Optimization of multiclass queueing networks with changeover times via the achievable region approach: Part I, the single-station case

We address the performance optimization problem in a single-stationmulticlass queueing network with changeover times by means of theachievable region approach. This approach seeks to obtainperformance bounds and scheduling policies from the solution of amathematical program over a relaxation of the system's performanceregion. Relaxed formulations (including linear, convex, nonconvexand positive semidefinite constraints) of this region are developedby formulating equilibrium relations satisfied by the system, withthe help of Palm calculus. Our contributions include: (1) newconstraints formulating equilibrium relations on server dynamics;(2) a flow conservation interpretation of the constraintspreviously derived by the potential function method; (3) newpositive semidefinite constraints; (4) new work decomposition lawsfor single-station multiclass queueing networks, which yield newconvex constraints; (5) a unified buffer occupancy method ofperformance analysis obtained from the constraints; (6) heuristicscheduling policies from the solution of the relaxations.

Optimization of multiclass queueing networks with changeover times via the achievable region method: Part II, the multi-station case

We address the problem of scheduling a multi-station multiclassqueueing network (MQNET) with server changeover times to minimizesteady-state mean job holding costs. We present new lower boundson the best achievable cost that emerge as the values ofmathematical programming problems (linear, semidefinite, andconvex) over relaxed formulations of the system's achievableperformance region. The constraints on achievable performancedefining these formulations are obtained by formulatingsystem's equilibrium relations. Our contributions include: (1) aflow conservation interpretation and closed formulae for theconstraints previously derived by the potential function method;(2) new work decomposition laws for MQNETs; (3) new constraints(linear, convex, and semidefinite) on the performance region offirst and second moments of queue lengths for MQNETs; (4) a fastbound for a MQNET with N customer classes computed in N steps; (5)two heuristic scheduling policies: a priority-index policy, anda policy extracted from the solution of a linear programmingrelaxation.

Assessing an intuitive condition for stability under a range of traffic conditions via a generalised Lu-Kumar network

We argue the importance both of developing simple sufficientconditions for the stability of general multiclass queueing networks and also of assessing such conditions under a range of assumptions on the weight of the traffic flowing between service stations. To achieve the former, we review a peak-rate stability condition and extend its range of application and for the latter, we introduce a generalisation of the Lu-Kumar network on which the stability condition may be tested for a range of traffic configurations. The peak-rate condition is close to exact when the between-station traffic is light, but degrades as this traffic increases.

A fast cross-entropy method for estimating buffer overflows in queuing networks

In this paper, we propose a fast adaptive importance sampling method for the efficient simulation of buffer overflow probabilities in queueing networks. The method comprises three stages. First, we estimate the minimum cross-entropy tilting parameter for a small buffer level; next, we use this as a starting value for the estimation of the optimal tilting parameter for the actual (large) buffer level. Finally, the tilting parameter just found is used to estimate the overflow probability of interest. We study various properties of the method in more detail for the M/M/1 queue and conjecture that similar properties also hold for quite general queueing networks. Numerical results support this conjecture and demonstrate the high efficiency of the proposed algorithm.

Coupling and monotonicity of queueing processes

The main purpose of this work is to give a survey of main monotonicity properties of queueing processes based on the coupling method. The literature on this topic is quite extensive, and we do not consider all aspects of this topic. Our more concrete goal is to select the most interesting basic monotonicity results and give simple and elegant proofs. Also we give a few new (or revised) proofs of a few important monotonicity properties for the queue-size and workload processes both in single-server and multi- server systems. The paper is organized as follows. In Section 1, the basic notions and results on coupling method are given. Section 2 contains known coupling results for renewal processes with focus on construction of synchronized renewal instants for a superposition of independent renewal processes. In Section 3, we present basic monotonicity results for the queue-size and workload processes. We consider both discrete-and continuous-time queueing systems with single and multi servers. Less known results on monotonicity of queueing processes with dependent service times and interarrival times are also presented. Section 4 is devoted to monotonicity of general Jackson-type queueing networks with Markovian routing. This section is based on the notable paper [17]. Finally, Section 5 contains elements of stability analysis of regenerative queues and networks, where coupling and monotonicity results play a crucial role to establish minimal suficient stability conditions. Besides, we present some new monotonicity results for tandem networks.

Parallel scheduling of multiclass M/M/m queues: Approximate and heavy-traffic optimization of achievable performance

We address the problem of scheduling a multiclass $M/M/m$ queue with Bernoulli feedback on $m$ parallel servers to minimize time-average linear holding costs. We analyze the performance of a heuristic priority-index rule, which extends Klimov's optimal solution to the single-server case: servers select preemptively customers with larger Klimov indices. We present closed-form suboptimality bounds (approximate optimality) for Klimov's rule, which imply that its suboptimality gap is uniformly bounded above with respect to (i) external arrival rates, as long as they stay within system capacity;and (ii) the number of servers. It follows that its relativesuboptimality gap vanishes in a heavy-traffic limit, as external arrival rates approach system capacity (heavy-traffic optimality). We obtain simpler expressions for the special no-feedback case, where the heuristic reduces to the classical $c \mu$ rule. Our analysis is based on comparing the expected cost of Klimov's ruleto the value of a strong linear programming (LP) relaxation of the system's region of achievable performance of mean queue lengths. In order to obtain this relaxation, we derive and exploit a new set ofwork decomposition laws for the parallel-server system. We further report on the results of a computational study on the quality of the $c \mu$ rule for parallel scheduling.

Estimating buffer overflows in three stages using cross-entropy

In this paper we propose a fast adaptive Importance Sampling method for the efficient simulation of buffer overflow probabilities in queueing networks. The method comprises three stages. First we estimate the minimum Cross-Entropy tilting parameter for a small buffer level; next, we use this as a starting value for the estimation of the optimal tilting parameter for the actual (large) buffer level; finally, the tilting parameter just found is used to estimate the overflow probability of interest. We recognize three distinct properties of the method which together explain why the method works well; we conjecture that they hold for quite general queueing networks. Numerical results support this conjecture and demonstrate the high efficiency of the proposed algorithm.

Using the EM Algorithm to Train Neural Networks: Misconceptions and a New Algorithm for Multiclass Classification

The expectation-maximization (EM) algorithm has been of considerable interest in recent years as the basis for various algorithms in application areas of neural networks such as pattern recognition. However, there exists some misconceptions concerning its application to neural networks. In this paper, we clarify these misconceptions and consider how the EM algorithm can be adopted to train multilayer perceptron (MLP) and mixture of experts (ME) networks in applications to multiclass classification. We identify some situations where the application of the EM algorithm to train MLP networks may be of limited value and discuss some ways of handling the difficulties. For ME networks, it is reported in the literature that networks trained by the EM algorithm using iteratively reweighted least squares (IRLS) algorithm in the inner loop of the M-step, often performed poorly in multiclass classification. However, we found that the convergence of the IRLS algorithm is stable and that the log likelihood is monotonic increasing when a learning rate smaller than one is adopted. Also, we propose the use of an expectation-conditional maximization (ECM) algorithm to train ME networks. Its performance is demonstrated to be superior to the IRLS algorithm on some simulated and real data sets.

Multiclass Classification of SRBCTs

A novel approach to multiclass tumor classification using Artificial Neural Networks (ANNs) was introduced in a recent paper cite{Khan2001}. The method successfully classified and diagnosed small, round blue cell tumors (SRBCTs) of childhood into four distinct categories, neuroblastoma (NB), rhabdomyosarcoma (RMS), non-Hodgkin lymphoma (NHL) and the Ewing family of tumors (EWS), using cDNA gene expression profiles of samples that included both tumor biopsy material and cell lines. We report that using an approach similar to the one reported by Yeang et al cite{Yeang2001}, i.e. multiclass classification by combining outputs of binary classifiers, we achieved equal accuracy with much fewer features. We report the performances of 3 binary classifiers (k-nearest neighbors (kNN), weighted-voting (WV), and support vector machines (SVM)) with 3 feature selection techniques (Golub's Signal to Noise (SN) ratios cite{Golub99}, Fisher scores (FSc) and Mukherjee's SVM feature selection (SVMFS))cite{Sayan98}.

QoS evaluation of bandwidth schedulersin IPTV networks offered SRD fluidvideo traffic: a simulation study

IPTV is now offered by several operators in Europe, US and Asia using broadcast video over private IP networks that are isolated from Internet. IPTV services rely ontransmission of live (real-time) video and/or stored video. Video on Demand (VoD)and Time-shifted TV are implemented by IP unicast and Broadcast TV (BTV) and Near video on demand are implemented by IP multicast. IPTV services require QoS guarantees and can tolerate no more than 10-6 packet loss probability, 200 ms delay, and 50 ms jitter. Low delay is essential for satisfactory trick mode performance(pause, resume,fast forward) for VoD, and fast channel change time for BTV. Internet Traffic Engineering (TE) is defined in RFC 3272 and involves both capacity management and traffic management. Capacity management includes capacityplanning, routing control, and resource management. Traffic management includes (1)nodal traffic control functions such as traffic conditioning, queue management, scheduling, and (2) other functions that regulate traffic flow through the network orthat arbitrate access to network resources. An IPTV network architecture includes multiple networks (core network, metronetwork, access network and home network) that connects devices (super head-end, video hub office, video serving office, home gateway, set-top box). Each IP router in the core and metro networks implements some queueing and packet scheduling mechanism at the output link controller. Popular schedulers in IP networks include Priority Queueing (PQ), Class-Based Weighted Fair Queueing (CBWFQ), and Low Latency Queueing (LLQ) which combines PQ and CBWFQ.The thesis analyzes several Packet Scheduling algorithms that can optimize the tradeoff between system capacity and end user performance for the traffic classes. Before in the simulator FIFO,PQ,GPS queueing methods were implemented inside. This thesis aims to implement the LLQ scheduler inside the simulator and to evaluate the performance of these packet schedulers. The simulator is provided by ErnstNordström and Simulator was built in Visual C++ 2008 environmentand tested and analyzed in MatLab 7.0 under windows VISTA.

Call admission control algorithms in OFDM-based wireless multiservice networks

Orthogonal frequency division multiplexing (OFDM) is becoming a fundamental technology in future generation wireless communications. Call admission control is an effective mechanism to guarantee resilient, efficient, and quality-of-service (QoS) services in wireless mobile networks. In this paper, we present several call admission control algorithms for OFDM-based wireless multiservice networks. Call connection requests are differentiated into narrow-band calls and wide-band calls. For either class of calls, the traffic process is characterized as batch arrival since each call may request multiple subcarriers to satisfy its QoS requirement. The batch size is a random variable following a probability mass function (PMF) with realistically maximum value. In addition, the service times for wide-band and narrow-band calls are different. Following this, we perform a tele-traffic queueing analysis for OFDM-based wireless multiservice networks. The formulae for the significant performance metrics call blocking probability and bandwidth utilization are developed. Numerical investigations are presented to demonstrate the interaction between key parameters and performance metrics. The performance tradeoff among different call admission control algorithms is discussed. Moreover, the analytical model has been validated by simulation. The methodology as well as the result provides an efficient tool for planning next-generation OFDM-based broadband wireless access systems.

On the Bleeding Edge : Debloating Internet Access Networks

As ever more devices are connected to the internet, and applications turn ever more interactive, it becomes more important that the network can be counted on to respond reliably and without unnecessary delay. However, this is far from always the case today, as there can be many potential sources of unnecessary delay. In this thesis we focus on one of them: Excess queueing delay in network routers along the path, also known as bufferbloat. We focus on the home network, and treat the issue in three stages. We examine latency variation and queueing delay on the public internet and show that significant excess delay is often present. Then, we evaluate several modern AQM algorithms and packet schedulers in a residential setting, and show that modern AQMs can almost entirely eliminate bufferbloat and extra queueing latency for wired connections, but that they are not as effective for WiFi links. Finally, we go on to design and implement a solution for bufferbloat at the WiFi link, and also design a workable scheduler-based solution for realising airtime fairness in WiFi. Also included in this thesis is a description of Flent, a measurement tool used to perform most of the experiments in the other papers, and also used widely in the bufferbloat community.

Contrasting Effects Of Land Use Intensity And Exotic Host Plants On The Specialization Of Interactions In Plant-herbivore Networks.

Human land use tends to decrease the diversity of native plant species and facilitate the invasion and establishment of exotic ones. Such changes in land use and plant community composition usually have negative impacts on the assemblages of native herbivorous insects. Highly specialized herbivores are expected to be especially sensitive to land use intensification and the presence of exotic plant species because they are neither capable of consuming alternative plant species of the native flora nor exotic plant species. Therefore, higher levels of land use intensity might reduce the proportion of highly specialized herbivores, which ultimately would lead to changes in the specialization of interactions in plant-herbivore networks. This study investigates the community-wide effects of land use intensity on the degree of specialization of 72 plant-herbivore networks, including effects mediated by the increase in the proportion of exotic plant species. Contrary to our expectation, the net effect of land use intensity on network specialization was positive. However, this positive effect of land use intensity was partially canceled by an opposite effect of the proportion of exotic plant species on network specialization. When we analyzed networks composed exclusively of endophagous herbivores separately from those composed exclusively of exophagous herbivores, we found that only endophages showed a consistent change in network specialization at higher land use levels. Altogether, these results indicate that land use intensity is an important ecological driver of network specialization, by way of reducing the local host range of herbivore guilds with highly specialized feeding habits. However, because the effect of land use intensity is offset by an opposite effect owing to the proportion of exotic host species, the net effect of land use in a given herbivore assemblage will likely depend on the extent of the replacement of native host species with exotic ones.

Complex networks: the key to systems biology

Though introduced recently, complex networks research has grown steadily because of its potential to represent, characterize and model a wide range of intricate natural systems and phenomena. Because of the intrinsic complexity and systemic organization of life, complex networks provide a specially promising framework for systems biology investigation. The current article is an up-to-date review of the major developments related to the application of complex networks in biology, with special attention focused on the more recent literature. The main concepts and models of complex networks are presented and illustrated in an accessible fashion. Three main types of networks are covered: transcriptional regulatory networks, protein-protein interaction networks and metabolic networks. The key role of complex networks for systems biology is extensively illustrated by several of the papers reviewed.