Learning in Small Cell Networks: A Social Interactive Model

In small cell networks, due to the small coverage of small cell access points (SAPs), handoffs may be executed frequently. Therefore, evaluating the utility that a user equipment (UE) can acquire from an SAP is of great significance. In this paper, different from traditional evaluation schemes, we propose a social interactive evaluation scheme. The UEs are allowed to share their local believes and fuse them in a non-Bayesian manner. One advantage of the scheme is that it allows UEs to evaluate an SAP that they do not connect to, based on which UEs can get prepared for handoff in advance. Both the theoretical analysis and simulation illustrate that UEs not connecting to an SAP is able to learn the real utility iteratively and accurately. Additionally, compared to UEs performing individual Bayesian estimation, UEs with the non-Bayesian scheme can learn the real utility faster if the signal cannot be observed in every iteration.