Network degradation effects on different codec types and characteristics of video streaming

DIGITAL IMAGE PROCESSING AND COMPUTER GRAPHICS VOLUME: 12 |NUMBER: 4 |2014 |SPECIAL ISSUE Ne wo k Deg ada ion E ec s on Di e en Codec Types and Cha ac e is ics o Video S eaming Ja osla FRNDA1, Lukas SEVCIK1, Mi osla UHRINA2, Mi osla VOZNAK1 1Depa men o Telecommunica ions, Facul y o Elec ical Enginee ing and Compu e Science, VSB–Technical Uni e si y o Os a a, 17. lis opadu 15, 708 00 Os a a-Po uba, Czech Republic 2Depa men o Telecommunica ions and Mul imedia, Facul y o Elec ical Enginee ing, Uni e si y o Zilina, Uni e zi na 8215/1, 010 07, Zilina, Slo akia ja osla[email p o ec ed], lukas.se [email p o ec ed], mi osla[email p o ec ed], mi osla . [email p o ec ed] Abs ac . Nowadays, he e is a quickly g owing de- mand o he ansmission o oice, ideo and da a o e an IP based ne wo k. Mul imedia, whe he we a e alking abou b oadcas , audio and ideo ansmis- sion and o he s, om a global pe spec i e is g owing exponen ially wi h ime. Wi h incoming eques s om use s, new echnologies o da a ans e a e con in- ually de eloping. Da a mus be deli e ed eliably and wi h he ewes losses a such high speed. Video qual- i y as pa o mul imedia echnology has a e y impo - an ole nowadays. I is in luenced by se e al ac o s, whe e each o hem can ha e many o ms and p ocess- ing. Ne wo k pe o mance is he majo deg ada ion e - ec ha in luences he quali y o esul ing image. Poo ne wo k pe o mance (lack o link capaci y, high ne - wo k load. . . ) causes da a packe losses o di e en deli e y ime o each packe . This wo k ocuses ex- ac ly on hese ne wo k phenomena. I examines he impac o di e en delays and packe losses on he qual- i y pa ame e s o iple play se ices, o e alua e he esul s using objec i e me hods. The aim o his wo k is o b ing a de ailed iew on he pe o mance o ideo s eaming o e IP-based ne wo ks. Keywo ds Delay a ia ion, packe loss, PSNR, SSIM, ideo quali y assessmen . 1. In oduc ion The g ow h o he In e ne ne wo k is using mo e and mo e esou ces o pe o mance analysis. This is simu- la ed using di e en models. This wo k compa es he pe o mance o he ne wo k om an expe imen al iew- poin . The objec i e o his wo k is o analyse a ious impac s on ansmi ed ideo, such as packe loss, ji - e , eo de ing. The wo k ocuses on he p esence o ideo and da a packe s in he ne wo k. I compa es he impac o da a loss and ou o o de da a. The esul s show whe he i is be e o ge he packe s in a di e en o de o comple ely los . I compa es s a ic and dynamic ideo; a ying quali y o he ansmi - ed ideo. We compa e he impac o he size o he ansmi ed da a. We deal wi h objec i e me hods o he e alua ion o he quali y o ideos in he wo ks. The e a e many a ibu es o he ideo image. These can be compa ed; he e o e, o measu e he exploi s o se e al o he mos well-known me hods o he e alua ion o image qual- i y. Each me hod has di e en p ocedu es and di e en me ic e alua ion sys em. We apply packe loss o e p ede e mined s eps on s eam. The indi idual objec- i e me hods will e alua e he cap u ed s eams. The aim o he pape is o e alua e he impac o loss du ing ansmission using di e en comp ession echnologies om se e al di e en pe spec i es. 2. S a e o he A The ecen ly g owing in e es in eal- ime se ice (such as audio and ideo) ans e h ough packe ne wo ks based on IP p o ocol has led o analyses o hese se - ices and hei beha io in such ne wo ks becoming mo e in ensi e. Logically, he g ea es emphasis is be- ing pu on he ans e o oice, since his se ice is he mos sensi i e o he o e all ne wo k s a us. Bu on he o he hand, ideo has become he majo i y pa o all da a a ic sen ia IP ne wo ks. In gene al, a ideo se ice is one-way se ice (excep e.g. ideo calls) so ne wo k delay is no such an impo an ac- o as in oice se ice. Dominan ne wo k ac o s ha in luence he inal ideo quali y a e especially packe loss, delay a ia ion and he capaci y o he ansmis- c 2014 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 377 DIGITAL IMAGE PROCESSING AND COMPUTER GRAPHICS VOLUME: 12 |NUMBER: 4 |2014 |SPECIAL ISSUE sion links [1]. Analysis o ideo quali y concen a es on he esis ance o ideo codecs o packe loss in he ne wo k, which causes a e ac s in he ideo [2], [3]. On he o he hand, a ew ac o s s ill lack, such as a complex iew o ideo pa ame e s on he inal ideo quali y. In ou p e ious wo ks [4], [7], we ocused on he quali y o he iple play se ice p edic ion model implemen a ion, whe e one pa was dedica ed o he quali y o he ideo se ice. The main mo i a ion behind his wo k is o ex end he men ioned compu a ional model and b ing a com- plex iew o all ideo pa ame e s like codec ype, cha - ac e and esolu ion, and hei in luence on nega i e ne wo k ac o s esis ance. In addi ion o packe loss, we ocused on ano he ne wo k dis up ion phenomenon called delay a ia ion (also known as ji e ). This phe- nomenon is e y o en o e looked due o de-ji e bu e implemen a ion on he ecei ing side, bu o be e p ocess o ne wo k si ua ion modeling and p edic ion, i is good o know how i in luences he inal ideo quali y. 3. Me hodology 3.1. Video P ocessing 1) Size o Digi al Image Da a The olume o digi al ideo da a is usually desc ibed in he e minology o bandwid h o ans e a e. Band- wid h o a classical digi al ideo ansmission wi hou comp ession is up o hund eds o Mbps. The amoun o da a o he pic u e signal is highe wi h an inc ease in esolu ion. The olume o da a is a majo p oblem in he ansmission, p ocessing, s o age and display o ideo in o ma ion. Digi al ideo compa ed wi h s a ic images is e y sensi i e o memo y needed sa ing [5]. S anda d ele ision b oadcas ing has a ame a e o a leas 25 ps ( ames pe second), [6]. I is su icien o he delay in pe cep ion o he human eye. E e y second o he mo ie a esolu ion 1080p (Full HD) o uncomp essed ideo can ake up o ens o megaby es o memo y. Video ypically con ains a la ge amoun o edundan da a. Those can be emo ed using he app op ia e comp ession algo i hms [5]. 2) Video T ansmission To ans e ideo iles, undamen ally un eliable p o o- cols a e used. The p inciple consis s in sending and e- cei ing da a wi hou eedback be ween he sende and he ecipien . Fac o s a ec ing he ideo ansmission a e: •La ency: This is he ime ha elapses be ween sending a message om he sou ce and adop ion o he des ina ion node. •Packe o de : Va iabili y in he packe deli e y ime o he des ina ion node causes inco ec o - de . •Packe loss: This is he a e age numbe o packe s ha a i e a he des ina ion node due o he s a e o he ne wo k. I is mos o en exp essed as a pe cen age. •Bandwid h: This exp esses he capaci y o he ansmission channel. •Delay: This is caused by o e c owding he packe queue on he ou going in e ace [5]. 3) Me hods o E alua ing he Quali y In he wo k, we used he objec i e me hods - PSNR and SSIM. Objec i e e alua ion me ic in ol es he use o he me ic’s compu a ional me hods, which o m a "sco e" o he quali y o he in es iga ed ideo. These me hods measu e he physical cha ac e is ics o he ideo signal, such as he ampli ude, iming and signal- o-noise a io. PSNR (Peak signal- o-noise a io) is he a io be- ween he maximum signal ene gy and noise ene gy. I is exp essed on a loga i hmic scale because di e - en signals ha e di e en dynamic anges. PSNR in decibels is de ined by he o mula [7]: PSNR = 10 log10 MAX2 I MSE [dB],(1) whe e MAX is he maximum alue ha he pixel can ake (e.g. 255 o 8-bi image) and MSE is he di e ence be ween wo gay-le el images o ideo se- quences. Technically, MSE e lec s he di e si y o he image, while PSNR exp esses i s iden i y. The s onges PSNR me hod is an easy and as calcula- ion, which is he eason why i is s ill used e y o en in scien i ic pape s al hough he co ela ion wi h he human pe cep ion is wo se han SSIM [6], [8]. The SSIM (S uc u al Simila i y Index) me hod in- cludes h ee componen s - he simila i y o he in en- si y, he co esponding con as and he co espond- ing s uc u e. The combina ion o hese h ee ac o s o ms one alue. This demons a es he quali y o he es ideo. This me hod di e s by e alua ing s uc- u al dis o ion and no e o a e. The main eason o his di e ence is cha ac e is ic o he human i- sual sys em. This pe cei ed dis o ion changes in he s uc u e o he ame much be e han he e o a e. Since he SSIM me hod achie es a good co ela ion o c 2014 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 378 DIGITAL IMAGE PROCESSING AND COMPUTER GRAPHICS VOLUME: 12 |NUMBER: 4 |2014 |SPECIAL ISSUE he subjec i e imp ession, a ing is de ined in he in- e al [0–1], whe e 0 ep esen s he wo s alue and 1 he bes one (iden i y), [7]. 3.2. Video Quali y E alua ion The aim o he measu emen was o simula e he e ec o packe loss and ji e o he ideo o ma s MPEG-2 and MPEG-4, o de e mine he impac on he esul - ing image using objec i e me hods o measu ing he quali y o he ideo and compa ing he esul s. We made measu emen s o one s a ic and wo dynamic ideos o 25 seconds. All he mo ies we e measu ed a a esolu ion o 720×576 (PAL), 1280×720 (HD) and 1920×1080 (Full HD). S a ic ideo was ep esen ed by TV news (slow mo ion), he i s dynamic ideo by a space shu le launch and he hi d ideo wi h he highes bi a e (60 Mbps) by an open sou ce anima ed mo ie called Big Buck Bunny. The whole p ocess o measu ing is shown in he Fig. 1 and Fig. 2. To e alu- a e he quali y we used he me hods SSIM and PSNR. SSIM co ela es be e wi h he pe cep ion o he hu- man eye [6]. We e alua ed hese me hods using MSU VQM Tools. As a i s s ep, we c ea ed a s eam in he VLC Playe . As o he ideo con en , s eaming p ocess RTP/UDP/IP me hod wi h MPEG2(TS) and H.264(MP4) was used. This b oadcas he ideo s eam on he local com- pu e in e ace. We cap u ed and sa ed he s eam o disk using ano he VLC Playe . We sa ed his ans e ideo and agged i as he o iginal ideo. Ou es ing scena ios e lec he si ua ion ha can ac ually hap- pen in he ne wo k. Especially he mobile ne wo ks capable o using IP a chi ec u e like UMTS and LTE each high alues o packe loss and delay a ia ion [10]. Fo he pu pose o se ings o ou scena ios, we used Linux ool called Ne em. Ne em p o ides Ne - wo k Emula ion unc ionali y o es ing p o ocols by emula ing he p ope ies o wide a ea ne wo ks. The cu en e sion emula es a iable delay, loss, duplica- ion and e-o de ing [9]. 1) Packe Loss We se he packe loss o 1 on he in e ace using Ne em and hen epea ed he whole measu emen . Then we epea ed his s ep o packe loss in inc emen s o 1 %, 2 %, ..., 10 %. 2) Ji e We se ha 25 % o packe s will be delayed ( esul s o ou p e ious wo k [4] showed ha app oxima ely 25 % o all a ic had di e en one-way delay). We epea ed he measu emen s o 10, 20, 30, 50, 75 and 100 ms delay a ia ion. By s eaming he ideos, we se he alue o he de-ji e bu e o 0 in VLC, so ha he delays we e eal. Fig. 1: Measu emen p ocedu e. Tab. 1: Pa ame e s o measu emen s. Used codecs MPEG-2, MPEG-4 - H.264 Video esolu ion [pixels] 720×576, 1280×720, 1920×1080 E alua ion me hods PSNR, SSIM Se ing packe loss on local in e ace: •# cqdisc add de lo oo ne em loss 1 %. Change packe loss on local in e ace: •# cqdisc change de lo oo ne em loss 2 %. This causes ha 2 % (i.e., 2 ou o 100) packe s a e andomly d opped. Videos o measu emen we e in hese o ma s, so we did no se any addi ional anscoding by c ea ing o cap u ing a s eam. Se ing packe delay on local in e ace: •# cqdisc add de e h0 oo ne em delay 10 ms eo de 75 % 50 %. In his example, 75 % o he packe s (wi h a co e- la ion o 50 %) will ge sen immedia ely, he o he s will be delayed by 10 ms. In ou case, co ela ion 50 % means ha he delayed pa o all da a a ic is oscil- la ed a ound a alue o 25 %. This se ing simula es he ne wo k pe o mance beha iou mo e exac ly. Change packe delay on local in e ace: •# cqdisc change de e h0 oo ne em delay 20ms eo de 75 % 50 %. c 2014 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 379 DIGITAL IMAGE PROCESSING AND COMPUTER GRAPHICS VOLUME: 12 |NUMBER: 4 |2014 |SPECIAL ISSUE 3) E alua ion o he Resul s By using he p og am MSU VQMT, we compa ed he o iginal sample and he es ed sample, which included damage caused by ou se ings. The p og am expo s esul s in o a CSV o ma , whe e we can ind he alue o e e y compa ed ame and he o al a e age alue o he whole ideo. Fig. 2: Compa ing s eam wi h o iginal ideo. 4. Resul s The esul s o he measu emen s e i ied ou p edic ion ha no only he ype o ideo codec has a deg ada ion impac on ideo quali y. On he o he hand, ideo esolu ion was no p o en as a signi ican pa ame e o ideo obus ness. The mos impo an ac o was shown o be he ideo code ype. Video codec H.264 (MPEG-4 Pa 10) is mo e p one o packe loss a e in he ne wo k in as- uc u e han he olde MPEG-2. Acco ding o he esul s o he s a ic ideo measu emen s, he e is no big di e ence be ween he esolu ions used. We de- ec ed a sligh dec ease in highe esolu ion. Du ing he s a ic scene, whe e changes we e e y slow, mainly he P and B ames con ained app oxima ely he same in o ma ion ega dless o he esolu ion used [8]. The i s es ed dynamic ideo achie ed wo se esul s han he s a ic ideo. Again, he di e ences be ween he used esolu ions we e small. All he GOP ames con ain mo e in o ma ion, so packe loss signi ican ly a ec s he pic u e dis o ion. Tha is he eason why dynamic ideo is gene ally mo e sensi i e o da a loss [5], [8]. The hi d ideo has a dynamic cha ac e , oo, bu a e y high bi a e compa ed o he p e ious wo ideos. Pe o mance o his ideo was e y poo . High bi a e means a lo o in o ma ion con ained in he I, B and P ames and i s loss causes signi ican deg ada ion o he ideo quali y. Fo a be e illus a ion, igu es o he numbe 3 and 4 demons a e he ideo quali y esul s o ull HD eso- lu ion. This pape ollows on om ou p e ious wo k [4] and ex ends he ideo p edic ion model ha was used he e. All hese men ioned esul we e p ocessed in o he ollowing eg essi e equa ions. Fig. 3: SSIM esul s o MPEG-2. Fig. 4: SSIM esul s o MPEG-4(H.264). 4.1. Slow-Mo ion Video 1) MPEG-2 SSIM = α(a+b·(X2)) + β(a+b·√X)+ +γ(a+b·(X2)).(2) 2) MPEG-4 SSIM = α a+b X!+β1 a+b·X+ +γ(exp(a+b·X)). (3) All he necessa y coe icien s a e p esen ed in Tab. 5. Because measu emen s we e pe o med on wo dy- namic ideos, he ollowing eg essi e equa ions ep e- sen a p edic ion model o bo h o hem. 4.2. Dynamic Video wi h O dina y and High Bi a e 1) MPEG-2 SSIM = α1 a+b·X+β(a+b·ln (X)) .(4) c 2014 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 380 DIGITAL IMAGE PROCESSING AND COMPUTER GRAPHICS VOLUME: 12 |NUMBER: 4 |2014 |SPECIAL ISSUE Tab. 2: S a ic ideo measu emen s esul s. Packe loss [%] PSNR ([dB])/SSIM 720×576 1280×720 1920×1080 MPEG-2 MPEG-4 MPEG-2 MPEG-4 MPEG-2 MPEG-4 129.20/0.914 23.84/0.805 26.25/0.942 17.87/0.742 27.45/0.905 17.63/0.774 228.73/0.876 15.05/0.688 25.84/0.894 17.36/0.730 23.54/0.885 16.94/0.776 323.320.877 13.21/0.52 22.02/0.839 14.43/0.709 20.86/0.876 15.70/0.673 422.89/0.8 13.90/0.529 20.03/0.830 15.13/0.713 21.28/0.821 13.69/0.648 523.74/0.77 12.72/0.514 14.29/0.743 12.64/0.555 16.69/0.820 14.59/0.450 620.36/0.721 11.34/0.499 14.36/0.726 11.36/0.539 14.84/0.812 11.92/0.362 717.98/0.669 12.12/0.486 13.08/0.707 10.34/0.534 15.53/0.800 12.44/0.255 815.44/0.536 10.33/0.485 12.43/0.660 9.37/0.483 15.53/0.740 10.86/0.225 915.68/0.481 9.96/0.463 13.90/0.658 8.46/0.444 15.51/0.569 11.64/0.205 10 13.40/0.452 8.55/0.362 13.08/0.638 9.77/0.421 12.46/0.481 10.33/0.184 Tab. 3: PSNR and SSIM esul s o i s es ed dynamic ideo. Packe loss [%] PSNR ([dB])/SSIM 720×576 1280×720 1920×1080 MPEG-2 MPEG-4 MPEG-2 MPEG-4 MPEG-2 MPEG-4 121.22/0.864 17.71/0.777 21.75/0.859 16.29/0.736 19.00/0.915 15.90/0.817 219.08/0.830 16.30/0.738 19.27/0.837 15.89/0.750 17.99/0.875 13.59/0.715 319.08/0.746 14.89/0.692 17.90/0.789 14.73/0.745 14.96/0.809 13.25/0.715 418.65/0.754 14.84/0.686 18.23/0.766 14.83/0.714 15.86/0.808 11.51/0.657 517.38/0.716 14.73/0.683 18.37/0.752 14.69/0.699 14.41/0.781 12.24/0.649 616.23/0.711 14.73/0.645 16.07/0.741 15.07/0.663 14.95/0.799 12.62/0.637 715.02/0.706 13.83/0.644 14.83/0.717 13.21/0.582 15.29/0.764 12.97/0.632 815.27/0.709 12.60/0.651 14.73/0.701 12.66/0.545 14.10/0.728 12.15/0.605 914.29/0.693 12.46/0.622 14.83/0.694 10.57/0.502 14.13/0.703 11.46/0.604 10 14.70/0.686 11.65/0.605 14.73/0.678 11.72/0.485 13.20/0.683 12.11/0.582 Tab. 4: PSNR and SSIM esul s o he second dynamic es ed ideo. Packe loss [%] PSNR ([dB])/SSIM 720×576 1280×720 1920×1080 MPEG-2 MPEG-4 MPEG-2 MPEG-4 MPEG-2 MPEG-4 127.0/0.923 21.32/0.695 27.63/0.927 16.10/0.530 24.72/0.806 15.56/0.589 226.51/0.9 17.72/0.551 22.79/0.765 15.75/0.511 22.41/0.737 15.16/0.501 321.59/0.714 15.36/0.446 22.56/0.745 15.30/0.422 22.0/0.694 14.42/0.487 421.26/0.718 15.28/0.447 21.43/0.7 14.92/0.421 22.01/0.699 13.8/0.474 520.73/0.705 15.33/0.441 21.10/0.664 14.04/0.428 19.84/0.614 13.37/0.446 619.97/0.668 15.21/0.405 21.10/0.597 12.05/0.393 18.72/0.589 13.02/0.458 720.07/0.618 14.28/0.338 19.15/0.566 11.46/0.372 18.22/0.564 12.17/0.429 819.14/0.575 14.41/0.333 17.90/0.531 10.91/0.353 17.59/0.558 12.09/0.425 917.97/0.562 11.08/0.303 17.65/0.528 10.91/0.352 18.28/0.552 11.09/0.387 10 17.81/0.493 10.95/0.326 17.29/0.511 10.87/0.326 18.07/0.545 10.70/0.364 Tab. 5: Coe icien s o s a ic ideo. Coe . MPEG-2 MPEG-4 (H.264) 720×576 1280×720 1920×1080 720×576 1280×720 1920×1080 a0.89957 1.08748 0.9216 0.146704 1.1027 0.08596 b−0.004924 −0.143973 −0.00389 0.528499 0.12312 −0.1839 α1 0 0 1 0 0 β0 1 0 0 1 0 γ0 0 1 0 0 1 2) MPEG-4 SSIM = α(a+b·ln (X)) + β1 a+b·X.(5) Table 6 and Tab. 7 con ain he coe icien s o hese wo equa ions. All eg essi e models desc ibed he e gained an R-squa e ac o (R2) highe han 90 %, which ep esen s a high le el o e aci y. The second g oup o measu emen s led o an analysis o he deg ada ion e ec o delay a ia ion – Ji e . The esul s o he pe o med es s unco e a c i ical bounda y o 20 ms. Abo e his alue, a signi ican educ ion o inal ideo quali y is obse ed. Due o he p ocess o decomp essing and p ocessing he ideo c 2014 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 381 DIGITAL IMAGE PROCESSING AND COMPUTER GRAPHICS VOLUME: 12 |NUMBER: 4 |2014 |SPECIAL ISSUE Tab. 6: Coe icien s o MPEG-2 dynamic ideos. Coe . Lowe bi a e dynamic ideo High bi a e dynamic ideo 720×576 1280×720 1920×1080 720×576 1280×720 1920×1080 a0.858125 0.875076 0.927705 0.9538 0.924954 0.819482 b−0.076882 −0.080159 −0.00389 0.094724 −0.179784 −0.1216 α0 0 0 1 0 0 β1 1 1 0 1 1 Tab. 7: Coe icien s o MPEG-4 (H.264) dynamic ideos. Coe . Lowe bi a e dynamic ideo High bi a e dynamic ideo 720×576 1280×720 1920×1080 720×576 1280×720 1920×1080 a0.783421 1.12248 0.800991 0.678364 1.8151 00.5818 b−0.073105 0.086301 −0.092020 −0.169198 0.122547 −0.0833 α1 0 0 0 0 0 β0 1 1 1 1 1 s eam on he end use side cos ing some ime, bo h codecs a e ole an o small delay a ia ion. Fig. 5: Resul s o delay a ia ion measu emen s, HD esolu ion. Fig. 6: Resul s o delay a ia ion measu emen s, Full HD eso- lu ion. The beha iou o dynamic ideos is app oxima ely on he same le el, wi h bigge di e ences be ween he MPEG-2 and MPEG-4 codec when he s a ic ideo was used. Typically in he eal wo ld, de-ji e bu e is used o elimina ion o his phenomenon, bu i is good o know how big a deg ada ion e ec on ideo quali y has been caused pa icula ly by he delay a ia ion. 5. Conclusion The aim o his wo k was o b ing a de ailed iew o he pe o mance o ideo s eaming o e an IP-based ne wo k. The measu ed esul s showed he ela ion be ween he ideo codec ype and bi a e o he i- nal ideo quali y. These esul s helped us o c ea e and ex end ou p e ious ma hema ical models o ideo s eaming beha iou . The second pa o he measu e- men s was dedica ed o ano he ad e se ne wo k im- pac on ideo quali y called Ji e . The esul s p o ed he impo ance o De-ji e bu e implemen a ion no only o oice se ices bu also o ideo s eaming se - ices. Ou u u e wo ks will ocus on wo di ec ions: Fi s ly, he new gene a ion o ideo codecs such as H.265 and VP9. Due o he limi a ions o ou e alua- ion MSU VQMT p og am, we a e cu en ly awai ing he o icial suppo o hese new ideo codecs. The second pa will be ocused on analysis o he impac o secu i y mechanisms, and on he enc yp ion algo- i hm implemen ed o QoS pa ame e s. Secu i y is a highly discussed opic nowadays, and p o ocols such as IPsec, VPN/SSL and SRTP a e becoming mo e and mo e equen ly used o secu e he con en o oice o ideo, so compu a ional ma hema ical models should handle his new si ua ion. Acknowledgmen This wo k was suppo ed by G an o he SGS No. SP2014/72 and was pa ially suppo ed by he Eu opean Regional De elopmen Fund in he IT4Inno a ions Cen e o Excellence p ojec (CZ.1.05/1.1.00/02.0070) and by he De elopmen o human esou ces in esea ch and he de elopmen o la es so compu ing me hods and hei applica ion in p ac ice p ojec (CZ.1.07/2.3.00/20.0072) unded by c 2014 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 382 DIGITAL IMAGE PROCESSING AND COMPUTER GRAPHICS VOLUME: 12 |NUMBER: 4 |2014 |SPECIAL ISSUE he Ope a ional P og amme Educa ion o Compe i- i eness. Re e ences [1] CHIKKERUR, S., V. SUNDARAM, M. REISSLEIN and L. J. KARAM. Objec i e Video Quali y Assessmen Me hods: A Classi- ica ion, Re iew, and Pe o mance Compa ison. IEEE T ansac ions on B oadcas ing. 2011, ol. 57, iss. 2, pp. 165–182. ISSN 0018-9316. DOI: 10.1109/TBC.2011.2104671. [2] YIM, C., A. C. BOVIK, M. REISSLEIN and L. J. KARAM. E alua ion o empo al a ia ion o ideo quali y in packe loss ne wo ks: A Classi- ica ion, Re iew, and Pe o mance Compa ison. Signal P ocessing: Image Communica ion. 2011, ol. 26, iss. 1, pp. 24–38. ISSN 0923-5965. DOI: 10.1016/j.image.2010.11.002. [3] FEAMSTER, N. and H. BALAKRISHNAN. Packe Loss Reco e y o S eaming Video. In: 12 h In e na ional Packe Video Wo kshop. Pi s- bu gh: ACM, 2002. [4] FRNDA, J., M. VOZNAK, J. ROZHON and M. MEHIC. P edic ion model o QoS o T iple play se ices. In: 2013 21s Telecommunica- ions Fo um Tel o (TELFOR). Belg ade: IEEE, 2013, pp. 733–736. ISBN 978-1-4799-1419-7. DOI: 10.1109/TELFOR.2013.6716334. [5] BOVIK A. Handbook o image and ideo p ocess- ing. 2nd ed. Bu ling on: Else ie Academic P ess, 2005. ISBN 01-211-9792-1. [6] UHRINA, M., J. HLUBIK and M. VACULIK. Co ela ion Be ween Objec i e and Subjec i e Me hods Used o Video Quali y E alua ion. Ad- ances in Elec ical and Elec onic Enginee ing. 2013, ol. 11, no. 2, pp. 135–146. ISSN 1804-3119. DOI: 10.15598/aeee. 11i2.775. [7] FRNDA, J., M. VOZNAK and L. SEVCIK. Ne - wo k Pe o mance QoS P edic ion. In: P oceed- ing o he Fi s Eu o-China Con e ence on In el- ligen Da a Analysis and Applica ions. Shenzen: IEEE, 2012, pp. 165–174. ISBN 978-3-319-07776- 5. DOI: 10.1007/978-3-319-07776-5_18. [8] UHRINA, M., J. HLUBIK and M. VACULIK. Impac o Comp ession on he Video Quali y. Ad ances in Elec ical and Elec onic Engi- nee ing. 2012, ol. 10, no. 4, pp. 251–258. ISSN 1804-3119. DOI: 10.15598/aeee. 10i4.737. [9] Ne em. Linux Founda ion [online]. 2009. A ail- able a : h p://www.linux ounda ion.o g/ collabo a e/ wo kg oups/ne wo king/ne em. [10] CHAN, M. C. and R. RAMJEE. TCP/IP pe - o mance o e 3G wi eless links wi h a e and delay a ia ion. In: P oceedings o he 8 h an- nual in e na ional con e ence on Mobile compu - ing and ne wo king - MobiCom ’02. New Yo k: ACM P ess, 2002, pp. 71–82. ISBN 1-58113-486- X. DOI: 10.1145/570645.570655. Abou Au ho s Ja osla FRNDA was bo n in 1989 in Ma in, Slo- akia. He ecei ed his M.Sc. om he VSB–Technical Uni e si y o Os a a, Depa men o Telecommu- nica ions, in 2013, and now he is con inuing his Ph.D. s udy a he same place. His esea ch in e es s include Quali y o T iple play se ices and IP ne wo ks. Lukas SEVCIK was bo n in 1989 in Cadca, Slo akia. He ecei ed his M.Sc in In o ma ics om he Facul y o Managemen Science and In o ma ics, Uni e si y o Zilina, in 2013. Now, he is con inuing his Ph.D. s udy a he Depa men o Telecommunica ions, VSB– Technical Uni e si y o Os a a. His esea ch in e es s include Quali y o T iple play se ices and IP ne wo ks. Mi osla UHRINA was bo n in 1984 in Zilina, Slo akia. He ecei ed his M.Sc and Ph.D. deg ees in Telecommunica ions om he Depa men o Telecommunica ions and Mul imedia, Uni e si y o Zilina, in 2008 and 2012, espec i ely. Nowadays, he is an assis an lec u e a he men ioned Depa men o Telecommunica ions and Mul imedia. His esea ch in e es s include audio and ideo comp ession, TV b oadcas ing (IPTV. DVB-T, DVB-H) and IP ne - wo ks. Mi osla VOZNAK is an associa e p o esso a he Depa men o Telecommunica ions, VSB– Technical Uni e si y o Os a a. He ecei ed his Ph.D. in elecommunica ions, and his disse a ion hesis was en i led “Voice a ic op imiza ion wi h ega d o speech quali y in ne wo ks wi h VoIP echnology” in 2002. Topics o his esea ch in e es s a e Nex Gene a ion Ne wo ks, IP elephony, speech quali y and ne wo k secu i y. c 2014 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 383