Wireless sensor networks (WSNs) technologies have attracted much attention to collect damage information in a natural disaster. WSNs to monitor temperature or humidity usually collect data once in some seconds or some minutes. Since structural health monitoring (SHM), meanwhile, aims to make a diagnosis for the state of a structure based on detected acceleration, WSNs are a promising technology to collect acceleration data. One concern to employ WSNs in SHM is to detect phenomena at a high sampling rate under energy-aware condition. In this paper, we describe a model for seismic acceleration monitoring, configured with multi-layer networks: WSNs, a wireless distribution system (WDS) and a database server, where the WDS is mainly operating in a wireless local area network (WLAN). Examining the performance in the test bed for the monitoring system, the results showed the system was capable of collecting acceleration at a rate of 100 sampling per second (sps) even in the fashion of intermittent operation, and capable of storing data into a database. We also suggest that the method using intermittent operation with appropriate sampling rate is effective in providing a long time operation for the system by considering in the response motion of a structure.