A Learning Automata-Based Cognitive Radio for Clustered Wireless Ad-Hoc Networks

In current wireless networks, the radio systems are regulated by a fixed spectrum assignment strategy. This policy partitions the whole radio spectrum into a fixed number of radio ranges, each exclusively assigned to a specific user. Such a spectrum assignment strategy leads to an undesirable condition under which some systems only use a small portion of the allocated spectrum while the others have very serious spectrum insufficiency. The learning automata-based cognitive radio which is proposed in this paper is a highly potential technology to address the spectrum scarcity challenges in wireless ad hoc networks. This paper proposes a learning automata-based dynamic frame length TDMA scheme for slot assignment in clustered wireless ad-hoc networks with unknown traffic parameters, where the intra-cluster communications are scheduled by a TDMA scheme, and a CDMA scheme is overlaid on the TDMA to handle an interference-free inter-cluster communication. In this method, each cluster-head is responsible for a collision-free slot assignment within the cluster and determines the input traffic parameters of its own cluster members. It then takes these traffic parameters into consideration for an optimal channel access scheduling in the cluster. The medium access control layer in each cluster is based on a time division multiple access (TDMA) scheme, in which each host is assigned a fraction of the TDMA frame proportional to its traffic load. The simulation experiments show the superiority of our proposed slotassignment algorithm over the existing methods in terms of the channel utilization, control overhead, and throughput, specifically, under bursty traffic conditions.