Krister Jacobsson

Window Flow Control: Macroscopic Properties from Microscopic Factors

This paper studies window flow control focusing on bridging the gap between microscopic factors such as burstiness in sub-RTT timescales, and observable macroscopic properties such as steady state bandwidth sharing and flow level stability. Using new models, we analytically capture notable effects of microscopic behavior on macroscopic quantities. For loss-based protocols, we calculate the loss synchronization rate for different flows and use it to quantitatively explain the unfair bandwidth sharing between paced and unpaced TCP flows. For delay-based protocols, we show that the ratios of round trip delays are critical to the stability of the system. These results deepen the fundamental understanding of congestion control systems. Packet level simulations are used to verify our theoretical claims.

Queue dynamics with window flow control

This paper develops a new model that describes the queueing process of a communication network when data sources use window flow control. The model takes into account the burstiness in sub-round-trip time (RTT) timescales and the instantaneous rate differences of a flow at different links. It is generic and independent of actual source flow control algorithms. Basic properties of the model and its relation to existing work are discussed. In particular, for a general network with multiple links, it is demonstrated that spatial interaction of oscillations allows queue instability to occur even when all flows have the same RTTs and maintain constant windows. The model is used to study the dynamics of delay-based congestion control algorithms. It is found that the ratios of RTTs are critical to the stability of such systems, and previously unknown modes of instability are identified. Packet-level simulations and testbed measurements are provided to verify the model and its predictions.

ACK-clock Dynamics in Network Congestion Control - An Inner Feedback Loop with Implications on Inelastic Flow Impact

The focus of this paper is the window mechanism in network congestion control, whereby packet acknowledgments control when new packets are being sent. This constitutes an inner control loop that so far has received little attention. We provide a novel model of this loop that bridges between the standard integrator link model and the more recent static link model. The model is in validation experiments shown to be accurate. It is also shown that as the amount of inelastic cross-traffic increases the dynamics of this inner loop becomes slower. This may influence overall performance and stability in scenarios with heavy inelastic flows

ACK-Clocking Dynamics: Modelling the Interaction between Windows and the Network

A novel continuous time fluid flow model of the dynamics of the interaction between ACK-clocking and the link buffer is presented. A fundamental integral equation relating the instantaneous flow rate and the window dynamics is derived. Properties of the model, such as well-posedness and stability, are investigated. Packet level experiments verify that this new model is more accurate than existing models, correctly predicting qualitatively different behaviors, for example when round trip delays are heterogeneous.

Implementation of provably stable maxnet

MaxNet TCP is a congestion control protocol that uses explicit multi-bit signalling from routers to achieve desirable properties such as high throughput and low latency. In this paper we present an implementation of an extended version of MaxNet. Our contributions are threefold. First, we extend the original algorithm to give both provable stability and rate fairness. Second, we introduce the MaxStart algorithm which allows new MaxNet connections to reach their fair rates quickly. Third, we provide a Linux kernel implementation of the protocol. With no overhead but 24-bit price signals, our implementation scales from 32 bit/s to 1 peta-bit/s with a 0.001% rate accuracy. We confirm the theoretically predicted properties by performing a range of experiments at speeds up to 1 Gbit/sec and delays up to 180 ms on the WAN-in-Lab facility.

Stability and robustness conditions using frequency dependent half planes

This paper presents a sufficient condition that establishes closed loop stability for linear time invariant dynamical systems with transfer functions that are analytic in the open right half complex plane. The condition is suitable for analyzing a large class of highly complex, possibly interconnected, systems. The result is based on bounding Nyquist curves by using frequency dependent half planes. It provides (usually non-trivial) robustness guarantees for the provably stable systems and generalizes to the multidimensional case using matrix field of values. Concrete examples illustrate the applications of the condition. From our condition, it is easy to derive a relaxed version of the classical result that the interconnection of a positive real and strictly positive real linear system under feedback is closed loop stable.

CLOSED LOOP ASPECTS OF FLUID FLOW MODEL IDENTIFICATION IN CONGESTION CONTROL

Abstract Fluid flow models have turned out to be instrumental for analysis and synthesis of primal/dual congestion control algorithms which rely on aggregated information from a network path. In particular stability has been analyzed using such models. In network congestion control, validation experiments will with necessity be performed in closed loop since the communication protocol has to be active. Guidelines on how such experiments should be carried out in practice has until now been lacking in the literature. Departing from the theory of modeling for control, we refine a fluid flow model by augmenting the customary model of transport latencies, link price and source control with estimator dynamics and sampling properties. The impact of cross-traffic and changes in network configuration is incorporated as well. Furthermore, we analyze, from a closed-loop perspective, how the network should be excited when validating such models. The resulting identification framework is used for validating the derived model using packet-level experimental data from NS-2 simulations.

Local Analysis of Structural Limitations of Network Congestion Control

Recently there have been a number of interesting contributions to the stability analysis of network congestion control based on fluid models. Here, we further this emerging analysis by studying the structural limitations that so called primal/dual congestion control algorithms impose. Such algorithms rely on aggregated information from a network path, e.g. TCP-Vegas use the aggregated queuing delay. We show through local analysis that this imposes certain limitations of feedback control. Viewed from the source side, the complementary sensitivity and the sensitivity functions are severely restricted when many sources share the same bottleneck. This impose that source control must be small enough to achieve suitable noise rejection. In addition, a specialized congestion control paradigm where all sources share a common time-base is analyzed. For this scenario the analysis facilitates significantly and robustness limitations towards configuration changes is observed.

UNBIASED BANDWIDTH ESTIMATION IN COMMUNICATION PROTOCOLS

Abstract Heterogeneous communication networks with their variety of application demands, uncertain time-varying traffic load, and mixture of wired and wireless links pose several challenging problem in modeling and control. In this paper we focus on bandwidth estimation and elucidate why estimates based directly on bandwidth samples are biased. Previously, this phenomenon has been observed but not properly explained, it seems. Standard techniques for bandwidth estimation are based on measurements of inter-arrival times of packets as the bandwidth is proportional to the inverse of the inter-arrival time. Two main classes of bandwidth estimators are analyzed wrt how variations in the inter-arrival times affect the estimates. It is shown that linear time-invariant filtering of instantaneous bandwidth estimates does not change the bias. In contrast to this, smoothing the inter-arrival-time samples does give a bias reduction which depends on the properties of the smoothing filter. Hence, with such approach, noise attenuation can be traded against tracking ability wrt changes in the actual bandwidth.