论文部分内容阅读
城市移动通信基站流量的准确预测对于关键基站的拥堵控制、基站新址的选择有着重要作用。基站流量数据不仅是区域的静态表现,同时也反映区域人员的流动特性。基站流量具有非线性混沌特性,而传统的线性时间序列方法比如自回归移动平均模型难以有效地捕获实际基站流量序列中复杂的非线性因素。同时,仅考虑单个基站时间序列而忽略邻近基站的影响并不能反映基站流量的动态特征。基于向量自回归模型(VAR)对大规模基站流量数据进行整体分析,将多响应变量预测问题转化为单响应变量预测模型,运用Lasso变量选择方法筛选目标基站的重要关联基站。实例表明,相对于传统预测方法,VAR-Lasso类方法不仅提高了基站流量的预测精度,同时也实现了大规模基站的实时预测。
Accurate prediction of urban mobile communication base station traffic plays an important role in the selection of new base stations for the congestion control of key base stations. Base station traffic data is not only the static performance of the area, but also reflects the mobility characteristics of the region. Base station traffic has nonlinear chaos characteristics, while the traditional linear time series methods such as autoregressive moving average model are difficult to effectively capture the complexity of the actual base station traffic sequence nonlinear factors. At the same time, considering only the single base station time series and ignoring the impact of neighboring base stations does not reflect the dynamic characteristics of the base station traffic. Based on vector autoregressive model (VAR), the large-scale base station traffic data is analyzed as a whole, the multiple response variable prediction problem is transformed into a single response variable prediction model, and the Lasso variable selection method is used to screen the important associated base stations of the target base station. The example shows that compared with the traditional prediction methods, the VAR-Lasso class method not only improves the prediction accuracy of base station traffic, but also realizes the real-time prediction of large-scale base stations.