Abstract:
We introduce and evaluate a feasible end-to-end congestion control algorithm for overcoming the severe deficiencies of TCP in IEEE 802.11 multihop wireless networks. Our ...Show MoreMetadata
Abstract:
We introduce and evaluate a feasible end-to-end congestion control algorithm for overcoming the severe deficiencies of TCP in IEEE 802.11 multihop wireless networks. Our approach, which we denote as TCP with Adaptive Pacing (TCP-AP), implements rate-based scheduling of transmissions within the TCP congestion window. The TCP source adaptively sets its transmission rate using an estimate of the current out-of-interference delay and the coefficient of variation of recently measured round-trip times. TCP-AP retains the end-to-end semantics of TCP and neither relies on modifications at the routing or the link layer nor requires cross-layer information from intermediate nodes along the path. As opposed to previous proposals that build on network simulators, we implement and evaluate our approach in a real wireless mesh test-bed comprising 20 nodes. In a comprehensive comparative performance study using our test-bed, we show that, depending on the current network state and traffic patterns, TCP-AP achieves up to 10 times more goodput than TCP NewReno, provides excellent fairness, and is highly responsive to changing network traffic conditions.
Published in: IEEE/ACM Transactions on Networking ( Volume: 19, Issue: 4, August 2011)