Techniques to improve tracking ability are presented for autonomous decentralized load tracking system. We propose a load tracking mechanism based on incomplete management of load information from subsystems. The system employs autonomous decentralized architecture, which enables online expansion and fault tolerance. Each subsystem asynchronously broadcasts load information to the data field shared by the subsystems. Load information in the data field is measured for a limited period of time to make the tracking ability efficient. The total system can be designed as a stable system to converge measurement errors. This paper proposes some techniques to improve tracking ability, including a feed forward technique and a nonlinear technique. The techniques are evaluated by two kinds of criteria by simulation.