Channel assignment is a challenge for distributed cognitive networks due to spectrum mobility and lack of centralized entity.We present a dynamic and efficient algorithm via conflict shifting,referred as Shifting-based Channel Assignment(SCA).In this algorithm,the system was modeled with a conflict graph,and users cannot assign the channels that primary users(legacy users) and neighbors already occupied.In order to eliminate the conflicts between neighbors efficiently,secondary users(unlicensed users) try to transfer them through a straight path to the boundary,where conflicts are easier to solve as there are less neighbors for boundary users.Actions in one shift are executed in slots,and users act in a synchronous and separated manner.As a result,some of the conflicting channels are avoid from directly abandoned,and for this,utility of the entire network can be improved.Simulation results show that the proposed algorithm can provide similar utility performance while obviously reducing the communication cost than bargaining-base algorithms.In small scale networks with low user mobility(under 20%),it reduces 50% of the communication overhead than the later. Channel assignment is a challenge for distributed cognitive networks due to spectrum mobility and lack of a centralized entity. We present a dynamic and efficient algorithm via conflict shifting, referred as Shifting-based Channel Assignment (SCA) .In this algorithm, the system was modeled with a conflict graph, and users can not assign the channels that primary users (legacy users) and neighbors already occupied. order to eliminate the conflicts between neighbors efficiently, secondary users (unlicensed users) try to transfer them through a straight path to the boundary, where conflicts are easier to solve as there are less neighbors for boundary users. Actions in one shift are executed in slots, and users act in a synchronous and separated manner. As a result, some of the conflicting channels are avoid from directly abandoned, and for this, utility of the entire network can be improved. Simulation results show that the proposed algorithm can provide similar utility performance while obviously reducing the communication cost than bargaining-base algorithms. small scale networks with low user mobility (under 20%), it reduces 50% of the communication overhead than the later.