Near-optimal opportunistic spectrum access in cognitive radio networks in the 5G and IoT era

We consider a cognitive radio network consisting of M primary users, M primary receivers, a secondary user with K channels and K secondary receivers. A channel is assigned to each primary user for transmitting data to a primary receiver. Secondary user is assumed as data backlogged. The secondary user has no direct information on states (good or bad) of M channels, or statistic of channel evolution process. Secondary user chooses K channels for transmission in each time slot. If secondary user senses good state for a chosen channel, it can transmit data to a secondary receiver in that time slot. Otherwise, secondary user does not use that channel and choose another channel for data transmission next time. This scheduling problem is tackled under average reward criterion. Uniforming Random Ordered Policy (UROP) is proposed, and shown to achieve nearly throughput-optimality. It performs much better than myopic policy.