CS 529 Multimedia Networking

Review

Digital Audio

• What is the tradeoff between sample size and fidelity? Between sample rate and fidelity?

• What uses of computer audio require a high sample rate and large sample size? Which do not? Why?

a

Digital Audio

• What are the sample rate and sample sizes for typical voice-quality audio?

• If typical encoding captures a wave form “as it is”, what does MP3 attempt to do? And how?

a

[RS75]

• What is energy?• What is the zero crossing rate?

• More importantly– how are they used?– why are they used?

[RS75]

• What is IMN?– How computed?

• What is IMX?– How computed?

Digital Video

• What are typical frame rates for videos?• What are some common resolutions for

videos (e.g., HD)?• How many bits are needed for “true color”

images?

QoS and Multimedia Applications

• Rank sensitivity (low, medium, high) for

Conversational Voice and DelayStreaming Video and DelayConversational Voice and CapacityStreaming Video and Capacity

Transport Protocols

• One reason TCP is “good” for multimedia applications? Example?

• One reason TCP is “bad” for multimedia applications? Example?

• One reason UDP is “good” for multimedia applications? Example?

• One reason UDP is “bad” for multimedia applications? Example?

Buffering

• What is a playout buffer?

• What is it used for?

• How big is it?

Skype

• What is the network architecture used for Skype?– What kinds of nodes are there?– How are the nodes connected?– Why are they connected this way?

HTTP Streaming

• How does HTTP work for streaming video?

• Where is the “intelligence” in an HTTP streaming system?

Multimedia Quality Assessment

• Broadly, what are two types of methods to assess multimedia performance?

• What is an example of each method for assessing voice quality?

Experiments in CS

• What are the main steps in the scientific process?

• Provide a Hypothesis for something in Multimedia Networking– How would you test it?

NetFlix

• Why isn’t frame loss a component of video quality for NetFlix?

• What are the components of NetFlix video quality?

RTP

• What is it?• Can I use it instead of TCP/UDP?• What are some of RTP’s major features?

SIP

• What is it? • What services does it provide?

– Registrar?– Proxy?

• How does it allow for a “mobile” caller?

Network Support for Multimedia?

• What are alternatives to “best effort”?• What is a challenge to “guaranteed” QoS?

Video Compression

• How is it similar to image compression?• How is it different?

• MPEG– What is a GoP?

[WS00]

• What is at least one implicit hypothesis the researchers test?

• How are subjective audio quality ratings provided?

• How are objective audio quality “costs” gathered?

[WS00]

• What is at least one implicit hypothesis the researchers test?– What are some of the independent variables?

• What tool do the researchers use to test audio?

[WS00]

• Provide order, best to worstQuiet – one recorded low volume, other normBad mic – one had low quality mic, other norm5% loss – both voices, with repetitionLoud – one recorded high volume, other norm20% loss – both voices, with repetitionEcho – one had open mic, no headsetReference – non-degraded

[HHH+12]

• Why is commercial streaming over HTTP confused?

• Why is commercial streaming over HTTP timid?

[RBV+12]

• Which is best?

• What else should be considered?

[LG14]

• For video messaging on smartphones, what are two main factors affect transmission speed?

• What are some suggested improvements by the authors?

[LG14]

• For video messaging on smartphones, what are two main factors affect transmission speed?

Answer: file size and location

• What are some suggested improvements by the authors?

Answer: Compress/transcode and distribute images among servers in cloud

[SKH02]

• What are the 3 fundamental limitations faced by MCGs?

[SKH02]

• What is the tension/tradeoff between responsiveness and consistency in MCGs?

• What is interest management? Describe how it works

• What is dead reckoning? Describe how it works

[SJNC13]

• What are the benefits of using cloud gaming instead of traditional gaming?

• What are the challenges?

[CFGS14]

What are the main messages from the graph based on [CFGS14]?

[CFGS14]

• Order these from least downstream bitrate to most:– Thin game– Live video– Pre-recorded video– Traditional game– Virtual environment

• What are the approximate bitrates for each?

[CFGS14]

Turbulence Summary

[SJ95]

• What is the E-Policy?• What is the I-Policy?

• When, in general, is the E-Policy good?• When, in general, is the I-Policy good?

• What is a talkspurt?

[SJ95]

• How does QM-120 work?

• What is QM-(600,2)? What is the motivation behind it?

• Describe one item of future work.– Why is it of interest?– Briefly, how would you approach doing it?

[WCCL09]

• What are the challenges proposed by the paper?– Hint: it considers buffer sizes

• What have the academics done about the challenges?

[WCCL09]

• Briefly describe the methodology the author uses to determine the “optimal” buffer size

[WCCL09]

• Briefly describe the methodology the author uses to determine the “optimal” buffer sizeAns: 1. Encode audio samples2. Simulate loss3. Add delay, dropping if needed4. Decode 5. Apply QoE measure – PESQ + E-model6. Repeat 1-5 for all buffer sizes, with highest QoE

being optimal

[WCCL09]

• Interpret the above graph• What are the main messages?

[PHH98]

• What are the client loss characteristics of the Mbone (and other P2P networks)?

• Why can TCP and retransmissions be used to repair loss in the Mbone?

[PHH98]

• What is forward error correction (FEC)?

• What are the two categories of FEC described by the authors? What are the main advantages and disadvantages of each?

[PHH98]

• Broadly, what are receiver based repair techniques? Why are they often called error concealment?

• What is a recommended repair technique for a non-interactive, multicast multimedia session? Why?

[HCH09]

2%

0%1%

0%

3%4%

5%6%

7%8%

9%10%

Significance of picture for Skype?

[HCH09]

• What’s is the optimal redundancy ratio for repair as defined by the authors?

[HCH09]

• What’s is the optimal redundancy ratio for repair as defined by the authors?Ans: Minimum amount of redundancy data needed to sustain same audio quality under different network conditions

[HCH09]

How did the author’s generate this graph? How to read it?

[HKBT01]

• What “alternate” service does ABE offer?– What is green traffic?– What is blue traffic?

[HKBT01]

• What does it mean to say “Green does not hurt Blue”?

• Based on below formula, what happens if green “cuts” in line? How do you fix it?

[HKBT01]

Using DSD, what is the order of the next 3 packets?

[HKBT01]

What is the significanceof this graph?

[LCK10]

• What general property about network bitrates and latency sensitivity does CHAP rely upon?

• What are credits as assigned by CHAP? How are they assigned? How are they “spent”?

[LCK10]

• Interpret the above graph• What are the messages for CHAP?

0 100 200 300 400 500 6000.0

0.5

1.0

FIFO (only qstat)FIFO (qstat + 1 HTTP downloads)

Server Ping (ms)

[MSM04]

• How does the above quote relate to motivation for the proposed work?

• Describe the authors’ variant on the “method of limits” to determine video quality

Recent IBM QoS policy says:“The priority for smooth video is higher than the priority for frame quality”

[MSM04]

• What was difference between acceptability of video for Palmtops versus Desktops?

[CC10]

• What are three primary ways latency affects players in network games?

• How much does latency affect players in network games?