Joint Latency and energy optimization in edge intelligence enabled internet of vehicles

Efficient task execution in edge intelligence-enabled Internet of Vehicles (IoV) must balance latency and energy consumption under dynamic communication and computation conditions. This paper proposes a joint latency--energy aware resource allocation framework in which vehicular tasks are either processed locally or partially offloaded to roadside units (RSUs). A unified model is developed to capture local computation, uplink transmission, edge execution, and their associated energy costs. On this basis, a weighted latency--energy optimization problem is formulated to jointly determine task partitioning, RSU association, channel allocation, and transmission power subject to latency and resource constraints. To solve the resulting mixed-integer non-linear problem, a decomposed iterative approach is introduced, combining task partition update, utility-driven RSU and channel assignment, and bisection-based power control. MATLAB-based simulations compare the proposed method with local-only, full-offloading, and greedy baselines. Results indicate that the proposed framework provides a better latency--energy trade-off, lowers the overall weighted cost, and improves deadline satisfaction, demonstrating its effectiveness for energy-aware vehicular edge computing.