1 / 16

VTube : Towards the Media Rich City Life with Autonomous Vehicular Content Distribution

VTube : Towards the Media Rich City Life with Autonomous Vehicular Content Distribution. Tom H. Luan * , Lin X. Cai * , Jiming Chen § , Xuemin (Sherman) Shen * , and Fan Bai † * University of Waterloo, ON, Canada § Zhejiang University, China † General Motors Corporation, MI, USA.

ofira
Download Presentation

VTube : Towards the Media Rich City Life with Autonomous Vehicular Content Distribution

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. VTube: Towards the Media Rich City Life with Autonomous Vehicular Content Distribution Tom H. Luan*, Lin X. Cai*, JimingChen§, Xuemin(Sherman) Shen*, and Fan Bai† *University of Waterloo, ON, Canada § Zhejiang University, China †General Motors Corporation, MI, USA

  2. Goal and Roadmap Roadmap: • Motivation of the research • Methodology – system model and algorithms • Simulation • Conclusion and Q&A Goal: To design a large-scalecost-effectiveinfrastructure-based content distribution network using off-the-shelve technologies VTube: Cost-effective Vehicular Content Distribution SECON’11

  3. Motivation: Dream v.s. Reality • The Dream: to download file contents for free at high speed • Contents such as video clips (like using YouTube), MP3s, HD images, flyers, magazines … • The Reality, the bandwidth is • Expensive: Rogers: 50 CAD for 500MB; AT&T: 60 USD for 5GB • Unreliable: pure vehicle-to-vehicle (V2V) or ad hoc content distribution is unreliable and far to be practical • Can we achieve large-scale content distribution in the urban area with low-cost? • Download everywhere in the city with no need to pay VTube: Cost-effective Vehicular Content Distribution SECON’11

  4. Components of the System • Roadside Buffers (RSB) • Low cost infrastructure installed on the roadside: buffer storage (e.g., 20GB) + wireless interface • Store contents and distribute to subscribed users • Vehicles • With caching storage and wireless interface • Download/upload contents from/to RSBs, and swap contents among each other • Transport contents cross the city (interested in the contents) (deployed by the bus company to distribute bus schedules) 50 USD + 30 USD (deployed by the store) (e.g., flyers) VTube: Cost-effective Vehicular Content Distribution SECON’11

  5. VTube: Cost-Effective Approach Share costs to build the content distribution infrastructure • RSBs are bought and deployed by distributed entities • like movie theatres, malls, hotels, schools, to distribute their own information content, like movie trailers, room pictures in hotels • RSBs distributedly deployed cross the city forms VTube, an integrated infrastructure for content distribution • Why would stores like to buy RSBs? Because they are • Cheap: Lump sum of 80 dollars on devises and no further fees • No extra charge on the bandwidth (not connected to Internet) • No messy cabling works (simply mount RSBs high on a pole and configure them using wireless communications) • Profitable: • Upload contents, like advertisements, flyers, …, and distribute to users on the street for free VTube: Cost-effective Vehicular Content Distribution SECON’11

  6. VTube: Autonomous System • Contents are • published by distributed entities through the RSBs which they have deployed • transported by vehicles and spread in the city to vehicles and RSBs at different locations • A distributedly and additively built network • Entities deploy their own RSBs without any coordination with others • New RSBs work distributedly and can be additively incorporated into the existing system • Design Problem: How to enable RSBs to distributedly and selectively cache contents from vehicular networks to minimize the average download delay of contents? VTube: Cost-effective Vehicular Content Distribution SECON’11

  7. Comparison with Related Works • Compared with femtocells and infostations [1, 2] • no connections to the Internet • focusing on content replications and buffer management with V2V (vehicle-to-vehicle) and V2R (vehicle-to-RSB) communications • built in a decentralized and additively manner by distributed entities • Compared with the V2V content distribution [3, 4] • RSB as the content injection points • focusing on the design of RSBs and can be incorporated with the V2V schemes proposed • can also be used to distribute contents to PDAs, tablets, etc • Compared with the existing infrastructure-based vehicular content distribution network [5, 6] • incorporate V2V communications in the content distribution • propose a detailed and practical system architecture VTube: Cost-effective Vehicular Content Distribution SECON’11

  8. Design of VTube: Model • RSBs: randomly deployed with equal buffer to store L files at most • : set of files published in the city • : portion of vehicles that subscribe to download file i • : portion of vehicles have file i stored in their buffer • : probability that a RSU stores file i in its buffer • : average download delay of file i • is function of and • : buffer storage of RSB given At any given time Design at each RSB to minimize the overall delay of content dissemination VTube: Cost-effective Vehicular Content Distribution SECON’11

  9. Design of VTube: Assumptions • A homogeneousnetwork • Files have equal size • RSBs are homogeneous with equal buffer storage and follow the same policy to store files • Each RSB selects a file i to store with the probability • How to evaluate ? • How to solve the optimization problem • How to design distributed algorithm Design at each RSB to minimize the overall delay of content dissemination VTube: Cost-effective Vehicular Content Distribution SECON’11

  10. Model of Average Download Delay • Each file is of equal size and chopped into k pieces • Randomly select a vehicle to examine the download delay of file i (time to download all k pieces of file i) • Model the file download of the vehicle by a 2D Markov process with state (X(t), Y(t)) • X(t): number of blocks downloaded • Y(t): Y(t) = 1, if the vehicle is inside a RSB, and 0, otherwise Download rate of file ioutside any RSBs Average download delay (mean first passage time) Rate to depart the coverage of an RSB Rate to drive into the coverage of an RSB Download rate of file iinside a RSB VTube: Cost-effective Vehicular Content Distribution SECON’11

  11. Centralized Content Distribution • Using the Markov model, we can derive the expression of download delay represented by the file availability at vehicles and RSBs • : portion of vehicles with file i stored in local buffer • : probability that a RSB store file i in its buffer • Apply this result, Chebyshev inequality and KKT to we have an optimal solution of as VTube: Cost-effective Vehicular Content Distribution SECON’11

  12. Decentralized Algorithm • Select file i to store in RSBs with probability • Random walk based approach • RSBs issue walkers to the vehicular networks • Walkers are forwarded among files stored at vehicles for TTL steps • The files with stopped walkers are select to be uploaded to RSBs ( ) VTube: Cost-effective Vehicular Content Distribution SECON’11

  13. Simulation Verification • Trace-driven Session level simulator coded in C++ • 1000 vehicles randomly selected in the Shanghai taxi trace within 24-hour period on Feb. 20th, 2010 • 200 RSBs are randomly deployed in a region of 50 30 km2 • (download demand of file i) follows the Zipf-like distribution VTube: Cost-effective Vehicular Content Distribution SECON’11

  14. Average Download Delay • Global: store files based on b* • Distributed algorithm: random walk based algorithm • Local greedy: RSB always select the files with the largest di to download utility average download delay of file i portion of vehicles subscribe to download file i VTube: Cost-effective Vehicular Content Distribution SECON’11

  15. Conclusions • While wireless networking and various personal electronics (e.g., PDA, tablet, smartphones) have proliferated in the past decade, low-cost content distributionis still unavailable • A content distribution infrastructure is necessary • Traditional cellular networks are expensive whereas the pure ad hoc networks are unreliable and not practical so far • We propose to build distributed infrastructure with shared expenses • VTube: Roadside Buffers + vehicular networks • Optimal content replication with central information • Distributed content replication using the random walk algorithm VTube: Cost-effective Vehicular Content Distribution SECON’11

  16. Q&A VTube: Towards the Media Rich City Life with Autonomous Vehicular Content Distribution Tom H. Luan*, Lin X. Cai*, Jimin Chen§, Xuemin (Sherman) Shen*, and Fan Bai† *University of Waterloo, ON, Canada §Zhejiang University, China †General Motors Corporation, MI, USA VTube: Cost-effective Vehicular Content Distribution SECON’11

More Related