An Energy-Efficient Adaptive GPS Scheduling Algorithm

The emergence of the Internet of Things (IoT) has created noticeable changes across different sectors, most notably in real-time monitoring, process automation, and intelligent decision-making. There are always several issues existing in the implementation of IoT solutions, and one of the most persistent is the use of electricity, particularly in portable and mobile IoT devices that are frequently carried around and used by the active human users. These devices are usually powered by batteries, and they are constantly used in changing environments, like while the user is walking, cycling, or commuting--all without expecting to have a charging point nearby at all times. Besides this, the excessive power consumption caused by sensors, processors, wireless modules, and continuous GPS usage serves to make the matter even worse. This paper introduces a novel machine learning-based adaptive GPS scheduling algorithm for energy-efficient human-centered IoT. The main idea of this system is to have two parts: one for predicting mobility and the other for optimizing power. It applies the technique of supervised learning to identify the user's activity modes—stationary, walking, or in-vehicle— through contextual parameters like speed, time, and location history. As predicted, the scheduler on the one side changes the GPS working condition to save power while at the same time keeping the location accurate. The experimental results show that the proposed method can save up to 58% of GPS-related energy consumption, and this reduction is achieved in opposition to conventional fixed-interval GPS sampling strategies. Thus, this adaptive scheduling method can be a desirable solution for users who are moving erratically, as in the case of health monitoring, personal navigation, and context-aware services.