Machine-to-machine communication becoming one of the research topics that get more attention these days. Machine-to-Machine itself can be define as an assortment technologies that being used to associate every systems with an aim to monitoring and control remotely without any intervention from human. One of the problem that may occur is the massive number of M2M devices that trying to attempt the resources in the same time. This condition may drives the network to have issues of overload and the random access. A deployment of a random access management and an efficient scheduling are needed to handle the issues of signaling overhead and the large gap of machine-to-machine devices requirement. In this dissertation, first we presented some related works about resource management and scheduling techniques that already exist as a base background of our works.
Our works propose a random access management system that applies coordination scheme for Random Access Channel (RACH) that can fairly reduce the signaling congestion and network overload that affect to better performance of random access probability. The signaling process and available uplink resources will be handled efficiently with a group classification of related M2M devices and the leader selection, meanwhile the scheme also will preventing unnecessary recurring data transmission that can reducing the signaling overhead.
This dissertation also present a slot allocation for scheduling M2M devices using Q-learning algorithm. Simulation results showed that the Q-learning algorithm can help maximizing the Quality of Service of M2M communication in terms of the better convergence time, success probability, and can ensure the interference that may happen on the M2M Networks.