Edge Computing Supported by 6G Network Slice for Behaviour Prediction of Autonomous Vehicles

The rapid advancements in Autonomous Vehicle (AVs) stance considerable challenges in accurately predicting vehicle behavior, especially in complex and dynamic environments. To address these challenges, the research proposes a behavior framework supported by edge computing and 6G network slicing, integrate with a novel Refined Spider Monkey Optimized Enhanced Decision Tree (RSMO-EDT) algorithm. The Local Using Edge Computing Framework, thus, assists in real-time data processing onboard the vehicle to foster the timely decision-making ability in a 6G network-slicing environment. Modeling trajectories, intentions, and near entities' actions of vehicles very precisely, the framework combines data from various sensors, including LIDAR, cameras, and radar, combined with vehicle dynamics. A pre-processing stage of sensor inputs in the form of min-max normalization ensures good-quality data feeding into the model, while the Refined Spider Monkey Optimized-Enhanced Decision Tree (RSMO-EDT) algorithm optimizes hyperparameters for better prediction accuracies. Considering all this, 6G network slicing implementation extends dynamic resource allocation functionality, supporting efficient edge computation. These are all indications brought on by simulations in an array of driving scenarios, including urban traffic, highways, and challenging intersections, indicating massive improvements over existing models in terms of prediction accuracy and reduced response time. This hence points toward providing a potential solution for the future AVs ensuring economical operation of systems through fast, accurate, and low latency behavior prediction.