With more devices and emerging data-intensive services, mobile data traffic grows explosively, which makes the energy efficiency issue in current and future wireless networks be a growing concern. Recently, the advantages of bringing caching scheme in wireless networks have been widely investigated. Although additional power consumption is incurred by deploying caching equipments, the traffic loading balance of radio access network (RAN) and total network delay will be improved. In this paper, we study the energy efficiency problem in future wireless network. Particularly, we give an energy-delay tradeoff algorithm by using the sleeping control and power matching scheme. In the proposed algorithm, we consider a N-based sleeping control scheme, where the base station (BS) is in the sleep state whenever there is no arriving user, and works again when N users arrive. Moreover, the proposed algorithm is suitable for multi-BS scenario; to simplify the analysis, we mainly consider the single case. Simulation results shows that the proposed algorithm can obviously reduce the network power and delay in heterogeneous network conditions. Moreover, we find that a larger cache size does not always obtain a less network cost, because more cache power is consumed as cache size increases.