Throughput Estimation For Wireless LANs Haynes Cheng Arnel

Throughput Estimation For Wireless LANs Haynes Cheng, Arnel Lim, Michael Weatherby University of British Columbia Department of Electrical and Computer Engineering Radio Science Lab III - METHODOLOGY I - OBJECTIVES 1. Indoor Path Loss Model • 2. Evaluate indoor path loss model parameters so that signal strength and carrier/interference (C/I) values can be calculated 1. Path Loss Models and RSSI Prediction 2. Throughput Estimation Static Throughput vs C/I Curves (Previously Measured) C/I Values from Path Loss Model and Simulation Indirect Throughput Estimation • 3. Infer throughput from signal strength and the measured effects of noise on performance 3. Direct throughput measurement • Establish test strategies for obtaining repeatable throughput field-measurements using minimal resources Indirect Method (STPA) • Since we are estimating throughput from calculated C/I, why not use real C/I? • Modify a UBC-developed mapping tool to obtain real signal and interference values for calculating C/I II - BACKGROUND & MOTIVATION 1. How can one predict or measure how well a wireless network will work under heavy load? IV - ACCOMPLISHMENTS 1. Path Loss Model of Indoor Environments (AP vs Client) • 2. Focus on Network Level Testing to Aid Network Administrators 2. Access Point Client • Cell Level • contention • MAC overhead Network Level • mutual interference • power settings • channel assignments 3. Collected and reduced detailed path loss data for multiple AP’s and clients (3 buildings, 40 access points, over 10 000 data points) Indirect Throughput Measurement Link Level • signal quality Throughput Estimation Implemented the indirect method based upon static throughput versus C/I curves Simple Throughput Prediction Algorithm (STPA) • Implemented a simple throughput prediction algorithm using Fatmapper, a UBC-developed mapping tool, as the data collection platform.