Complex and large scale systems and networks have evolved in many fields from communication data networks, banking, financial systems, bio-engineering, robotics, and of course control engineering. In control engineering the main driving force behind the expansion of large-scale complex systems has been the relatively cheap cost of installing advanced instrumentation hardware, the cheap cost of attaining high computing power and the desire to "see" and "monitor" as much of the process as possible.

As a result where as previously a typical small scale manufacturing process would be controlled by a few dozen loops, this has now risen to thousands of loops complemented with extensive data logging facilities. Classical or even modern control systems design applications are reaching their practical limits in their abilities to analyse and study these complex systems. In particular, classical techniques generally lack the framework to handle large scale systems and/or lead to exceptionally poor performance, and modern methods have a tendency to completely disregard model complexity and general overtly complex control solutions.

Evidently, the fact that still the overwhelming majority of industrial control systems run on a mixture of PI and decentralized control systems means that no viable alternative has yet been developed and a great amount of resources remain unutilized in these systems and networks. For example in a typical industrial control system, the controllers might be able to use only as little as 5% of the information collected from the process for control purposes.

Within industry, a large proportion of these complex and very large scale systems are found in resource and energy heavy industries such as petro-chemicals, pulp and paper, power generation etc.. Therefore in these cases, running the process inefficiently, not only leads to lesser profits but increased environmental pollution and energy consumption.

Numerous surveys have consistently shown that improved control will lead to more sustainable manufacturing, improved bottom line returns, efficiency gains and improved raw material yields.

Research into complex systems has not been very active until recent years. This has been the result of several factors. Previously the low cost of energy resulted in relatively healthy profit margins for the industry. This is compared to the current situation where a large section of heavy industry has been terminated in the UK, and a sizable portion are operating at a loss. Secondly, the environmental cost of manufacturing was not a high priority issue in the past and therefore the impact of inefficient processing was considered acceptable. Furthermore, expensive sensing and instrumentation equipment meant only a handful of information was collected from the process. Finally, system complexity considerations had gone largely ignored at the design stage because study of model complexity evokes combinatorial optimization problems which are extremely difficult to solve due to not having convenient mathematical features (such as convexity) There is an added layer of difficulty if in addition to model complexity, the performance is also being considered. Such hybrid problems yield mixed-integer-nonlinear programmes (MINLP) which are the most difficult type of optimization problems to solve. However, recent increases in energy costs, the elevated importance of the environment and the explosion in the number of currently operational complex systems has led to a particularly active research effort in this emerging field.

Consequently we propose a special issue to be published in IET Control Theory and Applications on the Control and analysis of large scale complex systems and networks. We believe this will make an interesting, timely and highly relevant special issue because not only is it addressing a challenging problem with cutting edge development in modeling, adaptation, optimization and analysis algorithms, but it is a highly interdisciplinary subject.

Previous studies have shown that Complex systems encountered in control, image processing, bio-engineering, cognitive science, machine learning, social intelligence, computer and data networks, traffic networks and financial systems, all share some common fundamental properties (such as having to deal with potentially significant non-linearities, non-stationarity, uncertainty and multi-variable interactions) and often the tools developed to study one particular class is applicable to the study of other types. This point has been repeatedly demonstrated in the course of multi-disciplinary events such as the IEEE International conferences on Networking, Sensing and Control, IEEE Swarm Intelligence Symposia, and the biennial International UKACC Control Conferences.

Original papers covering all theoretical and applied areas of complex control systems are invited. The proposed publication schedule of the Special Issue is given below:

Proposed timeline: Deadline for Submission of selected Expanded/Revised Contol2008 and other papers: 31 Dec 2008.

Authors to have received a 1st decision by:	30 Apr 2009
Final decision of acceptance:	30 Sep 2009
Online and Print publication:	Late 2009

Special Issue Guest Editors:

Dr. Amir Hussain
Stirling University
Dr Amin Nobakhti and Professor Hong Wang
The University of Manchester


IET Publishing Dept. contact:

Miss Joanna Lawrie: jlawrie@theiet.org IET CTA Editorial Assistant