SLBN: A Scalable Max-min Fair Algorithm for Rate-Based Explicit Congestion Control

The growth of the Internet has increased the need for scalable congestion control mechanisms in high speed networks. In this context, we propose a rate-based explicit congestion control mechanism with which the sources are provided with the rate at which they can transmit. These rates are computed with a distributed max-min fair algorithm, SLBN. The novelty of SLBN is that it combines two interesting features not simultaneously present in existing proposals: scalability and fast convergence to the max-min fair rates, even under high session churn. SLBN is scalable because routers only maintain a constant amount of state information (only three integer variables per link) and only incur a constant amount of computation per protocol packet, independently of the number of sessions that cross the router. Additionally, SLBN does not require processing any data packet, and it converges independently of sessions' RTT. Finally, by design, the protocol is conservative when assigning rates, even in the presence of high churn, which helps preventing link overshoots in transient periods. We claim that, with all these features, our mechanism is a good candidate to be used in real deployments.

Soporte de redes activas para un servicio integrado de multipunto fiable con control de congestión

Las limitaciones de las tecnologías de red actuales, identificadas en la Agencia de Proyectos de Investigación Avanzados para la Defensa (DARPA) durante 1995, han originado recientemente una propuesta de modelo de red denominado Redes Activas. En este modelo, los nodos proporcionan un entorno de ejecución sobre el que se ejecuta el código asociado a cada paquete. El objetivo es disponer de una tecnología de red que permita que nuevos servicios de red sean desarrollados e instalados rápidamente sin modificar los nodos de la red. Un servicio de red que se puede beneficiar de esta tecnología es la transmisión de datos en multipunto con diferentes grados fiabilidad. Las propuestas actuales de servicios de multipunto fiable proporcionan una solución específica para cada clase de aplicaciones, y los protocolos existentes extremo a extremo sufren de limitaciones técnicas relacionadas con una fiabilidad limitada, y con la ausencia de mecanismos de control de congestión efectivos. Esta tesis realiza propuestas originales conducentes a solucionar parte de las limitaciones actuales en el ámbito de Redes Activas y multipunto fiable con control de congestión. En primer lugar, se especificará un servicio genérico de multipunto fiable que, basándose en los requisitos de una serie de aplicaciones consideradas relevantes, proporcione diferentes clases de sesiones y grados de fiabilidad. Partiendo de la definición del servicio genérico especificado, se diseñará un protocolo de comunicaciones sobre la tecnología de Redes Activas que proporcione dicho servicio. El protocolo diseñado estará dotado de un mecanismo de control de congestión para que la fuente ajuste dinámicamente el tráfico inyectado a las condiciones de carga de la red. En esta tesis se pretende también profundizar en el estudio y análisis de la tecnología de Redes Activas, experimentando con dicha tecnología para proporcionar una realimentación a sus diseñadores. Dicha experimentación se realizará en tres ámbitos: el de los servicios y protocolos que puede soportar, el del modelo y arquitectura de las Redes Activas y el de las plataformas de ejecución disponibles. Como aportación adicional de este trabajo, se validarán los objetivos anteriores mediante una implementación piloto de las entidades de protocolo y de su interfaz de servicio sobre uno de los entornos de ejecución disponibles. Abstract The limitations of current networking technologies identified in the Defense Advance Research Projects Agency (DARPA) along 1995 have led to a recent proposal of a new network model called Active Networks. In this model, the nodes provide an execution environment over which the code used to process each packet is executed. The objective is a network technology that allows the fast design and deployment of new network services without requiring the modification of the network nodes. One network service that could benefit from this technology is the transmission of multicast data with different types of loss tolerance. The current proposals for reliable multicast services provide specific solutions for each application class, and existing end-to-end protocols suffer from technical drawbacks related to limited reliability and lack of an effective congestion control mechanism. This thesis contains original proposals that aim to solve part of the current drawbacks in the scope of Active Networks and reliable multicast with congestion control. Firstly, a generic reliable multicast network service will be specified. This service will be designed from the requirements of a relevant set of applications, and will provide different session classes and different types of reliability. Then, a network protocol based on Active Network technology will be designed such that it provides the specified network service. This protocol will incorporate a congestion control mechanism capable of performing an automatic adjustment of the traffic injected by the source to the available network capacity. This thesis will also contribute to a deeper study and analysis of Active Network technology, by experimenting with the technology in order to provide feedback to its designers. This experimentation will be done attending to three different scopes: support of Active Network for services and protocols, Active Network model and architecture, and currently available Active Network execution environments. As an additional contribution of this work, the previous objectives will be validated through a prototype implementation of the protocol entities and the service interface based on one of the current execution environments.

Contribution to the improvement of communication services in nanosatellite constellations using delay and disruption tolerant networking based architectures

Esta tesis se desarrolla dentro del marco de las comunicaciones satelitales en el innovador campo de los pequeños satélites también llamados nanosatélites o cubesats, llamados así por su forma cubica. Estos nanosatélites se caracterizan por su bajo costo debido a que usan componentes comerciales llamados COTS (commercial off-the-shelf) y su pequeño tamaño como los Cubesats 1U (10cm*10 cm*10 cm) con masa aproximada a 1 kg. Este trabajo de tesis tiene como base una iniciativa propuesta por el autor de la tesis para poner en órbita el primer satélite peruano en mi país llamado chasqui I, actualmente puesto en órbita desde la Estación Espacial Internacional. La experiencia de este trabajo de investigación me llevo a proponer una constelación de pequeños satélites llamada Waposat para dar servicio de monitoreo de sensores de calidad de agua a nivel global, escenario que es usado en esta tesis. Es ente entorno y dadas las características limitadas de los pequeños satélites, tanto en potencia como en velocidad de datos, es que propongo investigar una nueva arquitectura de comunicaciones que permita resolver en forma óptima la problemática planteada por los nanosatélites en órbita LEO debido a su carácter disruptivo en sus comunicaciones poniendo énfasis en las capas de enlace y aplicación. Esta tesis presenta y evalúa una nueva arquitectura de comunicaciones para proveer servicio a una red de sensores terrestres usando una solución basada en DTN (Delay/Disruption Tolerant Networking) para comunicaciones espaciales. Adicionalmente, propongo un nuevo protocolo de acceso múltiple que usa una extensión del protocolo ALOHA no ranurado, el cual toma en cuenta la prioridad del trafico del Gateway (ALOHAGP) con un mecanismo de contienda adaptativo. Utiliza la realimentación del satélite para implementar el control de la congestión y adapta dinámicamente el rendimiento efectivo del canal de una manera óptima. Asumimos un modelo de población de sensores finito y una condición de tráfico saturado en el que cada sensor tiene siempre tramas que transmitir. El desempeño de la red se evaluó en términos de rendimiento efectivo, retardo y la equidad del sistema. Además, se ha definido una capa de convergencia DTN (ALOHAGP-CL) como un subconjunto del estándar TCP-CL (Transmission Control Protocol-Convergency Layer). Esta tesis muestra que ALOHAGP/CL soporta adecuadamente el escenario DTN propuesto, sobre todo cuando se utiliza la fragmentación reactiva. Finalmente, esta tesis investiga una transferencia óptima de mensajes DTN (Bundles) utilizando estrategias de fragmentación proactivas para dar servicio a una red de sensores terrestres utilizando un enlace de comunicaciones satelitales que utiliza el mecanismo de acceso múltiple con prioridad en el tráfico de enlace descendente (ALOHAGP). El rendimiento efectivo ha sido optimizado mediante la adaptación de los parámetros del protocolo como una función del número actual de los sensores activos recibidos desde el satélite. También, actualmente no existe un método para advertir o negociar el tamaño máximo de un “bundle” que puede ser aceptado por un agente DTN “bundle” en las comunicaciones por satélite tanto para el almacenamiento y la entrega, por lo que los “bundles” que son demasiado grandes son eliminados o demasiado pequeños son ineficientes. He caracterizado este tipo de escenario obteniendo una distribución de probabilidad de la llegada de tramas al nanosatélite así como una distribución de probabilidad del tiempo de visibilidad del nanosatélite, los cuales proveen una fragmentación proactiva óptima de los DTN “bundles”. He encontrado que el rendimiento efectivo (goodput) de la fragmentación proactiva alcanza un valor ligeramente inferior al de la fragmentación reactiva. Esta contribución permite utilizar la fragmentación activa de forma óptima con todas sus ventajas tales como permitir implantar el modelo de seguridad de DTN y la simplicidad al implementarlo en equipos con muchas limitaciones de CPU y memoria. La implementación de estas contribuciones se han contemplado inicialmente como parte de la carga útil del nanosatélite QBito, que forma parte de la constelación de 50 nanosatélites que se está llevando a cabo dentro del proyecto QB50. ABSTRACT This thesis is developed within the framework of satellite communications in the innovative field of small satellites also known as nanosatellites (<10 kg) or CubeSats, so called from their cubic form. These nanosatellites are characterized by their low cost because they use commercial components called COTS (commercial off-the-shelf), and their small size and mass, such as 1U Cubesats (10cm * 10cm * 10cm) with approximately 1 kg mass. This thesis is based on a proposal made by the author of the thesis to put into orbit the first Peruvian satellite in his country called Chasqui I, which was successfully launched into orbit from the International Space Station in 2014. The experience of this research work led me to propose a constellation of small satellites named Waposat to provide water quality monitoring sensors worldwide, scenario that is used in this thesis. In this scenario and given the limited features of nanosatellites, both power and data rate, I propose to investigate a new communications architecture that allows solving in an optimal manner the problems of nanosatellites in orbit LEO due to the disruptive nature of their communications by putting emphasis on the link and application layers. This thesis presents and evaluates a new communications architecture to provide services to terrestrial sensor networks using a space Delay/Disruption Tolerant Networking (DTN) based solution. In addition, I propose a new multiple access mechanism protocol based on extended unslotted ALOHA that takes into account the priority of gateway traffic, which we call ALOHA multiple access with gateway priority (ALOHAGP) with an adaptive contention mechanism. It uses satellite feedback to implement the congestion control, and to dynamically adapt the channel effective throughput in an optimal way. We assume a finite sensor population model and a saturated traffic condition where every sensor always has frames to transmit. The performance was evaluated in terms of effective throughput, delay and system fairness. In addition, a DTN convergence layer (ALOHAGP-CL) has been defined as a subset of the standard TCP-CL (Transmission Control Protocol-Convergence Layer). This thesis reveals that ALOHAGP/CL adequately supports the proposed DTN scenario, mainly when reactive fragmentation is used. Finally, this thesis investigates an optimal DTN message (bundles) transfer using proactive fragmentation strategies to give service to a ground sensor network using a nanosatellite communications link which uses a multi-access mechanism with priority in downlink traffic (ALOHAGP). The effective throughput has been optimized by adapting the protocol parameters as a function of the current number of active sensors received from satellite. Also, there is currently no method for advertising or negotiating the maximum size of a bundle which can be accepted by a bundle agent in satellite communications for storage and delivery, so that bundles which are too large can be dropped or which are too small are inefficient. We have characterized this kind of scenario obtaining a probability distribution for frame arrivals to nanosatellite and visibility time distribution that provide an optimal proactive fragmentation of DTN bundles. We have found that the proactive effective throughput (goodput) reaches a value slightly lower than reactive fragmentation approach. This contribution allows to use the proactive fragmentation optimally with all its advantages such as the incorporation of the security model of DTN and simplicity in protocol implementation for computers with many CPU and memory limitations. The implementation of these contributions was initially contemplated as part of the payload of the nanosatellite QBito, which is part of the constellation of 50 nanosatellites envisaged under the QB50 project.

A distributed end-to-end overload control mechanism for networks of SIP servers.

The Session Initiation Protocol (SIP) is an application-layer control protocol standardized by the IETF for creating, modifying and terminating multimedia sessions. With the increasing use of SIP in large deployments, the current SIP design cannot handle overload effectively, which may cause SIP networks to suffer from congestion collapse under heavy offered load. This paper introduces a distributed end-to-end overload control (DEOC) mechanism, which is deployed at the edge servers of SIP networks and is easy to implement. By applying overload control closest to the source of traf?c, DEOC can keep high throughput for SIP networks even when the offered load exceeds the capacity of the network. Besides, it responds quickly to the sudden variations of the offered load and achieves good fairness. Theoretic analysis and extensive simulations verify that DEOC is effective in controlling overload of SIP networks.

Probe-based end-to-end overload control for networks of SIP servers

The Session Initiation Protocol (SIP) has been adopted by the IETF as the control protocol for creating, modifying and terminating multimedia sessions. Overload occurs in SIP networks when SIP servers have insufficient resources to handle received messages. Under overload, SIP networks may suffer from congestion collapse due to current ineffective SIP overload control mechanisms. This paper introduces a probe-based end-to-end overload control (PEOC) mechanism, which is deployed at the edge servers of SIP networks and is easy to implement. By probing the SIP network with SIP messages, PEOC estimates the network load and controls the traffic admitted to the network according to the estimated load. Theoretic analysis and extensive simulations verify that PEOC can keep high throughput for SIP networks even when the offered load exceeds the capacity of the network. Besides, it can respond quickly to the sudden variations of the offered load and achieve good fairness.