Evolution of bio-socially inspired strategies in support of dynamic spectrum access

Human and animal societies exhibit complex cognitive and social processes of coordination, cooperation, and competition among their members. Among other functions, these processes can facilitate fairer sharing of resources among community members and enhance individual survival outcomes. In this work, three bio-socially inspired models for secondary users of spectrum in cognitive radio networks are defined and compared to one other within an evolutionary framework. The proposed models reflect successively more sophisticated capabilities of secondary users in distributed spectrum access. The simplest of the three, blind channel access, is shown to be evolutionarily dominant when residual channel capacities are homogeneous. The second more advanced model assumes a capability to sense channel utilization; this model is shown to dominate when the channels have intermediate load and heterogeneous capacities. Finally, the most complex model (additionally) allows for social coalitions and within-group deference; this model is seen to dominate in high load heterogeneous resource settings. We explore the long term evolutionary pressures within societies whose members choose between these three schemes, with natural selection operating via a utility-based fitness function. Our research is based on systematic ns-3 simulation experiments of heterogeneous societies under a range of assumed channel conditions, population sizes, resource demands, and initial user attributes. Our results demonstrate that the secondary user population always evolves to adopt a unique and stable strategy, but that the winning strategy selected depends strongly on channel conditions. Our results further show that this kind of leaderless evolution leads to a significant 12-116% overall improvement in performance compared to systems in which a fixed strategy is deployed. In summary, we conclude that evolving bio-social behavioral models can be applied to great advantage in understanding dynamic environments such as those envisioned by distributed spectrum access.