Video On Interest.


128 views
Uploaded on:
Description
Video On Interest. Video on Interest. One video server Numerous video information Numerous customers Customer need to watch whenever. Presumptions. Consistent bitrate stream Flawless system transport. Unicast Arrangement. One channel for every customer No start-up inertness No customer cushion Low customer data transfer capacity
Transcripts
Slide 1

Video On Demand

Slide 2

Video on Demand One video server Many video information Many customers Client need to watch whenever NUS.SOC.CS5248-2010 Roger Zimmermann (situated to some degree on slides by Ooi Wei Tsang)

Slide 3

Assumptions Constant bitrate stream Perfect system transport NUS.SOC.CS5248-2010 Roger Zimmermann (situated to a limited extent on slides by Ooi Wei Tsang)

Slide 4

Unicast Solution One channel for each customer No start-up inertness No customer support Low customer transfer speed Large server transmission capacity Not adaptable NUS.SOC.CS5248-2010 Roger Zimmermann (situated partially on slides by Ooi Wei Tsang)

Slide 5

Multicast Solution Batching total customer demands serve utilizing multicast customers need to hold up No customer cradle Low customer transfer speed "Booking Policies for an On-Demand Video Server with Batching" Dan, Sitaram, Shahabuddin, IBM NUS.SOC.CS5248-2010 Roger Zimmermann (situated to some degree on slides by Ooi Wei Tsang)

Slide 6

Multicast Solution User-focused methodology Scheduling information in view of client solicitations Data-focused methodology Don\'t think about client Just communicate well known video NUS.SOC.CS5248-2010 Roger Zimmermann (situated to a limited extent on slides by Ooi Wei Tsang)

Slide 7

Multicast Solution Batching Staggered Broadcast NUS.SOC.CS5248-2010 Roger Zimmermann (situated to a limited extent on slides by Ooi Wei Tsang)

Slide 8

Staggered Broadcast Video C0 C1 C2 : NUS.SOC.CS5248-2010 Roger Zimmermann (situated to some degree on slides by Ooi Wei Tsang)

Slide 9

Staggered Broadcast 2 hour video 5 minutes holding up time Number of channels = 2 x 60/5 = 24 Required data transmission = 1.5Mbps x 24 = 36Mbps NUS.SOC.CS5248-2010 Roger Zimmermann (situated to some extent on slides by Ooi Wei Tsang)

Slide 10

Multicast Solution Batching Staggered Broadcast customers need to hold up No customer cushion Low customer data transmission Huge server transfer speed NUS.SOC.CS5248-2010 Roger Zimmermann (situated to some degree on slides by Ooi Wei Tsang)

Slide 11

Multicast Solution Batching Staggered Broadcast Periodic Broadcast NUS.SOC.CS5248-2010 Roger Zimmermann (situated to some extent on slides by Ooi Wei Tsang)

Slide 12

Periodic Broadcast Video C0 C1 C2 : NUS.SOC.CS5248-2010 Roger Zimmermann (situated to a limited extent on slides by Ooi Wei Tsang)

Slide 13

Pyramid Broadcast Video C0 C1 C2 : NUS.SOC.CS5248-2010 Roger Zimmermann (situated to some extent on slides by Ooi Wei Tsang)

Slide 14

Pyramid Broadcast Video C0 C1 C2 : NUS.SOC.CS5248-2010 Roger Zimmermann (situated to some extent on slides by Ooi Wei Tsang)

Slide 15

Analysis of Pyramid Broadcast Notations B : Total accessible transmission capacity B v : Bandwidth of video T v : Total length of every video K : Number of portions per video T i : Length of fragment i  : Factor in geometric arrangement NUS.SOC.CS5248-2010 Roger Zimmermann (situated to some extent on slides by Ooi Wei Tsang)

Slide 16

Channel Bandwidth playback time = T i i+1 D\'oh! Simply miss it! download time = T i+1 B v/B i Download time for section i+1 should be littler than T i for it to touch base in time. NUS.SOC.CS5248-2010 Roger Zimmermann (situated to some degree on slides by Ooi Wei Tsang)

Slide 17

Channel Bandwidth NUS.SOC.CS5248-2010 Roger Zimmermann (situated to some degree on slides by Ooi Wei Tsang)

Slide 18

 = 2 NUS.SOC.CS5248-2010 Roger Zimmermann (situated to some degree on slides by Ooi Wei Tsang)

Slide 19

Start-up Latency Worst case holding up time = NUS.SOC.CS5248-2010 Roger Zimmermann (situated to a limited extent on slides by Ooi Wei Tsang)

Slide 20

Optimal  T 1  2.5 NUS.SOC.CS5248-2010 Roger Zimmermann (situated to some degree on slides by Ooi Wei Tsang)

Slide 21

Storage Requirements NUS.SOC.CS5248-2010 Roger Zimmermann (situated to some extent on slides by Ooi Wei Tsang)

Slide 22

Pyramid Broadcast Large customer transfer speed ( KB v ) Huge customer cushion (70–80% T v ) NUS.SOC.CS5248-2010 Roger Zimmermann (situated to some degree on slides by Ooi Wei Tsang)

Slide 23

Permutation-based Pyramid Broadcast C0 C1 C2 NUS.SOC.CS5248-2010 Roger Zimmermann (situated to some degree on slides by Ooi Wei Tsang)

Slide 24

Channel Bandwidth playback time = T i i+1 D\'oh! Simply miss it! download time = T i+1 B v/B i X should be littler than T i for section i+1 to touch base in time. NUS.SOC.CS5248-2010 Roger Zimmermann (situated to some extent on slides by Ooi Wei Tsang)

Slide 25

Channel Bandwidth NUS.SOC.CS5248-2010 Roger Zimmermann (situated to some degree on slides by Ooi Wei Tsang)

Slide 26

Client Latency NUS.SOC.CS5248-2010 Roger Zimmermann (situated to a limited extent on slides by Ooi Wei Tsang)

Slide 27

Storage Requirement One channel at once Can respite and hold up NUS.SOC.CS5248-2010 Roger Zimmermann (situated to a limited extent on slides by Ooi Wei Tsang)

Slide 28

Storage Requirement stop k-1 k-1 k resume k Within time X, better not expend all information in support. X NUS.SOC.CS5248-2010 Roger Zimmermann (situated to some extent on slides by Ooi Wei Tsang)

Slide 29

Storage Requirement delay k-1 k-1 k resume Within time X, better not devour all information in cushion. X NUS.SOC.CS5248-2010 Roger Zimmermann (situated to some extent on slides by Ooi Wei Tsang)

Slide 30

Storage Requirement NUS.SOC.CS5248-2010 Roger Zimmermann (situated to a limited extent on slides by Ooi Wei Tsang)

Slide 31

Comparisons Carter, Long and Paris "Video on Demand Broadcasting Protocols" NUS.SOC.CS5248-2010 Roger Zimmermann (situated to some degree on slides by Ooi Wei Tsang)

Slide 32

Pyramid Broadcasting NUS.SOC.CS5248-2010 Roger Zimmermann (situated to some extent on slides by Ooi Wei Tsang)

Slide 33

Skyscraper Broadcasting Observations: stockpiling necessity is influenced by size of the biggest piece So, we should constrain the measure of the biggest lump! NUS.SOC.CS5248-2010 Roger Zimmermann (situated to a limited extent on slides by Ooi Wei Tsang)

Slide 34

Pyramid Skyscraper NUS.SOC.CS5248-2010 Roger Zimmermann (situated partially on slides by Ooi Wei Tsang)

Slide 35

Skyscraper Broadcasting Uses arrangement 1 2 5 12 25 52 … W NUS.SOC.CS5248-2010 Roger Zimmermann (situated to some degree on slides by Ooi Wei Tsang)

Slide 36

Skyscraper Example NUS.SOC.CS5248-2010 Roger Zimmermann (situated to some extent on slides by Ooi Wei Tsang)

Slide 37

Skyscraper Example NUS.SOC.CS5248-2010 Roger Zimmermann (situated to some extent on slides by Ooi Wei Tsang)

Slide 38

Comparisons Carter, Long and Paris "Video on Demand Broadcasting Protocols" NUS.SOC.CS5248-2010 Roger Zimmermann (situated to some extent on slides by Ooi Wei Tsang)

Slide 39

Other plans Pagoda Broadcasting 1 3 5 15 25 75 125 … Harmonic Broadcasting Equal fragment size, changes data transfer capacity! NUS.SOC.CS5248-2010 Roger Zimmermann (situated to a limited extent on slides by Ooi Wei Tsang)

Slide 40

Multicast Solution Batching Staggered Broadcast Periodic Broadcast Sending rate ≥ playback rate May require numerous channels Need extra customer cushion Need to hold up NUS.SOC.CS5248-2010 Roger Zimmermann (situated to some degree on slides by Ooi Wei Tsang)

Slide 41

Multicast Solution Batching Staggered Broadcast Periodic Broadcast Patching NUS.SOC.CS5248-2010 Roger Zimmermann (situated to some degree on slides by Ooi Wei Tsang)

Slide 42

Patching Time mcast unicast Client Request NUS.SOC.CS5248-2010 Roger Zimmermann (situated to a limited extent on slides by Ooi Wei Tsang)

Slide 43

Patching Window: W Time mcast Client Request NUS.SOC.CS5248-2010 Roger Zimmermann (situated to some extent on slides by Ooi Wei Tsang)

Slide 44

Grace Patching if W < B client support video[ W .. end] 30 minutes video 1 customer entry for each moment Total information conveyed = NUS.SOC.CS5248-2010 Roger Zimmermann (situated to some degree on slides by Ooi Wei Tsang)

Slide 45

Scenario 1: B = 15mins 30 minutes video 1 customer landing for each moment Total information conveyed = NUS.SOC.CS5248-2010 Roger Zimmermann (situated to a limited extent on slides by Ooi Wei Tsang)

Slide 46

Scenario 2: B = 5mins 30 minutes video 1 customer landing for each moment Total information conveyed = NUS.SOC.CS5248-2010 Roger Zimmermann (situated partially on slides by Ooi Wei Tsang)

Slide 47

Scenario 3: B = 2mins 30 minutes video 1 customer entry for every moment Total information conveyed = NUS.SOC.CS5248-2010 Roger Zimmermann (situated to a limited extent on slides by Ooi Wei Tsang)

Slide 48

Summary Batching (User Centered) Staggered Broadcast (Data Centered) Periodic Broadcast (Data Centered) Patching (True VOD) NUS.SOC.CS5248-2010 Roger Zimmermann (situated to some degree on slides by Ooi Wei Tsang)

Recommended
View more...