Lesson 11 (Video) Flashcards
What are some basic Video & Audio properties?
Video Properties: -High bit rate (100kbps to over 3Mbps) -Can be compressed Audio Properties: -Lower bit rate than video, but glitches in audio are generally more noticeable than with video -Can also be compressed
What are the 3 major categories of Multimedia Applications?
1) Streaming stored audio/video
2) Conversational voice and video over IP
3) Streaming live audio and video
How does VoIP work?
Encoding - Analog audio get encoded into a digital format. Uses “quantization” which rounds each sample’s value to a discrete number within a particular range.
Signaling - Takes care of how calls are set up and torn down. Ex: SIP (Session Initiation Protocol).
QoS (Quality of Service) metrics - Used to measure the quality of service.
What are the 3 major QoS metrics used in VoIP?
End-to-End Delay: Total delay “from mouth to ear”. Includes delay from:
- the time it takes to encode the audio
- the time it takes to put it in packets
- all the normal sources of network delay that network traffic encounters such as queuing delays
- “playback delay”, from the receiver’s playback buffer
- decoding delay, time it takes to reconstruct the signal
Jitter: When packets have varying delays. Problematic for VoIP because it interferes with reconstructing the analog voice stream.
Packet Loss: Ratio of packet loss. To minimize, VoIP uses UDP. There are 3 major methods to mitigate:
1) FEC (Forward Error Correction) - Sending redundant data along with main transmission.
2) Interleaving - Mixes chunks of audio together so that if one set of chunks is lost, the lost chunks aren’t consecutive.
3) Error concealment - “guessing” what the lost audio packet might be.
How does Video Streaming work?
1) Video content is first created
- Raw recorded content is typically at a high quality
2) Content is encoded
- Compressed using an encoding algorithm
3) Encoded content is secured using DRM and hosted on a server.
- Via CDN
4) End-users download the video content over the Internet
- Downloaded video is decoded and rendered on the user’s screen.
What is the most common type of Video compression?
Two types of compression:
1) Lossy: we can not recover the high quality video
- Most commonly used cause it gives higher bandwidth savings
2) Loss-less: the original video can be recovered
What are the two ways of achieving efficient Video compression?
1) within an image - pixels that are nearby in a picture tend to be similar (aka spacial redundancy)
Ex: The Joint Photographics Experts Group (JPEG) standard. The main idea behind JPEG is how to represent an image using components that can be further compressed.
2) across images - in a continuous scene, consecutive pictures are similar (aka temporal redundancy)
-I-frames and P-frames.
What are the 2 ways to leverage modifications of compression algorithms?
1) Output size of the video is fixed over time
- Also known as constant bitrate encoding (CBR)
2) Output size remains the same on an average but varies based on the underlying scene complexity
- Also known as variable bitrate encoding (VBR)
- Image quality is better with VBR but is more computationally expensive
- Typically for live streaming, where the video compression has to be done real-time, specialized/expensive hardware encoders are used by content providers.
Do content providers use UDP or TCP for video streaming?
Decoding falls apart if data is lost. For this reason, content providers use TCP for it’s reliability.
What application layer protocol is used for video delivery?
HTTP protocol
- The server is stateless and the intelligence to download the video will be stored at the client.
- Allows content providers to use CDN infrastructure
- Made bypassing middleboxes and firewalls easier as they already understood HTTP.
What is the Video streaming strategy for dealing with network and user device diversity?
- To address performance degradation caused by competing bandwidth, content providers encode their video at multiple bitrates chosen from a set of pre-defined bitrates.
- The video is chunked into segments which are usually of equal duration and each segment is then encoded at multiple bitrates and stored on a server.
- When the client requests a segment, it also specifies its quality. If there is a bandwidth degradation, the client requests a lower bitrate option.
- > This is known as bitrate adaptation.
- At the beginning of every video session, the client first downloads a “manifest” file over HTTP which contains all the metadata info about the video content and the associated URLs.
What is DASH?
When the client dynamically adjusts the video bitrate based on the network conditions and device type, this is known as Dynamic Streaming over HTTP or DASH where dynamic streaming signifies the dynamic bitrate adaptation.
What are the different signals that serve as input to a bitrate adaptation algorithm?
1) network throughput
- Bitrate adaptation using this signal is also known as rate-based adaptation.
2) video buffer
- Bitrate adaptation based on the video buffer is known as buffer-based adaptation.
-In practice, players do end up using both the network throughput and the video buffer together for bitrate adaptation.
What are the limitations of throughput-based bitrate adaptation?
- To have stall-free streaming, the buffer-filling rate should be > the buffer-depletion rate.
- It’s hard to know future bandwidth so previous chunk throughput is used to decide the bitrate of the next chunk.
What are the steps of a simple rate-based adaptation algorithm?
1) Estimation - Estimating future bandwidth by looking at the throughput of the last few downloaded chunks using a smoothing filter (eg moving average)
2) Quantization - The continuous throughput is mapped to a discrete bitrate. The maximum bitrate is selected that is less than the estimate of the throughput.
