Fernando Paganini
Universidad ORT Uruguay, School of Engineering, Faculty Member
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
This paper discusses network flow control from the perspective of multivariable feedback theory. We build on recent research on flow control based on pricing signals, which has cast these problems in terms amenable to mathematical... more
This paper discusses network flow control from the perspective of multivariable feedback theory. We build on recent research on flow control based on pricing signals, which has cast these problems in terms amenable to mathematical analysis, and more specifically we ...
Research Interests:
Research Interests: Computer Science, Distributed Computing, Congestion Control, Resource Allocation, Queueing theory, and 15 moreStability, Computer Network, Protocols, Network Capacity, Fluid flow, Steady state, High performance, Delays, Tcp, Decentralized Control, Dynamic Stability, Round Trip Time, Internet, Active Queue Management, and network congestion
This paper is aimed at designing a congestion control system that scales gracefully with network capacity, providing high utilization, low queueing delay, dynamic stability, and fairness among users. In earlier work we had developed... more
This paper is aimed at designing a congestion control system that scales gracefully with network capacity, providing high utilization, low queueing delay, dynamic stability, and fairness among users. In earlier work we had developed fluid-level control laws that achieve the first three objectives for arbitrary networks and delays, but were forced to constrain the resource allocation policy. In this paper we extend the theory to include dynamics at TCP sources, preserving the earlier features at fast time-scales, but permitting sources to match their steady-state preferences, provided a bound on round-trip-times is known.
Research Interests:
We consider distributed parameter systems where the underlying dynamics are spatially invariant, and where the controls and measurements are spatially distributed. These systems arise in many applications such as the control of platoons,... more
We consider distributed parameter systems where the underlying dynamics are spatially invariant, and where the controls and measurements are spatially distributed. These systems arise in many applications such as the control of platoons, smart structures, or distributed flow control through the fluid boundary. For optimal control problems involving quadratic criteria such as LQR, H2 and H1 , it is shown how to reduce the optimization to a family of problems over spatial frequency. We also show that optimal controllers have an inherent degree of decentralization, which leads to a practical distributed architecture. Under a more general class of performance criteria, a general result is given showing that optimal controllers inherit the spatial invariance structure of the plant.
Research Interests:
This paper concerns network congestion control, focusing on a recently proposed "primal-dual" control law that has been shown to give local stability and desirable equilibrium properties for networks of arbitrary capacity and a... more
This paper concerns network congestion control, focusing on a recently proposed "primal-dual" control law that has been shown to give local stability and desirable equilibrium properties for networks of arbitrary capacity and a fixed bound on delay.
Research Interests:
Research Interests:
Research Interests:
A breakthrough that has allowed the Internet to expand by five orders of magnitude in size and in backbone speed in the last 15 years was the invention in 1988 by Jacobson of an end-to-end congestion control algorithm in TCP (Transmission... more
A breakthrough that has allowed the Internet to expand by five orders of magnitude in size and in backbone speed in the last 15 years was the invention in 1988 by Jacobson of an end-to-end congestion control algorithm in TCP (Transmission Control Protocol) [1]. The algorithm is a distributed and asynchronous method to share network resources among competing users. TCP has been carrying more than 90% of the Internet traffic and Jacobson’s algorithm is instrumental in preventing the Internet from congestion collapse while the Web exploded in the 1990s. This algorithm, designed when most parts of the Internet could barely carry the traffic of a single uncompressed voice call, however, cannot scale to the future ultrascale networks that must be able to carry the traffic of 1.5 million concurrent voice calls. This is due to serious equilibrium and stability problems in high capacity long distance networks, and has led to doubts on whether the current TCP paradigm of end-to-end control co...
Research Interests:
Research Interests:
Research Interests:
We consider resource allocation questions for computing infrastructures with multiple server instances. In particular, the joint optimization of active service capacity, load balancing between clusters of servers, and task scheduling at... more
We consider resource allocation questions for computing infrastructures with multiple server instances. In particular, the joint optimization of active service capacity, load balancing between clusters of servers, and task scheduling at each cluster, under conditions of data locality which imply different service rates for different cluster locations.Building on previous work, we formulate a convex optimization problem, and use Lagrange duality to decompose it between the different decision variables. We include regularization terms from proximal methods to obtain continuous control laws for load balancing and scheduling, and optimize the remaining variables through primal-dual gradient dynamics. We prove convergence of the resulting control laws to the desired optimal points, and demonstrate its behavior by simulations.
Research Interests:
Dado el crecimiento de los sistemas peer-to-peer y en particular del protocolo BitTorrent para la distribución de contenido en redes, han surgido modelos que buscan comprender su funcionamiento y predecir su desempeño. En este trabajo... more
Dado el crecimiento de los sistemas peer-to-peer y en particular del protocolo BitTorrent para la distribución de contenido en redes, han surgido modelos que buscan comprender su funcionamiento y predecir su desempeño. En este trabajo realizamos estudios de simulación en NS-2 para verificar la validez de uno de estos modelos, que predice la operación alrededor de un equilibrio estable. Encontramos situaciones en las que el modelo predice los equilibrios adecuadamente, pero otras en los que se observan oscilaciones importantes, que pueden conducir a fallas de funcionamiento. A partir de estos resultados proponemos una variante del protocolo BitTorrent, tendiente a lograr mayor estabilidad. A través de simulaciones y del modelado matemático, evaluamos las mejoras de desempeño que ofrece esta variante.
Research Interests:
In this paper we analyze the hit performance of cache systems that receive file requests with general arrival distributions and different popularities. We consider timer-based (TTL) policies, with differentiated timers over which we... more
In this paper we analyze the hit performance of cache systems that receive file requests with general arrival distributions and different popularities. We consider timer-based (TTL) policies, with differentiated timers over which we optimize. The optimal policy is shown to be related to the monotonicity of the hazard rate function of the inter-arrival distribution. In particular for decreasing hazard rates, timer policies outperform the static policy of caching the most popular contents. We provide explicit solutions for the optimal policy in the case of Pareto-distributed inter-request times and a Zipf distribution of file popularities, including a compact fluid characterization in the limit of a large number of files. We compare it through simulation with classical policies, such as least-recently-used and discuss its performance. Finally, we analyze extensions of the optimization framework to a line network of caches.
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Ensuring predictability is a crucial goal for service systems. Traditionally, research has focused on designing systems that ensure predictable performance for service requests. Motivated by applications in cloud computing and electricity... more
Ensuring predictability is a crucial goal for service systems. Traditionally, research has focused on designing systems that ensure predictable performance for service requests. Motivated by applications in cloud computing and electricity markets, this article focuses on a different form of predictability: predictable allocations of service capacity. The focus of the article is a new model where service capacity can be scaled dynamically and service deferrals (subject to deadline constraints) can be used to control the variability of the active service capacity. Four natural policies for the joint problem of scheduling and managing the active service capacity are considered. For each, the variability of service capacity and the likelihood of deadline misses are derived. Further, the paper illustrates how pricing can be used to provide incentives for jobs to reveal deadlines and thus enable the possibility of service deferral in systems where the flexibility of jobs is not known to t...
Research Interests:
In recent years large strides have been taken in analytical approaches to network congestion control. These methods, originating in the academic community [1, 2, 3, 4, 5, 6, 7] are quickly evolving into viable alternatives to correct... more
In recent years large strides have been taken in analytical approaches to network congestion control. These methods, originating in the academic community [1, 2, 3, 4, 5, 6, 7] are quickly evolving into viable alternatives to correct deficiencies of current Internet ...
Research Interests:
Research Interests:
... (continued after index) Page 5. Geir E. Dullerud Fernando Paganini A Course in Robust Control Theory A Convex Approach With 36 Illustrations Springer Page 6. ... ISBN 0-387-98945-5 (hardcover : alk. paper) 1. Robust control. 2.... more
... (continued after index) Page 5. Geir E. Dullerud Fernando Paganini A Course in Robust Control Theory A Convex Approach With 36 Illustrations Springer Page 6. ... ISBN 0-387-98945-5 (hardcover : alk. paper) 1. Robust control. 2. Control theory. I. Paganini, Fernando G. II. Title. ...
Research Interests:
This paper studies state-space methods to compute the frequency domain conditions for robust H/sub 2/ performance under structured uncertainty, recently obtained by the author (1995, 1996). For uncertainty scalings which are constant or... more
This paper studies state-space methods to compute the frequency domain conditions for robust H/sub 2/ performance under structured uncertainty, recently obtained by the author (1995, 1996). For uncertainty scalings which are constant or spanned by fixed basis functions, the problem is reduced to a finite dimensional convex condition of the linear matrix inequality type.
Research Interests:
Research Interests:
Providing frequency regulation services in power networks has become an important part of network operation, traditionally carried out by fast responding generators. In this paper we consider regulation services from the demand side,... more
Providing frequency regulation services in power networks has become an important part of network operation, traditionally carried out by fast responding generators. In this paper we consider regulation services from the demand side, through a new actor in the power market: a demand aggregator that manages a large number of consumer loads. The aggregator exploits the deferability of certain loads to control the consumption profile and thus reduce regulation needs, or even provide regulation services to others. We analyze this control through macroscopic ODE models inspired by queueing systems, where a fluid state represents load quantities. Two versions are considered: a one-state model that tracks the entire load population, and a two-state version that separately tracks quantities of currently deferrable and non-deferrable loads. The control input is the fraction of deferrable loads that are active, and is controlled using a combination of feedforward for tracking of a reference signal, and feedback to reduce the impact of random fluctuations. The performance of such controllers is evaluated by simulation using regulation signals from real networks.
Research Interests:
Abstract Power control is considered as an efficient scheme to mitigate co-channel and multiple-access interference in cellular radio systems. Various approaches have been proposed in recent years to design power control algorithms. We... more
Abstract Power control is considered as an efficient scheme to mitigate co-channel and multiple-access interference in cellular radio systems. Various approaches have been proposed in recent years to design power control algorithms. We focus on the feedback ...