In this paper we investigate the impact of packet reordering on the performance of high-speed protocols. Our results show that even small fraction of packet reordering can severely impair the performance of these protocols. We then investigate the benefits of using delayed congestion response (TCP-DCR) with one of the high-speed protocols (LTCP). Our results indicate that the benefits in terms of avoiding performance degradation is significant, even at very high levels of packet reordering. In the absence of any packet reordering, the protocol behavior in terms of fairness among competing flows or impact on bottleneck link drop rates is remains unmodified. 1

The increase in link speeds, increased parallelism within routers and switches, QoS support and load balancing among links, all point to future networks with increased packet reordering. Unchecked, packet reordering will have a significant detrimental effect on the end-to-end performance, while resources required for dealing with packet reordering at routers and end-nodes will grow considerably. A formal analysis of packet reordering is carried out and Reorder Density (RD) metric is defined for measurement and characterization of packet reordering. RD captures the amount and degree of reordering, and can be used to define the reorder response of networks under stationary conditions. Properties of RD are derived, and it is shown that the reorder response of the network formed by cascading two subnets is equal to the convolution of the reorder responses of individual subnets. Packet reordering over the Internet is measured and used to validate the derivations. 1

This paper examines a new building block for next-generation networks: SNAPP, or Stateless Network-Authenticated Path Pinning. SNAPP-enabled routers securely embed their routing decisions in the packet headers of a stream of traffic, effectively pinning a flow’s path between sender and receiver. A sender can use the pinned path (even if routes subsequently change) by including the path embedding in later packet headers. This architectural building block decouples routing from forwarding, which greatly enhances the availability of a path in the face of routing misconfigurations or malicious attacks. To demonstrate the extreme flexibility of SNAPP, we show how it can support a wide range of applications, including sender-controlled paths, expensive route lookups, sender anonymity, and sender accountability. Our analysis shows that SNAPP’s overhead is low, and the system is easily implemented in hardware. We believe that SNAPP is a worthy addition to the network architect’s toolbox, enabling a variety of new designs and trade-offs.

Abstract—In this paper, we consider downward and upward vertical handovers in integrated wireless LAN and cellular networks, and address Wireless Profiled TCP (WP-TCP) premature timeouts due to step increase of round-trip time and false fast retransmit due to packet reordering. Specifically, we develop a mobile receiver centric loosely coupled cross-layer design, which is easy to implement and deploy, backward compatible with the Wireless Application Protocol version 2 (WAP 2.0) architecture, and robust in the absence of perfect cross-layer information. We propose two proactive schemes called RTT Inflation and RTT Equalization. The RTT Inflation scheme suppresses premature timeouts by carefully inflating retransmission timeout time and the RTT Equalization scheme prevents false fast retransmit by equalizing the round-trip delay experienced by all packets. We conduct extensive simulations to evaluate the performance in downward and upward vertical handovers. It is demonstrated that the proposed schemes significantly improve the performance in a wide range of network conditions. Index Terms — Cross-layer design, wireless profiled TCP, vertical handover, integrated wireless networks.

Commercially available routers typically have a monolithic operating system that cannot be easily tailored and upgraded to support new network protocols. This paper describes PromethOS, a modular router architecture based on Linux 2.4, which can be dynamically extended by plugin modules that are installed in the networking kernel. To install and configure plugins remotely we present a novel signaling protocol that establishes explicitly routed paths transiting selected nodes in a predefined order. Such paths can be non-simple, where a given node is being visited more than once.

Abstract—A sudden delay spike or reordering in the network can cause TCP to experience a loss event. Since loss is interpreted as a sign of congestion in TCP, this causes the protocol to reduce its sending rate. Several mechanisms for detecting and reacting to such spurious loss events have been proposed; each of them has some advantages and disadvantages. We extend this space with a new detection mechanism which complements the existing ones well. Furthermore, the mechanism is easy to implement because it only needs the sender to intelligently interpret feedback from ECN nonce. I.

I am fortunate to have had Professor Paul Amer as my Ph.D. advisor. His ability to ask the right questions and his attention to detail have never ceased to surprise me. He was always able to make time for me, even if I just walked into his office without any notice. His constant emphasis on clear communication and presentation have helped me significantly. Over the years, as I have gained a deeper appreciation of an advisor’s responsibilities, my respect for Prof. Amer has only grown. I have a lot to thank him for. I thank my dissertation committee members: Prof. Adarshpal Sethi, Prof. Stephan Bohacek, Prof. Phillip Conrad, and Randall Stewart for providing valuable direction and feedback. I specifically want to thank Randy for travelling great distances to be on my committee, and spending much time with me over the years. His inputs and insights have been invaluable. I have had the good fortune of interacting and working with some really wonderful people

Abstract – Packet reordering is an inevitable phenomenon on the Internet. An ideal metric for packet reordering should capture reordering accurately, provide insight into nature of reordering, and help in evaluation and analysis of reordering leading to mitigation of its adverse effects. Proposed metrics for

allows for the creation of networks in which data bursts are switched optically at each node, offering a greater degree of flexibility suitable for handling bursty Internet traffic. TCP-based applications account for a majority of data traffic in the Internet; thus understanding and improving the performance of TCP implementations over OBS networks is critical. Previously, several articles show that load-balanced routing improves loss-performance in OBS. In this paper, we identify the ill-effects of load-balanced OBS on TCP performance caused by false time-outs and false fastretransmits. We propose a source-ordering mechanism that significantly improves TCP throughput over a loadbalanced OBS network. Keywords Load-balancing · TCP · OBS 1

Abstract—Multi-interface mobile devices and multihomed residential Internet connections are becoming commonplace. However, standard transport protocols TCP and SCTP are unable to take advantage of several available paths so that the application using a single transport connection would receive the aggregate bandwidth of all paths. Multihoming and advanced security features make the Host Identity Protocol a good candidate to provide multipath data delivery. In this paper, we design and implement a multipath scheduler that distributes the incoming traffic among multiple available paths. Using Fastest Path First scheduling, packets from a single TCP connection could be spread to multiple paths with no reordering. Our simulations confirm effectiveness and TCP-friendliness of multipath transfer for a range of path bandwidths and in the presence of cross-traffic. 1