TheScienceofSowareand$ SystemDesign$ - Aaltodigi.aalto.fi/en/midcom-serveattachmentguid-1e712... ·...

8
The Science of So+ware and System Design (a.k.a. “Modelbased design”) Stavros Tripakis Aalto University and UC Berkeley Pitch talk, Aalto Digi Matchmaking Event, March 23, 2017

Transcript of TheScienceofSowareand$ SystemDesign$ - Aaltodigi.aalto.fi/en/midcom-serveattachmentguid-1e712... ·...

The  Science  of  So+ware  and  System  Design  

(a.k.a.  “Model-­‐based  design”)

Stavros  Tripakis    

Aalto  University  and  UC  Berkeley  

Pitch  talk,  Aalto  Digi  Matchmaking  Event,  March  23,  2017  

So+ware:  large,  complex,  costly

Tripakis   2  

GeFng  larger  and  larger,  more  complex,  more  costly

Need  a  scienHfic  approach  to  so+ware  development

Is  there  a  so+ware  science?

Is  so+ware  tesHng  the  answer? •  The  more  soDware,  the  more  bugs  

– Trial-­‐and-­‐error  – Too  expensive  – Does  not  scale  

• Bugs  found  too  late  • Not  reliable  enough  

Tripakis   3  RelaMve  cost  to  fix  them  

Origin  of  soDware  defects  

Tes$ng  shows  the  presence,  not  the  absence  of  bugs  Edsger  W.  Dijkstra,  1969  

It’s  not  just  about  so+ware,  it’s  about  systems!  (many  safety-­‐criHcal)

4  Courtesy  http://www.fastcodesign.com Thanks  to  Christos  Cassandras  for  recommending  this  video  

Tripakis  

AlternaHve/complementary  approach:  formal  methods

• Program  analysis  /  verificaMon:  prove  correctness  automaMcally  

•  Success  story  in  avionics,  defense,  hardware  (EDA),  and  soDware  industries  

Tripakis   5  

Tripakis   6  

April  2015  arMcle  

Our  work  at  Aalto

Tripakis   7  

•  Formal,  model-­‐based  approach  to  system  design  • Not  trial-­‐and-­‐error  

• MulMple  applicaMon  domains:  – Program  verificaMon  and  synthesis  

– System  monitoring  – Embedded  systems  (Simulink)  – Cyber-­‐physical  systems  – Security  

rcrs.cs.aalto.fi

come  to  our  demo!  

Thank  you  

8