Jay Aikat

Variability in TCP round-trip times

We measured and analyzed the variability in round trip times (RTTs) within TCP connections using passive measurement techniques. We collected eight hours of bidirectional traces containing over 22 million TCP connections between end-points at a large university campus and almost

The effects of active queue management on web performance

1

Simulating large-scale airborne networks with ns-3

million remote locations. Of these, we used over 1 million TCP connections that yield 10 or more valid RTT samples, to examine RTT variability within a TCP connection. Our results indicate that contrary to observations in several previous studies, RTT values within a connection vary widely. Our results have implications for designing better simulation models, and understanding how round trip times affect the dynamic behavior and throughput of TCP connections.

The Continued Evolution of Web Traffic

We present an empirical study of the effects of active queue management (AQM) on the distribution of response times experienced by a population of web users. Three prominent AQM schemes are considered: the Proportional Integrator (PI) controller, the Random Exponential Marking (REM) controller, and Adaptive Random Early Detection (ARED). The effects of these AQM schemes were studied alone and in combination with Explicit Congestion Notification (ECN). Our major results are: <ol>For offered loads up to 80% of bottleneck link capacity, no AQM scheme provides better response times than simple drop-tail FIFO queue management. For loads of 90% of link capacity or greater when ECN is not used, PI results in a modest improvement over drop-tail and the other AQM schemes. With ECN, both PI and REM provide significant response time improvement at offered loads above 90% of link capacity. Moreover, at a load of 90% PI and REM with ECN provide response times competitive to that achieved on an unloaded network. ARED with recommended parameter settings consistently resulted in the poorest response times which was unimproved by the addition of ECN.</ol.We conclude that without ECN there is little end-user performance gain to be realized by employing the AQM designs studied here. However, with ECN, response times can be significantly improved. In addition it appears likely that provider links may be operated at near saturation levels without significant degradation in user-perceived performance.We present an empirical study of the effects of active queue management (AQM) on the distribution of response times experienced by a population of web users. Three prominent AQM schemes are considered: the Proportional Integrator (PI) controller, the Random Exponential Marking (REM) controller, and Adaptive Random Early Detection (ARED). The effects of these AQM schemes were studied alone and in combination with Explicit Congestion Notification (ECN). Our major results are: For offered loads up to 80% of bottleneck link capacity, no AQM scheme provides better response times than simple drop-tail FIFO queue management. For loads of 90% of link capacity or greater when ECN is not used, PI results in a modest improvement over drop-tail and the other AQM schemes. With ECN, both PI and REM provide significant response time improvement at offered loads above 90% of link capacity. Moreover, at a load of 90% PI and REM with ECN provide response times competitive to that achieved on an unloaded network. ARED with recommended parameter settings consistently resulted in the poorest response times which was unimproved by the addition of ECN. We conclude that without ECN there is little end-user performance gain to be realized by employing the AQM designs studied here. However, with ECN, response times can be significantly improved. In addition it appears likely that provider links may be operated at near saturation levels without significant degradation in user-perceived performance.

Analysis of Topology Algorithms for Commercial Airborne Networks

Large-scale airborne networks, which connect airborne nodes with high-bandwidth communication links are being actively pursued commercially. We propose utilizing thousands of operational passenger and cargo aircraft as the principal components of an airborne network which could provide high-speed Internet to passengers on-board and on the ground. To simulate such a network we have augmented the ns-3 network simulator with a model for ingesting and processing aircraft position information, a steerable directional antenna model, a wireless point-to-point channel and associated net devices, and a distributed topology control application to manage the topology of the mesh network. We describe our implementation of these models and some tools for visualizing airborne networks. Using a simulation of a large airborne network, covering the United States, we perform experiments to evaluate the effectiveness of using the Optimized Link State Routing Protocol (OLSR) to route network traffic. Our simulations lead us to conclude that OLSR is likely not a good fit for our envisioned network.

Towards Traffic Benchmarks for Empirical Networking Research: The Role of Connection Structure in Traffic Workload Modeling

Over the last decade web content has evolved from relatively static pages often delivered by one or two servers, to websites rich with interactive media content served from numerous servers. This content change has affected the associated network traffic. Quantifying and analyzing these changes can lead to updated traffic models and more accurate web traffic simulations for testing new protocols and devices. In this work we analyze the TCP/IP headers in packet traces collected at various times over 13 years on the link that connects the University of North Carolina at Chapel Hill (UNC) to its ISP. We show that while the decade-old methodology for inferring web activity from these packet traces is still viable, it is no longer possible to infer all page boundaries given only the TCP and IP headers. We propose a novel method for segmenting web traffic into Activity Sections, in order to obtain comparable higher level statistics. Using these methods to analyze our data set, we describe trends in the HTTP request and response sizes, and a trend towards longer connection durations. We also show that the number of servers supporting web activity has increased, and present empirical evidence that suggests the number of unused connections has risen, likely due to new speculative TCP preconnect features of popular browsers.

Rethinking Security in the Era of Cloud Computing

Civilian Airborne Networks capable of providing network connectivity to users onboard aircraft and users on the ground may soon be viable. We propose a novel airborne network architecture consisting of commercial aircraft and ground station gateways inter-connected with Free-Space Optical Communications (FSOC) links to form a high-bandwidth mesh network. The use of directional FSOC links necessitates explicit topology control, where a protocol must manage which links will point at one another to form connections. The algorithm used by the topology control protocol to form topologies must have low computation time, and must compute topologies that are robust, inclusive, and contain short paths between nodes. We use FAA flight path data for the aircraft en route within the continental United States during a 24-hour period to analyze the properties of airborne mesh networks, and compare candidate topology algorithms. In our simulation an airborne network with ground stations was able to continuously connect over 98% of the air-craft into a mesh network using FSOC links. We propose two new topology algorithms (DCTRT and DCKruskal+Long) which are extensions to existing algorithms. DCKruskal+Long appears to perform best for the metrics measured.

Geographic Routing in Extreme-Scale Highly-Dynamic Mobile Ad Hoc Networks

Networking research would be well served by the adoption of a set of traffic benchmarks to model network applications for empirical evaluations; such benchmarks are common in many other areas of computing. While it has long been known that certain aspects of modeling traffic, such as round trip time, can dramatically affect application and network performance, there is still no agreement as to how such components should be controlled within an experiment. In this paper we advance the discussion of standards for empirical networking research by demonstrating how certain components of network traffic, such as the structure of application data exchanges within a TCP connection, can have a larger impact on the results obtained through experimentation than other dimensions of traffic such as round-trip time. Such findings point to the pressing need for traffic benchmarks in networking research. Through testbed experiments performed with synthetically generated network traffic from two very different traffic sources, and using several models of TCP connection structure, we demonstrate the strong effects of connection structure in traffic workload modeling on performance measures such as queue length at routers, number of active connections in the network, user response times, and connection durations.

Modeling, identifying, and simulating Dynamic Adaptive Streaming over HTTP

Cloud computing has emerged as a dominant computing platform for the foreseeable future, resulting in an ongoing disruption to the way we build and deploy software. This disruption offers a rare opportunity to integrate new approaches to computer security. The aggregating effect of cloud computing and the role of cloud providers as trust anchors can significantly benefit computing security.

The Effects of Traffic Structure on Application and Network Performance

In the near future extremely large scale mobile adhoc networks of thousands or tens of thousands of mobile nodes will be physically feasible and desirable for a host of applications. However, routing within these networks is challenging, especially at high data rates and when node movement is highly-dynamic. In this work we present Topology Aware Geographic Routing(TAG), a position-based routing protocol that strategically uses local topology information (when available) to make better local forwarding decisions, decreasing the number of hops required to deliver a packet when compared with other geographic routing protocols. In addition, TAG is able to reliably deliver packets even in topologies that violate the often used but unrealistic unit disk graph and quasi-static assumptions. We present empirical results from a variety of simulations, illustrating how TAG outperforms GOAFR+, GFG, and OLSR in both theoretical environments and in a simulated, real-world, continental-scale airborne network.