Forecasting Traffic Demand Experienced by Wireless Communication System with Application to Channel Allocation

摘要

A traffic demand to be experienced by a base station (BS) at a target time instant is forecasted by obtaining a long-term component of the traffic demand and predicting a short-term component thereof. To predict the short-term component, a service area of the BS is imaged at an observation-making time instant earlier than the target time instant. The number of people captured in the image and located in the service area is determined. The BS collects identifiers of user equipments (UEs), such as media access control (MAC) addresses, to determine the number of UEs present at the observation-making time instant. The short-term component is computed from the number of people and the number of UEs via a mathematical function determined by a genetic algorithm or a machine-learning algorithm according to historical records. The forecasted traffic demand facilitates channel allocation for the BS.