Constructing underlying logic trees can effectively reduce the large communication overhead for centralized top-

k

query.In this paper

an energy-balanced optimal support-tree is proposed to solve the problem of earlier death for nodes near sink in sensor networks without fixed sink.This method aggregates node energy

degree and communication overhead

and nodes with low energy cost are chosen to be dominating nodes for data processing based on connected dominating set.In each query

nodes take turns to be root by location-based ID

which effectively balances the energy cost.The simulation experiment shows that top-

k

query based on support-tree outperforms the exis

ting trees by similar query time

but lower average energy cost per round and balanced consumption among all sensor nodes after continuous top-

k

queries

which effectively prolong the network lifetime.