This document discusses maximizing the quality of monitoring (QoM) for multi-channel wireless infrastructure networks. It introduces the QoM metric defined as the expected number of active users monitored. It considers two models - user-centric which can distinguish users' activities, and sniffer-centric which only knows channel activity. For user-centric, devising algorithms attains a constant approximation ratio to the NP-hard optimization problem of maximizing QoM by assigning sniffers to channels. For sniffer-centric, stochastic inference schemes transform it into the user-centric domain to apply those algorithms. The proposed methods are evaluated using both synthetic and real-world traces showing effectiveness.