Dirceu Cavendish

ATM rate based congestion control using a Smith predictor: an EPRCA implementation

Presents a feedback control algorithm for ATM congestion control in which source rates are adjusted according to VC queue lengths at intermediate nodes along the path. The goal is to fill in the residual bandwidth, without exceeding a specified queue threshold. In order to obtain this, we propose a simple and classical proportional controller, plus a Smith predictor to overcome instabilities due to large propagation delays, as well as to avoid cell loss. We propose an effective EPRCA implementation in which each source computes its input rate based on the maximum VC queue length along the path. Theoretical and experimental results show that high throughput is achieved even with queue sizes independent of the round trip delay.

A control theoretical approach to congestion control in packet networks

In this paper, we introduce a control theoretical analysis of the closed-loop congestion control problem in packet networks. The control theoretical approach is used in a proportional rate controller, where packets are admitted into the network in accordance with network buffer occupancy. A Smith Predictor is used to deal with large propagation delays, common to high speed backbone networks. The analytical approach leads to accurate predictions regarding both transients as well as steady-state behavior of buffers and input rates. Moreover, it exposes tradeoffs regarding buffer dimensioning, packet loss, and throughput.

Routing and wavelength assignment in WDM mesh networks

We study routing and wavelength assignment (RWA) tasks in WDM mesh networks with wavelength conversion capabilities. For each lightpath request, we solve the routing and wavelength assignment sub-problems as a single problem, rather than separating them into two separate problems. We use four different optimization criteria: (1) minimum number of wavelength conversions; (2) minimum number of wavelengths used; (3) minimum hop count; (4) minimum use of scarce resources (trunking). Each of these problems is solved by a Dijkstra type of algorithm, executed sequentially on one request at a time. We also develop an integer linear programming (ILP) model with the objective of minimizing the number of wavelength conversions. The optimal ILP solution can only be obtained for small networks. We conduct a comprehensive simulation study of the performance of the algorithms for different network topologies and traffic scenarios.

CapStart: An Adaptive TCP Slow Start for High Speed Networks

In this document, we introduce Cap Start TCP, an adaptive Slow Start scheme that consistently achieves fast TCP file transfer times regardless of high speed network scenario. Once the TCP session is established, we estimate TCP session path capacity scenario, and tune the transport protocol to deliver fast transfer times. We demonstrate significant transaction performance improvements, of as much as three times faster completion times in transcontinental high speed network experiments for various capacity scenarios.

Distributed XML processing: Theory and applications

Basic message processing tasks, such as well-formedness checking and grammar validation, common in Web service messaging, can be off-loaded from the service providers own infrastructures. The traditional ways to alleviate the overhead caused by these tasks is to use firewalls and gateways. However, these single processing point solutions do not scale well. To enable effective off-loading of common processing tasks, we introduce the Prefix Automata SyStem - PASS, a middleware architecture which distributively processes XML payloads of web service SOAP messages during their routing towards Web servers. PASS is based on a network of automata, where PASS-nodes independently but cooperatively process parts of the SOAP message XML payload. PASS allows autonomous and pipelined in-network processing of XML documents, where parts of a large message payload are processed by various PASS-nodes in tandem or simultaneously. The non-blocking, non-wasteful, and autonomous operation of PASS middleware is achieved by relying on the prefix nature of basic XML processing tasks, such as well-formedness checking and DTD validation. These properties ensure minimal distributed processing management overhead. We present necessary and sufficient conditions for outsourcing XML document processing tasks to PASS, as well as provide guidelines for rendering suitable applications to be PASS processable. We demonstrate the advantages of migrating XML document processing, such as well-formedness checking, DTD parsing, and filtering to the network via event driven simulations.

Is it a good idea to design WDM networks to minimize the number of wavelengths used

In this paper, we study the routing and wavelength assignment (RWA) tasks in WDM mesh networks with wavelength conversion capabilities. We solve the routing and wavelength assignment sub-problems as a single problem, rather than separating them into two separate problems. We use two different optimization criteria: (1) minimize the number of wavelength conversions used; and (2) minimize the number of wavelengths used. Each of these problems is solved by using a method based on the Dijkstra algorithm, sequentially on one connection at a time. We study and comment on the performance of the algorithms for different network topologies and traffic scenarios.

Capacity and Congestion Probing: TCP Congestion Avoidance via Path Capacity and Storage Estimation

We introduce a transmission control protocol with a delay based congestion avoidance. The protocol design follows a control theoretical approach, uses session path estimators, and is adaptive and stable to a wide range of path characteristics and network conditions. We demonstrate the competitive performance of the protocol via open source based network experiments over a research network and the Internet.

Conservative Slow Start: Controlling Losses in Very High Speed Networks

In this paper, we address the problem of how TCP sessions ramp up their transmission windows in a controlled way. We introduce a conservative slow start scheme that reduces to the regular slow start when session path queues are empty, but slows down the control window increase speed as buffers start to build up. We show via event driven simulation and open source based high speed experimental testbed the effectiveness of our conservative slow start mechanism in reducing packet losses and consequent retransmissions.

MVA Modeling of Multi-core Server Distributed Systems

In this paper, we propose an extension to our previous MVA based methodology for estimating performance of transactions executed on multi-server systems for multicoreservers. The extension is based on the characterization of message processing service times for each server core under zeroloadconditions, and building an MVA model that accounts for each available core. Core utilization is characterized, as well as message routing probabilities within the multi-core machine. We illustrate the extended methodology on a prototype multi-server system.

Transmission Control Protocols Evaluation over Satellite Networks

We evaluate various TCP transport protocols on paths over a satellite link. More specifically, we evaluate Reno, Cubic, Compound TCP, Hybla, and a newly proposed Capacity and congestion probing TCP on file transfer applications. Our results show that a single transport protocol delivering best performance on various network scenarios does not yet exist.