Date of this Version
In this paper, we describe the long-term evolution of societies of secondary users in dynamic spectrum access networks. Such an understanding is important to help us anticipate future trends in the organization of large-scale distributed networked deployments. Such deployments are expected to arise in support of a wide variety of applications, including vehicular networks and the Internet of Things. Two new biologically-inspired spectrum access strategies are presented here, and compared with a random access baseline strategy. The proposed strategies embody a range of plausible assumptions concerning the sensing capabilities and social characteristics of individual secondary users. Considering these strategies as the basis of a game against the field, we use replicator dynamics within an evolutionary game-theoretic analysis to derive insights into the physical conditions necessary for each of the strategies to be evolutionarily stable. Somewhat surprisingly, we find that the physical channel conditions almost always uniquely determine which one of the three (pure) strategies is selected, and that no mixed strategy ever survives. We show that social tendencies naturally become advantageous for secondary users as they find themselves situated in network environments with heterogeneous channel resources. Hardware test-bed experiments confirm the validity of the analytic conclusions. Taken together, these results predict the emergence of social behavior in the spectrum access etiquette of secondary users as cognitive radio technology continues to advance and improve. The experimental results show an increase in the throughput of up to 90%, when strategy evolution is continuously operational, compared with any static strategy. We present use cases to envision the potential application of the proposed evolutionary framework in real-world scenarios.