As the elderly population grows, the prevalence of age-related health conditions such as cardiovascular diseases, cognitive decline, and mobility impairment is also increased. Among these health conditions, falls are considered one of the greatest threats to elderly individuals. A low-cost wearable fall detection system is designed, with the purpose of monitoring and detecting their activities. Three master modules were constructed, with each consisting of an inertial sensor, a microcontroller, and a power supply circuit block. The data were collected using IMU MPU6050 and preprocessed using the MCU ESP32. Each master module is also supplied using a 3.7V 1S LiPo battery. 18 healthy subjects, consisting of 13 males and 5 females, agreed to volunteer for the experiments. They were instructed to do 8 different activities, including non-fall (stand still, sit-to-stand, walk, and sleep position) and fall events (forward fall, sideways fall, and backward fall). Overall, the system showed a good performance using the Multilayer Perceptron (MLP) algorithm with an accuracy of 95.3% across all activities. While misclassification happens between classes, our system is still able to distinguish between non-fall vs. fall events with 100% accuracy. Cost analysis was also conducted; the overall cost for the three master modules in our proposed system is $65.4. This is cheaper than commercial fall detection systems and other related research, and our proposed system can also be used continuously. The system will alert caregivers to the immediate attention of elderly individuals.