Entry points set paths for contexts to be represented within systems, and for systems to support context activities.
Since entry points are meant to support interactions between contexts and systems, they must be set to support the targeted functional architecture. Based upon stereotyped contexts, interactions can be initiated by three types of active objects: human agents, mechanical devices, or digital systems:
- Human agents are live objects bearing native identities. Depending on organizations, they can be associated with individual responsibilities, access rights and authorizations. They interact with with systems through human interfaces using symbolic languages.
- Mechanical devices are physical artifacts bearing designed identities. They have no responsibilities and, deprived of symbolic capabilities, they cannot be associated with access rights and authorizations. They interact with systems through sensor/actuators using analog signals.
- Digital systems are symbolic artifacts bearing designed identities. They have no responsibilities but can be associated with access rights and authorizations. They interact with systems through channels using digital messages.
Also, connections between active context components can be characterized by the nature of flows and coupling capabilities; that will be mapped into sterotyped boundary objects for user user interactions, proxies, I/O devices.
As far as architecture is concerned, representation patterns for entry points should focus on constraints regarding couplings between contexts and system representations. For that purpose entry points should be characterized jointly by stereotyped active context objects, events, and roles, with communication channels set accordingly.
- Interactions with agents are symbolic. They convey messages, neutral regarding coupling, and requests, which modify execution states. They support logical identification of agents and mediated (asynchronous) mapping of representations.
- Interactions with information systems are also symbolic; in addition to messages and requests they may also signify changes in symbolic representations. They also support logical identification of systems and mediated (asynchronous) mapping of representations.
- Interactions with I/O devices are not symbolic and must therefore be supported by analog channels. Signals may or may not be associated with actual changes.
- Interaction with control systems devices are symbolic, supported by digital channels. They convey requests, messages, and actual state changes. They support physical identification of devices and immediate (synchronous) mapping of representations.
Stereotyped entry points may be mapped to user patterns and functional architecture.
From Entry Points to Functional Architecture
The aim of Architecture Driven System Modelling is to identify system supporting structures (“charpente” in French) from functional constraints. As a first step it may be possible to stereotype context elements:
- The system under consideration is first described as a control system with symbolic processing capabilities.
- The supervisory system support interactions with the supervisor and exchange messages with the control system.
- Parameters are updated from the workstation and read by the control unit which may also ask for application services and receive time events from notification services.
- Sensors send updates from context.
- Devices status are polled and their actuators activated.
Flows are then characterized as stereotyped data or events. Given stereotyped entry points, it will be possible to characterize system interfaces:
- User interfaces are symbolic entry points dedicated to specific roles to be played by agents whose identity and access rights may have to be checked. Combined with service requirements, those checks will entail specific architecture constraints and, as such, should be explicitly set by patterns.
- Digital channels, as user interfaces, deal with symbolic contents, but, contrary to user interfaces, those contents can be directly processed at system level. Whether symbolic representations are to be instantly updated or not is not relevant because triggering events don’t reflect actual states of context objects.
- Analog channels must provide for instant processing of symbolic contents into analog ones or vice versa. And to be of any use they must be coupled with instant update of symbolic representations.
Entry Points & Control Patterns
Entry points capability should support requirements regarding what is to be exchanged between context and system, for instance using the four-variable model (Parnas & Madey, 1995):
- Active coupling: monitored and controlled variables
- Passive coupling: input data and output results.
Those requirements should be unambiguously expressed when functional architecture is considered, especially regarding communication and control mechanisms.