Uncertainty modeling in information fusion challenges decision-making systems, with Dempster-Shafer Evidence Theory (DSET) as a key solution. Although Dempster’s Rule of Combination (DRC), DSET’s core, faces exponential complexity ( \(O(4^n)\) ) in CPU implementations, limiting scalability for high-dimensional frameworks. Heterogeneous computing improves efficiency, but FPGA-based methods still suffer from increased runtime with more elements in the framework. This study proposes a random signal-based DRC (RS DRC) using FPGA-accelerated parallel processing. When we fixing logical combination count \(\mathcal {N}\) , it reduces complexity from \(O(4^n)\) to linear O(n), resolving high-dimensional bottlenecks. Experiments show that with 1024 logical combinations, RS DRC fusion time stabilizes at 0.9811ms (true random numbers) and 0.0103ms (pseudo-random numbers) regardless of framework elements. Applied to UCI wine classification, it supports practical deployment of uncertain data processing in complex decision systems.
