A clustering prediction scheme for wireless cellular network

A centralized collaborative system between nodes and BSs is developed, and a new prediction mobility scheme is proposed with data mining and time series techniques. Based on the mobility prediction, bandwidth is reserved for the paths with the maximum support or the best confidence rule, so that the handoff calls´ service can be guaranteed. This new approach belongs to the direct group mobility (DGM) prediction scheme and is based on the tree path construction algorithm (TPCON) for each base station (BS). The nodes with DGM support provide the BSs with the important aggregate bandwidth information so that they can avoid the congestion for the handoff users´ sake. For finding the most popular group path, based on TPCON, clusters are constructed according to the various flows of the group mobility over an area of mobile stations. We focus on the center oriented clusters that are very crucial for bandwidth prediction purposes. An adaptive clustering algorithm creates the chain of activated cells at each time. A call admission control (CAC) algorithm is developed for each BS for minimizing the call dropping probability. This study deals with the system behavior only at exceptional congestion time periods (periodical events). Simulation results are provided