13 broadcast

67 %
33 %
Information about 13 broadcast
Entertainment

Published on December 23, 2007

Author: cooper

Source: authorstream.com

Making Friends with Broadcast:  Making Friends with Broadcast CMU 15-744 David Andersen Administrivia:  Administrivia Midterm Mean 66.5, Median 70, Stddev 13.7 Histo: 35-39 37 38 40-44 45-49 50-54 54 54 54 55-59 56 57 60-64 61 64 64 65-69 69 70-74 71 73 73 73 75-79 75 76 76 79 80-84 83 85-89 86 90-95 90 Correlation with PS1 scores: 0.7 This is a grad class. Expect As and Bs in the “normal” curve (stddev) If outlier, might want to talk with dga. Feedback Feedback:  Feedback Feedback #1 complaint: Post lecture notes earlier Answer: Okay! Second popularity group: Req. security topics Yes! Already planned; if suggestions, drop me a note. Security overview (problems, causes, challenges, definitions, packet floods, SYN floods, botnets, some defenses) DDoS control and traceback Worms Slides are sometimes hard to understand Will work on that. Many of them are brand new this semester Less “un-important” topics Need to clarify my emphasis. Every topic so far is important either because of practical impact, or because it’s intellectually important in terms of methods or the things that came from it, or because it illustrates open problems But very true that not everything is practical.  Thank you for the feedback! Back to Ad Hoc Networks:  Back to Ad Hoc Networks Recall that Transmissions interfere with many nodes, which constrains capacity of ad hoc nets Multiple receivers hear every transmission Delivery is probabilistic b/c of multipath interference [Roofnet sigcomm2005] Today’s papers: Past the cutting edge of what’s commonly used in wireless nets Will they be? We’ll see. 1) “Hop-over” overhearing:  1) “Hop-over” overhearing Observation 1: Best ETX/ETT path may have “overhearing”: What does p look like? If p > 0, can we take advantage of it when overhearing happens instead of having it interfere with C’s ability to talk concurrently? A B C 90% 90% 0 < p < 45% 2) Bidirectional Reception:  2) Bidirectional Reception Observation 2: When you Tx in a line, both sides can hear you: If sending from A  B  C A hearing (BC) is unwanted interference But we can turn it to our advantage A B C “packet” ExOR:  ExOR Let’s take advantage of the first observation, with an extra twist: Packets may hop over in a line Or may hop “sideways” as well Want to use the best route even if it goes off the “expected” best path Why ExOR might increase throughput (1):  Why ExOR might increase throughput (1) Best traditional route over 50% hops: 3(1/0.5) = 6 tx Throughput  1/# transmissions ExOR exploits lucky long receptions: 4 transmissions Assumes probability falls off gradually with distance src dst N1 N2 N3 N4 75% 50% N5 25% Slide Credit: Biswas & Morris Why ExOR might increase throughput (2):  Why ExOR might increase throughput (2) Traditional routing: 1/0.25 + 1 = 5 tx ExOR: 1/(1 – (1 – 0.25)4) + 1 = 2.5 transmissions Assumes independent losses N1 src dst N2 N3 N4 25% 25% 25% 25% 100% 100% 100% 100% Slide Credit: Biswas & Morris Design Choice:  Design Choice ExOR makes routing decision after packets have been received Lets you decide route based upon actual success instead of probability Requires a way of communicating to other nodes who actually received packet Priority ordering:  Priority ordering Goal: nodes “closest” to the destination send first Sort by ETX metric to dst Nodes periodically flood ETX “link state” measurements Path ETX is weighted shortest path (Dijkstra’s algorithm) Source sorts, includes list in ExOR header src N1 N2 N3 dst N4 Slide Credit: Biswas & Morris ExOR batching:  ExOR batching Challenge: finding the closest node to have rx’d Send batches of packets for efficiency Node closest to the dst sends first Other nodes listen, send remaining packets in turn Repeat schedule until dst has whole batch src N3 dst N4 tx: 23 tx: 57 -23  24 tx:  8 tx: 100 tx: 0 tx:  9 N1 N2 Slide Credit: Biswas & Morris ExOR: 2x overall improvement :  ExOR: 2x overall improvement Median throughputs: 240 Kbits/sec for ExOR, 121 Kbits/sec for Traditional Throughput (Kbits/sec) 1.0 0.8 0.6 0.4 0.2 0 0 200 400 600 800 Cumulative Fraction of Node Pairs ExOR Traditional Slide Credit: Biswas & Morris ExOR moves packets farther:  ExOR moves packets farther ExOR average: 422 meters/transmission Traditional Routing average: 205 meters/tx Fraction of Transmissions 0 0.1 0.2 0.6 ExOR Traditional Routing 0 100 200 300 400 500 600 700 800 900 1000 Distance (meters) Slide Credit: Biswas & Morris ExOR discussion:  ExOR discussion 2x improvement: Awesome! The cost? Look at Figure 6 in the paper. What’s the range of RTTs from src->N24? Up to 3.5 seconds. Ouch! Batching: Requires many pkts from srcdst Increases delay Interacts very poorly with TCP. (Translation: probably slower!) Solution: Proxy at edge, custom transport protocol across wireless network Awesome performance and nice design, but some serious deployment challenges Back to Bidirectional:  Back to Bidirectional When you Tx in a line, both sides can hear you: How do we make this work for us? A B C “packet” Coding with Bidirectional traffic:  Coding with Bidirectional traffic 4 “hops” in 3 transmissions: A B C Time 1: Pkt A->B Time 2: Pkt C->B Time 3: (Pkt C->B XOR Pkt A->B) Building it: COPE:  Building it: COPE Opportunistic listening (common to ExOR) Nodes listen all the time to all Tx (n.b. – would consume more power; assumption here is that you really want throughput) Periodic “reception reports” Tells neighbors what it’s heard Usually piggybacked on data Can also guess about reception using ETX What packets to code??:  What packets to code?? Node has some packets in Tx queue Which of them should it XOR together? Goal: max # of real packets delivered S.t. each nexthop can decode the real packet Let’s walk through an example… COPE-ing:  COPE-ing 1x2: C gets 2 1x3: C=3, A=1 1x3x4:  A B D C 3 4 1 3 1 4 Wants to send: 1 -> A 2 -> C 3 -> C 4 -> D Gain: Theory and Practice:  Gain: Theory and Practice Depends on topology. An “X”: middle node can xor 4 packets Each edge sends once Middle sends once = 8 pkts in 5 tx = 1.6 gain Max: ~2 But Overhead, loss, etc. Quirk: Coding+MAC gain:  Quirk: Coding+MAC gain Consider A-B-C topology w/out COPE: middle must send 2x as many packets If MAC is “fair”, middle only gets 1/3 of transmit time So packets build up and get lost BUT: TCP prevents too much packet buildup So this is achievable with UDP flows Can We COPE With It?:  Can We COPE With It? Overall: With symmetric, same-size UDP flows, COPE gain very nice With symmetric TCP, maybe 30%, but: COPE works best in a highly-loaded network (increases capacity) TCP performs very poorly with high loss rates! Requires lots of work to get TCP to work well over cope Fundamental and cool results, but may also need custom transport protocols to really use Could be great for software/multimedia/etc., dist over mesh network… If time permits:  If time permits TCP performance and wireless discussion Interesting RTS/CTS positive notes in COPE paper Application of cope and ExOR RTS-id Credits:  Credits Several of the ExOR slides (the pretty ones) are from “Opportunistic Routing in Multi-hop Wireless Networks”, Sanjit Biswas and Robert Morris, talk at SIGCOMM 2005. Aside: The Roofnet guys have a startup, Meraki (http://meraki.net/), doing mesh networks. They make cool little low-power mesh radios that implement many of the things we’ve read about in class. Fun stuff.

Add a comment

Related presentations

Related pages

YouTube

13.599.789 Aufrufe; vor 5 Tagen; 2:28. Nächstes Video; Jetzt wiedergeben; SWISS ARMY MAN Trailer 2 German Deutsch (2016) - Dauer: 2:28. KinoCheck . 270 ...
Read more

laurx.13 | YouNow LIVE

Folge laurx.13 live auf YouNow! Instagram: laurx.13. YouNow is the best way to broadcast live and get an audience to watch you. ...
Read more

Open Broadcaster Software - Index

Open Broadcaster Software is free and open source software for recording and live streaming. Stream to Twitch, YouTube and many other providers or record ...
Read more

Professional Cameras - Professional ... - 3D Broadcast Sales

New and used broadcast equipment sales, broadcast audio, editing, cameras and monitors. Solutions to all your audio and broadcasting equipment needs.
Read more

MEDIA BROADCAST: ENABLING MEDIA INNOVATION

MEDIA BROADCAST plant und verwirklicht Ihre Video-, Audio- oder Daten-Übertragung national und international über multimediale Übertragungsplattformen.
Read more

13: Broadcast - part 2 | Main quests - Antenna - Dying ...

Note - This quest appears in your journal while you are in the Old Town, i.e. as a result of one of the meetings with Troy. In order to complete its main ...
Read more

CBS TV Network Primetime, Daytime, Late Night and Classic ...

television, TV, video, CBS TV, Columbia Broadcast System, watch online video, watch tv, soap opera video, David Letterman, CSI, Big Brother, NCIS, The ...
Read more

13: Broadcast | Main quests - Old Town - Dying Light Game ...

Head towards the Northern part of the Old Town (point 5a on the map). Here, you need to find the stairs to the underground corridors. Open the left gate ...
Read more

Livestream | Broadcast & Watch HD Live Streaming Video Events

Livestream is the easiest way to broadcast your event live. Cloud streaming service includes playback on any device, DVR, analytics, ad insertion, CDN ...
Read more

13th Street Germany

Oktober, Mo - Mi ab 20:13. See more. Serien Scorpion: Staffel 2 Die neue Staffel auf 13th Street. Video Trailer: Scorpion - Staffel 2. 18 Fotos. Fotos / 09 ...
Read more