The rapid advancement of machine learning technologies has necessitated the optimization of models for real-time prediction at the edge, particularly in resource-constrained environments. This literature review examines key strategies including model compression, co-inference techniques, and hardware-runtime co-design to enhance the efficiency of deep neural network (DNN) inference. Recent studies highlight the importance of optimizing DNNs to meet the demands of edge computing, addressing challenges such as energy consumption, latency, and computational resource limitations. Furthermore, the integration of adaptive inference mechanisms and advanced communication protocols is essential for improving the overall performance of edge AI applications. This review aims to provide a comprehensive overview of the current state of research, identify gaps in existing methodologies, and propose future directions for enhancing edge inference capabilities