Cellular networks, used in wireless and mobile communications under multiple access conditions, depend upon good protocols to (a) allocate spectral bandwidth in the presence of interference, (b) control access among participating users and (c) support robust quality of service in a seamless manner. This proposal addresses a common underlying goal of resource management protocols, viz. the provision of guarantees on performance measured in terms of network indicators such as spectral efficiency, rate of call loss or call blocking, fraction of successful handoffs and the complexity of control mechanisms. Naturally, these performance guarantees are subject to the service quality factors and to the radio interference constraints in the modeled topology of the network.
The researcher proposes the design and analysis of new, distributed protocols to adaptively allocate spectrum resources, implement call admission and support handoffs while simultaneously trying to optimize one or more desireable performance factors, e.g. maximizing spectral utilization, minimizing interference, maintaining acceptable service quality etc. The work is further expected to focus on strategies that react quickly to changing local conditions and that try to avoid centralized planning and control as much as possible. Building upon encouraging preliminary results, this proposal will use a general approach that consists of formulating such protocol design problems as instances of online, approximation algorithms competing with unknown but optimal 'adversarial' protocols.
The overall goals of this work are two-fold. First the researcher will develop a suite of protocols in conjunction with realistic models that account for constraints as faithfully as possible, while guaranteeing theoretically derived bounds on performance. In addition, the work will be complemented by collaborations with undergraduate students to develop a comprehensive web-based visualization tool that can be used to validate the protocols through extensive empirical testing on both systhesized and real-world network traffic data.
|Effective start/end date
|10/1/99 → 3/31/03
- National Science Foundation: $83,727.00