23.08.2013 Views

Specification of Reactive Hardware/Software Systems - Electronic ...

Specification of Reactive Hardware/Software Systems - Electronic ...

Specification of Reactive Hardware/Software Systems - Electronic ...

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

6.7 Real-Time Modelling 225<br />

in the system. Accuracy is the aberration <strong>of</strong> a local clock from the global time. Timing<br />

requirements can be specified with a maximum, a minimum or both, depending on the<br />

needs.<br />

The requirements on events in the system can be defined relative or absolute in global<br />

or local time. Events can be the starting or stopping <strong>of</strong> an internal action or an execution<br />

in the system (process), the deliverance <strong>of</strong> a result, the appearance <strong>of</strong> the arrival <strong>of</strong> a<br />

stimulus or the appearance <strong>of</strong> a response at an output. The time between events can be<br />

specified as a duration (time interval). A time between an input and an output is <strong>of</strong>ten<br />

called a response time. Further the time between two input events or between two output<br />

events may be specified as a time interval. A special type <strong>of</strong> time interval is the time<br />

between the same type <strong>of</strong> events. This type <strong>of</strong> time interval can be defined as a repetition<br />

rate [HP88]. A repetition rate is a useful concept for inputs. A repetition rate for outputs<br />

can be specified as a reprocessing rate. In addition the character <strong>of</strong> the communication<br />

such as bursty traffic must be defined.<br />

Periodical processing can be specified as a reception rate on a processing request message.<br />

A processing time may be specified as a net processing time or as an elapse time, which<br />

is the maximum processing time including the delay caused by communication and<br />

waiting for resources etcetera.<br />

There are several options for adding timing requirements to a specification <strong>of</strong> a system.<br />

Some preliminary ideas about alternatives are the adding <strong>of</strong> requirements to:<br />

system boundaries;<br />

architecture module structure;<br />

implementation structure;<br />

message flows;<br />

object interfaces;<br />

cluster interfaces.<br />

For the moment we foresee the mapping <strong>of</strong> timing requirements onto clusters. Clusters<br />

already carry requirements as they represent various types <strong>of</strong> boundaries (see Section<br />

5.8). Mapping timing requirements onto clusters can be adequate because <strong>of</strong> the relation<br />

<strong>of</strong> timing and architecture. Boundaries formalise architecture and implementation<br />

structure design using clusters.<br />

The problems that designers in practice meet are very complex. Determination <strong>of</strong><br />

response-times <strong>of</strong> processes is hard. The communication between processes, the switching<br />

between processes and the waiting time for resources make timing calculations very<br />

difficult. This makes it very hard to design a complex real-time system adequately without<br />

special tools that support verification <strong>of</strong> the timing constraints. In addition the use<br />

<strong>of</strong> real-time operating systems will be useful. This implies that the properties <strong>of</strong> such a<br />

system part must be incorporated in a model.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!