September 14-19, 2003 | Douglas S. J. De Couto, Daniel Aguayo, John Bicket, Robert Morris
This paper presents the expected transmission count (ETX) metric, which finds high-throughput paths in multi-hop wireless networks. ETX minimizes the expected number of packet transmissions (including retransmissions) required to deliver a packet to its destination. Unlike the minimum hop-count metric, which chooses arbitrarily among paths of the same length, ETX accounts for link loss ratios, asymmetry in loss ratios between directions, and interference among successive links. Measurements from a 29-node 802.11b test-bed show that minimum hop-count performs poorly, often selecting routes with significantly lower throughput than the best available. ETX improves performance, especially for longer paths, by selecting routes with higher throughput. The paper describes the design and implementation of ETX for DSDV and DSR routing protocols, and evaluates its performance through experiments on the test-bed. ETX outperforms minimum hop-count by up to a factor of two for multi-hop routes. The paper also discusses the design of ETX, its implementation in DSDV and DSR, and the effects of various design decisions on performance. ETX is shown to be effective in improving throughput, even in the presence of asymmetric links and interference. The paper concludes that ETX is a promising metric for multi-hop wireless routing.This paper presents the expected transmission count (ETX) metric, which finds high-throughput paths in multi-hop wireless networks. ETX minimizes the expected number of packet transmissions (including retransmissions) required to deliver a packet to its destination. Unlike the minimum hop-count metric, which chooses arbitrarily among paths of the same length, ETX accounts for link loss ratios, asymmetry in loss ratios between directions, and interference among successive links. Measurements from a 29-node 802.11b test-bed show that minimum hop-count performs poorly, often selecting routes with significantly lower throughput than the best available. ETX improves performance, especially for longer paths, by selecting routes with higher throughput. The paper describes the design and implementation of ETX for DSDV and DSR routing protocols, and evaluates its performance through experiments on the test-bed. ETX outperforms minimum hop-count by up to a factor of two for multi-hop routes. The paper also discusses the design of ETX, its implementation in DSDV and DSR, and the effects of various design decisions on performance. ETX is shown to be effective in improving throughput, even in the presence of asymmetric links and interference. The paper concludes that ETX is a promising metric for multi-hop wireless routing.