Web search from a bus

Opportunistic connections to the Internet from open wireless access points is now commonly possible in urban areas. Vehicular networks can opportunistically connect to the Internet for several seconds via open access points. In this paper, we adapt the interactive process of web search and retrieval to vehicular networks with intermittent Internet access. Our system, called Thedu, has mobile nodes use an Internet proxy to collect search engine results and prefetch result pages. The mobile nodes download the pre-fetched web pages from the proxy. Our contribution is a novel set of techniques to make aggressive but selective prefetching practical, resulting in a significantly greater number of relevant web results returned to mobile users. In particular, we prioritize responses in the order of the usefulness of the response to the query, that allows the mobile node to download the most useful response first. To evaluate our scheme, we deployed Thedu on DieselNet, our vehicular testbed operating in a micro-urban area around Amherst, MA. Using a simulated workload, we find that users can expect four times as many useful responses to web search queries compared to not using Thedu's mechanisms. Moreover, the mean latency in receiving the first relevant response for a query is 2.7 minutes for our deployment; we expect Thedu to have even better performance in larger cities that have densely populated open APs.

[1]  W. Bruce Croft,et al.  A general language model for information retrieval , 1999, CIKM '99.

[2]  Donald F. Towsley,et al.  Study of a bus-based disruption-tolerant network: mobility modeling and impact on routing , 2007, MobiCom '07.

[3]  R. Manmatha,et al.  Modeling score distributions for combining the outputs of search engines , 2001, SIGIR '01.

[4]  S. Robertson The probability ranking principle in IR , 1997 .

[5]  Amanda Spink,et al.  Real life, real users, and real needs: a study and analysis of user queries on the web , 2000, Inf. Process. Manag..

[6]  W. Bruce Croft,et al.  A general language model for information retrieval (poster abstract) , 1999, SIGIR '99.

[7]  Andrei Broder,et al.  A taxonomy of web search , 2002, SIGF.

[8]  D.J. Goodman,et al.  INFOSTATIONS: a new system model for data and messaging services , 1997, 1997 IEEE 47th Vehicular Technology Conference. Technology in Motion.

[9]  Alex Pentland,et al.  DakNet: rethinking connectivity in developing nations , 2004, Computer.

[10]  Jörg Ott,et al.  Drive-thru Internet: IEEE 802.11b for "automobile" users , 2004, IEEE INFOCOM 2004.

[11]  Yang Zhang,et al.  CarTel: a distributed mobile sensor computing system , 2006, SenSys '06.

[12]  Srinivasan Keshav,et al.  Low-cost communication for rural internet kiosks using mechanical backhaul , 2006, MobiCom '06.

[13]  W. Bruce Croft,et al.  Indri : A language-model based search engine for complex queries ( extended version ) , 2005 .

[14]  Brian Gallagher,et al.  MaxProp: Routing for Vehicle-Based Disruption-Tolerant Networks , 2006, Proceedings IEEE INFOCOM 2006. 25TH IEEE International Conference on Computer Communications.

[15]  Cauligi S. Raghavendra,et al.  Spray and wait: an efficient routing scheme for intermittently connected mobile networks , 2005, WDTN '05.

[16]  R. Manmatha,et al.  Mobile distributed information retrieval for highly-partitioned networks , 2003, 11th IEEE International Conference on Network Protocols, 2003. Proceedings..

[17]  Jie Lu,et al.  Content-based retrieval in hybrid peer-to-peer networks , 2003, CIKM '03.

[18]  W. Bruce Croft,et al.  Distributed Information Retrieval For Disruption-Tolerant Mobile Networks , 2005 .

[19]  Donna K. Harman,et al.  The Text REtrieval Conference (TREC) , 1999, NTCIR.

[20]  Anders Lindgren,et al.  Probabilistic routing in intermittently connected networks , 2003, MOCO.

[21]  Jörg Ott,et al.  Bundling the Web: HTTP over DTN , 2006 .