An Efficient Power Management Mechanism for WiFi-Based Handheld Systems

We propose a system level dynamic power management scheme that considers interaction between CPU and WNIC, and types of workload to reduce energy consumption of handheld systems. The proposed scheme determines the clock speed (voltage) for the CPU and the power mode for WNIC according to the workload type. This mechanism works based on two observations: the consumed power of WNIC is reduced when the clock speed of CPU is increased, and the power modes of WNIC and the clock speed of CPU have different values based on the kind of workload. Experimental results show that on the average the proposed scheme reduces energy consumption by 8% for BE (best effort) workload, 3% for CBR (constant bit rate) workload, and 16% for interactive when compared to DVFS (dynamic voltage and frequency scaling) for CPU and DPM (dynamic power management) for WNIC