davidezanella/DS2-GossipingAppendOnlyLogs

Motion strategies

Closed this issue · 0 comments

simonedegiacomi commented

Currently we have only one motion strategy, which randomly chooses where a person should go. The location is a random point in the simulation grid. Once a person has reached the target point, it stays in that location for one tick and then picks another random point.

Ideas for other motion strategies:

Gaussian based on interests

Note: this idea works only in the transitive interests version.

The idea is that LANs are placed in point of interests such as libraries and coffees. As such, people visit those places at similar times as their friends. In this strategy, each person decides (1) which LAN to visit, and then (2) at which time.

  1. each person picks a LAN randomly, but each LAN has a probability of being picked proportional to the number of friends that went to that LAN is the last X ticks;
  2. each person decide the time by computing the mean of the time of the day in which its friends went to that LAN the previous time;

This means that we need to have the concept of a day (ex: 1 tick = 10 min, so 144 ticks = 1 day), and we need to bootstrap the timing and previous visited LAN somehow (randomly?)