Research on multi-sensor data fusion technique based on a novel associative memory system

This paper firstly proposes a novel high-order Associative Memory System based on the Newton's Forward Interpolation (NFI-AMS), which is capable of implementing error free approximations to multi-variable polynomial functions of arbitrary order. Secondly, a new multi-sensor data fusion method is presented based on the novel Associative Memory System, which mainly includes the architecture, interpolation algorithm and learning method. The advantages it offers over data fusion method based on CMAC-type AMS are: high precision of learning, much smaller memory requirement without the data-collision problem, and also much less computational effort for training and faster convergence rates than that attainable with conventional data fusion method based on multi-layer BP neural networks. Thirdly, a set of numerical simulations have been conducted, simulation results have shown that the novel data fusion method based on NFI-AMS is feasible and efficient.