Skip to content

Lab 5 — Stochastic multipath and the pheromone exponent β

Goal. See that AntHocNet forwards data stochastically over several next hops, and how one parameter, β, moves it from spreading to single-path. Time. 45 min. Site. The mesh world, seed 4, and the tuning bench.

← Classroom pack · Answer key

Background (5 min)

A node forwards a data packet to neighbour n with probability proportional to τₙ^β, where τₙ is the pheromone toward the destination through n (ant colony routing). The data exponent is betaData. The default is 20, the thesis's β₃ (configuration.md). Mission 5 of the Academy, Two roads, shows the split on this world.

1. Predict (5 min)

  1. At β = 1, will the first call's source use one next hop or several?
  2. At β = 20 (the default)? At β = 30?

2. Run (20 min)

  1. Open #mesh?seed=4. Select the first call's source and turn on the pheromone overlay. Run to t = 120 s.
  2. Watch which neighbours the source's data goes to. The strongest route overlay draws the best next hop, and the pheromone overlay draws every trail the source can choose from.
  3. In the tuning bench, set betaData to 1 and repeat. Then set it to 30 and repeat.

3. Explain (10 min)

  1. How many different next hops did the source use at each β?
  2. Spreading uses more paths. Name one benefit and one cost. Hint: reordering and delay.
  3. The thesis keeps β₃ at 20. What does that choice say about its authors' priorities?

4. Compare with ns-3 (5 min)

Read the betaData row of configuration.md. Which scenario showed the largest effect of β = 20, and on which metric?