Packet switched networks offer a particularly challenging research subject to the control community: the dynamics of a network buffer, their simplest component, are nonlinear and exhibit a saturation effect that cannot be neglected. In many practical cases, networks are made up of the interconnection of a large number of such basic elements. This gives rise to high dimensional nonlinear systems for which few general results exist today in the literature. Furthermore, these physical interconnections that may sometimes span a very long distance induce a transmission delay and the queues in intermediary nodes induce a buffering delay. Finding a model able to both take into account as much of this complexity as possible while being simple enough to be analysed mathematically and used for control purposes is the first objective of this thesis. To accomplish this goal, a so-called "fluid-flow model" based on fluid exchange between buffers is presented. Neglecting the transmission and propagation delays, this model concentrates on the dynamics of the buffer loads and is particularly well suited for a mathematical analysis. Throughout the work, a systematic system point of view is adopted in an effort to perform a rigorous analysis using tools from automatic control and dynamical systems theory. This model is then used to study a feedback control law where each node receives state information from its directly connected neighbours, hence referred to as hop-by-hop control. The properties of the closed-loop system are analysed and a global stability analysis is performed using existing results from the compartmental and cooperative system literature. The global mass conservation typically ensured by end-to-end control protocols is studied in the last chapter using, once again, a compartmental framework. Finally, a numerical study of a strategy combining the end-to-end and the hop-by-hop approaches is presented. It is shown that problems encountered with hop-by-hop control may then be successfully alleviated.

A Mobile Ad Hoc Network (MANET) is a shared wireless network without any infrastructure, consisting of mobile nodes connected by wireless links. The nodes are free to move and organize themselves arbitrarily. The nodes in the network are therefore depending on each other in order to communicate over multiple hops. Due to the physical characteristics of wireless networks, the channel is time-varying, which makes it hard to both predict and sustain a bit rate level. The nodes’ mobility causes topology changes, and further load and capacity variations. Traditional usage areas are battlefield and disaster areas, while new areas like extended network coverage and gaming are emerging. Quality of Service (QoS) is needed in every network in order to differentiate traffic with different performance requirements, e.g. voice and e-mail applications. Providing QoS in wireless environments with varying conditions is complex, and hard guarantees can not be given. Consequently, the aim is to give differentiated treatment to traffic with different performance requirements. In addition, we can not study the MANET without considering fixed networks. Communication with fixed networks is important, for example by accessing the Internet. In this thesis the Differentiated Services (DiffServ) architecture is applied and adapted to MANETs. Using the same QoS architecture will ease the transition between the wireless and wired domain. But the special characteristics of wireless networks require modifications to the original DiffServ architecture. In investigations there was found restrictions on the number of classes to use, and this number was dependent on the type of traffic in the network. A QoS architecture based on the DiffServ framework is proposed, with an admission control based on the concept of shadow classes, and Explicit Congestion Notification (ECN) to avoid congestion. New flows are tested in a shadow class before getting admission to the network and its designated class. The shadow class has the same scheduling properties as the designated class, but is differentiated by a higher drop probability in the buffers. Both the admission control and ECN are thus build on the same principle by controlling the load from probabilistic functions in the buffers, and are studied to find their individual and combined effects. In wireless environments the probability of a packet loss increases with the number of hops, which gives services an unpredictable performance for users. A predictable service, independent of number of hops, is provided by scheduling based on the path information; the packets are differentiated based on the number of hops made or left to make, increasing the predictability at the cost of performance.

This thesis demonstrates the performance of Probabilistic Early Response TCP (PERT), a new TCP congestion control, for video streaming. As a delay based protocol, it measures the delay at the end host and adjusts the congestion window accordingly. Our experiments show that PERT improves video delivery performance by decreasing the fraction of packets delivered late. Furthermore, our Linux live streaming test indicates that PERT is able to reduce the playback glitches, when high resolution video is delivered over a link with non-zero packet loss. In order to operate PERT at higher thoughputs, we design PERT to work with Acknowledgement (ACK) coalescing at the receiver. ACK coalescing makes data transfers burstier and makes it hard to estimate delays accurately. We apply TCP pacing to fix this issue, and validate its effectiveness in the aspects of throughput, packet loss and fairness. Our experiment results also show that PERT with Delayed ACK and Pacing is more friendly, and therefore more suitable when multiple traffic flows are competing for limited bottleneck bandwidth or sharing the same router buffer.

Many congestion control mechanisms have been proposed to solve the problems of a high loss rate and inefficient utilization of network resources in the present Internet. This problem is caused by competition between traffic flows while the network is congested. Differentiated Services (DiffServ) architecture permits the allocation of various levels of traffic resource requirements needed for Quality of Service (QoS). Random Early Detection (RED) is an efficient mechanism to pre-drop packets before actual congestion occurs, and it is capable of introducing a random early packet dropping scheme, and based on the queue length in reaching a certain degree of fairness for resource utilization. However, it still suffers from a lack of robustness among light traffic load, or in heavy traffic load using fixed RED parameters. In this dissertation, we modified the RED scheme and proposed a novel adaptive RED model, which we named the OURED model, to enhance the robustness of resource utilization so that it could be utilized in the DiffServ edge router. The OURED model introduces two additional packet dropping traces, one is Over Random Early Detection (ORED), which is used to speed up the dropping of packets when the actual rate is higher than the target rate, and the other one is the Under Random Early Detection (URED), used to slow down the packet dropping rate in the reverse situation. The simulation results show that OURED is not only more robust than MRED in resource utilization, but that it also can be implement efficiently in the DiffServ edge router. Another model proposed in this dissertation is the Age-Based packet discarding Traffic Conditioner. For the reason that the file sizes of on going flows are fairly disparate on the current network, we propose an ¡§Age-Based¡¨ packet discard scheme in the Traffic Conditioner of a gateway, to improve the performance of file transmission. The on going flows will be grouped to three classes of priority according to their ¡§age¡¨ as network congestion occurs and the simulation results show that the proposed model can work efficiently in most of the congestion conditions.

Abstract The throughput of a TCP connection is decided by the size of the congestion window. And cwnd increases when an acknowledgement arrives. It leads to that TCP has a bias against connections with long round-trip-time. For enhancing the fairness of TCP, we proposed a new scheme FTCP (Fair TCP). Unlike TCP, in FTCP congestion avoidance state, it compares its RTT with the standard RTT to adjust the increase amount of cwnd when an ACK arrives TCP sender. Therefore FTCP can keep the throughput increase rate of connections with different RTTs be the same. When FTCP enters timeout state, it sets appropriate slow start threshold by calculating the difference value of cwnd / 2 and the cwnd while standard connection achieves ssthresh. So that FTCP can eliminate the difference of throughput between connections with different RTT while leaving the slow start state. FTCP significantly improves the unfair bandwidth distribution between connections with different RTT. TCP connections over wireless links perform badly because of the unnecessary congestion control, inefficiency to burst packet loss, and long delay to slow down the cwnd recovery time. In proposed schemes, Snoop takes BS as a pivot point to cache the unacknowledged TCP packets. When errors occur in wireless link, Snoop retransmits the packets locally from BS instead of retransmitting these packets from sender. And Snoop shields off the duplicate ACKs caused by wireless errors to avoid sender triggering unnecessary congestion control. But Snoop adopts same retransmission style as TCP. It only retransmits one packet per continuous duplicate ACKs. Snoop recovers error packets more quickly and tolerates higher BER than TCP. But Snoop doesn¡¦t really solve the degraded performance problem of multiple errors of TCP. When the channel is the in a very bad quality, Snoop still performs badly. We proposed a new scheme, Csnoop (continuous snoop), extended from Snoop. When bursty errors happen in the wireless links, Csnoop retransmits one lost packets from the BS in first RTT and counts the number of ACKs that arrives BS to calculate the number of lost packets. And Csnoop retransmits these lost packets continuously. When local timeout happens, Csnoop infers that all packets were dropped and retransmits all packets cached in the buffer. Simulations show that Csnoop achieves better throughput compared to Snoop and TCP, especially for bad quality wireless links. Furthermore, Csnoop needs less buffer size to cache the unacknowledged packets at the base station than Snoop.

ATM (Asynchronous Transfer Mode) networks were deemed the best choice for multimedia communication. The traditional mode was replaced because ATM can provide varied traffic types and QoS (quality of service). Maintaining QoS, however, requires a flexible traffic control, including call admission control and congestion control. Traditional approaches fail to estimate the required bandwidth and cell loss rate precisely. To alleviate these problems, we employ AI methods to improve the capability of estimated bandwidth and predicted cell loss rate. This thesis aims to apply neural network techniques to ATM traffic control and consists of two parts. The first part concerns a neural-based call admission control, while the second part presents an intelligent congestion control for ATM networks. In the first part, we focus on the improvement of RBF (Radial basis function) networks and the design of a neural-based call admission control. RBF networks have been widely used for modeling a function from given input-output patterns. However, two difficulties are encountered with traditional RBF networks. One is that the initial configuration of a RBF network needs to be determined by a trial-and-error method. The other is that the performance suffers from some difficulties when the desired output has abrupt changes or constant values in certain intervals. We propose a novel approach to overcome these difficulties. New kernel functions are used for hidden nodes, and the number of nodes is determined automatically by an ART-like algorithm. Parameters and weights are initialized appropriately, and then tuned and adjusted by the gradient descent method to improve the performance of the network. Then, we employ ART-RBF networks to design and implement a call admission control. Traditional approaches fail to estimate appropriately the required bandwidth, leading to a waste of bandwidth or a high cell loss rate. To alleviate the problem, we employ ART-RBF networks to estimate the required bandwidth, and thus a new connection request can then be accepted or rejected. Because of the more accurate estimation on the required bandwidth, the proposed method can provide a better control on quality of service for ATM networks. In the second part, we propose a neural-fuzzy rate-based feedback congestion control for ATM networks. Traditional methods perform congestion control by monitoring the queue length. The source rate is decreased by a fixed rate when the queue length is greater than a predefined threshold. However, it is difficult to get a suitable rate according to the degree of traffic congestion. We employ a neural-fuzzy mechanism to control the source rate. Through learning, cell loss can be predicted from the current value and the derivative of the queue length. Then an explicit rate is calculated and the source rate is controlled appropriately. In summary, we have proposed improvements on architecture and performance of neural networks, and applied neural networks to traffic control for ATM networks. We have developed some control mechanisms which, through simulations, have been shown to be more effective than traditional methods.

This thesis presents two modern control methods to address the Internet traffic congestion control issues. They are based on a distributed traffic management framework for the fast-growing Internet traffic in which routers are deployed with intelligent or optimal data rate controllers to tackle the traffic mass. The first one is called the IntelRate (Intelligent Rate) controller using the fuzzy logic theory. Unlike other explicit traffic control protocols that have to estimate network parameters (e.g., link latency, bottleneck bandwidth, packet loss rate, or the number of flows), our fuzzy-logic-based explicit controller can measure the router queue size directly. Hence it avoids various potential performance problems arising from parameter estimations while reducing much computation and memory consumption in the routers. The communication QoS (Quality of Service) is assured by the good performances of our scheme such as max-min fairness, low queueing delay and good robustness to network dynamics. Using the Lyapunov’s Direct Method, this controller is proved to be globally asymptotically stable. The other one is called the OFEX (Optimal and Fully EXplicit) controller using convex optimization. This new scheme is able to provide not only optimal bandwidth allocation but also fully explicit congestion signal to sources. It uses the congestion signal from the most congested link, instead of the cumulative signal from a flow path. In this way, it overcomes the drawback of the relatively explicit controllers that bias the multi-bottlenecked users, and significantly improves their convergence speed and throughput performance. Furthermore, the OFEX controller design considers a dynamic model by proposing a remedial measure against the unpredictable bandwidth changes in contention-based multi-access networks (such as shared Ethernet or IEEE 802.11). When compared with the former works/controllers, such a remedy also effectively reduces the instantaneous queue size in a router, and thus significantly improving the queueing delay and packet loss performance. Finally, the applications of these two controllers on wireless local area networks have been investigated. Their design guidelines/limits are also provided based on our experiences.

Many computers and devices such as smart phones, laptops and tablet devices are now equipped with multiple network interfaces, enabling them to use multiple paths to access content over the network. If the resources could be used concurrently, end user experience can be greatly improved. The recent studies in MPTCP suggest that improved reliability, load balancing and mobility are feasible. The thesis presents a new multipath delay based algorithm, MPPERT (Multipath Probabilistic Early response TCP), which provides high throughput and efficient load balancing. In all-PERT environment, MPPERT suffers no packet loss and maintains much smaller queue sizes compared to existing MPTCP, making it suitable for real time data transfer. MP-PERT is suitable for incremental deployment in a heterogeneous environment. It also presents a parametrized approach to tune the amount of traffic shift off the congested path. Multipath approach is benefited from having multiple connections between end hosts. However, it is desired to keep the connection set minimal as increasing number of paths may not always provide significant increase in the performance. Moreover, higher number of paths unnecessarily increase computational requirement. Ideally, we should suppress paths with low throughputs and avoid paths with shared bottlenecks. In case of MPTCP, there is no efficient way to detect a common bottleneck between subflows. MPTCP applies a constraint of best single-path TCP throughput, to ensure fair share at a common bottleneck link. The best path throughput constraint along with traffic shift, from more congested to less congested paths, provide better opportunity for the competing flows to achieve higher throughput. However, the disadvantage is that even if there are no shared links, the same constraint would decrease the overall achievable throughput of a multipath flow. PERT, being a delay based TCP protocol, has continuous information about the state of the queue. This information is valuable in enabling MPPERT to detect subflows sharing a common bottleneck and obtain a smaller set of disjoint subflows. This information can even be used to switch from coupled (a set of subflows having interdependent increase/decrease of congestion windows) to uncoupled (independent increase/decrease of congestion windows) subflows, yielding higher throughput when best single-path TCP constraint is relaxed. The ns-2 simulations support MPPERT as a highly competitive multipath approach, suitable for real time data transfer, which is capable of offering higher throughput and improved reliability.

Um dos obstáculos para o uso disseminado do multicast na Internet global é o desenvolvimento de protocolos de controle de congestionamento adequados. Um fator que contribui para este problema é a heterogeneidade de equipamentos, enlaces e condições de acesso dos receptores, a qual aumenta a complexidade de implementação e validação destes protocolos. Devido ao multicast poder envolver milhares de receptores simultaneamente, o desafio deste tipo de protocolo se torna ainda maior, pois além das questões relacionadas ao congestionamento da rede, é necessário considerar fatores como sincronismo, controle de feedbacks, equidade de tráfego, entre outros. Por esses motivos, os protocolos de controle de congestionamento multicast têm sido um tópico de intensa pesquisa nos últimos anos. Uma das alternativas para o controle de congestionamento multicast na Internet é o protocolo ALMTF (Adaptive Layered Multicast TCP-Friendly), o qual faz parte do projeto SAM (Sistema Adaptativo Multimídia). Uma vantagem desse algoritmo é inferir o nível de congestionamento da rede, determinando a taxa de recebimento mais apropriada para cada receptor. Além disso, ele realiza o controle da banda recebida, visando à justiça e a imparcialidade com os demais tráfegos concorrentes. O ALMTF foi desenvolvido originalmente em uma Tese de doutorado e teve a sua validação no simulador de redes NS-2 (Network Simulator). Este trabalho tem como objetivo estender o protocolo para uma rede real, implementando, validando os seus mecanismos e propondo novas alternativas que o adaptem para esse ambiente. Além disso, efetuar a comparação dos resultados reais com a simulação, identificando as diferenças e promovendo as pesquisas experimentais na área. / One of the obstacles for the widespread use of the multicast in the global Internet is the development of adequate protocols for congestion control. One factor that contributes for this problem is the heterogeneity of equipments, enlaces and conditions of access of the receivers, which increases the implementation and validation complexity of these protocols. Due to the number (thousands) of receivers simultaneously involved in multicast, the challenge of these protocols is even higher. Besides the issues related to the network congestion, it is necessary to consider factors such as synchronism, feedback control, fairness, among others. For these reasons, the multicast congestion control protocols have been a topic of intense research in recent years. The ALMTF protocol (Adaptive Layered Multicast TCP-Friendly), which is part of project SAM, is one of the alternatives for the multicast congestion control in the Internet. One advantage of this algorithm is its ability to infer the network congestion level, assigning the best receiving rate for each receptor. Besides that, the protocol manages the received rate, aiming to achieve fairness and impartiality with the competing network traffic. The ALMTF was developed originally in a Ph.D. Thesis and had its validation under NS-2 simulator. The goal this work is to extend the protocol ALMTF for a real network, validating its mechanisms and considering new alternatives to adapt it for this environment. Moreover, to make the comparison of the real results with the simulation, being identified the differences and promoting the experimental research in the area.

Um dos obstáculos para o uso disseminado do multicast na Internet global é o desenvolvimento de protocolos de controle de congestionamento adequados. Um fator que contribui para este problema é a heterogeneidade de equipamentos, enlaces e condições de acesso dos receptores, a qual aumenta a complexidade de implementação e validação destes protocolos. Devido ao multicast poder envolver milhares de receptores simultaneamente, o desafio deste tipo de protocolo se torna ainda maior, pois além das questões relacionadas ao congestionamento da rede, é necessário considerar fatores como sincronismo, controle de feedbacks, equidade de tráfego, entre outros. Por esses motivos, os protocolos de controle de congestionamento multicast têm sido um tópico de intensa pesquisa nos últimos anos. Uma das alternativas para o controle de congestionamento multicast na Internet é o protocolo ALMTF (Adaptive Layered Multicast TCP-Friendly), o qual faz parte do projeto SAM (Sistema Adaptativo Multimídia). Uma vantagem desse algoritmo é inferir o nível de congestionamento da rede, determinando a taxa de recebimento mais apropriada para cada receptor. Além disso, ele realiza o controle da banda recebida, visando à justiça e a imparcialidade com os demais tráfegos concorrentes. O ALMTF foi desenvolvido originalmente em uma Tese de doutorado e teve a sua validação no simulador de redes NS-2 (Network Simulator). Este trabalho tem como objetivo estender o protocolo para uma rede real, implementando, validando os seus mecanismos e propondo novas alternativas que o adaptem para esse ambiente. Além disso, efetuar a comparação dos resultados reais com a simulação, identificando as diferenças e promovendo as pesquisas experimentais na área. / One of the obstacles for the widespread use of the multicast in the global Internet is the development of adequate protocols for congestion control. One factor that contributes for this problem is the heterogeneity of equipments, enlaces and conditions of access of the receivers, which increases the implementation and validation complexity of these protocols. Due to the number (thousands) of receivers simultaneously involved in multicast, the challenge of these protocols is even higher. Besides the issues related to the network congestion, it is necessary to consider factors such as synchronism, feedback control, fairness, among others. For these reasons, the multicast congestion control protocols have been a topic of intense research in recent years. The ALMTF protocol (Adaptive Layered Multicast TCP-Friendly), which is part of project SAM, is one of the alternatives for the multicast congestion control in the Internet. One advantage of this algorithm is its ability to infer the network congestion level, assigning the best receiving rate for each receptor. Besides that, the protocol manages the received rate, aiming to achieve fairness and impartiality with the competing network traffic. The ALMTF was developed originally in a Ph.D. Thesis and had its validation under NS-2 simulator. The goal this work is to extend the protocol ALMTF for a real network, validating its mechanisms and considering new alternatives to adapt it for this environment. Moreover, to make the comparison of the real results with the simulation, being identified the differences and promoting the experimental research in the area.