The problem of power allocation in cognitive radio networks plays an important role to improve the efficiency of spectrum utilization.However,most of previous works focus on the power allocation for secondary users in spectrum sharing overlay or spectrum sharing underlay,which needs to frequently handoff between the idle spectrum bands or considers the interference constraints in all spectrum bands respectively.In order to reduce the handoff and fully utilize the spectrum resource,we propose a new spectrum sharing paradigm which not only can just need to adjust the transmit power in spectrum bands instead of frequently bandoff between idle spectrum bands,but can fully utilize the spectrum resource as we only consider the interference power constraints in active spectrum bands rather than in all spectrum bands.Then based on this new spectrum sharing paradigm and the constraint conditions,we study the distributed power allocation for secondary users and formulate the optimization problem as a non-cooperative game problem,after that the variational inequality approach is used to solve this game problem and a Nash equilibria solution is got,finally simulation results are illustrated to demonstrate the performance of the proposed scheme.