The invention provides an intensive-network self-optimizing switching method and belongs to the technical field of radio communication. According to the architecture characteristics of intensiveness, hierarchy and
amorphism of an intensive heterogeneous
honeycomb network and switching characteristics of
high frequency and low performance, the intensive-network self-optimizing switching method includes firstly sensing passing-by rate,
moving speed, acceptance and idle degrees of a target
honeycomb of an MS (
mobile station), acquiring of tendency of the MS to the target
honeycomb by the passing-by rate and the speed based on FL (
fuzzy logic), and acquiring affinity of the honeycomb towards new users by the acceptance and the idle degrees, and self-adaptively adjusting switching parameters based on Q-Learning
algorithm by taking tendency and affinity as input and
resource utilization rate, call drop rate, switching
failure rate and ping-pong switching rate as instant awards, so as to optimize the switching
failure rate, the ping-pong switching rate and the call drop rate. Compared with the prior art, the intensive-network self-optimizing switching method can remarkably improve high-switching
failure rate, ping-pong switching rate and call drop rate of the MS under the intensive
heterogeneous network.