Network densification, the practice of deploying more radio access nodes into a geographical area, is being considered as a cost and bandwidth-effective method to increase wireless network capacity. Despite their benefits, widespread small-cell deployment also poses a tremendous challenge in terms of interference management. Interference mitigation techniques using directional antennas show great promise, allowing small-cell network nodes to focus energy only in the intended direction, thereby creating less interference between links and more potential for spatial reuse. Further, with the steering of beams, nodes can also suppress unwanted emission and interference, offering greater security.  Nevertheless, bringing these techniques to practice has been challenging due to (i) the difficulty of integrating directional antennas into existing wireless physical layer (PHY) and medium access control (MAC) stack of small cells, and (ii) the lack of robust antenna state selection techniques that can cope well with the wireless channel's stochastic nature and unpredictable variations in the operating environment of small cells.

To address these challenges, a joint research team from the University of Oulu and Drexel University have developed LinkPursuit, a novel learning protocol for distributed antenna state selection in directional cognitive small-cell networks. LinkPursuit utilizes reconfigurable antennas for directional transmission and reception. Further, it incurs low overhead and adapts quickly to environmental changes through probabilistic selection at each time step. The solution provides major advantages over omnidirectional transmission in terms of resilience to deliberate or unintentional interference from other nodes. When jammed continuously, LinkPursuit achieves a 12 dB anti-jam performance gain in link dropout prevention. It also achieves a 74% sum throughput improvement over omni-directional antennas when both base stations transmit concurrently at the same power levels and a 40% increase when they transmit with disproportionate power.