2. Mobile, ad hoc, no
global knowledge
Reaching a global goal
with local interactions
3. Local
measurements
but we can approximate global measurements
Good heuristics
(APF[1], GOSSIP[2])
can approximate
complicated protocols
quite well
[1] Vilmos Simon, László Bacsárdi, Sándor Szabó, Daniele Miorandi: “BIONETS: a new vision of opportunistic networks”, IEEE WRECOM 2007, Rome, Italy, 1-2 October, 2007
[2] Li Li Zygmunt J. Haas, Joseph Y. Halpern.: “Gossip-based ad hoc routing”, IEEE/ACM Transactions on Networking (TON), Volume 14 Issue 3, June 2006
4. Natural selection?
Genetic
programming!
- Mixing heuristic properties
- Adaptation to the environment
5. Custom
programming language
Heuristic = Event handlers
Event handler = list of instructions
Stack based
+ Genetic operators won't mess with the syntax
+ Simple
+ Designed specifically for the task
6. Heuristics
communicate with each other
disseminating messages
in the network
Nodes measure
heuristics' performance
+ reward useful messages
- punish duplicate ones
7. Populations
Renew
Periodically
- Each iteration means new, more adapted heuristics
- Heuristics evolve
- Heuristics wander in the system