Algorithm 1: Cross-layer optimization for low-power and lossy networks using iCPLA

while the device is on do set maximum retry limit set maximum backoff stages setCWmin = 0, CWmax = 31 set current reward = 0, $\Delta Q$ (s, a) = 0, $Q$(s, a) = 0 BE = MIN (n_collisions, CSMA_MAX_BE) measure CW using BE in (6) calculate $P_{coll}$ using CW in (7) calculate $\overline{P_{coll}}$ using (8) counter ++  if (counter = 5), then   counter = 1   $P_{coll}$ = $P_{coll\left(current\right)}$ end if $P_{coll}\left(N_i\right)=Avg\left(\overline{P_{coll}},P_{coll\left(current\right)}\right)$ if$(P_{coll\left(current\right)}<\overline{P_{coll}})$, then  reward = positive else reward = negative update reward table for r (s, a) update Q-values table according to (2) pick a random value to explore and exploit if (exploit), then find $mi{n}_{a}Q\left(s^{\prime },a^{\prime }\right)$ IP address else (explore) if (node = root node), then   root rank = 1 end if if (parent = null), then   rank = max path cost end if if (parent != null), then   $Rank\left(c_j\right)=Rank\left(p_i\right)+ran{k}_{increase}$   $ran{k}_{increase}$ = ($P_{coll}\left(N_i\right)$ end if if$(Rank\left(p_i\right)=0$, then   rank = base rank (128) end if return MIN $(Baserank+ran{k}_{increase})$ end while