1 / 17

Managing Chaotic Wireless Networks

Managing Chaotic Wireless Networks. David Wetherall University of Washington. Credits. UW/CSE wireless group John Zahorjan, Maya Rodrig, Charlie Reis, Ratul Mahajan, Ed Lazowska Intel Research Seattle Yatin Chawathe, Mike Chen, Brian Noble, Anthony Nicholson CMU

wan
Download Presentation

Managing Chaotic Wireless Networks

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Managing Chaotic Wireless Networks David Wetherall University of Washington

  2. Credits • UW/CSE wireless group • John Zahorjan, Maya Rodrig, Charlie Reis, Ratul Mahajan, Ed Lazowska • Intel Research Seattle • Yatin Chawathe, Mike Chen, Brian Noble, Anthony Nicholson • CMU • Srini Seshan et. al. for “Chaotic” wireless

  3. Context: WiFi boom deployments • Advantages: minimal infrastructure, mobility • Technologies: 802.11 a/b/g/n, and more • Deployments: hotspots, home networks measure cost-performance time

  4. Chaotic Wireless Networks Two consequences of growth: • Overlapping (dense) deployments • Compete for spectrum • Seen in urban areas today • That are independently managed • Limited visibility into the operation of others • Focus on their own goals; no central planning Chaotic = overlapping + independent Claim for chaotic environments: • Today’s 802.11 designs won’t fare well • There’s an opportunity to do much better

  5. CSE Allen Center, 3rd floor 10 802.11b/g APs per floor, managed as a whole: Channels 1, 6, 11 20/30 mW power Omni-directional antennas Site surveys, sparse deployments, and decoupling are the basis for good coverage and performance. A single 802.11 network architecture

  6. Adverse interactions across networks • 802.11 won’t degrade gracefully in chaotic environments • Ideally CSMA/CA would provide reasonable sharing … • Default channel assignments • Crowding, non-orthogonal effects • Static power management • Wastefully high, leads to interference • Rate diversity • High rate transmissions beat out by low rates • Hidden terminals and capture • Poor access from some locations • Other unlicensed band transmissions • Don’t argue with a phone/microwave … • Problems increase with density, load and heterogeneity

  7. A Band-aid: Dynamic tuning • One possible fix: dynamic parameter tuning • Each network measures environment and chooses the channel/power/routes that works best for it • Claim: dynamic tuning isn’t enough • Adaptation is necessary but not sufficient • There are opportunities to do much better

  8. Top: AP conflict on channel 1 (middle two blue nodes) can’t be resolved by individual tuning Bottom: but a solution needs only a simple switch between two adjacent nodes Pitfalls (1) – you can get stuck 1 5 4 3 2 6 1 5 4 3 2 6

  9. Pitfalls (2) – lack of stability • To avoid getting stuck, need to perturb the system, changing channels even without immediate gain • Q: What happens to in the configuration below? • A: Four APs dance around three channels • Tradeoff between stability and convergence 4 1 3 2

  10. Pitfalls (3) – power escalation • Suppose APs adjust their transmit power (between min and max) based on delivery to their clients • What will the APs on channel 1 do? • Higher power increases delivery/rate and interference • Potential tragedy of the commons (lose/lose) 4 1 3 2

  11. Opportunities for Coordination • Suppose we go beyond dynamic tuning … • Coordination = actively exchange information and make joint decisions (for selfish gain), rather than merely try to avoid each other • Claim: coordination beats tuning • Improves outcome for pitfalls • Enables new beneficial scenarios

  12. Relaying for coverage/performance • Relaying can leave both networks better off • Use of better links allows higher rates, • i.e., one transmission per packet at 1 Mbps vs. two transmissions per packet, each at 55Mbps 1 AP1 C1 55 55 55 55 1 AP2 C2

  13. Backup for reliability/performance • Use neighbor AP for backup Internet access • Or always use neighbor’s surplus for performance • Need to consider AUP issues Internet Internet X AP1 AP2 C1 C2

  14. Goal: Wireless internetworking • Wireless Internet = architecture that coordinates multiple individual wireless networks for better coverage, performance and reliability • Predictably allocate wireless and access bandwidths • Leave individual networks better off than before • Architectural roles: • Effectively allocate underlying resource • Allow new applications to flourish • IP does both; 802.11 does neither …

  15. Research agenda • How does wireless work in practice? • Predict what will happen for a given choice of channel/power/routes and protocols • What information is exchanged where? • Need to find small regions with tight coupling • How do we make this win-win? • Choices that benefit all players • How do we foster a grassroots effort? • Deployment incentives

  16. Questions?

  17. Aside: WiFi vs Cellular data • WiFi • Local = high bandwidth, inexpensive • Many islands of coverage • Cellular data • Wide-area = low bandwidth, expensive • Build-out for good coverage “away from home” • So concentrate on WiFi here • Potential for combining the best of both worlds

More Related