude,and motion partition,can be used to de?ne salient MBs.This can be used in conjunction with standard video formats, such as H.264/A VC,which provide inter and intra prediction modes to obtain improved coding performance[45].The MB partition information for the(i,j)th MB in the k th picture is de?ned as P k(i,j)and it is calculated by using rate-distortion optimization(RDO)for each MB.An example of MB partition information is showed in Fig.4(a)by using the 13th frame in the‘Soccer’test video clip.Small partition sizes are generally identi?ed with detailed regions or the edges of https://www.sodocs.net/doc/ee13142500.html,rge partition sizes are usually associated with monotonous,stationary image regions.For intra picture

HA et al.:PERCEPTUALLY UNEQUAL PACKET LOSS PROTECTION BY WEIGHTING SALIENCY AND ERROR PROPAGATION 1191

coding,a similar rule can be applied.Prediction modes that deploy a small block partition size are usually used to represent detailed https://www.sodocs.net/doc/ee13142500.html,rger partition sizes are used to coding homogeneous areas.

In addition to P k (i,j ),we utilize the velocity magnitude to de?ne salient areas.The motion intensity of (i,j )th MB in the k th picture is de?ned as I k (i,j )and it is calculated by I k (i,j )=

MV k x (i,j )2+MV k y (i,j )2where MV k x

(i,j )and MV k

y (i,j )represent the horizontal and vertical direction veloc-ity magnitude for the (i,j )th MB in the k th frame,https://www.sodocs.net/doc/ee13142500.html,ing P k (i,j )and I k (i,j ),candidate salient points can be selected.The detailed decision procedure is described in Fig.4(b).If P k (i,j )is small,it is probable that the MB contains details or information-bearing edges.Such MBs are taken to be part of salient regions.However,in the case that an MB is a part of the background,then P k (i,j )may not be large enough to report a possible saliency.Thus,after ?ltering out MBs using P k (i,j )we use the additional step of I k (i,j ).If I k (i,j )>0,the MB is regarded as part of a salient region.If I k (i,j )is too large,the viewer may not perceive such a rapid change,or might only obtain a limited amount of information [42],[44].This explains the use of a variable threshold on I k (i,j )for selecting salient points in the k th frame.The variability depends the global mean of velocity magnitudes in the frame (denoted by σk )which accounts for egomotion.Assuming that each frame is divided into M ×N

MBs,σk then is calculated as σk =1M ·N M ?1i =0 N ?1

j =0I k (i,j ).All other MBs excluding the selected salient MBs are treated as nonsalient MBs.Based on I k (i,j ),P k (i,j ),and σk ,we describe the decision algorithm for selecting salient points in Step 1and 2.

Step 1)Calculate I k (i,j ),P k (i,j )and σk for the (i,j )th

MB in the k th frame.

Step 2)We de?ne a binary function A k (i,j )as an indicator

of whether or not the (i,j )th MB in kth frame belongs to a salient region.A k (i,j )=1means a salient point,while A k (i,j )=0means others.

By using I k (i,j ),P k (i,j )and σk ,we determine whether A k (i,j )is 1or 0as follows:

A k (i,j )=

?

??

1,if (P k (i,j )

0

(5)

A result of the proposed algorithm is shown in Fig.4(c).

B.Foveation-Based Perceptual Weighting

Since video images are intended for human viewers,it is

presumed that the point of visual ?xation falls somewhere on the displayed video.It is also a reasonable assumption that visual input is dominated by the response of the cones (photopic vision),since the central dominant photoreceptors are highly responsive to bright objects,such as a glowing display monitor.The point on an object or monitor surface that projects light onto the center of the fovea,presuming that the gaze is ?xed,is termed a point of visual ?xation.At certain locations in the video stream where it is deemed likely that

Fig.4.Salient point selection for the 13th frame in the Soccer sequence.(a)A result of the MB block partition (P k (i,j ))obtained from H.264/A VC.(b)Salient MB selection algorithm.(c)Selected salient MBs.

the human gaze will fall at a given point in space and time,the video will be either represented at a higher resolution than other locations,and possibly given another kind of priority,such as increased error resilience.Such locations in the video stream will be referred to as foveation points .In the vicinity of a foveation point,the video is represented with a high-spatial resolution which falls off systematically away from the foveation point [except near other ?xation point(s)].In this way,the video presentation is made to have high resolution where the observers’visual ?xations are known or predicted to be placed.There has been useful work done on determining the visual resolution response (contrast sensitivity)as a function of the placement of the stimulus on the retina relative to the fovea,which is known as the retinal eccentricity [41],[42],[44].

For any given point x =(x 1,x 2)(pixels)in an image or video frame,the eccentricity (e )can be found by assuming

that the position of the foveation point x f =(x f 1,x f

2)(pixels)in the image plane and the viewing distance u from the eye to the image of size W (pixels)are known.The distance from x to x f is d ( x

, x f )= x 1?x f

1

2

+ x 2?x f

2

2

(pixels).The

eccentricity is e (u, x )=tan ?1(d ( x , x f

)

Wu

)[8].For a given eccentricity,e (u, x ),the local spatial cut-off frequency (cycle/degree)in x ,w c is de?ned in the sense that any higher frequency component beyond it is less visible or invisible.By setting the maximum possible contrast sensitivity to 1.0,w c is calculated as follows:

w c (e (u, x ))=e 2ln(1

CT 0

)α(e (u, x )+e 2) cycles degree

(6)

where CT 0is a minimum contrast threshold,e 2is a half-resolution eccentricity constant,and αis a spatial frequency

1192IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY,VOL.20,NO.9,SEPTEMBER2010 Fig.5.Distribution of normalized w n,k c(u)of the5th horizontal MBs for

W=1024and u=30cm.

decay constant.The?tting parameters given in[41]are

a=0.106,e2=2.3,and CT0=1/64.

In a displayed digital image,the maximum effective res-

olution is limited by the display’s visual resolution r(pix-

els/degree),which is approximately

r≈Wu

π

180

pixels

degree

.(7)

Based on the sampling theorem,the highest displayed Nyquist frequency is half the display resolution from(7)

w d(u)=r

2

≈Wuπ

360

cycles

degree

.(8)

Combining(6)and(8),the local foveal cutoff frequency for a given location x is

w c(u, x)=min(w c(e(u, x)),w d(u)).(9)

In addition,the cutoff frequency of the k th MB in the n th frame,w n,k c(u)can be expressed as the average value of the cutoff frequencies in the macroblock

w n,k c=avg(w c(u, x))(10)

where the x are the pixels in the k th MB.

As an example,Fig.5shows the distribution of normalized w n,k c(u)of the5th horizontal MBs for W=1024and u=30cm. It may be observed that larger weights are assigned to salient regions.These weights decrease as w n,k c(u)is decreased expo-nentially as a function of the distance from the foveation point. This process of foveation and weighting makes it possible to eliminate visual redundancies from nonsalient regions in order to improve coding ef?ciency.For brevity,w n,k c(u)is expressed by w n,k c to eliminate the dependence on u.Based on w n,k c,the perceptual weighting of the i th video packet of the j th BOP,μi,j can be calculated as

μi,j=

k∈S(p i,j)

w n,k c(11)

where p i,j is the i th video packet of the j th BOP and S(p i,j) is the set of MBs in p i,j.

TABLE I

Normalized Local Spatial Frequency,N s(m,n)in the4×4DCT

Domain

Item(m,n)0123

00.010.130.250.38

10.130.180.280.40

20.250.280.350.45

30.380.400.450.50

C.GOP-Level Hierarchical Weighting

To quantify the temporal propagation effects of packet loss on video quality,we use the length of the possible error propagation for each video packet.For example,a packet loss of the?rst frame causes a much more severe impact on the quality of the reconstructed sequence than a packet loss in one of the frames near the end ending.This simple method of assessing frame quality loss due to error propagation also has the virtue of simplicity and low complexity.Let f i,j andλi,j be the frame index in a GOP,and the length of error propagation of the i th video packet of the j th BOP,respectively.Then,λi,j is given by

λi,j=G+1?f i,j.(12)

Using(11)and(12),we then de?ne the perceptual weight of error propagation(PWEP),χi,j,which combines effects of both spatial and temporal video quality degradation

χi,j=μi,j·λi,j.(13)

V.Optimal FEC Assignment in PULP:Efficiency

and Fairness

The PWEP for each video packet can be obtained from (13)by using the spatial and temporal weighting principles outlined in the preceding.In order to minimize visual quality degradation as a function of the perceptual weighting,the proposed FEC assignment is adjusted as a function of the size of the salient region(s)and the channel status.

A.PWEP-FL(l)

For simplicity,denote the fairness level l as FL(l).The PWEP in(14)is speci?cally determined,as a function of the channel status,to achieve a desirable tradeoff between ef?ciency and fairness using the FL(l)algorithm.We term the proposed performance metric PWEP-FL(l).In the FL(l) algorithm,video packets having large LSB are protected from packet loss by adding more protection bits,and vice-versa. The lower the fairness level the algorithm obtains,the more unfair the video packets having a low LSB will be.Therefore it is important to carefully determine FL(l)to maintain an appropriate modicum of fairness for each channel state.Given the FL(l),a threshold on the cutoff frequency is determined. If w n,k exceeds the threshold,then the k th MB becomes a part of the salient region,and so on.

Fig.6(a)and(b)shows the mechanics of the proposed PULP framework as a function of the channel status.In this

HA et al.:PERCEPTUALLY UNEQUAL PACKET LOSS PROTECTION BY WEIGHTING SALIENCY AND ERROR PROPAGATION 1193

Fig.6.PULP framework.(a)Illustration of variation of the saliency size across the fairness level,FL(l ).(b)Distribution of perceptual weighting in the FL(l )algorithm as a function of the channel status.(c)Distribution of normalized PWEP-FL(l )for each BOP.

Fig.7.

For the 17th Stefan test sequence,shown are (a)contours of w n,k c ,(b)contours of ?w n,k c from w n,k

c ,(c)contours of v n,k from ?w n,k c .

example,the 17th frame of the "Stefan "test sequence is

utilized.In Fig.6(a),the lowest fairness level 0is assigned by setting FL(0).By maintaining the smallest salient region,it is possible to maximally protect perceptual quality within the salient region against a high-packet loss rate.For a moderate packet loss rate,the fairness level is increased by assigning an intermediate value of the LSB to,for example,FL(4),leading to enlargement of the salient region.Finally,at a low-packet loss rate,increased fairness can be assured by assigning a low threshold on the cutoff frequency.For example,for FL(9),the size of the salient region is noticeably larger.The LSB of those MBs lying within the salient region is set to the highest value of 0.5in the discrete frequency domain.To decide the size of the salient region(s),w n,k c from (10)is mapped onto a discrete level of frequency sensitivity,which varies with the frequency indices of the transform coef?cients,the coef?cient magnitudes,and the block luminances [3],[10],[11].Let m and n be the indices of 2-D transform coef?cients in a block.Then,the normalized local spatial frequency (cycle/degree)

can be expressed as N s (m,n )=1

√m 2+n 2where N s (m,n )is normalized by 0.5in [10].As shown in Table I,10values of N s (m,n )are used to control the size of the salient region(s).

For this purpose,w n,k c is quantized as a function of the value of N s (m,n ),so that the perceptual weighting of each MB is modi?ed and the size of the salient region(s)is controlled.

Let ?w n,k c be the quantized version of w n,k c ,where ?w n,k c

is mapped into the nearest discrete value of N s (m,n ).Fig.7(a)

and (b)shows the distribution of w n,k c and ?w n,k c .The values of

?w

n,k

c are mappe

d onto integer values denoted v n,k in th

e range [0,9],as shown in Fig.7(c).The relationship between v n,k

and ?w n,k c is de?ned by a weighting function ?n,k as follows:

?w n,k c =?n,k (v n,k )

(14)

where v n,k is an index used in Table I.If v n,k =9,the associated

MB obtains the highest discrete value ?w n,k c

=0.5.At the other extreme,if v n,k =0,then the lowest value ?w n,k c

=0.01is assigned to the MB.

Suppose that there are L fairness levels.For a given fairness level l ,the perceptual weighting of each MB is ?xed by

?n,k (?v n,k (l ))from (14),where ?v

n,k (l )indicates the modi?ed value of v n,k from the fairness level l ,which is calculated as

?v n,k (l )= L,v n,k ≥L ?l

v n,k +(L ?l ),otherwise

(15)

1194IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY ,VOL.20,NO.9,SEPTEMBER 2010

Fig.8.Variation of v n,k ,?v n,k (l ),and ?n,k (?v n,k (l ))when l =2.

where L =9in the implementation.This means that ?n,k (?v n,k (l ))is decreased in proportion to the distance from the center of the salient region,and is shifted as a function of the fairness level l .For example,when l =2,the variation of v n,k ,?v n,k (l ),and ?n,k (?v n,k (l ))is depicted in https://www.sodocs.net/doc/ee13142500.html,ing (14)and

(15),?v

n,k (2)is 9until v n,k is 7,and thereafter is decremented in units.?n,k (?v n,k (l ))shows the value of ?w n,k c

obtained from ?v

n,k (l ).Using the FL(l )algorithm,the perceptual weighting of the i th video packet of the j th BOP,which is denoted ?μi,j (l ),is found by including the fairness level l in μi,j in (11),and expressed using (14)and (15)as follows:

?μi,j (l )=

k ∈S (p i,j )

?n,k (?v n,k (l )).(16)

Finally,the performance metric PWEP-FL(l ),?χi,j (l )is found

by using (16)from (13)

?χi,j (l )=?μi,j (l )·λi,j .

(17)

The average value of PWEP-FL(l )for the FEC assignment

of the j th BOP is then calculated as

B j avg (l )

=1K j K j

i =1

?χi,j (l )(18)

where K j is the number of video packets in the j th BOP.

Fig.6(c)depicts the distribution of B j

avg as a function of

FL(l )with l =0,4,9.The slope of B j

avg for PWEP-FL(0)is steeper than for PWEP-FL(9)or PWEP-FL(4),since the size of the salient region(s)are reduced by setting a low-fairness level,when protecting the visual quality in a high-packet loss rate environment.As the fairness level is increased,then the

slope of B j

avg becomes reduced,when improving the visual quality of expanded salient region(s)in a low-packet loss rate environment.

B.Optimal FEC Assignment

Based on the weighting for each BOP in (18),the PULP FEC assigner performs an optimal FEC assignment to min-imize perceptual degradations,subject to a given protection redundancy and depending on the channel state.Expressed in terms of the packet loss rate in (18),and the average value of PWEP-FL(l )in (18),the spatio-temporal performance metric in (2)becomes

D ( F,

l )=J j =1

B j

avg (l )·γ(F j ).

(19)

Using the de?nition of D ( F,

l ),an optimal FEC assignment vector, F

?is found using a local hill-climbing search algorithm as in the following Steps 1–7.

Step 1)l curr and l prev represent the current and previous

fairness levels,respectively.Initially,l curr and l prev are set to 0.Thus,the smallest saliency region is initially used for searching the optimal FEC assignment as a function of the channel state.For given l curr and l prev ,the average distortions from (19)are denoted

as D ( F,

l curr )and D ( F,l prev )which are initially set to high values.

Step 2)Following compression of the video sequence

within a GOP,video packets are generated.The perceptual weights ?χi,j (l curr )are calculated by using (17).Each packet is then sorted to construct BOPs ordered by ?χi,j (l curr ).For a given K in RS(N,K ),the collection of J BOPs is partitioned as shown

in Fig.3(c).The value B j

avg (l curr )associated with BOP j is calculated to allow the assignment of FEC packets.

Step 3)The number of FEC packets for BOP j in the

face of a burst packet loss can be initially set to

be F init j = B ch ·H j J

i =1

H j where J is the

maximum number of BOPs.Then D ( F init ,l curr )is calculated by using (19).

Step

4)Next, F

best and F start are de?ned as the best FEC assignment in the GOP level and the starting point for the FEC assignment in the BOP level,respectively.

The algorithm seeks F

best at the GOP level.The initial values of F

best and D ( F best ,l curr )are F init and D ( F

init ,l curr ),respectively.Also,the initial value of F

start is set to a zero vector. F start is replaced by F best and the algorithm proceeds to Step 5.

Step

5)At the BOP level,the algorithm seeks the value of F

,denoted by F temp that achieves an optimal FEC assignment in the sense of minimizing D ( F

).F temp j ∈ F

temp is assumed to fall in the interval [? j , j ],where j is the search distance for BOP

j which is determined by j =max ε·B j

avg ,1 .This means that j I determined relative to the degree of the importance of the visual quality degradation from packet loss.If r is the determined value in the

interval [? j , j ],then F temp

j is updated as follows

F

temp = F start ,F temp j =F temp j +r .Step 6)We calculate D ( F

temp )=J j =1

B j avg ·γ(F temp j ).Step

7)If D ( F temp )

best )and F best are replaced by D ( F temp )and F

temp ,respectively.If D ( F temp )≥D ( F best ),the number of FEC packets in each BOP are adjusted to

minimize D ( F

temp )in the interval [? j , j ],then the algorithm returns to Step 5.If the search range falls outside the interval [? j , j ],then BOP j is shifted into the next BOP using j =j +1,and the algorithm returns to Step 5.If the index j of the BOP reaches the last value J ,the algorithm goes to Step 8.

Step

8)If F

best is equal to F start , F ?is replaced by F

best and the FEC assignment process goes to

HA et al.:PERCEPTUALLY UNEQUAL PACKET LOSS PROTECTION BY WEIGHTING SALIENCY AND ERROR PROPAGATION

1195

Fig.9.Performance comparison of average FSSIM and PSNR between PULP and conventional ULP algorithms for a FEC ratio of15%.(a)FSSIM of City.(b)PSNR of City.(c)FSSIM of Stefan.(d)PSNR of Stefan.

(e)FSSIM of Silent.(f)PSNR of Silent.(g)FSSIM of Soccer.(h)PSNR of Soccer.

the next step.Otherwise,the algorithm jumps to

Step4.

Step9)If D( F?,l curr)is lower than D( F?,l prev),we in-crease the fairness level by1:l curr+1.This means

that the size of the salient region is enlarged to adapt

to the channel state.Then,D( F?,l prev)is updated:

D( F?,l curr)and processing proceeds to Step2.Oth-

erwise,the FEC assignment process is terminated. It can be seen that the computational complexity of the proposed FEC assignment algorithm depends on the choice ofε.

VI.Simulation Results

In order to evaluate the performance of the proposed FEC scheme,extensive experiments under various test

conditions Fig.10.Performance comparison of average FSSIM and PSNR between PULP and conventional ULP algorithms for a FEC ratio of5%.(a)FSSIM of City.(b)PSNR of City.(c)FSSIM of Stefan.(d)PSNR of Stefan.(e)FSSIM of Silent.(f)PSNR of Silent.(g)FSSIM of Soccer.(h)PSNR of Soccer. were conducted.Four CIF video sequences City,Stefan, Silent,and Soccer,were used.The number of frames for each sequence is81and the frame rate is30f/s.The initial quantization parameter is set to be35.The videos were encoded using the H.264reference software[45].In the encoding con?guration,the RDO mode and the loop?lter were enabled.The content-based adaptive binary arithmetic coding option was enabled and variable block sizes with a search range of32were utilized for block motion estimation. The length of each GOP was selected to be15and the packet size1280bits.These sequences were encoded at a constant bit rate.A two-state Markov channel model described in[46]was used to model the packet loss with an average burst length of L B and an average packet loss rate of P B.The simulation parameters are shown in Table II.The error concealment scheme for H.264[8]in the reference software[45]was applied at the decoder side.The simulations were conducted

1196IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY,VOL.20,NO.9,SEPTEMBER2010

Fig.11.Performance comparison of frame-by-frame FSSIM between PULP and conventional ULP algorithms for a FEC ratio of5%–15%in City.(a)FEC ratio=15%and Pb=10%.(b)FEC ratio=15%and Pb=20%.(c)FEC ratio=5% and Pb=10%.(d)FEC ratio=5%

and Pb=20%.

TABLE II

Simulation Parameters

Sequence Name Stefan City Soccer Silent

FEC ratio(%)5and155and155and105and10 L B2213

P B(%)from5to20

K in RS(N,K)16

over the20different random channel loss patterns,and their results averaged.

The foveal weighting model was con?gured using a block size of4×4to evaluate?w n,k c.The parameterεin PULP was set to1.0.The simulation results were analyzed from two perspectives:objective perceptual quality and subjective perceptual quality.Fig.12.Performance comparison of frame-by-frame FSSIM between PULP and conventional ULP algorithms for a FEC ratio of5%–15%in Soccer.

(a)FEC ratio=15%and Pb=10%.(b)FEC ratio=15%and Pb=20%.(c)FEC ratio=5%and Pb=10%.(d)FEC ratio=5%and Pb=20%.

A.Objective Visual Quality Evaluation

We evaluated PULP algorithm over different channel states. Two fairness levels l=0and8were considered for high and low-packet loss rates.Due to the randomness of such a channel,100different runs of the simulation were conducted using different packet loss rates ranging from5%to20%.We investigated how well PULP adapted to channel variations as compared to other ULP schemes.

1)FEC-FL(0):Using FL(0)and PWEP-FL(0),the FEC

assignment was performed for pure ef?ciency.

2)FEC-FL(8):Using FL(8)and PWEP-FL(8),the FEC

assignment was performed to achieve intermediate per-formance between ef?ciency and fairness.

HA et al.:PERCEPTUALLY UNEQUAL PACKET LOSS PROTECTION BY WEIGHTING SALIENCY AND ERROR PROPAGATION 1197

Fig.13.Subjective quality comparison on the 15th frame of the Stefan test video clip (P B =5%).(a)Original video clip.(b)FEC-FL(0).(c)FEC-FL(8).(d)GRIP.(e)Equal FEC.

3)GOP and Resynchronization Integrated Protection (GRIP):The ULP scheme in [15]was performed using the length of error propagation as the performance metric,without using perceptual weights in each video packet.The resynchronization weighting scheme was not considered.

4)Equal FEC:The ULP scheme allocates FEC packets without considering either the packet loss rate or the perceptual signi?cance of error propagation.If FEC-FL(9)runs ULP without considering the packet loss rate in the FEC assignment,then this scheme becomes the same as the equal FEC.

To evaluate the quality of foveated video,the so-called foveal-peak signal-to-noise ratio (PSNR)was developed in [7].The foveal-PSNR is de?ned by weighting the LSB relative to the mean square error (MSE).This performance metric has been demonstrated to be a good objective quality measurement tool for predicting subjective quality of foveated images.Here,we introduce foveal-SSIM (FSSIM)in a manner similar to foveal-PSNR,but replacing the MSE with SSIM [29].To

compute FSSIM on each frame,the quantized LSB ?w n,k c

of the n th frame is applied to the SSIM index similar to [29]

FSSIM n = M

k =1SSIM (o n,k ,d n,k )·?w n,k c M k =1?w n,k

c (20)where M is the number of MBs in a frame,an

d o n,k and d n,k

are the k th matched MBs of the n th original and distorted frames.

The LSB varies for each window,frame and sequence over the spatial and temporal axes.Therefore,the ?nal score over the video is obtained using the simple pooling

FSSIM = T

n =1FSSIM n ·?w n sum n ?w n

sum (21)where T is the number of frames in the sequence,and ?w n sum = T n =1?w

n,k

c .In addition to conducting a perceptually relevant examina-tion using the FSSIM index,we also use the traditional (but perceptually questionable)average PSNR to evaluate the error.Figs.9an

d 10compar

e the average FSSIM and PSNR values for the conventional ULP and PULP algorithms,using the FEC ratios o

f 5%and 15%.At the low-packet loss rate,the FEC-FL(8)scheme exhibits higher FSSIM values than do FEC-FL(0),GRIP and Equal FEC by 0.01–0.05over all the test video frames.Relative to rate,a fair FEC assignment was performed for each BOP by enlargin

g the size of the salient

region,leading to improved objective quality.Conversely,at the high-packet loss rate,the FEC-FL(0)scheme achieved graceful degradation as measured by FSSIM in the range P B =15%–20%,while the GRIP and Equal FEC schemes resulted in steep degradation as measured by FSSIM.The gradual decrease of FSSIM in the proposed FEC scheme is due to the better protection from the packet loss,of those video packets having a large impact on visual quality,by maintaining a small salient region.On the other hand,the PSNR comparison shows that the GRIP scheme without visual weighting delivers higher PSNR values than does the proposed PULP algorithm.In particular,at a high-packet loss rate,since the smallest size of the salient regions is set by the proposed algorithm,the difference in PSNR values is largest within the feasible range of the packet loss rate.However,the subjective visual quality comparison in the next subsection makes it clear that the proposed FEC algorithm yields better visual quality than does the conventional FEC algorithm.

Figs.11and 12plot the frame-by-frame FSSIM using the FEC assignment scheme on the ?rst 80frames of the two test sequences,“City ”and “Soccer ”using FEC ratios of 5%and 15%.For each packet loss rate,it can be seen that the salient regions are better protected by FEC-FL(0)and FEC-FL(8)than those by the other protection schemes.Thus,spatial and temporal error propagation can be effectively alleviated using the proposed FEC assignment algorithm.B.Subjective Quality Comparison

To conduct subjective quality comparisons,we utilized a video test clip reconstructed using the benchmark methods with average packet loss rates of 5%and 20%,respectively.Fig.13shows the 15th frame of the “Stefan ”test video clip.Fig.13(a)is the original video clip,while Fig.13(b)–(e)are the reconstructed clips using FEC-FL(0),FEC-FL(8),GRIP,and Equal FEC,respectively,at a packet loss rate of P B =5%.It may be observed that improved subjective quality was delivered by the FEC-FL(8)scheme as compared to the GRIP,Equal FEC or FEC-FL(0)schemes.Thus,it may be deduced that fairness is more important than ef?ciency toward improving subjective video quality,by maintaining a wider range of salient regions given a low-packet loss rate.

Fig.14depicts the subjective quality comparison at the high-packet loss rate of P B =20%using the 15th frame of the “Soccer ”test video clip.Since FEC-FL(8)allocates FEC codes to video packets having a wide range of sizes of the salient region,noticeable degradations of subjective quality occur

1198IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY,VOL.20,NO.9,SEPTEMBER

2010

Fig.14.Subjective quality comparison on the15th frame of the Soccer test video clip(P B=20%).(a)Original video clip.(b)FEC-FL(0).(c)FEC-FL(8).

(d)GRIP.(e)Equal FEC.

in the reconstructed image.It is noticeable that FEC-FL(0) improves the subjective video quality more effectively than the fairness scheme at the high-packet loss rate.In the GRIP scheme,evident perceptual degradation occurs in the middle of the frame,owing to the lack of FEC codes.Conversely,FEC-FL(0)effectively inhibits perceptual degradations in areas of identi?ed perceptual importance,by allocating channel coding resource to those video packets.

VII.Conclusion

We proposed a new PULP algorithm appropriate for op-eration in a packet erasure network.To enable adaptation to the nonuniform resolution of the visual photoreceptors,we developed a simple and ef?cient performance metric,called PWEP.The proposed PULP scheme was developed based on two essential objectives,namely,enforcing ef?ciency and fairness across various channel states to improve visual quality. To mediate the tradeoff between ef?ciency and fairness,we proposed a FEC algorithm with a variable fairness level,FEC-FL(l)to allocate resources in order to manage the FEC codes for each video packet.The simulation results show that PULP algorithm achieves higher foveal-SSIM scores than conven-tional algorithms.It was demonstrated that PULP adapts well to dynamic channel environments,yielding good control of QoS,which is vital for achieving high quality and reliable video communication.

References

[1]M.R.Civanlar,A.Luthra,S.Wenger,and W.Zhu,“Introduction to

the special issue on sreaming video,”IEEE Trans.Circuits Syst.Video Tech.,vol.11,no.3,pp.265–268,Mar.2001.

[2]Y.Wang and Q.Zhu,“Error control and concealment for video com-

munication:A review,”Proc.IEEE,vol.86,no.5,pp.974–997,May 1998.

[3] B.Ciptpraset and K.R.Rao,“Human visual weighting progressive

image transmission,”in Proc.ICCS,Nov.1987,pp.1040–1044. [4] B.Girod,“What’s wrong with mean-square error?”in Digital Images

and Human Vision,A.B.Watson,Ed.Cambridge,MA:M.I.T.Univ.

Press,1993.

[5] A.B.Watson,J.Hu,and J.F.McGowan,III,“DVQ:A digital video

quality metric based on human vision,”J.Electron.Imag.,vol.10,no.

1,pp.1164–1175,Aug.1997.

[6]S.Lee,M.S.Pattichis,and A. C.Bovik,“Foveated video quality

assessment,”IEEE Trans.Multimedia,vol.4,no.1,pp.129–132,Mar.

2002.

[7]S.Lee,M.S.Pattichis,and A.C.Bovik,“Foveated video compression

with optimal rate control,”IEEE Trans.Image Process.,vol.10,no.7, pp.977–992,Jul.2001.

[8]Z.Wang and A.C.Bovik,“Embedded foveation image coding,”IEEE

Trans.Image Process.,vol.10,no.10,pp.1397–1410,Oct.2001.

[9]Z.Wang,L.Lu,and A.C.Bovik,“Foveation scalable video coding

with automatic?xation selection,”IEEE Trans.Image Process.,vol.12, no.2,pp.243–254,Feb.2003.

[10] C.Tang,“Spatiotemporal visual consideration for video coding,”IEEE

Trans.Multimedia,vol.9,no.2,pp.231–238,Feb.2007.

[11]J.Chen,J.Zheng,and Y.He,“Macroblock-level adaptive frequency

weighting for perceptual video coding,”IEEE Trans.Consumer Elec., vol.53,no.2,pp.775–781,May2007.

[12]K.Yang,C.C.Guest,K.E.Maleh,and P.K.Das,“Perceptual temporal

quality metric for compressed video,”IEEE Trans.Multimedia,vol.9, no.7,pp.1528–1535,Nov.2007.

[13]P.Frossard and O.Verscheure,“AMISP:A complete content-based error

resilient scheme,”IEEE Trans.Circuit Syst.Video Technol.,vol.11, no.9,pp.989–998,Sep.2001.

[14] A.E.Mohr,E.A.Riskin,and https://www.sodocs.net/doc/ee13142500.html,dner,“Unequal loss protection:

Graceful degradation of image quality over packet erasure channels through forward correction,”IEEE J.Select.Areas Commun.,vol.18, no.6,pp.819–828,Jun.2000.

[15]X.Yang,C.Zhu,Z.G.Li,X.Lin,and N.Ling,“An unequal packet loss

resilience scheme for video over the internet,”IEEE Trans.Multimedia, vol.7,no.4,pp.753–765,Aug.2005.

[16]J.Albanese,J.Edmonds,M.Ludy,and U.Horn,“Peioiry encoding

transmission,”IEEE https://www.sodocs.net/doc/ee13142500.html,rm.Theory,vol.42,no.6,pp.1737–1744,Nov.1996.

[17] F.Hartanto and H.R.Sirisena,“Hybrid error control mechanism

for video transmission in the wireless IP networks,”in Proc.Select.

Paper10th IEEE Workshop Local Metropolitan Area Netw.,2001, pp.126–132.

[18]M.Andronico and O.Verscheure,“Performance analysis of prior-

ity encoding transmission of MPEG video streams,”in Proc.Global Telecommun.Conf.,Nov.1996,pp.267–271.

[19]J.Kim,R.M.Mercereau,and Y.Altunbasak,“Error-resilient image

and video transmission over the internet using unequal error protection,”

IEEE Trans.Image Process.,vol.12,no.2,pp.121–131,Feb.2003.

[20]K.Stuhlmuller,M.Link,and B.Girod,“Robust internet video trans-

mission based on scalable coding and unequal error protection,”Signal Process.Image Comm.,vol.15,no.1,pp.99–94,1999.

[21]L.Zhuo,https://www.sodocs.net/doc/ee13142500.html,m,and L.Shen,“Channel-adaptive error protection

for streaming stored MPEG-4FGS over error-prone environments,”

IEEE Trans.Circuit Syst.Video Technol.,vol.16,no.5,pp.649–654, May2006.

[22]S.Lee,C.Podilchuk,V.Krishnan,and A.C.Bovik,“Foveation-based

error resilience and unequal error protection over mobile networks,”

Journal of VLSI Signal Processing Systems for Signal,Image and Video Technology,vol.34,pp.149–166,May2003(special issues on multimedia communications).

[23]L.Ning,Q.Zi-Yi,and M.Ai-Dong,“Perceptual optimized error-resilient

H.264video streaming system over the best-effort internet,”in Proc.

PDCAT,Dec.2005,pp.1039–1043.

[24]K.Stuhlmuller,N.Farber,M.Link,and B.Girod,“Analysis of video

transmission over lossy channels,”IEEE J.Select.Areas Commun.,vol.

18,no.6,pp.1012–1032,Jun.2000.

[25]X.K.Yang,C.Zhu,Z.G.Li,X.Lin,G.N.Feng,S.Wu,and N.Ling,

“Unequal loss protection for robust transmission of motion compensated video over the internet,”Signal Process.:Image Commun.,vol.18,no.

3,pp.157–167,Mar.2003.

[26]N.Feamster and H.Balakrishnan,“Packet loss recovery for streaming

video,”in Proc.IEEE Int.Packet Video Workshop,Apr.2002,pp.1–11.

[27]Z.Wang,A.C.Bovik,and L.Lu,“Why is image quality assessment so

dif?cult?”in Proc.IEEE Int.Conf.Acoust.Speech Signal Proc.,vol.4.

May2002,pp.3313–3316.

[28]Z.Wang and A.C.Bovik,Modern Image Quality Assessment.San

Mateo,CA:Morgan and Claypool Publishers,2006.

HA et al.:PERCEPTUALLY UNEQUAL PACKET LOSS PROTECTION BY WEIGHTING SALIENCY AND ERROR PROPAGATION1199 [29]Z.Wang,L.Lu,and A.C.Bovik,“Video quality assessment based on

structual distortion measurement,”Signal Process.Image Commun.,vol.

19,no.2,pp.121–132,Feb.2004.

[30]H.Zheng,C.Ru,C.W.Chen,and L.Yu,“Video transmission over

MIMO-OFDM system:MDC and space-time coding-based approaches,”

Advances Multimedia,vol.2007,article61491,pp.1–8,2007.

[31] C.H.Kuo,C.S.Kim,and C.C.J.Kuo,“Robust video transmission over

wideband wireless channel using space-time coded OFDM systems,”in

Proc.WCNC,vol.2.Mar.2002,pp.931–936.

[32]S.Wenger,Y.-K.Wang,and T.Schierl,“Transport and signaling of

SVC in IP networks,”IEEE Trans.Circuits Syst.Video Technol.,vol.17,

no.9,pp.1164–1173,Sep.2007.

[33]S.W.Yuk,M.G.Kang,B.C.Shin,and D.H.Cho,“An adaptive redun-

dancy control method for erasure-code-based real-time data transmission

over the internet,”IEEE Trans.Multimedia,vol.3,no.3,pp.366–374,

Sep.2001.

[34]Z.G.Li,C.Zhu,N.Ling,X.K.Yang,G.N.Feng,S.Wu,and F.Pan,

“A uni?ed architecture for real-time video-coding system,”IEEE Trans.

Circuit Syst.Video Technol.,vol.13,no.6,pp.472–486,Jun.2003.

[35]Y.Ming and W.H.Yuan,“A rate control scheme for H.264video

under low-bandwidth channel,”J.Zhejiang Univ.-Sci.A,vol.7,no.6,

pp.990–995,2006.

[36]Y.Liu,Z.G.Li,and Y.C.Soh,“A novel rate control scheme for

low-delay video communication of H.264/A VC standard,”IEEE Trans.

Circuit Syst.Video Technol.,vol.17,no.1,pp.68–78,Jan.2007.

[37]J.Lee,“Rate-distortion optimization of parameterized quantization ma-

trix for MPEG-2encoding,”in Proc.ICIP,Oct.1998,pp.383–386.

[38]H.Yin,“Adaptive quanitization in perceptual MPEG video encoders,”

in Proc.PCS,Apr.2006,pp.3–14.

[39] B.Girod,K.Stuhlmuller,M.Link,and U.Horn,“Packet loss resilient

internet video streaming,”in Proc.SPIE VCIP,Jan.1999,pp.833–844.

[40]T.Wiegand,G.J.Sullivan,G.Bj?ntegaard,and A.Luthra,“Overview

of the H.264/A VC video coding standard,”IEEE Trans.Circuit Syst.

Video Technol.,vol.13,no.7,pp.560–576,Jul.2003.

[41]W.S.Geisler and J.S.Perry,“A real-time foveated multiresolu-

tion system for low-bandwidth video communication,”in Proc.SPIE,

vol.3299.1998,pp.1–13.

[42]J.G.Robson and N.Graham,“Probability summation and regional

variation in contrast sensitivity across the visual?eld,”Vis.Res.,

vol.21,no.3,pp.409–418,1981.

[43]L.Itti,“Quantifying the contribution of low-level saliency human eye

motivement in dynamic scenes,”Vis.Cognit.,vol.12,no.6,pp.1093–

1123,Aug.2005.

[44]M.S.Banks,A.B.Sekuler,and S.J.Anderson,“Peripheral spatial

vision:Limited imposed by optics,photoreceptors and receptor pooling,”

J.Opt.Soc.Amer.,vol.8,no.11,pp.1775–1787,1991.

[45]H.264/AVC Software Coordination[Online].Available:

http://iphome.hhi.de/suehring/tml/index.htm

[46] E.O.Elliott,“A model of the switched telephone network for data

communications,”J.Bell.Syst.Tech.,vol.44,no.1,pp.98–109,

1965.

Hojin Ha received the B.S.degree in control and in-

strumentation engineering from Myongji University,

Yongin,Korea,in1998,the M.S.degree in control

and instrumentation engineering from Hanyang Uni-

versity,Ansan,Korea,in2000,and the Ph.D.degree

in electrical and electronic engineering from Yonsei

University,Seoul,Korea,in2009.

Since2000,he has been a Research Engineer

with the Digital Media and Communications Re-

search and Development Center,Samsung Electron-

ics,Suwon,Korea.His research interests include multimedia communications,multimedia signal processing,and peer-to-peer

networking.

Jincheol Park was born in Korea in1982.He

received the B.S.degree in information and elec-

tronic engineering from Soongsil University,Seoul,

Korea,in2006,and the M.S.degree in electrical

and electronic engineering from Yonsei University,

Seoul,Korea,in2008.He is currently pursuing

the the Ph.D.degree with the Wireless Network

Laboratory,Yonsei University.

His current research interests include wireless mul-

timedia communications and video quality assess-

ment.

Sanghoon Lee(M’05)was born in Korea in1966.

He received the B.S.degree from Yonsei Uni-

versity,Seoul,Korea,the M.S.degree from the

Korea Advanced Institute of Science and Technol-

ogy,Daejeon,South Korea,and the Ph.D.degree

from the University of Texas,Austin,all in elec-

tronic engineering,in1989,1991,and2000,respec-

tively.

From1991to1996,he was with Korea Telecom,

Seocho-gu,Seoul,Korea.In1999,he was with Bell

Laboratory,Lucent Technologies,Murray Hill,NJ, and worked on wireless multimedia communications.From2000to2002, he worked on developing real-time embedded software and communication protocols for3G wireless networks with Lucent Technologies.Since2003,he has been with the faculty of the Department of Electrical and Electronics En-gineering,Yonsei University,where he is currently an Associate Professor.His current research interests include4G wireless networks,3G W-CDMA/CDMA networks,multihop sensor networks,wireless multimedia communications, and image/video quality assessments.

Dr.Lee is an Associate Editor of Journal of Communications and Networks and IEEE Transactions on Image Processing

.

Alan Conrad Bovik(S’80–M’81–SM’89–F’96)re-

ceived the B.S.,M.S.,and Ph.D.degrees in electrical

and computer engineering from the University of

Illinois at Urbana-Champaign,Champaign,in1980,

1982,and1984,respectively.

He is currently the Curry/Cullen Trust Endowed

Professor with the University of Texas,Austin,

where he is the Director of the Laboratory for Image

and Video Engineering Center for Perceptual Sys-

tems.He has published over450technical articles

in his?elds and holds two U.S.patents.His current research interests include image and video processing,computational vision, digital microscopy,and modeling of biological visual perception.

Dr.Bovik is the author of the Handbook of Image and Video Processing (Amsterdam,The Netherlands:Elsevier,2005,2nd ed.)and Modern Image Quality Assessment(San Mateo,CA:Morgan and Claypool,2006).He has received a number of major awards from the IEEE Signal Processing Society, including the Education Award in2007,Technical Achievement Award in 2005,Distinguished Lecturer Award in2000,and Meritorious Service Award in1998.He is also a recipient of the Distinguished Alumni Award from the University of Illinois at Urbana-Champaign in2008,IEEE Third Millennium Medal in2000,and two journal paper awards from the International Pattern Recognition Society in1988and1993.He is a Fellow of the Optical Society of America,Society of Photo-Optical Instrumentation Engineers.He has held positions in numerous professional society activities,including the Board of Governors of the IEEE Signal Processing Society from1996to1998,the Editor-in-Chief of the IEEE Transactions on Image Processing from 1996to2002,an Editorial Board Member of Proceedings of the IEEE from 1998to2004,the Series Editor for Image,Video,and Multimedia Processing, Morgan and Claypool Publishing Company from2003to present,and the Founding General Chairman of the First IEEE International Conference on Image Processing,Austin,in1994.He is a registered Professional Engineer in the state of Texas and is a frequent consultant to legal,industrial,and academic institutions.

网络文明演讲稿3篇

本文目录络文明演讲稿络文明演讲稿英语之星的演讲稿: 络文明 计算机互联作为开放式信息传播和交流的工具，已经走进了我们的学习和生活。从它刚刚兴起直到现在的风靡一时，年青的我们凭着对新鲜事物特有的接受能力，一直都是它忠实的应用者，无论是学习、休闲还是交流，络都发挥了不可替代的作用。络是把“双刃剑”在给我们带来利处的同时，由于我们辨别是非的经验不足，一些络糟粕也随之侵袭着我们的心灵。到底怎么样才算是文明上 ?反过来络最大的文明又应该是什么? 其实，关于络所出现的问题，早已引起了家长、学校和社会的关注，年11月，团等8个单位发布了《全国青少年络文明公约》，提倡"要善于上学习，不浏览不良信息; 要诚实友好交流，不侮辱欺诈他人;要增强自我保护意识，不随意约会友;要维护络安全，不破坏络秩序;要有益身心健康，不沉溺虚拟时空。"只要我们正确健康地上，络就会成为我们学习知识、交流思想、休闲娱乐的重要平台。现在，我们不仅学校有电脑，而且很多家庭都有了电脑，那我们在使用络时该注意什么呢? 善于上学习，不浏览不良信息。现在人们对我们中学生上有一种普遍的看法：即不是玩游戏就是聊天。其实，上学习，天地宽广。在上学习，可以查关于学习的资料，也可以在bb 上与同学交流学习的经验。这样好的条件，何乐而不为呢?但凡是不良信息的站，不应该浏览;不健康的聊天室，都应该马上离开;如果不小心点击出了不健康的页面，应该马上关闭。 络语言尤其是对于正处于青春期与成长期的我们而言具有强烈的吸引力。青少年永远是时尚和流行的载体与传播者，络上的个性也在我们身上表现的淋漓尽致。络语言这些独特的表达方式，反映了我们的个性和联想能力。这从一个侧面反映了我们的特征——年轻的心态和不拘一格的想法。但是自从“第四媒体”互连出现以后，任何一个连接到互连的人都能够把自己的言论传播到络上去。络在带来发表言论，表达思想机会的同时，也导致了“滥言时代”。络的虚拟性，使的埋藏在人们心底的许多东西都迸发了出来，特别是许多不健康，不文明的东西，“反正都不认识现实的我，我想怎么说就怎么说”，许多人正是抱着这种想法，成为了不文明事物和行为的传播者，甚至是始作俑者。作为我们青少年应严格规范要求自己，坚决抵制上使用不文明用语，在络上不讲脏话不带粗口，做到时时使用络文明语言，在无限宽广的络天地里，倡导青少年络文明新风，为营造健康的络道德环境做己的努力。 上时文明守法是我们在场的每一位红岭人必需做到的。我们应遵守络道德规范，懂得基本的对与错、是与非，增强络道德意识，辨明上的善恶、美丑，在上不应随便发表不负责任的言论。前段时间在《高楼坠落玻璃块砸死过路小学生》一文登出后，福田某中学高三学生胡某发帖自称“肇事者”，经民和警方一路追踪发现竟是恶作剧。尽管只是场恶作剧，但是18岁的胡某还是为他的行为付出了代价。由于其发布帖子后，一方面造成了严重的社会影响，同时也干扰了警方对案件的正常侦查。根据《治安管理处罚法》的有关规定，警方依法对胡某采取了治安拘留。这也给我们敲响了警钟。 尊敬的老师，同学们： 大家好!

基于PacketTracer5.0构建虚拟校园网的基本功能

基于Packet Tracer5.0构建虚拟校园网的基本功能 摘要本文利用packet tracer 模拟器模拟校园网内局域网的建设，而且配置了校园网的常用功能，如三层交换机VLAN划分、交换机IP地址设置及远程登录功能、Web和DNS 服务器功能的实现。 关键词packet tracer 校园网功能 随着互联网的发展，各高校都有了自己的校园网，校园网基本功能的配置就成为网络管理人员的一项重要任务，本人利用packet tracer 软件模拟校园网的基本构架，详细介绍了校园网内如何配置VLAN和交换机的远程管理的配置,并对校园网内服务器做了一个简单的配置.希望本实验能给校园网络管理员带来一些帮助,而且本实验也可做为高职高等院校网络技术课的实训案例. 1：packet tracer 介绍 Packet Tracer 是由Cisco公司发布的一个辅助学习工具，为学习思科网络课程的初学者去设计、配置、排除网络故障提供了网络模拟环境。用户可以在软件的图形用户界面上直接使用拖曳方法建立网络拓扑，并可提供数据包在网络中行进的详细处理过程，观察网络实时运行情况。可以学习IOS的配置、锻炼故障排查能力。软件还附带4个学期的多个已经建立好的演示环境、任务挑战，该软件仿真度很高，受到业内的极高评价。 2：校园网拓扑图及Vlan 划分 本实验有一个三层交换机（校园网核心交换机），四个接入层交换机（每栋楼两个,一个做为楼栋主交换机,另外一个继连在主交换机上），四台主机，两台服务器（一台WEB服务器，一台DNS服务器），vlan10划分给A楼，取名为:Adonglou,IP段为222.17.193.0/26;Vlan20划分给B楼, 取名为:Bdonglou,IP段为：222.17.193.64/26;vlan30划分给服务器群，取名为:Fuwu,IP段为：222.17.192.0/26;vlan40划分给校园网内的交换机，取名为:SheBei,IP段为：222.17.199.0/26.

文明网络演讲稿4篇

文明网络演讲稿4篇Civilized network speech document 编订：JinTai College

文明网络演讲稿4篇 小泰温馨提示：演讲稿是在较为隆重的仪式上和某些公众场合发表的讲话文稿。演讲稿是进行演讲的依据，对演讲内容和形式的规范和提示，体现着演讲的目的和手段，用来交流思想、感情，表达主张、见解；也可以用来介绍自己的学习、工作情况和经验等等；同时具有宣传、鼓动、教育和欣赏等作用，可以把演讲者的观点、主张与思想感情传达给听众以及读者，使他们信服并在思想感情上产生共鸣。本文档根据演讲稿内容要求展开说明，具有实践指导意义，便于学习和使用，本文下载后内容可随意修改调整及打印。 本文简要目录如下：【下载该文档后使用Word打开，按住键盘Ctrl键且鼠标单击目录内容即可跳转到对应篇章】 1、篇章1：文明网络演讲稿 2、篇章2：网络文明演讲稿 3、篇章3：文明网络演讲稿 4、篇章4：英语之星的演讲稿网络文明 篇章1:文明网络演讲稿 遵守网络文明公约争做文明上网人

计算机互联网作为开放式信息传播和交流的工具，已经 走进了我们的学习和生活。从它刚刚兴起直到现在的风靡一时，年青的我们凭着对新鲜事物特有的接受能力，一直都是它忠实的应用者，无论是学习、休闲还是交流，网络都发挥了不可替代的作用。网络是把“双刃剑”在给我们带来利处的同时，由于我们辨别是非的经验不足，一些网络糟粕也随之侵袭着我们的心灵。到底怎么样才算是文明上网？反过来网络最大的文明又应该是什么？ 其实，关于网络所出现的问题，早已引起了家长、学校 和社会的关注，XX年11月，团中央等8个单位发布了《全国 青少年网络文明公约》，提倡"要善于网上学习，不浏览不良 信息；要诚实友好交流，不侮辱欺诈他人；要增强自我保护意识，不随意约会网友；要维护网络安全，不破坏网络秩序；要有益身心健康，不沉溺虚拟时空。"只要我们正确健康地上网，网络就会成为我们学习知识、交流思想、休闲娱乐的重要平台。现在，我们不仅学校有电脑，而且很多家庭都有了电脑，那我们在使用网络时该注意什么呢？ 善于网上学习，不浏览不良信息。现在人们对我们中学 生上网有一种普遍的看法：即不是玩游戏就是聊天。其实，网上学习，天地宽广。在网上学习，可以查关于学习的资料，也

基于packettracer智能校园网组建实验指导书

基于Packet tracer 组建校园网 -------实验指导书 一、IP地址划分 根据学校的部门数量划分，将学校分为以下几个VLAN: 二、拓扑图

三、配置 1. 交换机配置 核心交换机为Cisco 3560，将其配置为vtp Server, vtp domain 为senya。将图书馆、教学楼和实验楼等交换机配置为vtp Client，vtp domain为senya。这里以“中心交换机”和“服务器汇聚”交换机为例，讲解交换机的配置，其他交换机的配置可以参考“服务器汇聚”交换机。 第一步：中心交换机配置VTP Server Switch>en Switch# Switch#vlan database Switch(vlan)#vtp domain senya Switch(vlan)#vtp server Switch(vlan)#exit Switch#conf t Switch(config)#int fa 0/1 Switch(config-if)#switchport trunk encapsulation dot1q Switch(config-if)#switchport mode trunk Switch(config-if)#int fa 0/2 Switch(config-if)#switchport trunk encapsulation dot1q Switch(config-if)#switchport mode trunk Switch(config-if)#int fa 0/3 Switch(config-if)#switchport trunk encapsulation dot1q Switch(config-if)#switchport mode trunk Switch(config-if)#int fa 0/4 Switch(config-if)#switchport trunk encapsulation dot1q Switch(config-if)#switchport mode trunk Switch(config-if)#int fa 0/5 Switch(config-if)#switchport trunk encapsulation dot1q Switch(config-if)#switchport mode trunk Switch(config-if)#int fa 0/6 Switch(config-if)#switchport trunk encapsulation dot1q Switch(config-if)#switchport mode trunk Switch(config-if)#int fa 0/7 Switch(config-if)#switchport trunk encapsulation dot1q Switch(config-if)#switchport mode trunk Switch(config-if)#int fa 0/8 Switch(config-if)#switchport trunk encapsulation dot1q Switch(config-if)#switchport mode trunk Switch(config-if)# 注意：此处端口要处于开启状态 第二步：配置“服务器汇聚”交换机trunk链路，允许vlan标记的以太网帧通过该链

文明上网,从我做起(主题班会演讲稿)

男：同学们，在开班会之前，我们想问大家几个熟悉的问题：你的QQ号是多少？你的农场几级了？牧场呢？梦幻西游呢？ 女:现在随着中国社会的不断发展，人们的生活也逐渐富裕起来，网络也跟着在社会中流行，融入到了我们的生活中去，而且扩张的范围极大，尤其是对于中学生来说更是对“网络”爱不释手。 男：上个星期，我就问过几个中学生对告别网吧的看法，他们中的一些坚定的摇了头，并说出了一大堆上网的好处，而且反问我：“你不是也经常上网吗？”，于是，我产生了几个思考：告别网吧，是不是就等于告别了网络呢？告别网吧，这现实吗？我们的中学生在接触网络时，该如何把持自己呢？ 女：老师曾经说过，21世纪是知识经济的时代，是网络的时代，是人类数字化生存的时代，电脑和网络，是每一个学生都尽可能掌握的一门课程。在美国等西方国家，四五年级的学生都能够熟练地使用电脑、网络来查阅资料，学习知识。相比之下，让我们中学生告别网络的做法，我想在坐的同学没有一个会同意。 男：但我们的法律为什么又规定“禁止未成年人进入网吧” 呢？首先，我们不能把“网络”等同于“网吧”。第二，表面禁止的同时，深含对我们未成年人的身心保护。我们中

学生自制能力差，一旦迷恋网吧便不能自拔，导致学业无成，甚至是猝死网吧。这类的现象时有耳闻，如果事件发生，我们便总认为这是因无知而犯下的错误，但也为时已晚了。 女：“禁止未成年人进入网吧”是为了让我们的学生少犯或不犯相同的错误，让更多的人来关爱我们这些未成年人。 不接触网吧，我们同样可以接触网络，比如：通过家 庭的个人电脑、学校电脑室的电脑等。 男：当然，作为孩子的我，完全能够理解为什么有的人家里有电脑，但他还要去网吧。刚才也说到我们自制能 力差，只要用上电脑，便会觉得时间如白驹过隙，不 管用多久，都觉得不够。所以，每当感到刚玩了一会，就会听到父母说“你都玩了几个小时了！该去写写作业 了吧！”“你看看都几点了，还在玩？明天不用上学啊？” “你玩什么呢？这么聚精会神？要是学习也这样，还至 于几百多名吗”“你作业写完了也不能玩这么久啊！复习 去！”“作业写完了吧？来，别玩了，考考你学过的英文单 词。”“哎呀......忙了一天了，累死我了，别玩了，去把衣 服晾一下，然后扫一下地，再把碗刷了！”等。父母喋喋 不休的唠叨，真的可以让人发疯！但是大家有没有想 过，他们为什么要这样做？难道真的是因为看你不顺 眼、想拿你撒气？

CiscoPacketTracer教程

Cisco Packet Tracer实验讲义 实验1 模拟器的初步了解 1 模拟器界面 (1)设备选择区的网络设备：路由器、交换机、集线器、无线设备、线缆、终端设备、仿真广域网、自定义设备 （2）线缆： 自动选择：通用，不建议使用 控制线：用来连接计算机的COM 口和网络设备的Console口 直通线：使用双绞线两端采用同一种线序标准制作的网线，一般用来连接计算机和交换机、交换机与交换机、交换机与路由器 交叉线：使用双绞线两端采用不同线序标准制作的网线，一般用来连接计算机与计算机，计算机与路由器、路由器与路由器 光纤：连接光纤设备 电话线：连接调制解调器或者路由器的RJ-11端口的模块 同轴电缆 DCE/DTE串口线：用于路由器的广域网接入。需要把DCE串口线与一台路由器相连，DTE 串口线与另一台设备相连，但是Cisco Packet Tracer中，之需要选一根就可以了，选了DCE这一根线，则和这根线先连的路由器为DCE端，需要配置该路由器的时钟 （3）设备编辑工具箱 选择，更改布局，笔记，删除，查看，添加简单协议数据单元，增加复杂协议数据单元 2 PC配置方法 添加PC，单击，点击“配置”，可添加IP地址等 桌面菜单提供若干功能 3 路由器的配置模式 用户模式 enable:进入特权模式 configure terminal:进入全局配置模式 interface fastEthernet 0/1 进入端口查看模式

4 连接设备 设备添加好以后，选择相应的线缆，然后在要进行连线的网络设备上单击，会弹出如下图所示的端口选择界面，选中要进行连接的端口，再移动到另外一台设备，选中适当的端口就完成设备的连接了。 实验2 交换机的Vlan配置 1 交换机的基本配置 为设备命名:选中一个交换机加入图中，单击，进入CLI：

INTERNET演讲稿

so there are more and more people who have became internet users. do you know how many internet users in the world? let’s see the chart! 3 this is a questionnaire [?kw?st??n??r, ?kwest???ne?] (调查卷) about world internet users and population stats in 2009. the users has grown [gr?un] at a remarkable [ri?mɑ:k?bl] rate 。the rising line(上升的曲线） is not this ,not this ,is this. let us see the next pictures. 4 who is he？yes .our respectable premier [?premi?] wen .how old is he? a ha .this i dont know ,but i know he is very old , he is also surfing the internet and caring about current events and politics .so how we do ? 5 yes .the man may be watch a funny movie. the mother is telling her children learning some knowledge through the internet. in this picture .the people from different nations also learn the same thing-------how to use the internet. even the lovely dog is also want to use the internet. why are many people attracted by internet? now im not surprised to see the results, for i can find the reason just from my own life. 6 the internet changed my life; there is no doubt about that. to spend a part of our day on the internet is quite normal for many people. 7 second. the internet is a database ?deit?beis] full of latest information and offers me a lot of services. i can read the daily newspapers, movies and music from all over the world. 8 when i saw this news, my mood just like this picture—angry birds .how cruel the passed people are .ok, let me back to the topic .this is an online game--- angry birds. that’s another important part for internet .the online games are always attractive and challenging. in my opinion its more exciting to play with friends than playing alone. 9 third. to improve our english, through internet i often find professional english learning methods. is internet good or bad we should admit that there are so many advantages brought by the internet. firstly, the internet affords us lots of convenience. for example, we can shop, have meetings and even study on-line. furthermore, the internet has improved our working efficiency. we can contact colleagues on the other side of the world to talk about the working project via the internet easily. piles of files can be sent by e-mails with the help of the internet. in addition, the internet makes information conveyance much easier. just clipping “google”, related information will boom out explosively within several seconds. while applauding for those benefits the internet brings to us, we need to worry about disadvantages of the internet as well. to begin with, it is easy for the young to indulge themselves in the so-called “cyber romance”, which is full of dangers, cheatings

2、使用网络模拟器Packettracer

一、实验名称 使用网络模拟器Packettracer 二、实验目的 （1）掌握安装和配置网络模拟器软件Packettracer的方法 （2）掌握使用PacketTracer模拟网络场景的基本方法，加深对网络环境、网络设备和网络协议交互过程的理解。 三、实验内容和要求 （1）安装和配置网络模拟器； （2）熟悉PacketTracer模拟器； （3）观察与IP网络接口的各种网络硬件； （4）进行ping和tracerroute实验 （5）四、实验环境 四、操作方法与实验步骤 五、1）、安装网络模拟器：安装CISCO网络模拟器Packettracer，双击Packettracer 安装程序图标，进行安装，根据提示进行选择确认，可以顺利安装系统。步骤截图就不给出了，因为在实验之前就已经安装过了。 2）、使用Packettracer模拟器：对于其使用的掌握可以去阅读老师给的资料“Packettracer5.0全攻略上、下“。 3）、观察与IP网络接口的各种网络硬件进行 （1）、从Packettracer中打开路由器2620XM的物理设备视图，界面如下

太网接口，而且是要用光纤连接才能使用的接口（不知道翻译的对不对） 将其拖入设备，可以观察到面板硬件接口的情况如下： 自行测试该模块的适用使用场合：对于两台路由器我用除光纤以外的先连接添加的两个端口 会出现下面的错误提示

然后我再用光纤连接它们，发现可以连通，这就验证了NM-1FE-FXde适用的场合： 点击对NM-1FE-TX界面左下角出现对其的描述（备注：拖入设备的过程都和NM-1FE-FX 一样之后的模块就都省去拖入那个步骤了）：应该是提供了一个10/100Mbps的以太网接口，而且是要用交叉铜线才能使用的接口（不知道翻译的对不对） 自行测试该模块的适用使用场合：对于两台路由器我用除交叉铜线可以连通其他都不能连通，可以看出NM-1FE-TX的适用场合。 点击对NM-2FE2W界面左下角出现对其的描述：应该是提供了两个10/100Mbps的以太网接口，而且是要用交叉铜线连接才能使用的接口（不知道翻译的对不对） 自行测试该模块的适用使用场合：对于两台路由器由于使用NM-2FE2W给它们都添加了两个两个要用交叉铜线连接才能使用的10/100Mbps的以太网接口，我把一对接口用交叉铜线连接发现连接通了，而另一对则使用直连铜线连接发现连接不同，因此们我可以看出

精选请文明使用网络演讲稿3篇(最新)

网络语言尤其是对于正处于青春期与成长期的我们而言具有强烈的吸引力。青少年永远是时尚和流行的载体与传播者，网络上的个性也在我们身上表现的淋漓尽致。网络语言这些独特的表达方式，反映了我们的个性和******能力。这从一个侧面反映了我们的特征——年轻的心态和不拘一格的想法。但是自从“第四媒体”互连网出现以后，任何一个连接到互连网的.人都能够把自己的言论传播到网络上去。网络在带来发表言论，表达思想机会的同时，也导致了“滥言时代”。网络的虚拟性，使的埋藏在人们心底的许多东西都迸发了出来，特别是许多不健康，不文明的东西，“反正都不认识现实的我，我想怎么说就怎么说”，许多人正是抱着这种想法，成为了不文明事物和行为的传播者，甚至是始作俑者。作为我们青少年应严格规范要求自己，坚决抵制上网使用不文明用语，在网络上不讲脏话不带粗口，做到时时使用网络文明语言，在无限宽广的网络天地里，倡导青少年网络文明新风，为营造健康的网络道德环境做出自己的努力。 上网时文明守法是我们在场的每一位红岭人必需做到的。我们应遵守网络道德规范，懂得基本的对与错、是与非，增强网络道德意识，辨明网上的善恶、美丑，在网上不应随便发表不负责任的言论。前段时间在《高楼坠落玻璃块砸死过路小学生》一文登出后，福田某中学高三学生胡某发帖自称“肇事者”，经网民和警方一路追踪发现竟是恶作剧。尽管只是场恶作剧，但是18岁的胡某还是为他的行为付出了代价。由于其发布帖子后，一方面造成了严重的社会影响，同时也干扰了警方对案件的正常侦查。根据《治安管理处罚法》的有关规定，警方依法对胡某采取了治安拘留。这也给我们敲响了警钟。 其实网络只是给我们提供了一个自由发挥的平台。我们不能把网络中不文明事物和行为的出现怪罪于网络，我们所需要净化的不应该是网络，而应该是我们自己，要络方面的法律、法规知识。在提高修养的同时，还要不断检讨自己的行为。自觉抵制不文明事物和行为，并传播文明的事物和行为。在文明的网络中体会网络的乐趣。 同学们，网络在我们面前展示了一幅全新的生活画面，同时，美好的网络生活也需要我们用自己的美德和文明共同创造。让我们响应全国青少年网络文明公约的号召，从我做起，从现在做起，自尊、自律，上文明网，文明上网，真正做一名文明、健康的小网民。 尊敬的老师，同学们： 大家好! 我今天将要给大家演讲《网络——把锋利的双刃剑》。 众所周知，随着时代的时步，科技的更新，更多的新事物涌入我们的世界，其中也包括了互联网。自90年代后期互联网进入我国以来，网络就如同雨后春笋般，发展迅速，如今已遍布城乡。作为一种新的传媒手段，互联网在经济建设和人们的各项活动中起着不可忽视的作用，同时也为青春少年的学习.交流以及

文明上网演讲稿【五篇】

文明上网演讲稿【五篇】 【篇一】文明上网演讲稿 各位领导，各位老师，同学们： 大家早上好！ 我今天演讲的题目是《告别痴迷，文明上网》。 现在随着中国社会的不断发展，人们的生活也逐渐富裕起来，网络也跟着在社会中流行，融入到了我们生活中去，而且扩张的范围很大，尤其是对于中学生来说更是对“网络”爱不释手。为此，上个星期，老师叫我写有关网络文明的演讲稿，我想就写“告别网吧，做文明健康中学生”为题的演讲稿吧，于是我就中学生告别网吧一事专门调查了几位同学，他们中的一些坚定的摇了头，并说出了一大堆上网的好处，而且反问了我：“你也不是经常上网吗？”，于是，产生了我的几个思考：告别网吧，是不是就等于告别了网络？告别网吧现实吗？我们的中学生在接触网络时，该如何把持自己呢？ 老师说过，21世纪是知识经济的时代，是网络时代，是人类数字化生存的时代，电脑和网络，是每一个学生都必须掌握的一门课程。在美国等西方国家，四五年级的学生都能够熟练地使用电脑、网络、查阅资料，学习知识。相比之下，让我们中学生告别网络的做法，我想在座的同学没有一个会同意的。但我们法律为什么又规定“禁止未成年人进入网吧”？首先，我们不能把“网络”等同于“网吧”。不接触网吧，我们同样可以接触网络。比如通过：家庭个人电脑、学校电脑室等，当然，这方面的开发和利用，有待学校、老师、家长和我们同学的共同认识基础上逐步实现。第二，表面禁止的同时，深含对我们未成年人的身心保护。我们中学生迷恋网吧，不能自拔，导致学业无成，甚至是猝死网吧的现象时有耳闻，一旦事件发生，我们总认为这是因无知而犯下的错，但也为时已晚。“禁止未成年人进入网吧”是为了让我们的学生少犯或不犯同样的错误，让更多的人来关爱我们这些未成人。 但是，目前那些黑网吧，像一个个美丽的陷阱，使许多同学丢魂失魄、丧失意志、无心学习、前途废弃。据调查，学生上网80%以上是打游戏，15%左右是交友聊天，真正查询资料用于学习的为数极少。有13%以上的男生很喜欢上网或迷恋上网，达到了严重影响学习的地步。一到寒暑假，学生们更是肆无忌惮，不

文明上网演讲稿

理性面对网络诱惑 尊敬的领导、老师，亲爱的同学们： 大家早上好！ 我是来自八（6）班的贺靖，今天，演讲的题目是：理性面对网络诱惑。 非洲原野上有一种花，它色彩斑斓，芳香扑鼻，路过的飞虫往往经不起这 种诱惑，扑上去贪婪地吮吸，并不知道这它分沁的是一种毒液。很快，花的黏 液将贪吃的飞虫牢牢地粘住，花瓣悄悄的合拢，把这些小虫都关闭起来，变成 了腹中之物。学习、生活中，我们每个人都经历着大大小小的诱惑。迷茫时， 网络中虚拟世界的五光十色是一种诱惑；疲惫时，网络游戏的惊险刺激也是一 种诱惑。 互联网拓宽了我们的求知途径，为我们打开了认识世界的一扇窗，更为我 们创造了一个展现自我个性的空间，当然也存在着传播不健康信息、提供不文 明服务等严重危害社会的问题，对学生们身心健康的危害尤其严重。有调查显示，11.4%的学生有长时间玩电子游戏的习惯。随着学习阶段的上升，长时间玩电子游戏的比例呈上升趋势。网络既能造就人才，也能毒害人类，有的同学因 沉迷网络，导致学习成绩一落千丈；更有同学因长时间玩游戏，视力严重下降；还有同学受不良信息的影响，走上犯罪的道路。 每次放假，父母在家里备好了饭菜，对我们的归来翘首以盼。有些同学却 三五成群地在网吧污浊的空气里挥汗如雨，忘乎所以。虚拟世界在一步步隔断 我们的亲情，减退我们的交流能力。更严重的是，许多同学沉迷于网络游戏， 放假的时间全泡在网络游戏里，荒废了学业，浪费了青春。一旦上瘾，甚至会 不顾一切翻墙出校，最后被学校开除，留下终身的遗憾与忏悔。 所有这些情景，令人心痛令人警醒。我们中小学生是祖国的未来，家庭的 希望，是最具科技意识和创新能力的一代。如果在上网时，我们不抵制不良信息、不理性面对网络游戏、不能够做到文明上网和健康上网，那网络不仅网住 了我们的心，更会毁掉我们的一生！ 为此，为践行“文明上网，理性上网，拒绝网络诱惑”的行动，我提倡： 1、利用自动关机软件或者闹钟来提示和控制自己的上网时间。让外力成为一副让人清醒的良药。 2、详细记录下自己每次上网所做的事，看看哪些是有益的，哪些是在浪费时间，从中改变自己的不良习惯，形成不看无用信息，抵制网络游戏的习惯。 3、在学校或在家里上网时，自觉遵守网络文明，上网时不使用侮辱、谩骂性的语言聊天。不轻易和网友约会，自觉遵守互联网规范，自觉抵制不良网络 信息的侵害。 4、我们上网要做到“三上”和“三不上”。“三上”即不进营业性网吧，不进色情、垃圾网站，不沉迷网络游戏。“三上”是健康上网，把网络作为获 取知识的园地；文明上网，听听网络专家的视频课堂，拓宽我们的求知途径， 正确利用这个展现自我个性的空间；绿色上网，熟悉上网的安全通道。 同学们，如果你能正确使用网络，可以学到比课本更多的知识；如果正确 使用网络，你会获得一个更广阔的发展空间。同学们，请让我们告别网络不良 行为，学会辨明是非，提高自身控制力，理性面对网络诱惑，让人类的科技发 明助力我们成长，让我们的学习更上一层楼！ 我的发言完毕，谢谢大家！

利用packet tracer设计校园网 毕业设计

利用packet tracer设计校园网 专业：通信工程 学生：指导老师： 摘要 随着网络技术和计算机多媒体的不断发展与普及，很大程度的加快了校园网建设和应用的步伐，网络技术人才的需求量也在不断增加，如今在校园网中，由于网络应用的大幅度增加，网络变得越来越的拥挤造成了冲突不断产生，网络受到的威胁性也随之增大，管理难度日益加增大，学校对于校园网的要求越来越高，校园网的基本功能的配置成为了网络管理人员的一项重要任务，得到各个学校的重视。 校园网的核心设备是带有路由功能的三层交换机，同时核心设备连接着路由器和汇聚交换机，汇聚交换机下面是接入层交换机。本文利用Cisco packet tracer 软件模拟一个完整的校园网络的基本构架，详细介绍了校园网内如何利用三层交换机实现VLAN的互通。并对当今校园网的形式、情况以及不足做分析。本文给出了较详细的网络设备功能实现的配置命令，实现了具有基本功能的校园网。同时也展示了Packet Tracer 软件的高度仿真性，将Packet Tracer 引入到教学中，脱离了以前全理论知识的抽象难懂，激发学生学习的兴趣和积极性。 关键词：校园网 VLAN 三层交换技术

Using packet tracer design of campus network Major: Communication Engineering Student: Supervisor: Abstract With the continuous development of network technology and computer multimedia and popularization, greatly accelerated the pace of the construction of campus network and application. In the campus network, network application is growing, the network becomes more and more crowded, conflicts arise, the threat of the network is increasing, the management difficulty is increased, the school is more and more high requirements for the campus network, the basic function of campus network configuration has become an important task for the network management personnel, each school's attention. The core equipment of the campus net is the three layer switch with routing function, at the same time, the core equipment connected to the router and the aggregation switch, aggregation switches the access layer switch. The basic framework of this paper by using Cisco Packet Tracer software to simulate a complete campus network, introduces in detail how to use the three layer switchto achieve interoperability of VLAN campus network. And on the campus network,and do not form analysis. This paper gives the detailed implementation ofnetwork equipment configuration commands, has realized the basic function ofcampus network. At the same time also showed a high level simulation of PacketTracer software, Packet Tracer will be introduced into the teaching, from the previous full theoretical knowledge to understand the abstract, arouse students' learning interest and enthusiasm. Key words：Campus network VLAN The three layer switching technology

网络文明演讲稿范文

网络文明演讲稿范文 尊敬的老师，同学们： 大家好! 我将要给大家演讲《网络——把锋利的双刃剑》。 众所周知，随着时代的时步，科技的更新，更多的新事物涌入我们的世界，其中也包括了互联网。自90年代后期互联网进入我国以来，网络就如同雨后春笋般，发展迅速，如今已遍布城乡。作为一种新的传媒手段，互联网在经济建设和人们的各项活动中起着不可忽视的作用，同时也为青春少年的学习.交流以及娱乐提供了更为广阔的天地。但是对于为成年人来说，网络则是一把锋利的“双刃剑”，一朵带刺的玫瑰，很多青少年正是因为过度沉迷于网络，沉迷于网吧，缺乏自我保护意识和对危险的预防能力，对其学业.健康和思想都造成了巨大的危害。 由此可见，网络可以造就人才，也可以把人才引向无底深渊，以下是我总结的三点上网的危害： 一.网络对青少年的人生观.世界观和价值观形成的构成潜在威胁。

二.网络使许多青少年沉溺于虚拟的世界脱离现实，也使一些青少年荒废学业。很多人把游戏称作“电子海络因”是不无道理的。 三.互联网中的不良信息和网络犯罪对青少年的身心健康和安全构成危害和威胁。 许多学生因此耽误了自己的学习.废弃了美好的前程。据一项调查显示，上网学生超过80%是在打游戏， 15%左右是在交友聊天，真正查询资料用于学习的人数少之又少。 沉迷于网络的学生，别说是上课学习，就连吃饭睡觉也能省就省，一天到晚迷迷糊糊，欲罢不能，可换来的是什么呢?在此，我呼吁那些还于网络游戏，沉侵在虚幻甜蜜中的同学，当你替身意意正酣的时候，你可曾想到焦急的父母在正在等待你回家的消息?当你陶醉于刀光剑影的时候，你可曾想辛苦的老师为你耕耘的身影?当你于新识的网友高谈阔论的'时候，你可曾想到同学间真挚的话语才是真正的友情?当你沉溺在那些略带灰色的网络文学中的时候，你可曾想到现实世界正在渐渐离你远去?同学们，快从虚幻的世界解脱出来吧，一样最宝贵的东西——时间，正从你点击鼠标，轻巧键盘的手指尖悄悄溜走，无法挽回。同学们，一个人的青春能有几年?莫待白了少年

基于cisco packet tracer企业网络规划与设计

泰州师范高等专科学校 TaiZhou Teachers College 10 届毕业设计（论文） 课题：基于cisco packet tracer 企业网络规划与设计 学生姓名：佟浩 学号： 10533002 系部：数理信息学院 专业：计算机应用企业信息管理 指导教师：朱晔

目录 摘要-----------------------------------------------------------------------------------------------------------------3 关键词--------------------------------------------------------------------------------------------------------------3 绪论-----------------------------------------------------------------------------------------------------------------4 课题背景-----------------------------------------------------------------------------------------------------------4 cisco packet tracer（思科软件）-------------------------------------------------------------------------4相关理论知识------------------------------------------------------------------------------------------------------5企业网的规划与设计---------------------------------------------------------------------------------------------5核心三层交换机(Core SW)配置VLAN-------------------------------------------------------------------------5启用DHCP服务 路由器(Router)实现NAT(网络地址转换)和端口映射功能---------------------------------------------6配置静态路由,使路由器和三层交换机互相连同----------------------------------------------------------6配置应用服务器---------------------------------------------------------------------------------------------------6实现无线上网------------------------------------------------------------------------------------------------------6配置实现------------------------------------------------------------------------------------------------------------6 Core SW交换机实现基于端口的VLAN划分------------------------------------------------------------------6配置Core SW交换机各VLAN接口IP地址并启用路由功能------------------------------------------------7配置路由器Router端口地址----------------------------------------------------------------------------------7静态路由配置------------------------------------------------------------------------------------------------------7配置Router启用NAT及端口映射-----------------------------------------------------------------------------8 结束语---------------------------------------------------------------------------------------------------------------8