Search

Orientador: Christian Rodolfo Esteve Rothenberg Dissertação (mestrado) - Universidade Estadual de Campinas, Faculdade de Engenharia Elétrica e de Computação Resumo: Por muitos anos as Redes Definidas por Software (SDN) têm revolucionado o comportamento das redes de computadores, dando aos administradores das mesmas a possibilidade de programar o plano de controle da rede. No entanto, a implantação de soluções SDN deu aos pesquisadores espaço para novos desafios, com o objetivo de atualizar nossas redes por meio de uma programação mais detalhada do plano de dados. O P4 é uma Linguagem de Domínio Específico (DSL) para expressar como os pacotes são processados em uma plataforma de rede programável. Considerando o objetivo de permitir a programação P4 com alto desempenho, o Multi-Architecture Compiler System for Abstract Dataplanes (MACSAD) utiliza o projeto open source OpenDataPlane (ODP) para fornecer APIs específicas, permitindo a interoperabilidade entre hardwares diferentes e minimizando a sobrecarga dos mesmos. O MACSAD é um compilador que aproveita a simplicidade da linguagem P4 e a flexibilidade das APIs do ODP para trabalhar em diferentes plataformas, mantendo o alto desempenho. Assim, o MACSAD pode ser chamado de um "sistema de compilador unificado de alto desempenho", considerando que ele pode executar o mesmo programa P4 em múltiplos hardwares com alta performance. Este projeto tem como objetivo adicionar suporte VXLAN ao MACSAD, integrá-lo a um controlador SDN, fazer uma análise de throughput, latência e da distribuição do balanceador de carga através de múltiplos polinômios. Assim, para alcançar essa integração, faremos uma implementação P4 VXLAN com uma abordagem SDN para preencher as tabelas através de um controlador simples. Por fim, faremos uma análise dos métodos de balanceamento de carga, principalmente através de funções Checksum e CRC para uma avaliação de desempenho de todo o sistema. Dessa forma, utilizaremos o Network Function Performance Analyzer (NFPA) e Open Source Network Tester (OSNT) para efetuar os testes de throughput e latência, gerando diferentes tipos de tráfego para an'alise de performance de nosso programa P4 definido na aplicação do plano de dados Abstract: For some years, Software-Defined Networking (SDN) has been revolutionizing the networking landscape, giving administrator users the possibility to program the network control plane. However, the deployment of SDN solutions gave researchers space to new challenges, aiming to upgrade our networks to new levels through deeper data plane programmability. The Programming Protocol-Independent Packet Processors (P4) is a Domain Specific Language (DSL) to express how packets are processed on a programmable network platform. Considering the objective to allow P4 programmability with high performance, the Multi-Architecture Compiler System for Abstract Dataplanes (MACSAD) uses the OpenDataPlane (ODP) Open Source project to provide specific Application Programming Interfaces (APIs), enabling the interoperability between different hardwares and minimizing the overhead. The MACSAD is a compiler that takes advantage of the P4 language simplicity and ODP APIs flexibility to work on different platforms, but still maintaining high performance. Thus, MACSAD can be called as a "unified compiler system with high performance", considering that it can execute the same P4 program on multiple targets with high throughput. This project aims to add Virtual eXtensible Local Area Network (VXLAN) support to MACSAD,integrate it with an SDN controller, evaluate the throughput, latency and the Load balance distribution through multiple polynomials. Thus, to achieve this integration we will make a P4 VXLAN implementation and an SDN approach to populate the tables through a simple controller. Finally, we will analyze different load balancing polynomials, mainly through Checksum and CRC functions and a performance evaluation of the whole system, to perform the last one we will take advantage of Network Function Performance Analyzer (NFPA) and Open Source Network Tester (OSNT), generating different types of traffic to benchmark our P4-defined dataplane application Mestrado Engenharia de Computação Mestre em Engenharia Elétrica Funcamp 35789-17, 78064/2018

Orientador: Christian Rodolfo Esteve Rothenberg Dissertação (mestrado) - Universidade Estadual de Campinas, Faculdade de Engenharia Elétrica e de Computação Resumo: Dentre as novas tendências em programação de dataplane dentro de SDN (Software Defined Networking) destacam-se os esforços para prover um suporte multi-plataforma dotado de alta definição das informações que são processadas pelo pipeline do plano de dados. No entanto, alguns desafios ainda persistem, como a necessidade de um plano de dados programável ou a adoção de uma abstração de programação independente de protocolo. Como forma de mitigar tais problemas, verifica-se que a Linguagem Específica de Domínio~(DSL) Programming Protocol-Independent Packet Processors~(P4) desponta como uma tendência emergente para expressar como os pacotes são processados pelo plano de dados de uma plataforma de rede programável. De modo independente e em paralelo, constata-se que o projeto OpenDataPlane~(ODP) cria um conjunto de plataformas abertas de Application Programming Interfaces~(APIs) projetado para o plano de dados de rede. Isso posto, tem-se que o Multi-Architecture Compiler System for Abstract Dataplanes~(MACSAD) surge como uma abordagem para convergir P4 e ODP em um processo de compilação convencional, arquivando a portabilidade dos aplicativos de plano de dados sem afetar as melhorias de desempenho do alvo. O MACSAD pode integrar a API do ODP e o P4, reunindo-os e definindo um plano de dados programável em um sistema de compilador unificado. Este trabalho tem como objetivo adicionar o suporte do Longest Prefix Match~(LPM) do IPv4/IPv6 ao MACSAD, integrado com as APIs do ODP e à programação P4, oferecendo recursos de planejamento de dados de alto desempenho. O suporte ao LPM proposto para o MACSAD combina o algoritmo de lookup e a biblioteca da API do ODP com o suporte à tabela MACSAD, para criar uma base de encaminhamento completa usada no processo do LPM. A implementação do IPv4 adapta o atual algoritmo de lookup do ODP para trabalhar com o MACSAD. A implementação de lookup IPv6, atualmente não suportada pelo ODP, é uma extensão do suporte IPv4 que é desenvolvido usando o mesmo algoritmo adaptado a uma chave de 128 bits. A pesquisa IPv4 e IPv6 usa uma base de árvore binária para executar o lookup do LPM. Para a avaliação de desempenho do suporte ao LPM, utilizamos uma ferramenta geradora de tráfego Network Function Performance Analyzer~(NFPA) que permite gerar diferentes tipos de tráfego no MACSAD. Cabe ainda destacar, como uma contribuição lateral deste trabalho, o desenvolvimento da ferramenta geradora de pacote BB-Gen, já com lançamento open source. Resultados experimentais mostram que é possível atingir um throughput de 10G com tamanhos de pacotes de 512 bytes ou superiores Abstract: New trends in dataplane programmability inside Software Defined Networking~(SDN) are in efforts to bring multi-platform support with a high definition of the information that is processed by the dataplane pipeline. However, some challenges are still present, as the necessity of a programmable dataplane or a protocol independent programming abstraction. The Programming Protocol-Independent Packet Processors~(P4) Domain Specific Language (DSL) is an emerging trend to express how the packets are processed by the dataplane of a programmable network platform. In parallel, OpenDataPlane~(ODP) project creates an open-source, cross-platform set of Application Programming Interfaces~(APIs) designed for the networking data plane. Multi-Architecture Compiler System for Abstract Dataplanes~(MACSAD) is an approach to converge P4 and ODP in a conventional compilation process, achieving portability of the dataplane applications without affecting the target performance improvements. MACSAD can integrate the ODP API and the P4, bringing them together and defining a programmable dataplane across multiple targets in a unified compiler system. This work aims at adding IPv4/IPv6 Longest Prefix Match~(LPM) support to MACSAD integrated with ODP APIs and P4 programmability delivering high-performance dataplane capabilities. The proposed LPM support for MACSAD combines the lookup algorithm and the ODP API library with MACSAD table support, to create a complete forwarding base used in the LPM process. The IPv4 implementation adapts the current ODP lookup algorithm to work with MACSAD. IPv6 lookup implementation, currently not supported by ODP, is an extension of the IPv4 support, developed using the same algorithm adapted to a 128-bit key. IPv4 and IPv6 lookup use a binary tree base, to perform the LPM lookup. For the performance evaluation of the LPM support, we use a traffic generator tool Network Function Performance Analyzer~(NFPA) that allows generating different types of traffic across MACSAD. A side contribution on this front we developed and released open source the BB-Gen packet crafter tool. Experimental results show that it is possible to reach a throughput of 10G with packets sizes of 512 Bytes and above Mestrado Engenharia de Computação Mestre em Engenharia Elétrica

Energy efficiency is of paramount importance in designing low-power wireless sensor nodes. Approximate computing is a new circuit-level technique for reducing power consumption. However, the gain in power by applying this technique is achieved at the cost of computational errors. The impact of such inaccuracies in the circuit level of a radio transceiver chip on the performance of Wireless Sensor Networks has not yet been explored. The applicability of such low-power chip design techniques depends on the overall energy gain and their impact on the network performance. In this paper, we analyze various inaccuracy fields in a radio chip, and quantify their impact on the network performance, in terms of packet latency, goodput, and energy per bit. The analysis is supported by extensive network simulations. The outcome can be used to investigate in which WSN application scenarios such power reduction techniques at circuit level can be applied, given the network performance and energy consumption requirements. © 2018 Association for Computing Machinery.

Energy efficiency is of paramount importance in designing low-power wireless sensor nodes. Approximate computing is a new circuit-level technique for reducing power consumption. However, the gain in power by applying this technique is achieved at the cost of computational errors. The impact of such inaccuracies in the circuit level of a radio transceiver chip on the performance of Wireless Sensor Networks has not yet been explored. The applicability of such low-power chip design techniques depends on the overall energy gain and their impact on the network performance. In this paper, we analyze various inaccuracy fields in a radio chip, and quantify their impact on the network performance, in terms of packet latency, goodput, and energy per bit. The analysis is supported by extensive network simulations. The outcome can be used to investigate in which WSN application scenarios such power reduction techniques at circuit level can be applied, given the network performance and energy consumption requirements. © 2018 Association for Computing Machinery.

Wireless redundant networks are expected to play a fundamental role to backhaul dense LTE networks. In these scenarios, backpressure-based routing strategies such as BP-MR can exploit the network redundancy. In this paper, we perform an exhaustive performance evaluation of different TCP variants over an LTE access network, backhauled by various routing protocols (including per-packet and per-flow BP-MR variants and a static alternative, OLSR) over two different wireless topologies: a regular mesh and an irregular ring-tree topology. We compare the performance of different TCP congestion control algorithms based on loss (NewReno, Cubic, Highspeed, Westwood, Hybla, and Scalable) and delay (Vegas) under different workloads. Our extensive analysis with ns-3 on throughput, fairness, scalability and latency reveals that the underlying backhaul routing scheme seems irrelevant for delay-based TCPs, whereas per-flow variant offers the best performance irrespective of any loss-based TCP congestion control, the most used in the current Internet. We show that BP-MR per-flow highly reduces the download finish time, if compared with OLSR and BP-MR per-packet, despite showing higher round-trip-time.

Emerging cloud-based applications, running in geographically distributed data centers (DCs), generate new dynamic traffic patterns which claim for a more efficient management of the traffic flows. Geographically distributed DCs interconnection requires automatic and more dynamic provisioning and deletion of end-to-end (E2E) connectivity services, through heterogeneous network domains. Each network domain may use a different data transport technology but also a different control/management system. The fast development of Software Defined Networking (SDN) and the interworking with current control plane technologies such as Generalized Multi-protocol Label Switching (GMPLS), demand orchestration over the heterogeneous control instances to provide seamless E2E connectivity services to external applications (i.e. Cloud Computing applications). In this work, we present different orchestration architectures based on the SDN principles which use the Path Computation Element (PCE) as a fundamental component. In particular, a single SDN controller orchestration approach is compared with an orchestration architecture based on the Application Based Network Operations (ABNO) defined within the International Engineering Task Force (IETF), in order to find the potential benefits and drawbacks of both architectures. Finally, the SDN IT and Network Orchestration (SINO) platform which integrates the management of Cloud Computing infrastructure with the network orchestration, it is used to validate both architectures by evaluating their performance providing two inter-DC connectivity services: E2E connectivity and Virtual Machine (VM) migration. (C) 2016 Elsevier B.V. All rights reserved.

In this paper, a new multi-hop routing protocol (called ORACE-Net) for disaster and emergency networks is proposed. The proposed hierarchical protocol creates an ad-hoc network through body-To-body (B2B) communication between rescue members and the command center. The on-body coordinators establish forward routes towards a command center, whereas, the command center builds reverse routes towards these nodes. The Routing tables are optimized based on real-Time end-To-end Link Quality Estimation (LQE) metrics (i.e., end-To-end signal strength level and end-To-end hop count). We evaluate our proposed protocol with other widely used protocols in the disaster context which are covered by the routing classes (i.e., Reactive, Proactive and Geographic-based). The evaluation is based on a realistic disaster mobility trace. The results show that the proposed protocol outperforms the other studied protocols in terms of packet reception rate and energy consumption. The proposed ORACE-Net protocol increases the body-To-body network lifetime and reliability. This publication was made possible by NPRP grant #[6 1508 2 616] from the Qatar National Research Fund (a member of Qatar Foundation). Scopus

In this paper, we consider a data-flow management mechanism for data center networks, in which a centralized controller called arbiter manages data flows. We propose a simple algorithm for the arbiter to distribute flows over different time points and paths, in a preemptive scheduling and traffic load-balancing manner. The proposed algorithm is based on table-driven resource reservation, in which states of all the links in a data-center network are registered in a single table, and its information is updated whenever a new flow-request arrives at the arbiter. We evaluate the performance of the proposed algorithm through simulation experiments, investigating bit allocation rate and flow allocation rate, under different flow-size distributions. Numerical results show that the proposed algorithm can achieve high bit allocation rate without collisions. It is also shown that the proposed algorithm can allocate many flows within a small time interval even when the variance of flow-size is large., APCC 2016 : 22nd Asia-Pacific Conference on Communications , Aug 25-27, 2016 , Yogyakarta, Indonesia

From the perspective of resource allocation and routing, this paper aims to save as much energy as possible in data center networks. We present a general framework, based on the blocking island paradigm, to try to maximize the network power conservation and minimize sacrifices of network performance and reliability. The bandwidth allocation mechanism together with power-aware routing algorithm achieve a bandwidth guaranteed tighter network. Besides, our fast efficient heuristics for allocating bandwidth enable the system to scale to large sized data centers. The evaluation result shows that up to more than 50% power savings are feasible while guaranteeing network performance and reliability. NPRP grant No.6-718-2-298 from the Qatar National Research Fund. Scopus

This paper investigates OpenFlow performance, focusing on how the delay between switch and OpenFlow controller can impact the performance of a network. We use open-source controllers that support network virtualization to evaluate how such delay impacts ICMP, TCP and UDP traffic. We compare the controllers’ flow set-up strategies and we conduct several experiments to compare their TCP and UDP performance. In ad-dition, we introduce a new metric to measure UDP packet losses at the beginning of the flow. The results of the measurements indicate that there are large differences in performance between controllers, and that performance depends on switch-controller delay and flow set-up strategy. QC 20141209

The characterization of flows duration and their handover metrics is of paramount importance for the design and resource management of mobile networks. Elastic flows duration characterization is more involved than for streaming sessions, as it depends on the network load. In this work, the distribution of flows duration is characterized by a phase type distribution. The model is then applied to determine the handover probability under different cell residence time and flow size distributions. Simulation results confirm that the proposed model yields very accurate results under realistic assumptions. © 2012 IEEE., This work was supported by the Spanish Government through projects TIN2008-06739-C04-02 and TIN2010-21378-C02-02.

In order for service providers to provide their users high quality services in the home network, Quality of Service (QoS) provisioning is needed to protect premium services. In this paper, we describe how a Universal Plug-and-Play (UPnP) based home network architecture solves this problem in a heterogeneous home network. We outline how it both relieves the end user from troublesome configuration and still offers control to the service provider. We particularly present performance assessment results for UPnP-QoS v3, based on a fully operational experimental implementation. The quantitative measurement results are further used in extensive simulations demonstrating acceptable response times and clear QoS admission differentiation.

There has been a spectacular growth in the use of wireless networks in recent times and consequently, adapting TCP to the wireless networks is a hot topic of current research. However, most of the existing works proposed for this problem have been designed for specific wireless networks, or they necessitate changes at either the receiver or the intermediate nodes, or at both, because of which their deployment becomes difficult. In this work, we propose a TCP variant which works over both multi-hop ad hoc wireless networks as well as single-hop (last-hop) wireless networks, like Wireless LANs, cellular, and satellite networks. We use a learning based method to dynamically change the congestion window size according to the network conditions. Our protocol does not rely on any explicit feedback from the network and requires only sender-side modifications. Through extensive simulations we show that our protocol achieves the desired goals of performance improvement in goodput, reduction in packet loss, and fairness to the competing flows. To the best of our knowledge, this is the first unified solution for both single-hop and multi-hop wireless networks. � 2009 Springer Science+Business Media, LLC.

Future wireless communication systems are expected to provide more stable and higher data rate transmissions in the whole network. WiMAX is one of the technologies to cover diverse geographic regions. WiMAX covers the high dense urban areas and provide access over large geographic regions in remote rural area. However, in both scenarios it is challenging to cover the entire service area. In both cases relays are foreseen to extend the range of the BS and to increase the system throughput allowing for a cost-efficient deployment and service. As multi-hop/relay has complex interactions between the routing and medium access control decisions, the extent to which analytical expressions can be used to explore its benefits are limited. Consequently simulations tend to be the preferred way of assessing the performance of relays. In this paper, we explain, how to deploy fixed-relay and mobile relay by using computer simulation in WiMAX system and we consistently observed that the throughput is increased and the outage is decreased in the relay-augmented WiMAX network, which is converted to range extension without any capacity penalty, for the realistic range of values of the propagation and other system parameters investigated.