When Cellular and Wi-Fi Meet in a Small Cell Network

The ever-increasing mobile data consumption has strained mobile networks, leading operators to upgrade their network infrastructure by means of low-power small cell base stations (femto-, picocells, and Wi-Fi), aiming at offloading macrocell traffic and achieving cell splitting gains. As future small cell base stations (SCBSs) will be multi-mode, i.e., capable of transmitting simultaneously on both licensed and unlicensed bands, a cost-effective integration of both radio access technologies (RATs) is crucial to cope with peak data traffic and users' heterogeneous requirements. In this paper, the novel framework of cross-system learning is proposed, whereby SCBSs self-organize and judiciously steer their traffic across different RATs, cells and frequency bands. Leveraging Wi-Fi, SCBSs learn the probability distribution function of their transmission strategy over the licensed spectrum, while offloading delay-tolerant traffic to Wi-Fi. The developed framework is validated in a Long-Term Evolution (LTE) simulator overlaid with Wi-Fi hotspots. Remarkably, it is shown that the cross-system learning-based approach outperforms a number of benchmark algorithms and traffic steering policies.