Currently, a set of DC switching power supply is shared by multiple operators in a communication base station, and there is a demand for accurate metering of DC power consumption of each operator in this scenario. This paper studies the problems of Hall sensors in DC metering applications. A reduced-rank square root Kalman filter algorithm is designed to efficiently solve the zero-drift problem existing in Hall sensors. A nonlinear fitting algorithm is introduced to efficiently solve the temperature drift problem existing in Hall sensors. An innovative multi-circuit DC metering device based on an open end Hall sensor is designed, which integrates Hall sensors with metering circuits. It can achieve dynamic measurement of 16-user DC electrical energy and real-time monitoring of voltage and current. And the eccentricity problem of Hall sensors is solved through innovative structural design of limit structures. The experimental results show that the accuracy of the DC metering device can reach the accuracy standard of 1.0 level.