Cross-Layer Rate Control in Wireless Networks with Lossy Links: Leaky-Pipe Flow, Effective Network Utility Maximization and Hop-by-Hop Algorithms

Abstract

Due to multi-path fading and co-channel interference, wireless links are lossy in nature. As a result, the data rate of a given flow becomes "thinner and thinner" along its routing path, and the data rate received successfully at the destination node ( <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">the</i> <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">effective</i> <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">rate</i> ) is typically lower than the transmission rate at the source node ( <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">the</i> <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">injection</i> <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">rate</i> ). In light of this observation, each flow is treated as a "leaky-pipe" model in this study. Moreover, we introduce the notion of "effective utility" associated with the effective rate (not the injection rate) for each flow, and explore rate control mechanisms through effective network utility maximization (ENUM). We focus on two network models: (1) ENUM with link outage constraints with a maximum error rate at each link; (2) ENUM with path outage constraints where there exists an end-to-end outage requirement for each flow. For both problems, we explicitly take into account the "thinning" feature of data flows and devise distributed hop-by-hop rate control algorithms accordingly. Our numerical examples corroborate that higher effective network utility and better fairness among effective flow rates can be achieved by the ENUM algorithms than the standard NUM.