This algorithm expands all prefixes to three kinds of lengths according to the prefixes distribution of route tables.And several Hash functions are selected by using the maximum entropy criterion method proposed in this paper

then map the expanded prefixes to different levels of three Hash tables.In process of IP routing lookup

the algorithm dynamically calculates the searching cost using the hit rate of the three Hash tables

and then adjusts its searching sequence accordingly.The algorithm supports increasing update

and is easy to be implemented in software or pipelined hardware.The experiment demonstrates that it needs less than 3.7 M bytes for the real route tables of 128K prefixes

and averagely it needs only 1.1 memory accesses for each lookup and few memory accesses for each update.