A theoretical framework for simulated annealing

Simulated Annealing has been a very successful general algorithm for the solution of large, complex combinatorial optimization problems. Since its introduction, several applications in different fields of engineering, such as integrated circuit placement, optimal encoding, resource allocation, logic synthesis, have been developed. In parallel, theoretical studies have been focusing on the reasons for the excellent behavior of the algorithm. This paper reviews most of the important results on the theory of Simulated Annealing, placing them in a unified framework. New results are reported as well.     

A theoretical framework for simulated annealing

Simulated Annealing has been a very successful general algorithm for the solution of large, complex combinatorial optimization problems. Since its introduction, several applications in different fields of engineering, such as integrated circuit placement, optimal encoding, resource allocation, logic synthesis, have been developed. In parallel, theoretical studies have been focusing on the reasons for the excellent behavior of the algorithm. This paper reviews most of the important results on the theory of Simulated Annealing, placing them in a unified framework. New results are reported as well.This research was sponsored by SRC and DARPA monitored by SNWSC under contract numbers N00039-87-C-012 and N00039-88-C-0292.     

A neural gain scheduling network controller for nonholonomicsystems

We propose a neural gain scheduling network controller (NGSNC) to improve the gain scheduling controller for nonholonomic systems. We derive the neural networks that can approximate the gain scheduling controller arbitrarily well when the sampling frequency satisfies the sampling theorem. We also show that the NGSNC is independent of the sampling time. The proposed NGSNC has the following important properties: 1) same performance as the continuous-parameter gain scheduling controller; 2) less computing time than the continuous-parameter gain scheduling controller; 3) good robustness against the sampling intervals; and 4) straightforward stability analysis. We then show that some of nonholonomic systems can be converted to equivalent linear parameter-varying systems. As a result, the NGSNC can stabilize nonholonomic systems     

A framework for opportunistic scheduling in wireless networks

We present a method, called opportunistic scheduling, for exploiting the time-varying nature of the radio environment to increase the overall performance of the system under certain quality of service/fairness requirements of users. We first introduce a general framework for opportunistic scheduling, and then identify three general categories of scheduling problems under this framework. We provide optimal solutions for each of these scheduling problems. All the proposed scheduling policies are implementable online; we provide parameter estimation algorithms and implementation procedures for them. We also show how previous work by us and others directly fits into or is related to this framework. We demonstrate via simulation that opportunistic scheduling schemes result in significant performance improvement compared with non-opportunistic alternatives.     

A framework for reinforcement-based scheduling in parallelprocessor systems

Task scheduling is important for the proper functioning of parallel processor systems. The static scheduling of tasks onto networks of parallel processors is well-defined and documented in the literature. However, in many practical situations a priori information about the tasks that need to be scheduled is not available. In such situations, tasks usually arrive dynamically and the scheduling should be performed on-line or “on the fly”. In this paper, we present a framework based on stochastic reinforcement learning, which is usually used to solve optimization problems in a simple and efficient way. The use of reinforcement learning reduces the dynamic scheduling problem to that of learning a stochastic approximation of an unknown average error surface. The main advantage of the proposed approach is that no prior information is required about the parallel processor system under consideration. The learning system develops an association between the best action (schedule) and the current state of the environment (parallel system). The performance of reinforcement learning is demonstrated by solving several dynamic scheduling problems. The conditions under which reinforcement learning can used to efficiently solve the dynamic scheduling problem are highlighted     

A theoretical framework for an intelligent design catalogue

Product catalogues constitute a valuable source of information for engineers engaged in design activities. Unfortunately, these catalogues provide only limited support to engineers in the earlier, conceptual stages of design. This research proposes the intelligent design catalogue consisting of a virtual design environment linked to catalogues of standard components. Engineers develop their design concepts within the virtual environment and refer to the catalogues as these concepts are refined. The selected components are assembled within the design environment. The intelligent design catalogue provides search aids as well as assessment tools. The theoretical framework draws on several engineering areas. Manufacturing demonstrates how process plans can be developed in a virtual environment independently of the machines on the shop floor just as products can be conceptually designed independently of the standard components available. The standard components themselves can be grouped borrowing from classification schemes of group technology. Object-oriented programming (OOP) provides an environment for the development of the software that runs the intelligent design catalogue. As the objects of OOP parallel standard components, OOP also serves as a design paradigm after which the catalogue can be modelled. Design theory suggests frameworks for developing a (semi-) hierarchical structure for cataloguing parts.     

A theoretical framework for multiple neural network systems

Multiple neural network systems have become popular techniques for tackling complex tasks, often giving improved performance compared to single network systems. For example, modular systems can provide improvements in generalisation through task decomposition, whereas multiple classifier and regressor systems typically improve generalisation through the ensemble combination of redundant networks. Whilst there has been significant focus on understanding the theoretical properties of some of these multi-net systems, particularly ensemble systems, there has been little theoretical work on understanding the properties of the generic combination of networks, important in developing more complex systems, perhaps even those a step closer to their biological counterparts. In this article, we provide a formal framework in which the generic combination of neural networks can be described, and in which the properties of the system can be rigorously analysed. We achieve this by describing multi-net systems in terms of partially ordered sets and state transition systems. By way of example, we explore an abstract version of learning applied to a generic multi-net system that can combine an arbitrary number of networks in sequence and in parallel. By using the framework we show with a constructive proof that, under specific conditions, if it is possible to train the generic system, then training can be achieved by the abstract technique described.     

A small gain framework for networked cooperative force-reflecting teleoperation

For cooperative force-reflecting teleoperation over networks, conventional passivity-based approaches have limited applicability due to nonpassive slave–slave interactions and irregular communication delays imposed by networks. In this paper, a small gain framework for stability analysis design of cooperative network-based force reflecting teleoperator systems is developed. The framework is based on a version of weak input-to-output practical stability (WIOPS) nonlinear small gain theorem that is applicable to stability analysis of large-scale network-based interconnections. Based on this result, we design a cooperative force-reflecting teleoperator system which is guaranteed to be stable in the presence of multiple network-induced communication constraints by appropriate adjustment of local control gains and/or force-reflection gains. Experimental results are presented that confirm the validity of the proposed approach.     

MFHS: A modular scheduling framework for heterogeneous system

Current innovative distributed architectures, proposing on-line services, involve more and more computing resources. From a provider point of view, the platform management leads to challenging problematic relating to resource allocation, which involve different kind of quality of service parameters, the provider has to focus on to keep his platform reliable and efficient. MFHS is a modular generic framework, which can be adapted to any distributed computing environment. Structured in modules, MFHS allows to discover the existing computing resources in terms of computing performance, network throughput and disk I/O speeds (Resources Discovery module) and to predict how the experiment should behave (P_i value). As the setting up of real experiments is often complex, MFHS allows: to make theoretical experimentation (based on models), to use any kind of distributed emulators, or to deploy experiments on real-experimental platforms. In this article, these three environments are used to highlight the reliability of MFHS (measured P_i=90% against 94% for the predicted P_i). Deployment and scheduling studies have also been achieved using an experimental Cloud based on OpenStack while Emulab test-bed has been used as emulator. During experiments, four QoS parameters are taken into account (Resources Monitoring module): energy consumption, cost, resource utilization, and makespan. These studies also includes a new heuristic called MMin, based on Max-Min and Min-Min algorithms. Experimentation section, proposes a detailed comparative analysis of these algorithms in terms of QoS results, while the abilities of the proposed heuristic MMin regarding the makespan metric is shown.     

A framework for delivery scheduling in the poultry industry

This paper presents a framework for scheduling the collection of broiler chickens from poultry houses (SCPH) and delivering them to abattoirs, which is especially suitable for small- and medium-size businesses. Many such companies in Brazil draw up SCPH manually. Computational implementation provides a deterministic simulator for the process. By using this tool, companies can adopt scheduling strategies taking into account the proper supply, the holding-time and the heat stress of the birds. The final objective is to reduce losses due to dead on arrival (DOA) at the abattoir and weight loss by excessive fasting. The application considers a 15-day schedule of an abattoir, the first two days of which are presented in detail. The use of the simulator allows greater control of the events and enables smaller poultry stocks at the abattoir. This better control and the hiring of at least one more team of bird catchers would lead to a significant improvement in the current collection system. On a conservative estimate, gains would result in, at least, an extra 49 metric tons annually of live chicken for processing at the abattoir studied.     

A simulation framework for priority scheduling on heterogeneous clusters

Executing heterogeneous workloads with different priorities, resource demands and performance objectives is one of the key operations for today’s data centers to increase resource as well as energy efficiency. In order to meet the performance objectives of diverse workloads, schedulers rely on evictions even resulting in waste of resources due to lost executions of evicted tasks. It is not straightforward to design priority schedulers which capture key aspects of workloads and systems and also to strike a balance between resource (in)efficiency and application performance tradeoff. To explore large space of designing such schedulers, we propose a trace-driven cluster management framework that models a comprehensive set of system configurations and general priority-based scheduling policies. In particular, we focus on the impact of task evictions on resource inefficiency and task response times of multiple priority classes driven by Google production cluster trace. Moreover, we propose a system design as a use case exploiting workload heterogeneity and introducing workload-awareness into the system configuration and task assignment.     

Combining classifiers: A theoretical framework

The problem of classifier combination is considered in the context of the two main fusion scenarios: fusion of opinions based on identical and on distinct representations. We develop a theoretical framework for classifier combination for these two scenarios. For multiple experts using distinct representations we argue that many existing schemes such as the product rule, sum rule, min rule, max rule, majority voting, and weighted combination, can be considered as special cases of compound classification. We then consider the effect of classifier combination in the case of multiple experts using a shared representation where the aim of fusion is to obtain a better estimate of the appropriatea posteriori class probabilities. We also show that the two theoretical frameworks can be used for devising fusion strategies when the individual experts use features some of which are shared and the remaining ones distinct. We show that in both cases (distinct and shared representations), the expert fusion involves the computation of a linear or nonlinear function of thea posteriori class probabilities estimated by the individual experts. Classifier combination can therefore be viewed as a multistage classification process whereby thea posteriori class probabilities generated by the individual classifiers are considered as features for a second stage classification scheme. Most importantly, when the linear or nonlinear combination functions are obtained by training, the distinctions between the two scenarios fade away, and one can view classifier fusion in a unified way.     

Hypertext functionality: A theoretical framework

This article presents a theoretical framework useful for understanding the functionality of Hypertext systems in terms of their ability to satisfy cognitive requirements for expressing concepts and their relationships. This theoretical framework is derived from a re‐interpretation of Guilford's theory of the Structure of Intellect.     

A theoretical framework for designing human-centered automotive automation systems

Increasingly sophisticated and robust automotive automation systems are being developed to be applied in all aspects of driving. Benefits, such as improving safety, task performance, and workload have been reported. However, several critical accidents involving automation assistance have also been reported. Although automation systems may work appropriately, human factors such as drivers errors, overtrust in and overreliance on automation due to lack of understanding of automation functionalities and limitations as well as distrust caused by automation surprises may trigger inappropriate human–automation interactions that lead to negative consequences. Several important methodologies and efforts for improving human–automation interactions follow the concept of human-centered automation, which claims that the human must have the final authority over the system, have been called. Given that the human-centered automation has been proposed as a more cooperative automation approach to reduce the likelihood of human–machine misunderstanding. This study argues that, especially in critical situations, the way control is handed over between agents can improve human–automation interactions even when the system has the final decision-making authority. As ways of improving human–automation interactions, the study proposes adaptive sharing of control that allows dynamic control distribution between human and system within the same level of automation while the human retains the final authority, and adaptive trading of control in which the control and authority shift between human and system dynamically while changing levels of automation. Authority and control transitions strategies are discussed, compared and clarified in terms of levels and types of automation. Finally, design aspects for determining how and when the control and authority can be shifted between human and automation are proposed with recommendations for future designs.

     

Selecting operating points for discrete-time gain scheduling

Gain scheduling is a popular approach for nonlinear control system design. A controller is obtained by designing a set of controllers at operating points and then linearly interpolating controller values between them. However, little guidance has been provided in the literature for the selection of operating points. We use interval mathematics and a classical synthesis design approach to determine a near minimal set of design points and assess the quality of a gain scheduled controller. A sufficient condition for the assignment of the system closed loop poles is developed, and an algorithm for selecting the operating points is provided. An example is given to demonstrate the approach.     

Human performance in industrial scheduling: A framework for understanding

Planning and scheduling activities have a significant impact on the performance of manufacturing enterprises. Throughout the 1980s there was a belief that computer‐led solutions would “solve” complex industrial planning and scheduling problems. However, in most manufacturing organizations, planning and scheduling still require significant human support to ensure effective performance. Although the contribution of these human resources is often highly valued, we are only beginning to develop a coherent body of knowledge that can contribute toward the successful integration of human and computer‐based planning and scheduling systems. Here we examine the state of knowledge in this domain and identify the need for field investigations. We present a framework to facilitate research in human and organizational issues in planning and scheduling in manufacturing. A structured and detailed set of research questions is developed to underpin field studies. The framework focuses on understanding the scheduling environment, the process of scheduling, and related performance issues. The application of the framework is illustrated using our own field studies, where a number of specific research questions of practical importance have been identified: what scheduling is, who carries it out, what influences scheduling practice and performance, how schedulers actually schedule, what makes a good scheduler and schedule, and what support is needed. The framework makes a valuable contribution to advancing knowledge in an area of real practical benefit to contemporary manufacturing industry. © 2001 John Wiley & Sons, Inc.     

A model-based gain scheduling approach for controlling the common-rail system for GDI engines

The progressive reduction in vehicle emission requirements have forced the automotive industry to invest in research for developing alternative and more efficient control strategies. All control features and resources are permanently active in an electronic control unit (ECU), ensuring the best performance with respect to emissions, fuel economy, driveability and diagnostics, independently from engine working point. In this article, a considerable step forward has been achieved by the common-rail technology which has made possible to vary the injection pressure over the entire engine speed range. As a consequence, the injection of a fixed amount of fuel is more precise and multiple injections in a combustion cycle can be made. In this article, a novel gain scheduling pressure controller for gasoline direct injection (GDI) engine is designed to stabilise the mean fuel pressure into the rail and to track demanded pressure trajectories. By exploiting a simple control-oriented model describing the mean pressure dynamics in the rail, the control structure turns to be simple enough to be effectively implemented in commercial ECUs. Experimental results in a wide range of operating points confirm the effectiveness of the proposed control method to tame efficiently the mean value pressure dynamics of the plant showing a good accuracy and robustness with respect to unavoidable parameters uncertainties, unmodelled dynamics, and hidden coupling terms.     

A real-time scheduling framework for embedded systems with environmental energy harvesting

Real-time scheduling refers to the problem in which there is a deadline associated with the execution of a task. In this paper, we address the scheduling problem for a uniprocessor platform that is powered by a renewable energy storage unit and uses a recharging system such as photovoltaic cells. First, we describe our model where two constraints need to be studied: energy and deadlines. Since executing tasks require a certain amount of energy, classical task scheduling like earliest deadline is no longer convenient. We present an on-line scheduling scheme, called earliest deadline with energy guarantee (EDeg), that jointly accounts for characteristics of the energy source, capacity of the energy storage as well as energy consumption of the tasks, and time. In order to demonstrate the benefits of our algorithm, we evaluate it by means of simulation. We show that EDeg outperforms energy non-clairvoyant algorithms in terms of both deadline miss rate and size of the energy storage unit.