<p>Internet of Things (IoT) devices use advanced sensors to support many applications related to monitoring of health metrics, transportation systems, and industrial parameters. A major challenge in such IoT networks is that the local task processing in IoT devices can lead to increased latency, energy consumption, network congestion, and resource competition. This challenge can be addressed by fog computing as it presents a promising solution by extending cloud capabilities closer to the edge of the network, and therefore, facilitating localized data processing. Addressing the challenge of task allocation and offloading for minimal energy consumption, we propose an effective algorithm for task offloading in D2D-assisted IoT networks that use the stable matching technique, i.e., Gale–Shapley. The D2D communication between the IoT devices and fog nodes (FNs) (in the surrounding) can be established to support IoT devices for task offloading. In particular, we develop novel preference profiles for IoT devices and FNs based on the energy consumption and utilize Gale–Shapley to match IoT devices with FNs. The main idea is to optimize offloading policies to reduce the energy consumption. Through simulations, it is demonstrated that the Gale–Shapley matching reduces the overall energy consumption by about 40–300% compared to the other compared schemes at transmission power of 0&#xa0;dB.</p>