Microseismic activity is an important phenomenon to investigate the physical properties of fault systems. Detecting these weak signals requires high‐sensitivity seismic networks and noise suppression algorithms. High‐sensitivity networks generally require dense station coverage, high deployment effort, special field environment, and high funding cost, which are not suitable for rapid response for aftershock monitoring. To investigate aftershock activities of the 2017 6.5 Jiuzhaigou earthquake, we designed and deployed an experimental seismic network (September–December 2017) near the source region. This network design is named as array of small arrays (AsA), which is composed of nine small aperture subarrays; thus, waveform coherency within each subarray can be used to enhance seismic signals. We conducted a theoretical analysis of linearly stacking and geometric mean envelope (GME) weighting algorithms and found that the GME algorithm could significantly suppress impulsive noise signal present at a single station. We tested detection performance of different algorithms, including template matching, linear stacking, local similarity, and GME, with synthetic and real data. These tests show that the GME algorithm and local similarity outperforms other algorithms by more than 10% of detection completeness. The GME algorithm achieves similar detecting performance as the local similarity algorithm, although it has much higher calculation efficiency. Our preliminary test of detection performance shows that the AsA design and the GME algorithm serve as a promising and efficient approach to monitor microseismic signals.