pimrc 05 B

75 %
25 %
Information about pimrc 05 B
Education

Published on June 18, 2007

Author: FunnyGuy

Source: authorstream.com

A Comparative Measurement Study of the Workload of Wireless Access Points in Campus Networks:  A Comparative Measurement Study of the Workload of Wireless Access Points in Campus Networks Maria Papadopouli Assistant Professor Department of Computer Science University of North Carolina at Chapel Hill (UNC) This work was partially supported by the IBM Corporation under an IBM Faculty Award 2004 Coauthor & collaborator:  Coauthor andamp; collaborator Felix Hernandez-Campos Department of Computer Science University of North Carolina at Chapel Hill (UNC) Roadmap:  Roadmap Motivation andamp; Research Objectives Main Contributions Data Acquisition Performance Analysis Related work Future work Motivation:  Motivation Optimization of the performance of wireless networks Better capacity planning andamp; load balancing Support of applications with real-time constraints Models for simulation studies Research Objectives:  Research Objectives Analyze traffic characteristics at each AP Total number of bytes andamp; packets (sent andamp; received) Different time scales Number of associations and roaming operations Impact of building types Contrast traffic models andamp; distributions from large production wireless networks (UNC andamp; Dartmouth) Data Acquisition:  Data Acquisition 729-acre campus with 26,000 students, 3,000 faculty, 9,000 staff Diverse environment 14,712 unique MAC addresses 488 APs (Cisco 1200, 350, 340 Series) SNMP polling every AP every 5minutes using a non-blocking library calls Tracing period of 63 days Data cleaning follows … Aggregate Traffic Load Analysis:  Aggregate Traffic Load Analysis Dichotomy among APs (1/2) APs with uploaders & APs with downloaders:  Dichotomy among APs (1/2) APs with uploaders andamp; APs with downloaders Dichotomy among APs (2/2)APs with uploaders & APs with downloaders:  Dichotomy among APs (2/2) APs with uploaders andamp; APs with downloaders Modeling Aggregate Sent & Received Traffic:  Modeling Aggregate Sent andamp; Received Traffic Lognormal distribution Significant heavier tail than lognormal Two lognormals with different parameters 5-Minute Traffic Load:  5-Minute Traffic Load Dartmouth APs receive has a lighter body (75% andlt; 100KB) UNC has two regions: 40% of intervals with very light traffic another close to sent distribution Similarity between Dartmouth andamp; UNC [500KB, 2MB] during most active intervals AP sent traffic load: AP receive traffic load: Total Number of Mcast/Bcast Packets Received vs. Sent by AP:  Total Number of Mcast/Bcast Packets Received vs. Sent by AP Total Number of Unicast Packets Sent vs. Received by AP:  Total Number of Unicast Packets Sent vs. Received by AP Summary of Contributions (1/2):  Summary of Contributions (1/2) Similarities between UNC vs Dartmouth traces Log-normality is prevalent Light traffic load but with long tails Majority of APs send andamp; receive packets of small size Prominent dichotomy among APs APs dominated by uploaders AP dominated by downloaders APs with average large sent andamp; small receive packet APs with small sent andamp; large receive packet Substantial number of non-unicast packets Conclusions (2/2):  Conclusions (2/2) Heavy distribution of associations/roaming operations Daily averages between 170-4,310 Weak correlation with traffic load Need more efficient association mechanisms Future Research Plan :  Future Research Plan Analysis of packet headers for better understanding the uploading behavior Formal parametric models for traffic load Spatial correlations of APs and classification of APs based on various parameters (traffic characteristics, number of associations, distinct clients) Topological properties of wireless infrastructure Analysis of traces from diverse set of testbeds andamp; contrast of their traffic models UNC-FORTH data repository of traces andamp; benchmarks More Info:  More Info http://www.cs.unc.edu/~maria http://www.ics.forth.gr/mobile/ maria@cs.unc.edu Thank You!

Add a comment

Related presentations

Related pages

A Comparative Measurement Study of the Workload of ...

1 A Comparative Measurement Study of the Workload of Wireless Access Points in Campus Networks Felix´ Hernandez-Campos´ Maria Papadopouli Emails ...
Read more

InvitedPaper - Vodafone Chair Mobile Communications ...

InvitedPaper Gerhard Fettweis, ... (64-QAM, g=1.05, `=5 ... frequency offset,fl in Proc. IEEE PIMRC 2002, vol. 2, pp. 561Œ566, Sept. 2002. [26] ...
Read more

Synchronization in OFDM-based WLAN with Transmit and ...

Synchronization in OFDM-based WLAN with Transmit and Receive Diversities ... B. Frequency Synchronization in SISO-OFDM ... and the PAPR is 4.05 dB. 3) ...
Read more

Smart Antenna Techniques and Their Application to Wireless ...

1 12/05/03 Slide 1 Tuesday, September 13, 2005 Smart Antenna Techniques and Their Application to Wireless Ad Hoc Networks Jack H. Winters September 13, 2005
Read more

Call for Papers

PIMRC’05 The 16th Annual IEEE International Symposium on Personal Indoor and Mobile Radio Communications Maritim Congress Hotel, Berlin, Germany
Read more

www.ics.forth.gr

@inproceedings{papa05comp, author ="Felix Hernandez-Campos and Maria Papadopouli", title="A Comparative Measurement Study of the Workload of Wireless ...
Read more

CODING LIMITS FOR SHORT RANGE WIRELESS INFRARED ...

CODING LIMITS FOR SHORT RANGE WIRELESS INFRARED TRANSMISSION ... P. O. Box 10 05 65, 98684 Ilmenau, Germany Phone: +49 (3677) 69-2619, Fax: +49 ...
Read more

K. Sayrafian-Pour and D. Kaspar, “Indoor Positioning ...

... (PIMRC’05), Berlin, Vol. 4, September 2005, pp. 2722-2726. ... JOURNAL NAME: Wireless Sensor Network DOI: 10.4236/wsn.2010.24051 Sep 05, 2014;
Read more

Indoor Localization with Probability Density Functions ...

INDOOR LOCALIZATION WITH PROBABILITY DENSITY FUNCTIONS ... B given a location loc: P(meas A ... Full Paper for PIMRC'05: ...
Read more