State-of-the-art video-text retrieval (VTR) methods typically involve fully fine-tuning a pre-trained model (e.g. CLIP) on specific datasets. However, this can result in significant storage costs in practical applications as a separate model per task must be stored. To address this issue, we present our pioneering work that enables parameter-efficient VTR using a pre-trained model, with only a small number of tunable parameters during training. Towards this goal, we propose a new method dubbed Multimodal Video Adapter (MV-Adapter) for efficiently transferring the knowledge in the pre-trained CLIP from image-text to video-text. Specifically, MV-Adapter utilizes bottleneck structures in both video and text branches, along with two novel components. The first is a Temporal Adaptation Module that is incorporated in the video branch to introduce global and local temporal contexts. We also train weights calibrations to adjust to dynamic variations across frames. The second is Cross Modality Tying that generates weights for video/text branches through sharing cross modality factors, for better aligning between modalities. Thanks to above innovations, MV-Adapter can achieve comparable or better performance than standard full fine-tuning with negligible parameters overhead. Notably, MV-Adapter consistently outperforms various competing methods in V2T/T2V tasks with large margins on five widely used VTR benchmarks (MSR-VTT, MSVD, LSMDC, DiDemo, and ActivityNet). Codes will be released.

Graphs have been widely used in image processing and understanding tasks. We introduce a novel graph generation model which greatly reduces the size of the traditional pixel-based graph. Based on the generated graph, we propose two feature extraction methods which utilize spectral graph information, and apply the features to image. Experiments show that our proposed oscillatory image heat content and weighted heat content spectrum features are more robust to small distortions and changes of viewpoint than the image heat content feature we proposed previously. The features are also capable of capturing important image structural information of the image and perform well alone or in combination with other low-level image features.

The main duty of key management is to keep cryptographic keys in secret. However, it is difficulty to quantitatively assess that how well does a key management scheme protect the keys. In this paper, we propose to use reliability theory, which was mainly used to evaluate performance persistence for engineering systems, to estimate the performance of key management schemes. The reliability analysis leads to counter-intuitive results such as the widely deployed periodic key update scheme is ineffective when key thefts are possible. The analysis also shows that using password with an electronic security token for authentication is a strong security measure in the beginning but is unreliable in the long run. In general, the reliability analysis demonstrates that current key management schemes focus too much on postponing the first key theft from occurring but lack of considerations on quickly recovering stolen keys. In the later part of this paper, we discuss possible directions that may improve the reliability of key management schemes.

In this paper, we propose a complex network based low-level feature for image retrieval systems based on the graph representation of the image and the mathematical theory of diffusion over manifolds. We show that the proposed image feature is invariant to non-structural changes on images and performs well in hand written digits classification task. We also show that performance of image retrieval with existing low-level features could be improved by combining with the proposed feature.

There has been a great deal of interest recently in load-balancing switches due to their simple architecture and high forwarding bandwidth. Nevertheless, the mis-sequencing problem of the original load-balancing switch hinders the performance of underlying TCP applications. Several load-balancing switch designs have been proposed to address this mis-sequencing issue. They solve this mis-sequencing problem at the cost of either algorithmic complexity or special hardware requirements. In this paper, we address the mis-sequencing problem by introducing a three-stage load-balancing switch architecture enhanced with an output load-balancing mechanism. This three-stage load-balancing switch achieves a high forwarding capacity while preserving the order of packets without the need of costly online scheduling algorithms. Theoretical analyses and simulation results show that this three-stage load-balancing switch provides a transmission delay that is upper-bounded by that of an output-queued switch plus a constant that depends only on the number of input/output ports, indicating the same forwarding capacity as an output-queued switch.

This paper proposes a semi-random backoff (SRB) method that enables resource reservation in contention-based wireless LANs. The proposed SRB is fundamentally different from traditional random backoff methods because it provides an easy migration path from random backoffs to deterministic slot assignments. The central idea of the SRB is for the wireless station to set its backoff counter to a deterministic value upon a successful packet transmission. This deterministic value will allow the station to reuse the time-slot in consecutive backoff cycles. When multiple stations with successful packet transmissions reuse their respective time-slots, the collision probability is reduced, and the channel achieves the equivalence of resource reservation. In case of a failed packet transmission, a station will revert to the standard random backoff method and probe for a new available time-slot. The proposed SRB method can be readily applied to both 802.11 DCF and 802.11e EDCA networks with minimum modification to the existing DCF/EDCA implementations. Theoretical analysis and simulation results validate the superior performance of the SRB for small-scale and heavily loaded wireless LANs. When combined with an adaptive mechanism and a persistent backoff process, SRB can also be effective for large-scale and lightly loaded wireless networks.

We present a new virtual indexing method for estimating host connection degrees for high speed links. It is based on the virtual connection degree sketch where a compact sketch of network traffic is built by generating the associated virtual bitmaps for each host. Each virtual bitmap consists of a fixed number of bits selected randomly from a shared bit array by a new method for recording the traffic flows of the corresponding host. The shared bit array is efficiently utilized by all hosts since its every bit is shared by the virtual bitmaps of multiple hosts. To reduce the “noise” contaminated in a host's virtual bitmaps due to sharing, we propose a new method to generate the “filtered” bitmap used to estimate host connection degree. Furthermore, it can be easily implemented in parallel and distributed processing environments. The experimental and testing results based on the actual network traffic show that the new method is accurate and efficient.

Packet losses in the network have a considerable performance impact on transport-layer throughput. For reliable data transfer, lost packets require retransmissions and thus cause very long delays. This tail of the packet delay distribution causes performance problems. There are several approaches to trading off networking resources up-front to reduce long delays for some packets (e.g., forward error correction, network coding). We propose packet pacing as an alternative that changes traffic characteristics favorably by adding intentional delay in packet transmissions. This intentional delay counters the principle of best effort but can reduce the burstiness of traffic and improve overall network operation - in particular in network with small packet buffers. As a result, pacing improves transport-layer performance, providing a tradeoff example where small amounts of additional delay can significantly increase connection bandwidth. We present a Queue Length Based Pacing (QLBP) algorithm that paces network traffic using a single queue and that can be implemented with small computational and memory overhead. We present a detailed analysis on delay bounds and the quantitative impact of QLBP pacing on network traffic. Through simulation, we show how the proposed pacing technique can improve connection throughput in small-buffer networks.

802.11e HCCA (Hybrid coordination function Controlled Channel Access) exhibits good QoS provisioning for constant bit rate (CBR) video streams in a single collision domain. However, its performance degrades significantly for variable bit rate (VBR) video streams particularly in multi-collision domains. In addition, HCCA has the disadvantage of high complexity. In this paper, we introduce a deterministic backoff (DEB) method into the HCCA mechanism, which achieves virtual polling via carrier sense on the wireless channel. DEB inherits the merit of HCCA that achieves near collision-free channel access in single collision domain. It also exhibits robust performance in multi-collision domains. Simulation results show that DEB supports well both CBR and VBR video streams.

For the optical packet-switching routers to be widely deployed in the Internet, the size of packet buffers on routers has to be significantly small. Such small-buffer networks rely on traffic with low levels of burstiness to avoid buffer overflows and packet losses. We present a pacing system that proactively shapes traffic in the edge network to reduce burstiness. Our queue length based pacing uses an adaptive pacing on a single queue and paces traffic indiscriminately where deployed. In this work, we show through analysis and simulation that this pacing approach introduces a bounded delay and that it effectively reduces traffic burstiness. We also show that it can achieve higher throughput than end-system based pacing.

This paper introduces a set of low-complexity algorithms that when coupled with link layer retransmission mechanisms, strengthen wireless communication security. Our basic idea is to generate a series of secrets from inevitable transmission errors and other random factors in wireless communications. Because these secrets are constantly extracted from the communication process in realtime, we call them dynamic secrets. Dynamic secrets have interesting security properties. They offer a complementary mechanism to existing security protocols. Even if the adversary exploits a vulnerability and steals the underlying system secret, security can be automatically replenished. In many scenarios, it is also possible to bootstrap a secure communication with the dynamic secrets.

Security becomes a prominent issue for mobile cloud computing since valuable information moves to the cloud. An important security problem is how the users identify themselves to the cloud. If an attacker is capable of faking or stealing user credentials, such as passwords and digital certificates, current security measures are not sufficient to safe-guard user's valuable information in the cloud. This paper proposes a set of light weight algorithms to generate dynamic credential to defend against such powerful attackers. Many communication randomness, like the user mobility, were commonly believed harmful for security. In this paper, they are converted to helpful elements to generate dynamic credentials. Dynamic credential introduces interesting security properties.

QoS provisioning for video streaming over 802.11 home networks is challenging due to the tight bandwidth and delay constraints. Currently, both 802.11e HCCA and EDCA have their disadvantages when delivering video over 802.11 networks. EDCA exhibits degraded QoS performance for increased number of stations, while HCCA has high complexity, and also suffers from the impact of inter-cell interference. In this paper, we propose a new backoff mechanism which is fundamentally different from traditional random backoff methods in 802.11. The mechanism achieves resource reservation by reusing one or multiple time slots for transmission in consecutive backoff cycles. The basic idea is borrowed from R-ALOHA, but several modifications are made to allow it to work in the context of CSMA network. An additive increase multiplicative decrease based algorithm is proposed to control resource allocation for video streaming in the 802.11 network. Theoretical analysis and numerical simulations are conducted to validate the effectiveness of the method and the algorithm.

In recent years, a series of power saving (PS) protocols has been proposed in the family of 802.11 standards to save energy for mobile devices. To evaluate their performance, many works have been carried out on testbeds or simulation platforms. However, till now, there is a lack of accurate theoretical models to analyze the performance for these protocols. In an effort to fill this gap, we present a Markov chain-based analytical model in this paper to model these PS protocols, with its focus on multicast services in wireless LANs. The proposed analytical model successfully captures the key characteristic of the power saving system: the data delivery procedure starts periodically at the previously negotiated time, but ends at a rather random time with its distribution depending on the end time of data delivery in the last delivery period as well as the arrival rate of incoming traffic. In the situations with light to moderate traffic loads and under the Poisson assumption for incoming traffic, the amount of data delivered between consecutive delivery periods possesses the Markov property, which builds up our Markov chain-based model. For incoming traffic with long-range dependence (LRD), a multistate Markov-Modulated Poisson Process (MMPP) is used to approximate the traffic, making the analytical model valid in more general cases. We verify our model by simulations on ns2 and the results show that the model can faithfully predict the performance of these PS protocols over a wide variety of testing scenarios.

In this paper, we propose a semi-random backoff (SRB) mechanism to combat network collisions for 802.11 DCF/EDCA in wireless LANs. SRB is fundamentally different from traditional random backoff methods because it provides a form of resource reservation at no extra cost. The key idea of SRB is to set the backoff counter to a deterministic value upon a successful transmission. This allows a station to reuse a time slot in consecutive backoff cycles without colliding with other stations, thus achieving resource reservations. In cases of failed transmissions due to network collisions or channel errors, SRB automatically reverts to the standard random backoff method. SRB can be readily applied to both 802.11 DCF and 802.11e EDCA networks with minimum modification to the existing DCF/EDCA implementations. Theoretical analysis and simulations are conducted to evaluate this new method and our results validate its good performance in small to middle-sized 802.11 wireless LANs.

Recently there has been a great deal of interest in load-balancing switches due to their simple architecture and high bandwidth. In this paper we propose a three-stage load- balancing switch along with output load-balancing to address the mis-sequencing problem. We show that our proposed scheme provides a delay guarantee bounded by the delay of an OQ switch with the same input traffic plus a constant while achieving 100% throughput for admissible traffic with (sigma, rho) -upper constraint.

This paper investigates the benefit of network coding for TCP traffic in a wireless mesh network. We implement network coding in a real 802.11a wireless mesh network and measure TCP throughput in such a network. Unlike previous implementations of network coding in mesh networks, we use off-the-shelf hardware and software and do not modify TCP or the underlying MAC protocol. Therefore, our implementation can be easily exported to any operational wireless mesh network with minimal modifications. Furthermore, the TCP throughput improvement reported in this paper is due solely to network coding and is orthogonal to other improvements that can be achieved by optimizing other system components such as the MAC protocol. We conduct extensive measurements to understand the relation between TCP throughput and network coding in different mesh topologies. We show that network coding not only reduces the number of transmissions by sending multiple packets via a single transmission but also results in a smaller loss probability due to reduced contention on the wireless medium. Unfortunately, due to asynchronous packet transmissions, there is often little opportunity to code resulting in small throughput gains. Coding opportunity can be increased by inducing small delays at intermediate nodes. However, this extra delay at intermediate nodes results in longer round-trip-times that adversely affect TCP throughput. Through experimentation, we find a delay in the range of 1 ms to 2 ms to maximize TCP throughput. For the topologies considered in this paper, network coding improves TCP throughput by 10% to 85%.

It is generally believed harmful to have transmission errors in the wireless communications. The high decoding complexity of dense parity check codes is unfavorable. This paper proposes to apply these two "negative" facts to enable the secrecy sharing with the information theoretical security. We claim that the secrecy sharing is always possible if the wiretap channel is not error-free, regardless of the main channel performance. Particularly, the proposed secrecy sharing protocol can provide provable and testable security using the existing wireless technologies.

As many people rely on e-mail communications for business and everyday life, Internet e-mail worms constitute one of the major security threats for our society. Unlike scanning worms such as Code Red or Slammer, e-mail worms spread over a logical network defined by e-mail address relationships, making traditional epidemic models invalid for modeling the propagation of e-mail worms. In addition, we show that the topological epidemic models presented by M. Boguna, et al. (2000) largely overestimate epidemic spreading speed in topological networks due to their implicit homogeneous mixing assumption. For this reason, we rely on simulations to study e-mail worm propagation in this paper. We present an e-mail worm simulation model that accounts for the behaviors of e-mail users, including e-mail checking time and the probability of opening an e-mail attachment. Our observations of e-mail lists suggest that an Internet e-mail network follows a heavy-tailed distribution in terms of node degrees, and we model it as a power-law network. To study the topological impact, we compare e-mail worm propagation on power-law topology with worm propagation on two other topologies: small-world topology and random-graph topology. The impact of the power-law topology on the spread of e-mail worms is mixed: E-mail worms spread more quickly on a power-law topology than on a small-world topology or a random-graph topology, but immunization defense is more effective on a power-law topology.

Recently there is an increasing interest in using directional antennas and multi-channel for ad-hoc wireless networks. The two techniques can also be used for infrastructure local area networks. In this paper, we present the architecture and the MAC layer design for a multi-channel multi-sector directional antenna wireless local area network (MCMSDA WLAN). This paper has three key components. First, we describe the network architecture and propose a TDMA-based MAC layer. Second, we formulate the channel time allocation problem as an optimization problem and introduce a Lagrangian relaxation based load-balancing algorithm as a solution. The simulation study shows that the algorithm obtains good sub-optimal solution very quickly. Finally, we propose two methods to accelerate the load-balancing process further

In defending against various network attacks, such as distributed denial-of-service (DDoS) attacks or worm attacks, a defense system needs to deal with various network conditions and dynamically changing attacks. Therefore, a good defense system needs to have a built-in "adaptive defense" functionality based on cost minimization-adaptively adjusting its configurations according to the network condition and attack severity in order to minimize the combined cost introduced by false positives (misidentify normal traffic as attack) and false negatives (misidentify attack traffic as normal) at any time. In this way, the adaptive defense system can generate fewer false alarms in normal situations or under light attacks with relaxed defense configurations, while protecting a network or a server more vigorously under severe attacks. In this paper, we present concrete adaptive defense system designs for defending against two major network attacks: SYN flood DDoS attack and Internet worm infection. The adaptive defense is a high-level system design that can be built on various underlying nonadaptive detection and filtering algorithms, which makes it applicable for a wide range of security defenses

We present a game-theoretic study of the selfish behavior of TCP users when they are allowed to use multiple concurrent TCP connections so as to maximize their goodputs or other utility functions. We refer to this as the TCP connection game. A central question we ask is whether there is a Nash equilibrium in such a game, and if it exists, whether the network operates efficiently at such a Nash equilibrium. Combined with the well known PFTK TCP model (1998), we study this question for three utility functions that differ in how they capture user behavior. The bad news is that the loss of efficiency or price of anarchy can be arbitrarily large if users have no resource limitations and are not socially responsible. The good news is that, if either of these two factors is considered, efficiency loss is bounded. This may partly explain why there will be no congestion collapse if many users use multiple connections.

In this paper, we study the interaction between overlay routing and traffic engineering (TE) in a single autonomous system (AS). We formulate this interaction as a two-player non-cooperative non-zero sum game, where the overlay tries to minimize the delay of its traffic and the TE's objective is to minimize network cost. We study a Nash routing game with best-reply dynamics, in which the overlay and TE have equal status, and take turns to compute their optimal strategies based on the response of the other player in the previous round. We prove the existence, uniqueness and global stability of Nash equilibrium point (NEP) for a simple network. For general networks, we show that the selfish behavior of an overlay can cause huge cost increases and oscillations to the whole network. Even worse, we have identified cases, both analytically and experimentally, where the overlay's cost increases as the Nash routing game proceeds even though the overlay plays optimally based on TE's routing at each round. Experiments are performed to verify our analysis.

Routing optimization is used to find a set of routes that minimizes cost (delay, utilization). Previous work has addressed this problem for the case of a known, static end-to-end traffic matrix. In the Internet, it is difficult to accurately estimate a traffic matrix, and the constantly changing nature of Internet traffic makes it costly to maintain optimal routing by responding to traffic changes. Thus, it is of interest to maintain a set of routes that are "good" for a number of different possible traffic scenarios. In this paper, we explore ways to find an optimal set of routes with multiple traffic matrices to minimize expected cost. We focus on two general approaches, source-destination routing and destination routing. In the case of source-destination routing, we extend existing methods with a single traffic matrix to solve the optimization problem with multiple traffic matrices: we extend the convex optimization solution methods for a single traffic matrix to the multiple traffic matrix case; we also extend the gradient-based solution methods for a single traffic matrix to the multiple traffic matrix case. However, the multiple traffic matrix case requires many more control variables. In the case of destination routing, we encounter many more differences from the single traffic matrix case. The loop-free property, which is valid for the single traffic matrix case, is no longer valid for the multiple traffic matrix case, and it is difficult to extend existing methods for a single traffic matrix to solve the optimization problem with multiple traffic matrices. We show that it is NP-complete even to determine the feasibility of multiple traffic matrices. We thus propose and evaluate a heuristic algorithm for this case.

After many Internet-scale worm incidents in recent years, it is clear that a simple self-propagating worm can quickly spread across the Internet and cause severe damage to our society. Facing this great security threat, we need to build an early detection system that can detect the presence of a worm in the Internet as quickly as possible in order to give people accurate early warning information and possible reaction time for counteractions. This paper first presents an Internet worm monitoring system. Then, based on the idea of "detecting the trend, not the burst" of monitored illegitimate traffic, we present a "trend detection" methodology to detect a worm at its early propagation stage by using Kalman filter estimation, which is robust to background noise in the monitored data. In addition, for uniform-scan worms such as Code Red, we can effectively predict the overall vulnerable population size, and estimate accurately how many computers are really infected in the global Internet based on the biased monitored data. For monitoring a nonuniform scan worm, especially a sequential-scan worm such as Blaster, we show that it is crucial for the address space covered by the worm monitoring system to be as distributed as possible.

Email worms constitute one of the major Internet security problems. We present an email worm model that accounts for the behaviors of email users by considering email checking time and the probability of opening email attachments. Email worms spread over a logical network defined by email address relationship, which plays an important role in determining the spreading dynamics of an email worm. Our observations suggest that the node degrees of an email network are heavy-tailed distributed. We compare email worm propagation on three topologies: power law, small world and random graph topologies; and then study how the topology affects immunization defense on email worms. The impact of the power law topology on the spread of email worms is mixed: email worms spread more quickly on a power law topology than on a small world topology or a random graph topology, but immunization defense is more effective on a power law topology than on the other two.

In this paper, we present a time series-based approach for managing power in mobile processors and disks that see multimedia workloads. Since multimedia applications impose soft real-time constraints, a key goal of our approach is to reduce energy consumption of multimedia applications without degrading performance. We present simple statistical techniques based on time series to dynamically compute the processor and I/O demands of multimedia applications and present techniques to dynamically vary the voltage settings and rotational speeds of mobile processors and disks, respectively. We implement our approaches in the Linux kernel running on a Sony Transmeta laptop and in a trace-driven simulator. Our experiments show that, compared to the traditional system-wide CPU voltage scaling approaches, our technique can achieve up to a 38.6% energy saving while delivering good performance to applications. Simulation results for our disk power management technique show a 20.3% reduction in energy consumption without any significant performance loss when compared to a traditional disk power management scheme.

After the Code Red incident in 2001 and the SQL Slammer in January 2003, it is clear that a simple self-propagating worm can quickly spread across the Internet, infects most vulnerable computers before people can take effective countermeasures. The fast spreading nature of worms calls for a worm monitoring and early warning system. In this paper, we propose effective algorithms for early detection of the presence of a worm and the corresponding monitoring system. Based on epidemic model and observation data from the monitoring system, by using the idea of "detecting the trend, not the rate" of monitored illegitimated scan traffic, we propose to use a Kalman filter to detect a worm's propagation at its early stage in real-time. In addition, we can effectively predict the overall vulnerable population size, and correct the bias in the observed number of infected hosts. Our simulation experiments for Code Red and SQL Slammer show that with observation data from a small fraction of IP addresses, we can detect the presence of a worm when it infects only 1% to 2% of the vulnerable computers on the Internet.

We first study the impact of traffic burstiness on the accuracy of timed-stepped-simulation (TSS). Then we propose a compensation scheme to expand TSS's working ranges.

In this paper we present analytical techniques that can be used to better understand the behavior of malware, a generic term that refers to all kinds of malicious software programs propagating on the Internet, such as e-mail viruses and worms. We develop a modeling methodology based on Interactive Markov Chains that is able to capture many aspects of the problem, especially the impact of the underlying topology on the spreading characteristics of malware. We propose numerical methods to obtain useful bounds and approximations in the case of very large systems, validating our results through simulation. An analytic methodology represents a fundamentally important step in the development of effective countermeasures for future malware activity. Furthermore, we believe our approach can help to understand a wide range of "dynamic interactions on networks", such as routing protocols and peer-to-peer applications.

No abstract available.

The call stack of a program execution can be a very good information source for intrusion detection. There is no prior work on dynamically extracting information from the call stack and effectively using it to detect exploits. In this paper we propose a new method to do anomaly detection using call stack information. The basic idea is to extract return addresses from the call stack, and generate an abstract execution path between two program execution points. Experiments show that our method can detect some attacks that cannot be detected by other approaches, while its convergence and false positive performance is comparable to or better than the other approaches. We compare our method with other approaches by analyzing their underlying principles and thus achieve a better characterization of their performance, in particular on what and why attacks will be missed by the various approaches.

Since CDN simulations are known to be highly memory-intensive, in this paper, we argue the need for reducing the memory requirements of such simulations. We propose a novel memory-efficient data structure that stores cache state for a small subset of popular objects accurately and uses approximations for storing the state for the remaining objects. Since popular objects receive a large fraction of the requests while less frequently accessed objects consume much of the memory space, this approach yields large memory savings and reduces errors. We use bloom filters to store approximate state and show that careful choice of parameters can substantially reduce the probability of errors due to approximations. We implement our techniques into a user library for constructing proxy caches in CDN simulators. Our experimental results show up to an order of magnitude reduction in memory requirements of CDN simulations, while incurring a 5-10% error.

The Code Red worm incident of July 2001 has stimulated activities to model and analyze Internet worm propagation. In this paper we provide a careful analysis of Code Red propagation by accounting for two factors: one is the dynamic countermeasures taken by ISPs and users; the other is the slowed down worm infection rate because Code Red rampant propagation caused congestion and troubles to some routers. Based on the classical epidemic Kermack-Mckendrick model, we derive a general Internet worm model called the two-factor worm model. Simulations and numerical solutions of the two-factor worm model match the observed data of Code Red worm better than previous models do. This model leads to a better understanding and prediction of the scale and speed of Internet worm spreading.

We use a previously developed nonlinear dynamic model of TCP to analyze and design active queue management (AQM) control systems using random early detection (RED). First, we linearize the interconnection of TCP and a bottlenecked queue and discuss its feedback properties in terms of network parameters such as link capacity, load and round-trip time. Using this model, we next design an AQM control system using the RED scheme by relating its free parameters such as the low-pass filter break point and loss probability profile to the network parameters. We present guidelines for designing linearly stable systems subject to network parameters like propagation delay and load level. Robustness to variations in system loads is a prime objective. We present no simulations to support our analysis.

In this paper we study a previously developed linearized model of TCP and active queue management (AQM). We use classical control system techniques to develop controllers well suited for the application. The controllers are shown to have better theoretical properties than the well known RED controller. We present guidelines for designing stable controllers subject to network parameters like load level propagation delay etc. We also present simple implementation techniques which require a minimal change to RED implementations. The performance of the controllers are verified and compared with RED using ns simulations. The second of our designs, the proportional integral (PI) controller is shown to outperform RED significantly.

Data communication networks have been experiencing tremendous growth in size, complexity, and heterogeneity over the last decade. This trend poses a significant challenge to the design of scalable performance evaluation methodologies. In this paper we propose time-stepped hybrid simulation (TSHS) to deal with the scalability issue faced by traditional packet-level discrete event simulation methods. TSHS is a framework that offers the user the flexibility to choose the simulation time scale so as to trade off the computational cost of the simulation with its fidelity. Simulation speedup is achieved by evaluating the system at coarser time-scales. The potential loss of simulation accuracy when fine-time-scale behavior is evaluated at a coarser time-scale is studied both analytically and experimentally.

151-160

Practical sequential estimation algorithms inherently involve "iterated conditioning"; that is, the estimate at any algorithm step is conditioned on information passed from the previous step. Except in special cases, standard probability theory cannot deal with operations like iterated conditioning, so it is difficult even to formulate algorithms that make optimal use of the available information. The previously developed product space conditional event algebra (PS-CEA) allows for the definition of "conditional events" as regular events in a probability space, and thus for the extension of Bayesian analysis to cases of iterated conditioning. We describe the PS-CEA approach and give an example of its use in deriving iterated image estimators when the information passed from one stage to the next is the current image estimate.

Diffusion networks (i.e., networks with stochastic differential neuron equations) are attractive as means of implementing segmentation algorithms because of their potential for fast, parallel processing. However, the complexity of the required neuron circuitry makes many analog implementations impractical. In this paper, we consider a digital version of a diffusion network proposed by Wong (1991). The analog neuron equations are first discretized into versions driven by binary noise sequences. The discretized equations are then implemented with digital differential analyzers (DDAs) to produce a digital diffusion network. While much work remains to be done to finalize the digital network design, preliminary simulation test results of the digital network applied to image segmentation are encouraging.

Performance index functions with integer parameters are important in the evaluation and analysis of computer systems, communication networks and general distributed and parallel processing systems. These functions are often difficult to evaluate for large integers. On the other hand, most of these performance functions have nice properties such as monotonicity, convexity, analyticity, as well as obtainable asymptotic behavior. These properties make it possible to use rational approximants to extrapolate the performance functions. In this paper we present the rational approximants for several example systems. Our numerical results suggest that this approach could lead to a promising tool for many applications.

In this paper we present a new approach to the design of multihop lightwave networks with connectivity patterns that can be dynamically reconfigured. We introduce a practical algorithm that efficiently reconfigures the connectivity diagram by reassigning wavelengths to best fit the current traffic pattern. A newly developed stochastic optimization algorithm is used to obtain the mapping of a regular structure into a WDM star, while optimizing the system performance measures. The optimization tool is generated and can be used with any objective function which is a function of the traffic patterns, the system control and the system configuration. The results show that this is a practical tool and the resulting structures have a superior system performance.>

An approach to the design of multihop lightwave networks with connectivity patterns that can be dynamically reconfigured is presented. A practical algorithm that efficiently reconfigures the connectivity diagram by reassigning wavelengths to best fit the current traffic pattern is introduced. A stochastic optimization algorithm is used to obtain the mapping of a regular structure into a wavelength-division-multiplexed (WDM) star, while optimizing the system performance measures. The optimization tool is general and can be used with any objective function which is a function of the traffic patterns, the system control, and the system configuration. The results show that this is a practical tool and that the resulting structures have a superior system performance.>