tailieunhanh - Model-Based Design for Embedded Systems- Part 5

Model-Based Design for Embedded Systems- P5: This book contains information obtained from authentic and highly regarded sources. Reasonable efforts have been made to publish reliable data and information, but the author and publisher cannot assume responsibility for the validity of all materials or the consequences of their use. | 16 Model-Based Design for Embedded Systems a u A min au 0 pu 0 p1 pu az1 A min a1 0 pu 0 p1 p1 p u A max inf Pu À - a1 À 0 A X p 1 A sup p1 A - au À 0 X A In the examp1e system of Figure the processing semantics of the tasks Pi P2 P3 and P4 can be mode1ed with abstract GPCs. Fina11y 1et us consider a component that is used for event stream shaping. A greedy shaper component GSC with a shaping curve r de1ays events of an input event stream such that the output event stream has r as an upper arriva1 curve. Additiona11y a greedy shaper guarantees that no events are de1ayed 1onger than necessary. Typica11y greedy shapers are used to reshape bursty event streams and to reduce g1oba1 buffer requirements. If the abstract input event stream of a GSC with the shaping curve r is represented by the tup1e of arriva1 curves a1 au then the output of the GSC can be mode1ed as an abstract event stream with arriva1 curves aGsC au 0 r Gsc a1 0 ff0ff Note that a greedy shaper does not need any computation or communication resources. Thus the transfer function of an abstract GSC considers on1y the ingoing and the outgoing event stream as we11 as the shaping curve r. More detai1s about greedy shapers in the context of MPA can be found in 19 . In the examp1e system of Figure the semantics of the shapers Si and S2 can be mode1ed with abstract GSCs. System Performance Model In order to ana1yze the performance of a distributed embedded p1atform it is necessary to bui1d a system performance mode1. This mode1 has to represent the hardware architecture of the p1atform. In particu1ar it has to ref1ect the mapping of tasks to computation or communication resources and the schedu1ing po1icies adopted by these resources. To obtain a performance mode1 of a system we first have to mode1 the event streams that trigger the system the computation and communication resources that are avai1ab1e and the processing components. Then we have to interconnect the arriva1 and service inputs and .