Systems-NET

Although the benefits from adopting a systems engineering approach across most sectors is evident to many industrialist and academics, many senior managers are reluctant to incorporate a systems methodology into their processes, as they are yet to appreciate the benefits.

The first grand challenge is therefore to communicate case study material where systems engineering tools have been successfully applied. However, a persuasive argument for the adoption of systems engineering should take into account details of costs for the transition and preparation of skilled systems engineers.

Design of  complex systems can originate vast amount of information that includes the inherent design and aspects of systems operation during a long service life, which can extend in some cases to several decades.  The development of a system that allows searching, storing and securing information, including methods of annotating the information to permit rapid retrieval and interpretation of data is very necessary.

Most complex systems have a lifespan that extends two or three decades. During this time the information related to the system can grow exponentially. The information can involve the original design, the decisions taken, the final product and in-service data. At a small-scale level such information could also include diagrams, photographs, minutes, and associated emails.  A particular problem is to consider that certain decisions in a system are taken against a specific requirement or context but over a period of time the original basis for a design may be forgotten or lost as personal moves or retires. For this reason it is very important to develop methods to store and manage information so that it is possible to access it at the lowest possible level as well as at a high level of abstraction to satisfy a particular query.  Several challenges within this context are: Managing access to sensitive commercial data in a product development enterprise, the effective use of distributed databases and the creation of an electronic diagnostic consultant that is able to search the data for relevant information.

Modelling and simulation support the design of systems allowing testing of solutions and assessing systems performance. This is particularly true where system testing on real equipment is not cost effective or where additional data is required. The third grand challenge is concerned with developing modelling and simulation techniques to allow systems analysts to concentrate on critical systems parameters/behaviours and evaluate their potential results

As the complexity and the scale of a system increases, so does the problem of developing an accurate and validated model. When dealing with complex systems involving humans in the loop, the mayor weakness is that there is not a comprehensive model of human behaviour or cognitive performance. Our ability to build and use complex models has been enhanced by the use of systems engineering techniques such as hierarchical decomposition, object-oriented modelling and programming, graphical depiction of systems behaviour, visual analytics, model based systems engineering, and agent based modelling. Multi-paradigm and multi-scale modelling at different levels of abstraction also constitute a definite advancement in many application domains. However challenges remain in many instances, for example in the case of systems of systems (SOS); these are neither designed from scratch nor reach a particular end state so they challenge the systems community in terms of whether it is possible or even practical to undertake extensive testing.

Modelling and simulation constitute a specialised area requiring high levels of knowledge and skill within the application domain. As systems become more complex and multi-disciplinary, the challenge extends to requiring the support of domain and subject matter experts.  The key challenge of any model and simulation activity is how accurately it represents the system under investigation. The verification, validation and assurance of models and simulations require confidence in them. Some of the specific challenges in this area are: shortened system design and development time-scale, model representation of the human within the context of a system model, comprehensive modelling frameworks, and complete model driven engineering environments.

Modelling and simulation require the correct environment to allow the systems engineer to investigate the impact of complex interactions taking place throughout the life of a system. The fourth Grand Challenge is therefore to create an environment in which systems models can start as abstractions that later evolve into a full representation of the system. In order to achieve this all the models in a system must be compiled into an executable code and downloaded into the required target hardware. Model Systems Base Engineering MSBE is a step in this direction but at this time very few of its tools provide a traceable route to catalogue a system evolution from requirements to implementation. 

An example of this challenge is illustrated by UK Defence Industrial Strategy, which has introduced the requirement for capability based acquisitions rather than only equipment and performance specifications. This has had a profound effect on the modelling and simulation tools used for taking design decisions in the associated engineering systems. The areas of research for this challenge include the development of heterogeneous modelling tools which encapsulate human performance data, the creation of open architecture models that allow integration and incorporation of legacy system models from across different vendor tools, the generation of executable code from model based representation so that transformation tools can support systems from requirements to implementation and the development of tools that support design-model-evaluate-redesign-model…build processes.

The increase level of the integration of products/processes  in the manufacturing industry has become extremely complex requiring an inter-disciplinary team to devise a system from design to in-service deployment. The cost of a malfunctioning system can be staggering lingering to the aftermath of a failure. Unfortunately, some systems faults are not detected until the moment they happen in service and then only after an incident occurs. The fifth Gran Challenge is therefore to avoid errors in systems by designing them for reliability from the outset with an emphasis on Verification, Validation and Assurance.

As systems become more complex their reliability becomes a major concern for manufactures. The need for VV&A cannot be overemphasised as design faults escalate costs of production. There are several examples that illustrate unanticipated problems in industry:  In January 2010, Toyota announced recalls of approximately 5.2 million vehicles for the pedal entrapment/floor mat problem, and a further 2.3 million vehicles for an accelerator pedal issue. Since January 2012 the following products were recalled for a variety of operational failures: Lexus RX400h, Crown Upright Fan Heater, Renault Kangoo, Saltrock Scooter, Sterimar nasal sprays, BMW Mini Cooper, Mitsubishi ASX, Koni Shock absorber, Isuzu Trooper 3.0 Diesel, MAN Trucks, Volvo Xc60, Ford Fiesta and Poly Implant Prostheses Silicone Breast Implants.  These faults result in both financial and reputational costs to the companies involved. The inevitable complexity of modern products makes it difficult to understand where in design undesirable emergent behaviour appears or simply, oversights in the design process result. There are several important factors that need to be addressed including functional correctness issues such as quality, and many non-functional issues such as dependability. Unless the system has been designed for reliability from the outset a single bit error can have catastrophic consequences for the whole system.

Over recent years systems have grown from single user to include geographically distributed multi-user systems. The advent of high bandwidth networks has further enable data connectivity between hardware and software systems around the world.  However some of these connections are unregulated and can lead to security breaches. As the scale of the systems grow the chances of failure increase too. The sixth Grand Challenge is therefore the design of ultra-scalable heterogeneous systems.

Large-scale systems present a significant challenge in terms of architecture, security and privacy of the data. Ultra scale systems need to be open in order to evolve into larger and more complex systems and include rapidly evolving technology.  The architecture of such systems must support from one to many hundreds of nodes with the required level of security and integrity. The Internet is a good example; its adoption has led to a phenomenal plethora of uncontrolled/unregulated connected computing resources resulting in a system that is prone to attack. On the other hand grid computing although based on similar network architectures is not as vulnerable because security certificates have been built in from the outset. As large systems have emerged there are growing examples of cases where they have failed to deliver. For example, the National Programme for IT in England was designed to reform the way the National Health Service (NHS) uses information, and hence improves services and the quality of patient care. The estimate cost of the programme was £12.7 billion and required substantial organisational and cultural change if it was to be successful. At the outset of the program the aim was the implementation of the systems by 2010 but latest forecasts suggest it is likely to take until 2015.

Cloud computing has recently appeared as a computing architecture in which dynamically scalable and often distributed computing resources are loosely-coupled to create a virtual computer comprising a cluster of networked computers acting in concert to process extremely large tasks as a service over the internet. It is anticipated that the demand for networked assets could increase further because of the number of invisible computers embedded in everyday objects around us. Many of these will have the capability to communicate with each other, which will in turn create more complex systems. Such systems present new challenges and prompt the following questions: What are the key features of an infrastructure capable of supporting ultra-scale network enabled systems? Will ever be possible to test and validate the whole system?

Autonomous systems replace human based systems and are beneficial in many respects such cost effectiveness, infrastructure requirements and ability to perform without emotion. However, it is necessary to ensure this type of systems act responsibly. The seventh Grand Challenge is about ensuring that autonomous systems interact with each other for their individual and collective good, in particular as they evolve into ultra-scalable systems. 

A true autonomous system has little if any reliance on human operators. Such system incorporate disciplines including artificial intelligence, distributed computing, object oriented systems, software engineering, economics, sociology and organisational science. Currently the field of autonomous and multi-agent systems is receiving a great deal of attention and represent a rapidly expansive area of research and development.  However this challenge is particularly concerned with ultra-scalable autonomous systems that can exist from a single autonomous system to a large civilisation of autonomous systems working co-operatively. The form and function of the individuals in the civilisation need not to be all the same and diversify according to need. Clearly autonomous agents and multi-agent systems require effective ways of analysing, designing and forming more complex systems.  The design and construction of ultra-scalable autonomous systems should be such that they are able to automatically reconfigure to tackle new tasks and evolve with technology. Each individual autonomous system in the ultra-scale must be fully aware of its environment and undertake task cooperatively with neighbouring systems. The ultimate goal for such systems is to respond intelligently to an event, in the same fashion as humans do.