Difference between revisions of "Checklist and Best Practice Examples for Designing Self-Organizing Systems"
From Self-Organization Wiki
(New page: * Clearly specify goals and performance metrics * Check the following options: ** Applying a reference design **** Biologically-inspired **** From other disciplines ** Mathematica...) |
(→Design Process) |
||
| (16 intermediate revisions by 2 users not shown) | |||
| Line 1: | Line 1: | ||
| + | = Design Process= | ||
| + | Difference w.r.t. traditional system design | ||
| + | * a top-down approach should produce local rules | ||
| − | + | ==1. Clearly specify== | |
| + | * goals | ||
| + | * performance metrics | ||
| + | * constraints (limited resources, like energy) | ||
| + | ==2. Check the following options== | ||
| + | * Applying a reference design | ||
| + | ** Biologically-inspired | ||
| + | ** From other disciplines | ||
| + | * Mathematical model | ||
| + | * Trial and error (test the resulting global behavior from different types of local interactions) | ||
| + | ** Manual trial and error | ||
| + | ** Evolutionary approaches | ||
| + | ==3. Iterate== | ||
| + | * Refine goals | ||
| + | * Integrate more constraints into the model | ||
| − | * | + | =Best Practice Example= |
| + | * Define an idealized system (e.g. no delays, simple topologies, perfect and simple environment), where still no entity has global view | ||
| + | * Find a simple distributed solution for the given goal | ||
| + | * Gradually enhance the system model and solution to make it more realistic | ||
| − | + | =Checklist= | |
| − | + | Issues to be kept in mind: | |
| − | + | * Does the proposed solution meets the performance goals? | |
| − | + | * Did you consider all important constraints? Is it feasible to implement in practice? | |
| − | * | + | * Is the proposed solution scalable? |
| − | + | * Is the proposed solution robust to faults? | |
| − | + | * Did you compare the solution to external control solutions and to known theoretical bounds? | |
| − | + | * Are there any potential deadlocks or other impediments? | |
| − | |||
Latest revision as of 14:21, 29 July 2010
Contents
Design Process
Difference w.r.t. traditional system design
- a top-down approach should produce local rules
1. Clearly specify
- goals
- performance metrics
- constraints (limited resources, like energy)
2. Check the following options
- Applying a reference design
- Biologically-inspired
- From other disciplines
- Mathematical model
- Trial and error (test the resulting global behavior from different types of local interactions)
- Manual trial and error
- Evolutionary approaches
3. Iterate
- Refine goals
- Integrate more constraints into the model
Best Practice Example
- Define an idealized system (e.g. no delays, simple topologies, perfect and simple environment), where still no entity has global view
- Find a simple distributed solution for the given goal
- Gradually enhance the system model and solution to make it more realistic
Checklist
Issues to be kept in mind:
- Does the proposed solution meets the performance goals?
- Did you consider all important constraints? Is it feasible to implement in practice?
- Is the proposed solution scalable?
- Is the proposed solution robust to faults?
- Did you compare the solution to external control solutions and to known theoretical bounds?
- Are there any potential deadlocks or other impediments?
