Learning from experts in cognitive radio networks: The docitive paradigm

In this paper we introduce the novel paradigm of docition for cognitive radio (CR) networks. We consider that the CRs are intelligent radios implementing a learning process through which they interact with the surrounding environment to make self-adaptive decisions. However, in distributed settings the learning may be complex and slow, due to interactive decision making processes, which results in a non-stationary environment. The docitive paradigm proposes a timely solution based on knowledge sharing, which allows CRs to develop new capacities for selecting actions. We demonstrate that this improves the CRs' learning ability and accuracy, and gives them strategies for action selection in unvisited states. We evaluate the docitive paradigm in the context of a secondary system modeled as a multi-agent system, where the agents are IEEE 802.22 CR base stations, implementing a real-time multi-agent reinforcement learning technique known as decentralized Q-learning. Our goal is to solve the aggregated interference problem generated by multiple CR systems at the receivers of a primary system. We propose three different docitive algorithms and we show their superiority to the well know paradigm of independent learning in terms of speed of convergence and precision.