On Tue, Dec 23, 2008 at 12:48:04PM +0000, Matthew Toseland wrote: > On Tuesday 23 December 2008 03:14, vive wrote: > > Simulation: > > 0. Load the two networks, each of size 4000. All nodes are set to be pure > darknet. > > 1. Connect the networks by joining pairs of random nodes, where one node is > picked from each network. I tried this with a cut of 10,100,1000,10000 links. > > 2. Nodes randomize their positions and swap for a while to stabilize the > locations (successful swaps get more uncommon after this) to start from > a "stable" state > > 3. 1000 keys are inserted by random nodes into the network (both sink and > cache). Nodes have a store of 50 keys and a cache of 50 keys. Both are LRU as > normal. > > Actually we use pseudo-random replacement now. This probably doesn't strongly > affect the results.
Is this documented? I was unable to find it. It would be great to have a place on the wiki where the heuristics related to 1. Routing 2. Caching/Storing are documented. Maybe at least a short description of each. This way, simulations can be made more realistic and design decisions made more clear. Right now the simulations do routing with backtracking as well as storing/caching according to the sinking strategy (store) and LRU (both store and cache). One could of course read the code ... but it may also be hard to get the high-level description right-away from many details of lower-level code. :-) Is it possible that someone who has insight into the details of the heuristics could write a short description? Cheers, /vive -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 195 bytes Desc: not available URL: <https://emu.freenetproject.org/pipermail/devl/attachments/20090105/bfaf3ce9/attachment.pgp>
