Introduction

The SOCRADES integrated project will create new methodologies, technologies and tools for the modelling, design, implementation and operation of networked systems made up of smart embedded devices. Achieving enhanced system intelligence by co-operation of smart embedded devices pursuing common goals is relevant in many types of perception and control system environments. In general, such devices with embedded intelligence and sensing/actuating capabilities are heterogeneous, yet they need to interact seamlessly and intensively over a (wired or wireless) network.

The middleware technologies to be developed in this project will be based on the Service-Oriented Architecture (SOA) paradigm, will encompass both wired and wireless networking technologies, and will provide open interfaces enabling interoperability at the semantic level. A SOCRADES service will be a software component encapsulating device-specific functionality. This functionality is advertised to the outside world, so as to be located and invoked by other networked devices and/or applications without the latter being aware of how the functionality is implemented.

SOCRADES will exploit results of the SIRENA project, which pioneered usage  of the SOA paradigm for device-level communications.

Rationale and Overall approach

The focus is set to communication between and integration of heterogeneous embedded systems and devices, with particular emphasis on platform independence, real-time requirements, robustness and security. The underlying tendency is that the increasing availability of affordable, high-performance, low-power electronic components allows incorporating unprecedented horsepower into ever-tinier components. At the same time, Internet and (wired or wireless) Ethernet technologies are emerging as the basic carriers for interconnecting electronic devices. These technologies can be leveraged to build advanced functionality into embedded devices, thus enabling new distributed application paradigms based on interconnected "smart devices" with a high level of autonomy. This applies to many types of embedded devices, whether used in industrial automation systems, automotive electronics, telecommunications equipment, building controls, home automation, telemetry, medical instrumentation, etc. The SOCRADES project will operate in this sense of general applicability across a broad range of application domains, while using as its application cornerstone one of the most prominent embedded systems domains, viz. manufacturing and process automation.

The umbrella paradigm of SOCRADES is called "collaborative automation". The aim is to effectively utilise this paradigm, and to develop the corresponding tools and methods, so as to achieve flexible, re-configurable, scalable, interoperable network-enabled collaboration between decentralised and distributed embedded systems. Applying device-level SOA, it is expected to contribute to the creation of an open, flexible and agile environment, by extending the scope of the collaborative architecture approach through the application of a unique communications infrastructure, down from the lowest levels of the device hierarchy up into the manufacturing enterprise's higher-level business process management systems. Business application systems increasingly benefit from the adoption of SOA, allowing to flexibly compose components of heterogeneous software systems across traditional system boundaries and to swiftly adopt business processes to changing requirements. Therefore, the device platform has to be very flexible, using both wireless and wireline communication media, scalable embedded systems, and appropriate engineering methods and tools.

The SOCRADES technical approach is to create a service-oriented ecosystem where networked systems are composed from smart embedded devices interacting with the physical environment and with the enterprise environment, pursuing well-defined system goals. Taking the granularity of intelligence to the device level allows intelligent system behaviour to be obtained by composing configurations of devices that introduce incremental fractions of the required intelligence.

This approach favours adaptability and rapid reconfigurability, as re-programming of large monolithic systems is replaced by reconfiguring loosely coupled embedded units. In turn, this allows meeting business demands not foreseen at the time of design.

From a functional perspective, the focus will be on managing the vastly increased number of intelligent devices and mastering the associated complexity. From a run-time infrastructure viewpoint, the focus will be on a new breed of very flexible real-time embedded devices (wired/wireless) that are fault-tolerant, reconfigurable, safe and secure. Auto-configuration management is a new challenge that will be addressed through basic plug-and-play and plug-and-run mechanisms.

Objectives and expected results

A key goal of SOCRADES is to specify a service-oriented framework for device-level infrastructures, where system intelligence is achieved by intelligent physical agents embedded in smart devices.

To this end, the following developments will be undertaken:

• Development of a comprehensive device-level SOA infrastructure -- based on the Devices Profile for Web Services (DPWS) -- for encapsulating intelligence and sensing or actuating skills as services, as well as to specify associated frameworks for management and orchestration of device-level services.
• Definition of a methodology for describing services with semantic mark-up that can be interpreted and processed by agents (Semantic Web Services), for the discovery, selection and composition of resources.
• Specification of a framework for service-enabled intelligent physical agents.

Wireless technologies can significantly facilitate deployment and reconfiguration by eliminating the need for installing and maintaining cabling, reducing both costs and time. The actual shortfall of industrial adoption of wireless technology is due to its lack of maturity, failure to provide real-time performance and lack of reliability metrics comparable to those of wired networks. Therefore, a major project objective is to specify new wireless communication protocols that provide the required reliability, safety, security and real-time parameters for embedded devices. Associated to this objective is the specification of middleware that encapsulates both the mechanisms to offer specific QoS provisions and the underlying wireless technologies. Furthermore, mobile sensor/actuator devices will be considered in terms of location-awareness.

SOCRADES is driven by major global players of the European IT business. In view of their expanding business and marketing strategies, the project anticipates achieving groundbreaking future applications, in particular process automation and advanced control, based on seamless integration of wired and wireless networks combined with a universal communications infrastructure.

The use of the SOA paradigm at the device level enables the adoption of a unifying technology for all levels of the enterprise, from sensors and actuators up to enterprise business processes. This will lead to information being available "on demand" and allow business-level applications to use high-level information for such purposes as diagnostics, traceability and performance indicators -- resulting in increased overall equipment effectiveness and business agility.

This figure illustrates the adoption of a uniform service-oriented communication infrastructure at all levels of an industrial enterprise, down from the "shop floor" up to the "top floor". At the device level, it allows devices to expose high-level service interfaces, using protocols of the same Web Service family as those employed by higher-level business processes. This evolution is made possible by the unprecedented and ever-increasing horsepower that can be incorporated into very tiny and cheap processing components. Harnessing this computing power, devices of all kinds gain more and more intelligence. In the industrial automation domain, driving intelligence away from centralised controllers into the devices enables entirely new automation architectures, in which devices become autonomous units, capable of making decisions based on local information and of communicating directly with peer devices, with higher-level devices and/or applications and even with manufactured pieces of equipment. Devices may discover each other dynamically through plug-and-play mechanisms that may be further enriched at the semantic level so as to allow a device to reason and infer the skills and services offered by other devices. Devices may be composed into higher-level composite devices, machines or subsystems, with a composite device encapsulating and coordinating the operations of the lower-level devices using orchestration or choreography techniques, whereby the services of the lower-level devices are aggregated into services exposed by the composite device.

The shift from centralised to decentralised operations enables component-based automation architectures, where a component is an autonomous unit consisting of automation (sensing/actuating) resources with its own computing resources (processor, memory, network interface, electronic interfaces to the automation resources) and control software (operating system, communication protocol stack, application programs). Such a component provides self-contained functionality, exposed through its service interface, all the rest of its properties being opaque to the outside. This approach favours reuse (across designs of different installations) and flexibility (replacement of one implementation by another, reconfiguration of an installation).

The SOCRADES project integrates two complementary technological approaches:

• One that focuses on device-centric functionality at the lowest level of the embedded device hierarchy, i.e. sensors and actuators. Major concerns at this level are to ensure that networked devices behave in a flexible manner -- as regards properties like fault-tolerance, reconfigurability, safety, security and real-time behaviour -- despite the peculiarities of the network.
• One that adopts a service-oriented view in order to cut across, not only all levels of the embedded device hierarchy, but also the higher-level business processes of which the device-level processes are a constituent part. Major objectives in this context are to allow devices to directly communicate, both amongst themselves and with applications, to aggregate devices into subsystems and to integrate devices and subsystems with enterprise-level applications -- all through the same service-oriented high-level communications infrastructure.

As illustrated by the figure, these two orthogonal approaches meet at the sensor/actuator level.

At this level, the emergence of wireless sensor/actuator networks poses new challenges with respect to the robustness and reliability of communications links. Such issues are addressed by a device-level middleware , hiding the peculiarities of the underlying network -- whether wired or wireless -- to the applications using the networked embedded devices. Different solutions for communication, reliability, safety, security and real-time behaviour in the context of wireless sensor and actuator networks will be investigated and developed. The approach will be to consider a special class of control applications as well as taking a holistic view including the communication link and network in the design. By doing this, the middleware will provide services for fault-tolerant communication for control purposes at the device level.

The device-level application platform layer above this middleware makes use of the services offered by the device-level middleware. An open framework for integrating ad-hoc wired or wireless networks of embedded devices into a service-oriented communications infrastructure will be elaborated, bridging application-level functionality and device-level functionality through a common, unifying technological approach based on the SOA paradigm. Web Services technology constitutes the preferred SOA implementation vehicle. A Web Service is platform-independent and can communicate with and/or be aggregated with other Web Services. As each service encapsulates its own complexity, scalability becomes a built-in feature. Additionally, manageability and maintainability are greatly enhanced, especially as each device presents a high-level management interface in order to facilitate configuration, monitoring, fault diagnosis, etc.

Enterprise-level middleware provides functionality like service orchestration and choreography, knowledge-based service discovery, agent-based service interactions and service repository management. In this perspective, the project will create and experiment mechanisms required for service orchestration -- i.e. the sequencing and synchronised execution of several services -- adaptable to usage inside resource-constrained devices. It is further intended to develop and experiment a lightweight service-oriented agent framework based on Web Services, as well as a framework based on semantically enriched Web Services, enabling knowledge-based interactions between embedded devices and agents and thus facilitating the process of networking devices and agents together.

The enterprise-level application platform is situated at the level of enterprise information systems. This platform is intended to be an extended version of an existing business process management system. It is an important part of the project insofar as it will allow demonstrating how the use of a uniform service-oriented communications infrastructure opens unprecedented perspectives of enabling low-level embedded devices to become fully participatory members of their global business environment and may lead to very innovative business process management structures characterised by a high degree of agility and adaptability to change.

Automatic control is a central component of any modern process and manufacturing industry. The information flows between sensor, actuator and control nodes have traditionally been hard-wired asynchronous communication. Over the last decade, there has been a transition to communication buses, such as fieldbus and Ethernet technology, in these control systems. Currently there is a major drive to take the next step in this evolution by moving to wireless communication. More efficient and lower costs for installation and commissioning are important factors. There is also a large potential for major technological advances due to increased flexibility and mobility, which may lead to totally new system designs. For example, in mobile robotic systems, autonomy is growing and multi-robot systems and co-ordination with Automated Guided Vehicles (AGVs) is becoming a reality.

In this context, solutions for the harsh environmental conditions of automation installations are of high relevance. Wireless industrial communications based on WLAN and IEEE 802.15 standards are in the focus of this kind of research and development. In particular, wireless sensor/actuator networks (WSN) are to be closely investigated, as they will definitely foster the mobility and flexibility required in industrial communication. The natural features of wireless technologies enable greater opportunities for reconfiguration/upgrading, maintenance and fault tolerance. Given the existence of lots of legacy wired industrial communication systems, suitable schemes for transitioning between the wired and wireless approaches have to be elaborated. Topics to be addressed include:

• Node architecture, sensor integration and the interface between sensors and the network.
• Wireless network topology, self-configuration, self-management, routing, scalability.
• Communication technologies for Wireless Sensor/Actuator Networks (WSN) in industrial environment (e.g. IEEE 802.15.4, 802.15.4a or ZigBee).
• Power supply for the network infrastructure as well as for the sensor itself.
• New services based on the use of WSNs.

Wireless technology is however subject to major uncertainties with respect to availability and reliability. Wireless communication is not only sensitive to interference from background noise, but also to competing wireless networks. If wireless communication finds its way into industrial communication, which is very likely, we must cope with the situation where several different wireless networks will compete for the shared media. A possible solution would be to use standards and have friendly co-existence of various wireless protocols. However, this is unlikely to happen as many different wireless protocols will be used to meet different kinds of requirements, making it virtually impossible to achieve compatibility. Instead, any new technology needs to have a built-in robustness and adaptation to cope with the major uncertainties and variations the wireless communication presents.

Traditional control design is based on ideal assumptions concerning the amount, type and accuracy of the data flow that can be circulated across the control system. Inherent limitations of control performance are considered with respect to bounds on controller resources, such as actuator authority and sensor dynamics. Unfortunately, real implementations often invalidate assumptions on ideal transmission and networking, and as a result, the system's closed-loop performance can be severely affected.

Our goal is therefore to address the challenges of networked control over a wireless link and to incorporate the functionality into a service-oriented architecture. The basis for our work is a networked control approach, taking a holistic view on control over wireless networks. By combining expertise in automatic control, communication systems and wireless sensor networks, integrated and adaptive solutions will be developed and tested for achieving suitable control performance despite communication uncertainties and variations.

Our objectives include:

• Development of architecture for fault tolerant control;
• Control under uncertain sensor and actuator communication;
• Wireless communication for control purposes;
• Implementation of a component, which embodies the developed control and communication theories and algorithms and provides a DPWS interface;
• Demonstration of the feasibility of the technology.
• The fundamental cornerstones of the envisioned component are (see figure):
• The component service interface, offering high level services to the system
• Networked control over wireless sensor and actuator links and networks
• A middleware for closed-loop control application development
• Control application engineering and development methods to retain traditional deployment paradigms (e.g. PID).

It is vitally important to be able, reliably and repeatably, to construct and compose distributed embedded systems that can meet and adapt readily to ever changing user requirements. Such systems need to be generally applicable to a broad spectrum of application domains as described in section B1 and yet be capable of easy and precise tailoring to specific applications. To address this need an engineering environment will be created aiming at the support for configuration and optimisation of networked control facilities as well as the provision of device management, configuration and lifecycle support services for distributed embedded devices in both wired and wireless automation systems. This engineering system will be specified to meet the user applications requirements for the domains covered by the project.

The objective will be not only to support application design, simulation and monitoring of real-time distributed automation components from the control perspective (control dimension) but also to support the integration of these devices with higher-level business process systems (enterprise dimension), with supply chain partners (value/supply-chain dimension) and within a lifecycle engineering context (lifecycle dimension), i.e., adopting a four-dimensional approach inspired by ARC's CMM model.

Support for the configuration and optimisation of networked control capabilities will be combined with device management, configuration and lifecycle support services. End-users will assist in exploring these new directions. In particular the aim is to create an effective tool-set, the benefit and value of which are clear within the user's business context. This tool-set will support a well-integrated modular approach to real-time control system design, deployment, maintenance, diagnostics and enterprise integration. For example, a machine builder may require support to:

• implement a machine control system;
• integrate this automation system seamlessly with the end-users business system functionality, e.g., MES, ERP, scheduling, data archiving;
• integrate and share component-specific and process-related information with supply-chain partners throughout the lifecycle, from initial simultaneous engineering through maintenance to reuse and decommissioning.

All the above need to be done in a manner that is amenable to unforeseen changes at any stage in the system's lifecycle, as business needs dictate. Whilst control and business systems do in fact share many common features, they have, to date, almost always been developed separately in relative isolation from one another using very different approaches, which has led to inherently poor integration and difficulty in coping with change. The SOCRADES project intends to bring these domains together around a common service-oriented paradigm and associated infrastructure.

An integrated engineering environment is needed to support the multi-perspective needs of different classes of application users in an efficient manner. The approach adopted will be to identify reusable, configurable components, the aim being to mask complexity, maximise reuse and build domain specific libraries of configurable components and associated services, minimising the need for new custom components for each new application. Previous studies in several industrial sectors have shown that a relatively small library of components, specifically tailored to the needs of a given automation user, could meet 80-90% of their needs. In particular support for internal functionality needs to be provided in terms of component behaviour and also support for external functionality in terms of how components, each offering specific service capabilities are composed, and at the higher level orchestrated together, to meet the overall application need.

Our objective is to enable the automated description, simulation, composition, testing, and verification of Web Services for embedded systems relevant to the application domains studied. This technology will also be applied in tools for commissioning and diagnosis. It is the innovative middleware that has a key role to play in SOCRADES, in particular to:

• functionally bridge the gap between application programs and the lower-level hardware and software infrastructure in order to co-ordinate how parts of applications are connected and how they interoperate;
• enable and simplify the integration of components developed by multiple technology suppliers;
• provide a common reusable accessibility for functionality and patterns that formerly were placed directly in applications but that in actuality are application independent and need not be developed separately for each new application.

The development of many of the engineering environment's capabilities is thus intimately associated with the configuration of the middleware at both device and enterprise levels, since it is the middleware itself that offers the potential to hold much of the traditional application functionality, greatly reducing application complexity, e.g., through support for logic, error handling, maintenance and diagnostics.

Application pilots

Coming soon ...