Abstract: Blockchain shows great potential in wireless network scenarios due to its security features. It establishes trust and consistency between system nodes without the trusted central authority. However, blockchain systems tend to consume a lot of energy, as many rounds of communication are required for consensus. In some wireless networks, the limited capacity of node batteries can quickly lead to node power depletion and offline, which can affect consensus performance. In addition, when the network is large, the huge energy consumption can also limit the scalability of the blockchain. Therefore, a low-energy-consumption sharding algorithm for RAFT consensus is designed to minimize the energy overhead in this paper. This algorithm restricts nodes to specific shards based on their geographic location, thereby avoiding nodes participating in a global consensus. Meanwhile, this paper also proposes an energy consumption estimation method to simplify the energy calculation of sharded wireless blockchain networks. The simulation results show that the proposed algorithm is effective. The energy consumption can be reduced by 98.36%, and the minimum error of the estimation method is only 0.4%.