Recently, IoT-based sensor networks for monitoring environmental parameters have attracted research and development groups due to their numerous advantages. However, these battery-powered devices experience the difficulties of periodic recharge or replacement. Therefore, to provide a sustainable and continuous power supply, piezoelectric energy harvesters can be used where the wired power connection is not feasible due to inaccessible setup or hazardous environment. The efficiency of piezoelectric converters is determined by their geometrical configuration and the materials used. This paper addresses the analytical modeling of a unimorph piezoelectric harvester and the optimization of harvester geometry to improve the output power and resonance frequency. The harvester geometry is modeled using Finite Element Method (FEM), and the RADAR plot is used to optimize the harvester width for maximum output. The FEM-optimized model is experimentally tested with the help of a controlled vibratory system designed and developed for this purpose. The optimized model is a dumbbell-shaped harvester with 390[Formula: see text] more open circuit output voltage and 9.6[Formula: see text] reduced resonant frequency compared to a rectangular harvester. The maximum power of 59.9[Formula: see text][Formula: see text] is harvested across 1[Formula: see text]M[Formula: see text] load with a power density of 1[Formula: see text]mWcm-3 at 141[Formula: see text]Hz resonant frequency.