Non-intrusive appliance load monitoring is a preferred technique to provide appliance-level energy usage information without deploying submeters and benefits smart building technologies such as energy management and home automation. Load signature construction, an indispensable design step for identifying residential appliances, is still an open challenge in non-intrusive appliance load monitoring. To improve the identification accuracy of the appliance loads with only on/off states, this paper constructs an admittance-based load signature called resolution-enhanced admittance (REA). With the voltage and current signal serving as the input and output respectively, the aggregate of on-off appliance loads is interpreted as the transfer function of a linear time-invariant system, namely the load admittance. Different from existing load signatures, the phase information of load admittances is accurately estimated by computing fundamental and harmonic frequencies in high resolution. To verify the feasibility and effectiveness of the proposed load signature, a laboratory testbed is built to generate datasets with high sampling rate for subsequent signature construction and load identification. The experimental results show that REA improves the accuracy of non-intrusive load monitoring and reflects more intrinsic properties of appliance loads compared to state-of-art load signatures. (C) 2018 Elsevier B.V. All rights reserved.