Enhancing Interactive Web Applications in Hybrid Networks (“thedu”)

Slides:



Advertisements
Similar presentations
U NIVERSITY OF M ASSACHUSETTS, A MHERST Department of Computer Science R3: Robust Replication Routing in Wireless Networks with Diverse Connectivity Characteristics.
Advertisements

U NIVERSITY OF M ASSACHUSETTS, A MHERST Department of Computer Science R3: Robust Replication Routing in Wireless Networks with Diverse Connectivity Characteristics.
Aruna Balasubramanian, Ratul Mahajan Arun Venkataramani, Brian N Levine, John Zahorjan Interactive WiFi Connectivity from Moving Vehicles University of.
Augmenting Mobile 3G Using WiFi Sam Baek Ran Li Modified from University of Massachusetts Microsoft Research.
FindAll: A Local Search Engine for Mobile Phones Aruna Balasubramanian University of Washington.
Aruna Balasubramanian Department of Computer Science University of Massachusetts Amherst Architecting Protocols to Improve Connectivity.
Dynamic Adaptive Streaming over HTTP2.0. What’s in store ▪ All about – MPEG DASH, pipelining, persistent connections and caching ▪ Google SPDY - Past,
Measurement, Modeling, and Analysis of a Peer-2-Peer File-Sharing Workload Presented For Cs294-4 Fall 2003 By Jon Hess.
1 Web Server Performance in a WAN Environment Vincent W. Freeh Computer Science North Carolina State Vsevolod V. Panteleenko Computer Science & Engineering.
A Mobile Infrastructure Based VANET Routing Protocol in the Urban Environment School of Electronics Engineering and Computer Science, PKU, Beijing, China.
1 Network Measurements of a Wireless Classroom Network Carey Williamson Nuha Kamaluddeen Department of Computer Science University of Calgary.
Web Caching Schemes1 A Survey of Web Caching Schemes for the Internet Jia Wang.
1 Prefetching the Means for Document Transfer: A New Approach for Reducing Web Latency 1. Introduction 2. Data Analysis 3. Pre-transfer Solutions 4. Performance.
Beneficial Caching in Mobile Ad Hoc Networks Bin Tang, Samir Das, Himanshu Gupta Computer Science Department Stony Brook University.
Exploiting Content Localities for Efficient Search in P2P Systems Lei Guo 1 Song Jiang 2 Li Xiao 3 and Xiaodong Zhang 1 1 College of William and Mary,
1 Probabilistic Models for Web Caching David Starobinski, David Tse UC Berkeley Conference and Workshop on Stochastic Networks Madison, Wisconsin, June.
Implementing ISA Server Caching. Caching Overview ISA Server supports caching as a way to improve the speed of retrieving information from the Internet.
Web Caching Schemes For The Internet – cont. By Jia Wang.
Application Layer  We will learn about protocols by examining popular application-level protocols  HTTP  FTP  SMTP / POP3 / IMAP  Focus on client-server.
Personalized Ontologies for Web Search and Caching Susan Gauch Information and Telecommunications Technology Center Electrical Engineering and Computer.
Augmenting Mobile 3G Using WiFi Aruna Balasubramanian Ratul Mahajan Arun Venkataramani University of Massachusetts Microsoft Research.
UMass DieselNet: A Disruption-Tolerant Network Testbed John Burgess Department of Computer Science UMass Amherst John Burgess Department of Computer Science.
Augmenting Mobile 3G Using WiFi Aruna Balasubramanian Ratul Mahajan Arun Venkataramani University of Massachusetts Microsoft Research.
Niranjan Balasubramanian Aruna Balasubramanian Arun Venkataramani University of Massachusetts Amherst Energy Consumption in Mobile Phones: A Measurement.
1 Enabling Secure Internet Access with ISA Server.
By Ravi Shankar Dubasi Sivani Kavuri A Popularity-Based Prediction Model for Web Prefetching.
COGNITIVE RADIO FOR NEXT-GENERATION WIRELESS NETWORKS: AN APPROACH TO OPPORTUNISTIC CHANNEL SELECTION IN IEEE BASED WIRELESS MESH Dusit Niyato,
Multimedia and Mobile communications Laboratory Augmenting Mobile 3G Using WiFi Aruna Balasubramanian, Ratul Mahajan, Arun Venkataramani Jimin.
Aruna Balasubramanian Brian Neil Levine Arun Venkataramani University of Massachusetts, Amherst Enhancing Interactive Web Applications in Hybrid Networks.
Overview Goal: video streaming in vehicular networks via WiFi Compelling usage scenarios –Gas stations and local shops deploy APs to provide video and.
Web Cache Replacement Policies: Properties, Limitations and Implications Fabrício Benevenuto, Fernando Duarte, Virgílio Almeida, Jussara Almeida Computer.
Disruption Tolerant Networks Aruna Balasubramanian University of Massachusetts Amherst 1.
Sharing Information across Congestion Windows CSE222A Project Presentation March 15, 2005 Apurva Sharma.
Disruption Tolerant Networks Aruna Balasubramanian University of Massachusetts Amherst 1.
Hour 7 The Application Layer 1. What Is the Application Layer? The Application layer is the top layer in TCP/IP's protocol suite Some of the components.
Energy Consumption in Mobile Phones: A Measurement Study and Implications for Network Applications REF:Balasubramanian, Niranjan, Aruna Balasubramanian,
PPWEB: A Peer-to-Peer Approach for Web Surfing On the Go Ling-Jyh Chen, Ting-Kai Huang Institute of Information Science, Academia Sinica, Taiwan Guang.
HUAWEI TECHNOLOGIES CO., LTD. Page 1 Survey of P2P Streaming HUAWEI TECHNOLOGIES CO., LTD. Ning Zong, Johnson Jiang.
Adaptive Web Caching CS411 Dynamic Web-Based Systems Flying Pig Fei Teng/Long Zhao/Pallavi Shinde Computer Science Department.
Internet Real-Time Laboratory Arezu Moghadam and Suman Srinivasan Columbia University in the city of New York 7DS System Design 7DS system is an architecture.
Web Image Retrieval Re-Ranking with Relevance Model Wei-Hao Lin, Rong Jin, Alexander Hauptmann Language Technologies Institute School of Computer Science.
Module 9: Implementing Caching. Overview Caching Overview Configuring General Cache Properties Configuring Cache Rules Configuring Content Download Jobs.
LANGUAGE MODELS FOR RELEVANCE FEEDBACK Lee Won Hee.
ECEN “Internet Protocols and Modeling”, Spring 2012 Course Materials: Papers, Reference Texts: Bertsekas/Gallager, Stuber, Stallings, etc Class.
Department of Computer Science Aruna Balasubramanian, Brian Neil Levine, Arun Venkataramani DTN Routing as a Resource Allocation Problem.
Dissertation Proposal Aruna Balasubramanian Department of Computer Science, University of Massachusetts, Amherst Architecting Protocols To Enable Mobile.
Efficient P2P Search by Exploiting Localities in Peer Community and Individual Peers A DISC’04 paper Lei Guo 1 Song Jiang 2 Li Xiao 3 and Xiaodong Zhang.
Performance of Web Proxy Caching in Heterogeneous Bandwidth Environments IEEE Infocom, 1999 Anja Feldmann et.al. AT&T Research Lab 발표자 : 임 민 열, DB lab,
Implementing ISA Server Caching
Search Engine using Web Mining COMS E Web Enhanced Information Mgmt Prof. Gail Kaiser Presented By: Rupal Shah (UNI: rrs2146)
Aruna Balasubramanian, Yun Zhou, W Bruce Croft, Brian N Levine and Arun Venkataramani Department of Computer Science, University of Massachusetts, Amherst.
Development of a QoE Model Himadeepa Karlapudi 03/07/03.
09/13/04 CDA 6506 Network Architecture and Client/Server Computing Peer-to-Peer Computing and Content Distribution Networks by Zornitza Genova Prodanoff.
On the Placement of Web Server Replicas Yu Cai. Paper On the Placement of Web Server Replicas Lili Qiu, Venkata N. Padmanabhan, Geoffrey M. Voelker Infocom.
An Analysis of Internet Content Delivery Systems 19 rd November, 2007 Youngsub CSE, SNU.
Mixus Daniel Mittelman Tomer Brosh
Energy-Aware Opportunistic Mobile Data Offloading
Delay-Tolerant Networks (DTNs)
Energy Consumption in Mobile Phones
Live Customer Support Solution
Department of Computer Science University of Calgary
Computer Communication & Networks
08/03/14 Energy Consumption in Mobile Phones: A Measurement Study and Implications for Network Applications REF:Balasubramanian, Niranjan, Aruna Balasubramanian,
Efficient Evaluation of k-NN Queries Using Spatial Mashups
Augmenting Mobile 3G Using WiFi
Performance Evaluation of Computer Networks
Performance Evaluation of Computer Networks
Internet: Quality of Service Mechanisms at Application Level
PPWEB: A Peer-to-Peer Approach for Web Surfing On the Go
Client/Server Computing and Web Technologies
Presentation transcript:

Enhancing Interactive Web Applications in Hybrid Networks (“thedu”) Authers Aruna Balasubramanian, Brian Neil Levine, Arun Venkataramani Department of Computer Science University of Massachusetts Amherst {arunab, brian, arun}@cs.umass.edu Presenting Ido Barkan, Kfir Damari Department of Communication System Engineering Ben Gurion University

Mobile Internet Users Options for mobile users: Comparison 3G Wifi Price (one pay -> monthly) Availability “free” connectivity (for workers)

Application space

Questions “1) How can we enable system support for interactive web applications to tolerate disruptions in connectivity from mobile nodes?” “2) Can opportunistic mobile-to-mobile (m2m) transfers enhance application performance over only using APs, and if so, under what conditions and by how much?”

You’re sitting in the bus – overhearing two people talking about mobicom www.google.com

Understanding the problem Many user refresh Tcp session creation overhead More then one HTTP request (Springfield) Bottom line – It will take time!

Thedu Goals & Design Thedu is designed to enable quick and easy access to web search applications from moving vehicles. Robustness: Tolerate disruptions (interactive searches need persistent connectivity) Low-cost deployability: The system should be low-cost minimal change to existing infrastructure or software Extending access: extend the Internet’s reach to rural or developing regions with limited access points to the internet

A few facts about searching 75% of users do not view more than the top 20 web pages most users are interested in at most 5 results per query IR - information retrieval (IR) techniques

Part 1 – m2i traffic Aggressive prefetching Prioritization protocol to allocate limited bandwidth to the most useful web pages Thedu proxy only returns the top few web pages (based on their relevance probability)

Aggressive prefetching

Prioritization Query-type classification(homepage, content and service) Prioritize web pages across different queries - normalization algorithm is based on the Kullback-Liebler divergence and Markov chains

relevance probability We then estimate the relevance probability using Bayes’ rule. P(score|rel) - denote the probability of score given the document is relevant P(score|nonrel) the probability of score given the document is not relevant.

What about helping others? Mostly top 5 queries needed

Part 2 - M2M ROUTING m2m routing can leverage other mobile nodes needs. m2m routing is DTN like. (DTN- delay & disruption tolerant network) Question. If a mobile node has pending web pages download- should it still download web responses for others??? Apparently- YES. Why? Most of the utility resides in the first few web pages.

Client prioritization If node Y has a connection to X it is prioritizing pages to it by U(ri)=P(ri)Qx(ri) (page is relevant and will reach X in dead line). Qx(ri) is estimated statistically (Poisson dist.). Cache everything you route (popularity, aging).

M2M basics Assumptions: Demand is infinite Responses are of unit size Experiment runs for unit time Trying to compute the m2m benefit (noted G) to system in relation to Searching for the maximum added m2m BW by searching for the amount of nodes that benefit from it.

Conclusion: m2m is beneficial. WHY? (intuition) Concave utility function. It is better for nodes with good AP reception to download responses for others than downloading their own. Workload skew- practically some nodes has lower demands from others and they may support other’s demand m2m routing is only beneficial when mobile nodes meet each other in the same frequency they meet AP.

Test cases (from theorem to practice) Deployment Trace-Driven Simulation

Deployment – What? DieselNet in Amherst MA 40 buses total ~21 buses are on the road bus is equipped with linux computer 802.11b radio ,constantly scan for APs&buses Server-side proxy (Thedu & Stateless)

Deployment – How? Why Indri? Some numbers predefined user queries & data-set human relevance judgments assigns a relevance score Some numbers Queries per hour: 10 per bus Deadline: 30 min queryIDs: allows query repetitions

Deployment - results Thedu returned 4.5 times more pages Number of relevant pages is 4 times larger Thedu returned at least one relevant page for twice as many queries With Thudu 90% of the time, a relevant page is received within 5 min Mean delay receiving first relevant web page is 2.7 min Mean delay receiving all relevant web pages is 2.3 min

Trace-Driven Simulation Traces were collected from vehicles to: Study affect of AP density on Thedu. Evaluate the benefits of m2m routing. Evaluate the perormance of Thedu for web browsing. Buses scans for APs and other buses. AP found- trying to connect. Bus found- connect till radio is out of range.

Trace-Driven Simulation Buses constantly exchange random data to measure capacity of connection (except when operational AP in range) Pinging is done to measure connectivity time. Logging: m2i: SSID, location, duration. m2m: bytes transferred 20 days of data logged.

Trace-Driven Simulation Each node simulates queries (Poisson dist.) Proxy retrieves top 20 web pages. Compute number of relevant responses. M2i BW is estimated to 205KBps. Town was divided to grids

m2m routing 151 open AP. In order to evaluate m2m benefit only 5 AP’s were chosen in random to fulfill a 1:1.3 m2m-to-m2i ratio. 3 variations: (i) no m2m routing. (ii) m2m routing. (iii) routing + caching. Only 20% of queries repeats itself (conservative).

Results summary m2i improvement with Thedu dramaticlly m2m benefit Queries received relevant result with a mean delay of 2.3 minutes (0.55 minutes in areas with high AP density) m2m benefit M2M with caching improves the number of relevant responses by up to 58% But the mean delay is significantly high Practicality for interactive applications is questionable

Consultations “1) How can we enable system support for interactive web applications to tolerate disruptions in connectivity from mobile nodes?” Proxy-based architecture (to tolerate disruptions) Aggressive fetching (to maximize bandwidth utilization) Smart prioritizing (reduce needed bandwidth) “2) Can opportunistic mobile-to-mobile (m2m) transfers enhance application performance over only using APs, and if so, under what conditions and by how much?” Mostly where there aren’t many APs. Caching improves performance drastically Result mean delay makes traffic irrelevant for human applications m2m is not cost-effective in most Urban environments today

???????? pink-bus-burning-man