|
Resumo do trabalho
|
A IPv6 over the TSCH mode of IEEE 802.15.4e (6TiSCH) network provides IPv6 connec-
tivity through links IEEE 802.15.4 governed by Time Slotted Channel Hopping (TSCH).
Essentially, TSCH intends to provide low energy consumption and high reliability by sche-
duling communication resources and the channel hopping mechanism. However, 6TiSCH
does not define the policies to build and maintain the communication schedule. The com-
ponent responsible for creating such a schedule is called a Scheduling Function (SF). SFs
are a fundamental part of the 6TiSCH architecture and, therefore, have been intensely
studied by the research community. Several works proposed different scheduling schemes,
each with particular characteristics and trade-offs. A possible approach for the scheduling
problem is the utilization of Artificial Intelligence (AI), which allows for a good estima-
tion of a scheduling policy for dynamic networks with unpredictable traffic. Although
several studies have been done to compile and compare different schedule strategies, no
study has systematically compared AI-bases schedulers with traditional schedulers. This
work introduces the methodology used to produce such a study. Another significant
contribution of this work is the proposal of two new scheduling algorithm strategies:
a Q-Learning-based scheduling algorithm, discussed in detail, and a schedule based on
aggregating multiple scheduling functions.
|
|---|
