Recent advances in computing and network technologies have contributed much to the successful handling of certain problems in biology, physics, economics, etc. that until recently were thought too difficult to be analysed. These complex systems problems tend to share a number of interesting properties. For example, they have many components that interact in some interesting way and these components or agents may be similar or differ in important characteristics. The systems are dynamic in nature, interact with their environments and adapt their internal structures as a consequence of such interaction. A key feature of such a system is that the nonlinear interactions among its components can lead to interesting emergent behaviour.
The overall aim of this special issue is to bring together the latest/innovative knowledge and advances in handling complex systems, which may depend largely on methods from artificial intelligence, mathematics, statistics, operational research and engineering. There were more than 60 papers submitted to this special issue, which covered various aspects of complex systems, including nonlinear dynamics, time-series analysis, dynamic systems, cellular automata, artificial life, evolutionary computation, game theory, neural networks, multi-agents and heuristic search methods. With a rigorous review process, eight papers have been selected that provide solutions, or early promises, to modelling, analysis and control problems of real-world complex systems, such as multi-network evolutionary systems, integrated human-engineered systems, uncertain nonlinear delayed systems, semiotics motivated project planning systems, software refactoring systems, resource discovery grid systems, brain control systems and collaborative design systems.
Multiple neural network systems can often solve complex problems more effectively than their monolithic counterparts. Modular neural networks tackle a complex problem by decomposing it into simpler subproblems and then solving them. Unlike the decomposition in modular neural networks, a neural network ensemble usually includes redundant component nets and is often inspired by statistical theories. In the first paper, “Multi-network evolutionary systems and automatic decomposition of complex problems” by Khare et al., different types of problem decompositions are presented, and the suitability of various multi-network systems for different decompositions are discussed. A classification of various multi-network systems, in the context of problem decomposition, is obtained by exploiting these differences. Then a specific type of problem decomposition, which gives no information about the subproblems and is often ignored in literature, is discussed in detail and novel modular neural network architecture for problem decomposition is presented. Finally, a co-evolutionary model is presented, which is used to design and optimize such modular neural networks with sub-task specific modules.
Reliability, performance and enhanced operating range of large-scale human-engineered systems (e.g. multi-national business, space exploration and military command and control) are of paramount importance to successful completion of their missions. The operation of these missions can be viewed as an interconnected complex dynamical system. The complexity in such systems emerges as a consequence of nonlinear, non-stationary, multi-timescale and uncertain dynamics of the mutually interacting subsystems. In the second paper “Integrated decision and control of human-engineered complex systems” by Tolani et al. the authors formulate a comprehensive control and health management strategy for human-engineered complex dynamical systems to achieve high performance and reliability over a wide range of operation. This is done by a hierarchically structured decision and control system that synergistically combines the technologies from several systems-theoretic disciplines, such as probabilistic robust control, damage mitigating control, discrete event supervisory, decision and control and health and usage monitoring.
Most complex dynamical systems are nonlinear to some degree and often include time delays as intrinsic components; in addition, the systems may be subject to unknown perturbations. Sometimes the states of the system cannot be measured directly and, therefore, in this case, it may be desirable to obtain estimates of such states through the design of appropriate observers. In the third paper “Observer design for complex systems in the presence of uncertainties, nonlinearities and distributed delays” by Goodall and Wang, the problem of nonlinear, full-order, state observer design for a class of complex uncertain continuous-time systems with time-varying, multiple, discrete and distributed, time delays is investigated. A sufficient condition, in terms of a quadratic matrix inequality, is presented, which ensures the existence of full-order observers. In addition, it is shown that, provided the matrix inequality condition is satisfied, the zero state is globally exponentially stable, under the error dynamics of the observation process. Under additional hypotheses, the gain of the observers is characterized in terms of an arbitrary orthogonal matrix.
A large and complex project may be regarded as a system that is constrained within a temporal period and at times involves multiple organizations. An organization is a system consisting of people, activities, processes, information, resources and goals. Understanding and modelling such a project and its interrelationship with relevant organizations are essential for organizational project planning. In the fourth paper “Modelling complex systems for project planning: a semiotics motivated method” by Liu et al., the problem articulation method is introduced as a semiotic method for organizational infrastructure modelling. This method offers a suite of techniques, which enables the articulation of the business, technical and organizational requirements, delivering an infrastructural framework to support the organization. It works by eliciting and formalizing (e.g. processes, activities, relationships, responsibilities, communications, resources, agents, dependencies and constraints), and mapping these abstractions to represent the manifestation of the “actual” organization. Many analysts forgo organizational modelling methods and use localized ad hoc and point solutions, but this is not amenable for organizational infrastructures modelling. A case study of the infrared atmospheric sounding interferometer is used to demonstrate the applicability of the proposed method, and to examine its relevancy and significance in dealing with the innovation and changes in the organizations.
The majority of information systems these days engender a high level of complexity through the extent of possible inputs to testing, required processing and consequent outputs. In fact, complexity permeates every level of this model for an information system. Complexity thus has a direct effect on the extent to which a system needs to be tested, through those inputs. Complexity also inhibits the ease with which a system can be modified since more time needs to be devoted to assessment of change complexity and resulting tests. Reduction of complexity is the goal of every developer when initially developing a system and, as importantly, after the system has been developed and inevitable changes are made. In the fifth paper “Understanding the complexity of refactoring in software systems: a tool-based approach” by Advani et al. the authors analyse an automated technique for extracting the typical changes made to various Java systems over different versions of its lifetime. The goal is to identify areas of change where complexity can be examined more thoroughly and aid thus given to the developer when maintaining systems. A generic tool is developed specifically for this task and the results show new and promising insights into the way systems behave as they evolve. In particular, the complex refactorings are relatively rare compared with more simple refactorings.
A grid is defined as a coordinated resource sharing and problem-solving environment in dynamic, multi-institutional virtual organizations. It is a multifaceted system with many components and innovative features. Next-generation grid systems are heading for globally collaborative, service-oriented and live information systems, which have the hallmarks of complex systems, namely, adaptation, self-organization and emergence. No one designed the whole grid or the metabolic-like processes within users and resources. In such a complex system, resource discovery is a critical activity. In the sixth paper “An adaptive social network-inspired approach to resource discovery for the complex grid systems” by Gao et al., the authors apply the principles and concepts in social networks to design a decentralized, survivable, and adaptive resource discovery approach in complex grid systems. It is shown via simulation that the proposed approach can form relationship among clusters and significantly improve the discovery performance. It can also adapt well to different resource distributions and user request patterns, and survive from the changes of dynamic environments, including variable-biased user requests and agent amounts as well as partial failure of the agents.
Attention is arguably the brain's highest control system, functioning as a filter on lower-level activations (up to semantic level) to allow through the filter activations from those stimuli or motor responses of value in solving tasks, and represses distracters. In the seventh paper “Attention as the control system of the brain” by Taylor, the details of the control system involved, including an inverse model controller for the signal for the movement of the focus of attention to a particular goal, and a forward model for predicting the attended state of the world are discussed, and their firm basis on experimental data is described. The results of various simulations are then considered. How value maps and the interaction of attention and emotion occur is discussed next. Further, executive powers are shown to be able to be developed on the basis of attention control. Finally, the manner in which a reference copy may help in understanding the control basis of consciousness is outlined.
Today, economic globalization is creating competitive pressures on industry to minimize the time to bring products to market. Competing suppliers, designers, manufacturers and customers form a link via internet or traditional media in new product development where the dynamics and uncertainty rule out any centralized control. In a complex collaborative design system, participants are intricate organizational, technological and financial meta-systems operating under dynamic market conditions and uncertain business circumstances. Collaborative design system is characterized by very high complexity and is typically heterogeneous and very dynamic, involving complex interactions among many humans, applications, services and devices. In the last paper “Analysis and control of complex collaborative design systems” by Qin and Sun, a novel method is proposed for modelling the complexity of collaborative design systems based on its analysis. The interaction and interfacing properties among many components of a complex design system are analysed from different viewpoints, and a complexity model for collaborative design is established accordingly. In order to simplify the complexity and improve the performance of collaborative design, a general solution of decomposing a whole system into sub-systems and using unified interface mechanism between them is proposed. This proposed solution is tested via a case study, and it is shown that the proposed solution is both significant and practical.
This special issue is a timely reflection of the research progress in the area of analysis and control for complex systems. Finally, I would like to acknowledge all authors for their efforts in submitting high-quality papers. I am also very grateful to the reviewers for their thorough and on-time reviews of the papers. Last, but not least, my deepest gratitude goes to Professor George Klir (editor-in-chief) and Monika Fridrich (editorial assistant) of International Journal of General Systems for their consideration, help and advice.