No abstract

In recent years cheap Internet access has become increasingly popular within the industrialized countries. Especially in the age of Web 2.0, where popular Internet applications begin to dominate computer use, ubiquitous network access is more important than ever. A new class of networks aim to satisfy the demand for high bandwidth, low-cost, and ubiquitous Internet access: wireless mesh networks (WMNs). However, in order to reach a greater market penetration a fundamental factor must yet to be addressed: the ease of use. Except for small networks, manual configuration is infeasible, so robust autoconfiguration mechanisms are needed. To address this shortcoming of complicated setup procedures we introduce and evaluate a novel autoconfiguration protocol for WMN: the Dynamic WMN Configuration Protocol (DWCP). It deals with the assignment of unique addresses, the management of free and assigned addresses, the autonomous reaction to failures, and features support of conventional Dynamic Host Configuration Protocol (DHCP) clients. It is specifically suitable for large installations and was deployed and evaluated in a real testbed.

During the last decades TCP and the networks it is used in steadily evolved. To aid further development it is crucial to give researchers measurement tools so they can evaluate and analyze their TCP modifications in real world network environments.In this paper we introduce the new measurement tool flowgrind. Unlike existing measurement tools, flowgrind's distributed architecture allows for an easy setup of complex scenarios. Besides the usual application perceived metrics it can also measure core variables from the operating system's TCP implementation thus enabling the researchers to analyze and understand the interactions between TCP and the underlying network. I. MOTIVATIONOver the last two decades the Internet evolved from a small and well defined research network, to a global and essential infrastructure for hundreds of millions of users. In todays Internet, all kind of application and services are used over all kind of networks, a diversity never foreseen by the original designers of the TCP/IP protocol suite. Devices are connected over low bandwidth voiceband modems, high delay satellite links, lossy wireless multi-hop networks or high speed optical fiber. And the most used transport protocol TCP is used for financial transactions, interactive command line sessions, video streaming and DVD sized file downloads. Of course the TCP of 1981 as standardized in RFC 793, had to be heavily extended and modified (and still struggles) to support this vast diversity of networks and applications.As a result a modern TCP implementation on the one hand has to cope with a huge number of permutations to find the best possible denominator of TCP variants for two communicating end-hosts, but also includes various tunables to adjust its complex behavior. While networks, applications and TCP stacks steadily evolved, measurement tools to evaluate and analyze the performance of TCP in a network did fall behind. This gives rise to the need for new performance measurement tools capable of providing metrics needed to analyze the effects of any new algorithms, while allowing an easy setup for complex scenarios in new types of edge networks like wireless mesh networks (WMNs).The remainder of this paper is structured as follows: In Section II existing network performance measurement tools are discussed. Section III describes our new measurement tool flowgrind in detail. Some example measurements performed in our WMN testbed are presented in Section IV. Section V concludes the work. II. RELATED WORK A. IperfA relatively simple tool, based on a client-server architecture, to measure goodput is Iperf [1]. To perform a measurement, an Iperf client connects to an Iperf server and exchanges a few test parameters, after which the bulk data transfer starts. Unfortunately, Iperf can only test against one single server at a time. Iperf measures unidirectional TCP and UDP goodput, either sent from the client to the server or vice versa. If testing the TCP goodput, data can optionally be sent in parallel with multiple connections to the same se...

No abstract

Disruptions in end-to-end path connectivity, which last longer than one retransmission timeout, cause suboptimal TCP performance. The reason for this performance degradation is that TCP interprets segment loss induced by long connectivity disruptions as a sign of congestion, resulting in repeated retransmission timer backoffs. This, in turn, leads to a delayed detection of the re-establishment of the connection since TCP waits for the next retransmission timeout before it attempts a retransmission. This document proposes an algorithm to make TCP more robust to long connectivity disruptions (TCP-LCD). It describes how standard ICMP messages can be exploited during timeout-based loss recovery to disambiguate true congestion loss from non-congestion loss caused by connectivity disruptions. Moreover, a reversion strategy of the retransmission timer is specified that enables a more prompt detection of whether or not the connectivity to a previously disconnected peer node has been restored. TCP-LCD is a TCP senderonly modification that effectively improves TCP performance in the case of connectivity disruptions. Status of This Memo This document is not an Internet Standards Track specification; it is published for examination, experimental implementation, and evaluation. This document defines an Experimental Protocol for the Internet community. This document is a product of the Internet Engineering Task Force (IETF). It represents the consensus of the IETF community. It has received public review and has been approved for publication by the Internet Engineering Steering Group (IESG). Not all documents approved by the IESG are a candidate for any level of Internet Standard; see Section 2 of RFC 5741. Information about the current status of this document, any errata, and how to provide feedback on it may be obtained at http://www.rfc-editor.org/info/rfc6069.