Navi-Based Distributed Adaptive Clustering and Estimation Over Multitask Networks

The distributed multitask learning (MTL) are widely studied under the area of wireless sensor networks, Internet of Things, and Internet of Vehicles in recent years. However, when faced with real scenarios, the presence of unknown clustering information and intricate network environment drives the degradation of existing distributed MTL algorithms. To tackle the issue, the navi-based distributed adaptive clustering and estimation algorithm is proposed. This algorithm operates within a hybrid system: one subsystem dedicated to clustering, another to estimation, and the navigation subsystem (N-Sub), which enhance the overall system accuracy by guiding the other subsystems through convex combinations. In addition, two extra schemes are designed to improve the accuracy and convergence speed of the overall system. The mean and mean-square performance of the proposed algorithm is analyzed theoretically. Simulations demonstrate the effectiveness of the proposed algorithm.