Near real time identification of predicted wireless communication congestion is based on a relationship between a number of key performance indicator (KPI) trend prediction values and corresponding KPI trend thresholds. Each KPI trend prediction value is calculated based on a number of past intervals and a number of KPIs corresponding to each past interval of the number of past intervals utilizing linear regression. A mobile communication network node, such as an enhanced node B, may identify predicted wireless communication congestion and notify mobile devices served by the node. Mobile devices served by the node, based on the notification, may restrict background application data communication and/or take appropriate action to contribute to a reduction in and/or elimination of the congestion.