Data-Driven Small Cell Placement Optimization With Users' Differential Privacy For Wireless Ngns
Jingyi Wang,Xinyue Zhang,Wenjun Xu,Qixun Zhang,Zhiyong Feng,Miao Pan
DOI: https://doi.org/10.1109/ICDCS.2019.00156
2019-01-01
Abstract:In the coming fifth generation (5G) or beyond 5G next generation networks (NGNs), the small cell deployment is a promising solution to meet the ever increasing demands of mobile devices, and the proliferation of wireless services. The low power base station (BS), such as femtocell BS, is a cost-effective and environmental friendly substitution for the power-hungry macrocell BS. One potentially effective way to deploy those small cells is to use two-tier NGN architecture, where the first-tier carrier can authorize the second-tier carrier's access to users' transmission information database (e.g., uplink/downlink service demands), and thereafter the second-tier carrier can decide how to place small cell BSs according to the mobile users' requirements locally. However, the second-tier carriers/operators for small cell placement may not be trustworthy, and the NGN users' data privacy might be compromised. To address this issue, we integrate differential privacy (DP) preserving techniques into data-driven optimization, and propose a novel scheme that not only preserves the privacy of NGN users' transmission information, but also maximizes the revenue of small cell deployment. Briefly, differential private noises are intentionally added into the users' transmission information database. Based on queries, the second-tier carrier can aggregate a given set of users' differentially private historical data, estimate the users demands, and formulate the data-driven revenue maximization problem. Given the stochastic programming optimization formulation, we develop feasible solutions and conduct extensive simulations with real-world transmission datasets (i.e., transmission data collected hourly from 3072 4G eNBs deployed in several southern cities of China in 2015) to verify the effectiveness of the proposed scheme.