An improved sensing data cleaning scheme for object localization in edge computing environment.

This article focuses on improving Radio Frequency Identification (RFID) data cleaning within an edge computing environment to enhance object localization accuracy. It proposes a two-level cleaning scheme: a tag-level method using an adaptive sliding window that dynamically adjusts based on tag motion state and read rate variations, and a reader-level method that estimates tag numbers via Chebyshev's inequality, optimizes unequal time slot parameters, and employs Markov chain-based cyclic control to reduce tag collisions. Experimental results using real-world datasets demonstrate that the combined approach reduces redundant and missing data, decreases collision time slots, and improves tag recognition rates and processing efficiency compared to existing algorithms. The study highlights the benefits of integrating RFID data cleaning with edge computing to address challenges posed by environmental interference and large-scale tag deployments.