Bandwidth Estimation and Management for Mobile, Wireless Networks

Period of Performance: 12/22/2000 - 06/22/2001


Phase 1 SBIR

Recipient Firm

Scientific Systems Company, Inc.
500 West Cummings Park Array
Woburn, MA 01801
Principal Investigator


Bandwidth management in mobile, wireless networks is different and more difficult than in landline networks due to lower data rates, mobility, interference, and channel variability, which makes bandwidth dynamic instead of fixed over time. Furthermore, bandwidth allocations affect more users through increased noise levels or collision interference at neighboring nodes and inefficient bandwidth allocations lower network performance through unnecessary pre-emption blocking. To solve these problems we propose innovative machine learning and analysis methods to obtain bandwidth estimates that account for mobility and variable channel conditions. We use these estimates to determine information and protocol steps for a full suite of robust, effective bandwidth management protocols supporting allocation, channel sharing, reservation, and Quality of Service. We build on our work in designing a WIreless NEtwork Simulation (WINES) that provides simple mobility and channel models along with flexible, parameterizablemodels of many proposed mobile, wireless routing methods. We will be assisted in this effort by Sonia Fahmy of Purdue University, an expert in IP and ATM approaches to bandwidth reservation, allocation, and Quality of Service. We will also be assisted by BBN Technologies, who developed the Internet and have an extensive track record in development of innovative mobile wireless technologies.This effort will provide effective means for bandwidth allocation for mobile, wireless networks (and other IP networks for which the problem are simpler) operating in harsh conditions where users contend forscarce bandwidth. Immediate applications include military networks and civilian emergency communications networks. Bluetooth, a new standard for low power wireless communications will result in new applications for mobile, self organizing, wireless networks that must adapt to and carry out prescribed functions in widely variable conditions. The methods developed here will prove applicable to those networks and help propel their widespread acceptance for industrial, home, and personal applications.