In this thesis, a new model for the slow start phase based on the discrete evolutions of congestion window is developed, and we integrate this part into the improved TCP steady state model for a better prediction performance. Combining these short and steady state models, we propose an extensive stochastic model which can accurately predict the throughput and latency of the TCP connections as functions of loss rate, round-trip time (RTT), and file size. We validate our results through simulation experiments. The results show that our model?s predictions match the simulation results better than the Padhye and Cardwell's stochastic models, about 75% improvement in the accuracy of performance predictions for the steady state and 20% improvement for the short-lived TCP flows.

Sensor network congestion avoidance and control primarily aims to reduce packet drops while maintaining fair bandwidth allocation to existing network flows. The design of a congestion control algorithm suited for all types of applications in sensor networks is a challenging task due to the application-specific nature of these networks. With numerous sensors transmitting data simultaneously to one or more base stations (also called sinks), sensor nodes located near the base station will most likely experience congestion and packet loss. In this thesis, we propose a novel distributed congestion avoidance algorithm which calculates the ratio of the number of downstream and upstream nodes. This ratio value (named Characteristic ratio) is used to take a routing decision and incorporate load balancing while also serving as a pointer to the congestion state of the network. Available queue sizes of the downstream nodes are used to detect incipient congestion. Queue characteristics of candidate downstream nodes are used collectively to implement both congestion avoidance and fairness by adjusting the node's forwarding rate and next hop destination. Such an approach helps to minimize packet drops, improve energy efficiency and load balancing. In cases of severe congestion, the source is signaled to reduce its sending rate and enable the network recovery process. This is essentially a transport layer algorithm and would work best with a multi-path routing protocol and almost any MAC layer standard. We present the design and implementation of the proposed protocol and compare it with the existing avoidance protocols like Global rate control and Lightweight buffering. Our simulation results show a higher packet delivery ratio with greater node buffer utilization for our protocol in comparison with the conventional mechanisms.

Ingeniero Civil Eléctrico / Las telecomunicaciones han evolucionado a un nivel tal que prácticamente se volvieron parte de la cotidianidad de la sociedad. La cantidad de aplicaciones y servicios disponibles así como la facilidad de acceso para gran parte de la población, ya sea por los bajos costos o por incluso gratuidad de conexión en lugares específicos, hacen que los requisitos de capacidad de transferencia de datos crezcan por lo tanto un óptimo uso de la capacidad del canal de transmisión es una necesidad. Esta necesidad es el foco de este trabajo. En este trabajo se propone un algoritmo que aborda la problemática a nivel de la capa de transporte. La capa de transporte utiliza el protocolo TCP para la transferencia segura e íntegra de los datos. Dado el funcionamiento de este protocolo, en donde se requiere confirmación de cada paquete recibido para la transmisión de datos (ACK), el Round-Trip Time (RTT) o retardo afecta drásticamente la velocidad de transmisión porque los paquetes en vuelo (que están siendo transferidos en el canal) ya que retrasan el crecimiento de la ventana de congestión. Este problema se acentúa particularmente en las redes de high Round-Trip Time (BDP) que corresponden a redes que tienen una alta capacidad de transferencia y RTT altos provocando que la cantidad de datos en vuelo sea elevado. El uso de TCP en las redes de alto BDP hace que las transmisiones sean muy dependientes del retardo y su mejor utilización de recursos corresponde a un área de estudio muy activa dentro de los algoritmos de evasión de congestión. En este trabajo se propone un método para encontrar la capacidad del canal de telecomunicaciones usando información que está disponible en la capa de transporte. El método consiste en utilizar una métrica basada en la distancia (secuencial o temporal) entre dos pérdidas y con esta información estimar la capacidad del canal. De esta forma se puede obtener un nivel de conocimiento sobre la red explícito en lugar de tantear mediante probing, método empleado por la mayoría de los protocolos existentes. Con la capacidad de canal estimada se calcula la ventana de congestión óptima teniendo varios factores en consideración como: throughput, TCP-friendliness y evasión de congestión. Además de conocer el valor de la capacidad del canal es necesario transmitir con ese conocimiento, por lo tanto también se propone la implementación de dos algoritmos de evasión de congestión en entorno a Linux. Las pruebas entregan resultados alentadores en donde se logra mantener la ventana de congestión muy cerca y establemente a la capacidad encontrada del canal. Esto se refiere principalmente a que las desviaciones estándar de la ventana de congestión de los módulos implementados tienen valores bajos y son alrededor de cuatro veces más pequeñas que las entregadas por el algoritmo Reno. Como resultado de este trabajo se publicaron en dos conferencias (Chilecon, IEEE Latincom) y se planea enviar a un journal.

Reliable and efficient data transfer on the Internet is an important issue. Since late 70's the protocol responsible for that has been the de facto standard TCP, which has proven to be successful through out the years, its self-managed congestion control algorithms have retained the stability of the Internet for decades. However, the variety of existing new technologies such as high-speed networks (e.g. fibre optics) with high-speed long-delay set-up (e.g. cross-Atlantic links) and wireless technologies have posed lots of challenges to TCP congestion control algorithms. The congestion control research community proposed solutions to most of these challenges. This dissertation adds to the existing work by: firstly tackling the highspeed long-delay problem of TCP, we propose enhancements to one of the existing TCP variants (part of Linux kernel stack). We then propose our own variant: TCP-Gentle. Secondly, tackling the challenge of differentiating the wireless loss from congestive loss in a passive way and we propose a novel loss differentiation algorithm which quantifies the noise in packet inter arrival times and use this information together with the span (ratio of maximum to minimum packet inter arrival times) to adapt the multiplicative decrease factor according to a predefined logical formula. Finally, extending the well-known drift model of TCP to account for wireless loss and some hypothetical cases (e.g. variable multiplicative decrease), we have undertaken stability analysis for the new version of the model.

現今我們所住的世界不再只是一個單純的有線網路，隨著WLAN的無線區域網路的存取點越來越普遍，每個家庭使用有線網路做為後端網路然後使用無線區域網路的存取點作為最後一哩的情況也隨著變的常見了，所以現在的網路世界不再只是一個單純的有線網路而是一個有線跨無線的異質性無線網路了，TCP一個我們所常用的傳輸層的協定當遇到異質性無線網路時已經被證實會對封包遺失的原因有誤判，在無線傳輸時封包遺失的理由包括訊號品質不好以及用戶端之間在無線介質中的競爭，因此我們設計了一個具有跨層考量的EATCP-Assisted模組來幫助傳送端收集媒體存取控制層的資訊並使的傳送端在調整競爭視窗上面更有效率，在EATCP中有兩大部分：第一部份是估算階段，第二階段就是EATCP的擁塞控制機制，藉由EATCP-Assisted模組所收集到的wirelessRTT以及utilization我們可以有效的估算出wiredBW以及wirelessBW，並藉由這兩個參數我們可以準確的將封包遺失的原因從頻道衰減以及用戶端之間的競爭中分別出來，因此一個適當的競爭視窗就可以藉由這些資訊來調整出來。 / Recently, with the growth of WLAN, the world we live today is no longer a pure wired network, it’s a heterogeneous wireless network. The TCP that we commonly used has been proven to have misjudgements of packet loss in heterogeneous wireless networks. We design a cross-layer architecture called EATCP-Assisted module to help the sender collect MAC layer information, and adjust the congestion window more efficiently. In EATCP, there are two important parts: the first is estimation phase; the second is congestion control algorithm. By collecting the wirelessRTT, and utilization we can estimate the wiredBW and wirelessBW. The EATCP will distinguish the causes of packet loss from network congestion, channel fading, or contention between wireless clients. Thus, appropriate congestion window adjustment can be done accordingly. The simulation results show that our EATCP outperforms other versions of TCP.

The Master´s thesis deals with the field of converged networks and their effective utilization, focusing specifically on telematic networks. For modelling of a specific application, the field of vehicle telematics has been chosen. The thesis consists of two series of simulations performed in Opnet Modeler network simulator. The first theoretic part gave an overview of protocols used in converged networks, with a more detailed description of routing protocols. The attention also centred on the mechanisms for assuring the quality of service in IP. The subsequent practical part consisted in creating the simulation of a converged telematic network in Opnet Modeler network simulator, configuration of scenarios for assuring QoS using RSVP and DSCP protocols and the simulation itself. The results of the simulations for both scenarios were compared and represented in graphic form. The second theoretic part described queueing disciplines, as well as congestion control and congestion avoidance mechanisms. The corresponding practical part compared FIFO, PQ and WFQ queueing algorithms within the same simulated converged telematic network, using DSCP protocol scenario, which gave much better results in the first series of simulations. Again, the results of the simulations were compared and represented in graphic form.

With a new United States Army initiative to exploit commercially developed information technology, there is a heightened interest in using Internet protocols over the military's geosynchronous satellite links. TCP is the dominant Internet protocol used for reliable data exchange, but its own design limits performance when used over long delay network links such as satellites. Initially this research set out to compare TCP with another proposed protocol, the Satellite Transport Protocol (STP). However through a series of tests, we found that STP does not fulfill its claims of increased throughput over TCP and uncovered a flaw in STP's founding research. In addition, this thesis proposes and demonstrates novel performance enhancing techniques that significantly improve transport protocol throughput.

Content of master’s thesis is description field of Sliding window and it’s expansion algorithms, witch are Slow start, Congestion avoidance, Fast Retransmit and Fast Recovery algorithm. Thereinafter is described creation of model in Opnet Modeler’s simulation area. In this simulation area was analyzed reactions of average transfer speed onto variance of data size, lost ratio, latency in short and long time slot and variance of receiver’s buffer size. In last section of this document is method design witch makes it possible of transfer speed control through the use of receiver’s buffer size dynamic setting.