Learning to Thrive in a Leasing Market: An Auctioning Framework for Distributed Dynamic Spectrum Leasing (D-DSL)

Recently proposed dynamic spectrum leasing (DSL) for dynamic spectrum sharing (DSS) presumes that there is a reward for primary users for accepting secondary activity. The reward primary users received in previous DSL proposals, however, were considered to be proportional to the secondary interference level. In this paper, we propose a new way to encourage primary users to lease their spectrum: The secondary users indicate their willingness to spend a portion of their transmit powers to relay the primary signals. Thus, the primary reward is the power saving achieved due to asymmetric cooperation. We develop a repeated first-price-auction model to realize the proposed DSS scheme in a completely distributed implementation. A reinforcement learning algorithm is proposed for allowing all users to learn their bids in subsequent auctions. Our results show that the proposed Distributed-DSL framework leads to efficient spectrum utilization and, in particular, the proposed learning helps to improve secondary throughput.