Large-Scale Full WiFi Coverage: Deployment and Management Strategy Based on User Spatio-Temporal Association Analytics.
Feng Lyu,Liang Fang,Guangtao Xue,Hua Xue,Minglu Li
DOI: https://doi.org/10.1109/jiot.2019.2933266
IF: 10.6
2019-01-01
IEEE Internet of Things Journal
Abstract:Full WiFi coverage becomes more and more prevalent in corporate places, such as university, big mall, airport, and so forth. To achieve full WiFi coverage in a wide area is costly due to the large-scale AP deployment spending and considerable operating expenditure. However, with limited literature available, how to deploy and manage those APs in an efficient and economical way, is still unknown for system providers. To bridge this gap, in this article, we first collect large-scale AP usage data in our campus WiFi system, which contains over 8000 APs and serves more than 40 000 active end-users in the area of 3.0925 km(2). After mining large-scale spatio-temporal user associations, we obtain several key insights as follows. First, Idle Phenomenon prevails throughout the trace, in which large portion of APs are wasted without any user association. Second, AP usages in different buildings have very distinct characteristics in terms of user association and traffic consumption. Third, diurnal usage patterns are very obvious not only at singe AP level but also at the building and the whole system level. Many deployment and management strategies can benefit from these insights, e.g., heterogeneous AP deployment and intelligent AP management. Among them, we then propose an intelligent large-scale AP management scheme, called LAM, to dynamically control large-scale APs (ON or OFF) for energy saving and meanwhile without loss of WiFi coverage. In LAM, based on history association records, the user load of each AP is predicted by machine learning algorithms, and those APs whose idle durations are longer than the length of the predefined time window, will be switched off during the duration. We conduct extensive trace-driven experiments to demonstrate its efficacy; on average, more than 70% of power consumption can be markedly saved with over 92% of WiFi coverage guaranteed, which is able to save empirical $59 000 per year just for our system.