Reliable Service Node Set Selection and Task Offloading Strategy in Edge-Enabled Robot Swarms via Dynamic Interference and Link Reliability Models

Abstract

This paper proposes a collaborative optimization strategy of service node selection (NSS) and task unloading based on dynamic interference sensing and link reliability for edge computing driven robot cluster environment. The core innovation lies in the construction of the interference decision model with the dynamic distance proportional coefficient K, the combination of SINR physical model accuracy and protocol model efficiency, and the significant reduction of the computational complexity from O(n³) to O(n²). At the same time, it pioneered the reliability modeling method of series parallel system (SPS), which integrates node computing capability and link reliability generation performance indicators to achieve multi-path redundancy and fault tolerance. The simulation results show that the strategy has excellent characteristics in key performance indicators. The system reliability reaches 95.8%, 13.6%/10.7% higher than the Min-Hop / Path Prediction algorithm, the average task completion time can be reduced to 1.82 seconds, and the resource utilization rate reaches 92.3%. In addition, its lightweight design meets the stringent constraints of the on-board unit (OBU). Under the 128 node scale, the TCT increases by only 18%, and the reliability of the 30m/s high-speed mobile scene remains more than 90%, providing an efficient and reliable edge computing solution for the highly dynamic robot cluster.